-
Notifications
You must be signed in to change notification settings - Fork 90
OSSM-2369: [federation] Introduces federation deployment (#585) #699
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
s.XDSServer.InitGenerators(e, args.Namespace, s.internalDebugMux) | ||
|
||
// Initialize workloadTrustBundle after CA has been initialized | ||
if err := s.initWorkloadTrustBundle(args); err != nil { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Is the order here significant?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
What do you suggest? Do you think the current order is not correct?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It is different than what the current upstream has so I wonder if there's a significance here. But I lack the knowledge if that was a deliberate decision. Clearly, tests work both ways, so from this limited viewpoint, it's irrelevant.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
No I think the changes above this line are not needed. Only the one that adds the federation struct
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ok I will restore the order so it might result in less merge work in the future.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
After a closer look, I think we need s.initIstiodCerts
call first, so the order matters. The Federation Controller relies on it, so the watches should be triggered first. I will put the comment in the code.
@@ -72,7 +72,7 @@ func (t Bundle) ReplaceTrustDomainAliases(principals []string) []string { | |||
principalsIncludingAliases = append(principalsIncludingAliases, t.replaceTrustDomains(principal, trustDomainFromPrincipal)...) | |||
} else { | |||
authzLog.Warnf("Trust domain %s from principal %s does not match the current trust "+ | |||
"domain or its aliases", trustDomainFromPrincipal, principal) | |||
"domain or its aliases: %v", trustDomainFromPrincipal, principal, t.TrustDomains) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Worth pushing upstream?
Could you add 6e3a698 to this PR? |
Converting back to draft to only run jobs I need until fixed. |
// See the License for the specific language governing permissions and | ||
// limitations under the License. | ||
|
||
package aggregate |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I thought keeping extension func
s in their own file would make it easier to carry them forward and minimize merging effort. Not entirely convinced about the file name though.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
generally I like the idea, but in this case I'm not even sure why we needed a separate func to begin with- it's only called in one place afaict
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yes, but without it, we are not able to construct imported sets properly, as the types do not match and we are not able to get hostname et al for importedService
. Unless we move a callee part to the same package we can't really unpeel aggregate registry entry [1] for this purpose.
I don't like this kind of argument, but... it's been done like that before except for the new file :) Do you have some alternative solution in mind?
istio/pilot/pkg/serviceregistry/aggregate/controller.go
Lines 427 to 432 in 46ba364
func (c *Controller) Unwrap(registry serviceregistry.Instance) serviceregistry.Instance { | |
if registryEntry, ok := registry.(*registryEntry); ok { | |
return registryEntry.Instance | |
} | |
return registry | |
} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The code in question assumes we can cast the serviceregistry.Instance
istio/pkg/servicemesh/federation/discovery/controller.go
Lines 193 to 199 in 46ba364
if federationRegistry, ok := registry.(*federationregistry.Controller); ok { | |
c.Logger.Debugf("updating settings for ServiceMeshPeer %s", instance.Name) | |
federationRegistry.UpdatePeerConfig(instance) | |
federationRegistry.UpdateImportConfig(importConfig) | |
} else { | |
return fmt.Errorf("registry %s is not a Federation registry (type=%T)", instance.Name, registry) | |
} |
I was thinking if we could use registry.Provider() == provider.Federation
for that purpose, but I'm not sure if we only foresee providing one registry from the whole Federation
feature moving forward. It also does not solve the problem of being able to call .Update...()
functions I'm afraid.
maistra#699) * [federation] Introduces federation deployment (maistra#585) * [federation] Initial federation implementation * MAISTRA-2194 Add server/client code for Federation Service Discovery v1 * MAISTRA-2195 Implement /watch endpoint * MAISTRA-2293 add CRD and controller for federating meshes * MAISTRA-2294 create CRD for federation ServiceExport (maistra#324) * MAISTRA-2294 update example VirtualService resources for ratings and mongodb (maistra#333) * [federation] MAISTRA-2295 create CRD for federation ServiceImport (maistra#336) Signed-off-by: rcernich <rcernich@redhat.com> * [misc] Use objects and clients from maistra/api repo - Remove local objects and clients - Update Makefile * [federation] MAISTRA-2309 create CRD for FederationStatus (maistra#348) Signed-off-by: rcernich <rcernich@redhat.com> * [federation] Federation fixes and improvements MAISTRA-2423 update federation api to v1 Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2424 minor updates to federation api Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2427 configure locality info on imported services Signed-off-by: rcernich <rcernich@redhat.com> Cherry-pick multi-root support (maistra#387) * Update go-control-plane to v0.9.9 * Support multiple roots Squashed commit, contains: - MAISTRA-2325 Distribute trust bundles over SDS - MAISTRA-2390 Push trust bundle updates through xDS (maistra#357) MAISTRA-2425 move spec.security.certificateChain to ConfigMap reference; add ability to specify ports for service and discovery (maistra#392) Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2426 move FederationStatus into MeshFederation (maistra#393) Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2513 federation API refinements Signed-off-by: rcernich <rcernich@redhat.com> [federation] MAISTRA-2237 Encrypt service discovery traffic (maistra#411) MAISTRA-2610 Prefix federation discovery endpoints with /v1/ (maistra#422) MAISTRA-2297 Support updates of federation resources (maistra#417) MAISTRA-2375: Do not create automatic routes for Federation Gateways Remove a redundant call `setHostname()` is already being called within `NameForService()` see https://github.com/maistra/istio/blob/21ee900cf8825711f70d88dc97afcf6862ed2626/pkg/servicemesh/federation/common/namemapping.go lines 83, 120, 129 Remove techPreview.meshConfig from PoC example It's set by default now. MAISTRA-2611 Fix deletion of service exports to federated mesh (maistra#421) Fix test MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil (maistra#437) * MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil * Fix test MAISTRA-2682 Fix watch mechanism in federation (maistra#439) Previously, no events were read from the watch response, because the read started with an endless loop that waited for data to be available in the decoder's buffer. This never happened, because the buffer is only written to when you call decoder.Decode(); this function was never called because the code waited for the buffer to have data. MAISTRA-2683 Properly close incoming watch connections when shutting down (maistra#440) Log actual error returned by pollServices() (maistra#441) Previously, instead of the actual error, only the following error message was logged: "expected condition not met". MAISTRA-2439: Prevent federation from exporting services that are not visible to the federation gateway (maistra#432) By taking into consideration the service annotation `networking.istio.io/exportTo`. This annotation restricts where this service is visible: https://istio.io/latest/docs/reference/config/annotations/ If a service is not reachable from the federation gateway namespace due to this annotation, it should not be exported. MAISTRA-2617: Do not watch all namespaces in Extensions controller (maistra#425) When using MemberRoll, we should rely on it to provide the list of namespaces to watch. If not using it, defaults to command line arguments. This fixes an istiod startup error as seen in the logs: ``` github.com/maistra/xns-informer/pkg/informers/informer.go:204: Failed to watch *v1.ServiceMeshExtension: failed to list *v1.ServiceMeshExtension: servicemeshextensions.maistra.io is forbidden: User "system:serviceaccount:i1:istiod-service-account-basic" cannot list resource "servicemeshextensions" in API group "maistra.io" at the cluster scope ``` * Remove package export and extensions Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix creating discovery.Controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix calling nil ResourceManager Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Remove panicing from AppendNetworkGatewayHandler Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [misc] OSSM-774 Fix flaky TestStatusManager (maistra#456) This adds a little sleep to our unit tests for the StatusManager, because without it, we're running into the issue that we're updating a ServiceMeshPeer's status very quickly, and in some cases it might be that the last change has not been propagated when we're generating the patch for the next status change, which can lead to failures. This can happen in the real world, but you would need to change a ServiceMeshPeer's status within a few milliseconds, I doubt that it affects users. It would also be fixed with the next status update. For those reasons, I'm only fixing it in the test, with a Sleep() call. * Refactor manager_test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-1150 Fix flaky TestStatusManager unit test (maistra#478) Co-authored-by: Marko Lukša <marko.luksa@gmail.com> * OSSM-1252 Fix federation status updates (maistra#512) * Copy federation privileges from base to istio-discovery * Remove unnecessary ServiceMeshExtensions CRD Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add model.NetworkGatewaysHandler to federation controller to implement AppendNetworkGatewayHandler Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [federation] MAISTRA-2640 Add federation integration test (maistra#447) * Fix building federation test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add package gogo from maistra-2.2 to temporarily fix TbdsGenerator Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Disable configuring remote cluster in federation deployment Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [federation] OSSM-1128 Fix federation (maistra#480) * Fix SecretCacheClient Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Send initial XDS request for trust bundle from proxy Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Disable using EndpointSlices to fix error on getting federation-egress endpoints Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Remove unused serviceMeshExtensionController Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix lint errors Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Update maistra CRDs Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * fedration_cp_version_update (maistra#521) * OSSM-1529 Improve federation example script (maistra#522) * OSSM-1529 Improve federation example install.sh Previously, the script would fall back to using nodeports when the load balancer IP wasn't set. This meant that if the provision of the load balancer took too long, the SMCPs would be misconfigured and you had to run the install script again. With this change, the script now waits for the load balancer IP to appear. It never falls back to using node ports, because they never really worked (the nodes' hostnames typically aren't FQDN and the node ports are typically protected by firewalls). If the user wants to expose the federation ingresses in a different way, they can now set the environment variables MESH1_ADDRESS, MESH1_DISCOVERY_PORT, and MESH2_SERVICE_PORT (likewise for MESH2) and run the script. * Update Federation example README * Better "Waiting for load balancer" message * OSSM-1211 Fix federation locality failover issues (maistra#561) Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Signed-off-by: rcernich <rcernich@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: maistra-bot <57098434+maistra-bot@users.noreply.github.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <xuyuanlin_00@hotmail.com> * fix: removes deprecated gogo protobuf conversion * fix: goimport format * fix(lint): removes unused funcs * fix(lint): removes deprecated io/ioutil * fix(lint): disables staticcheck for federation tests it requires at least two clusters to make sense * fix(lint): use anypb.UnmarshalTo instead of ptypes * fix: no need to exclude grpcgen_test.go it seems to be fixed in v1.39 see: grpc/grpc-go#4476 * chore(backoff): aligns backoff dependency with v4 used by upstream * chore: reverts removed blank line - irrelevant for merge * chore(revive): adds explanation why json:inline is skipped from linting * OSSM-1962: Use EndpointSlices instead of Endpoints in federation controller (maistra#614) Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-2049: Fix handling ServiceAccounts in federation controller (maistra#627) * Fix collecting empty or repeated ServiceAccounts in federation controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Collect ServiceAccounts in sorted order Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-1093 Shorten exported resource name (maistra#653) * Shorten exported resource name * Fix import createResourceName + unit tests * Rearrange unit tests + renaming function * Gen and lint * Rearrange unit tests + renaming function Gen and lint * Fix minor changes * Error message RFC 1123 * Reorganize structs in TestStatusManager Instead of having three arrays (events, expectedStatuses, and assertions), we now only have a single array, where each entry is a triplet containing the event, the expected status and the assertion. This allows you to see the event and its expected effects together and not have to scroll up and down, matching the indexes of the three arrays. * OSSM-2193 Fix flaky TestStatusManager See comment in https://issues.redhat.com/browse/OSSM-2193 to understand why this change fixes the test. * fix: runs make gen * chore: explains why staticcheck linter is disabled for federation_test * OSSM-728 Configuration scripts for Federation on Z and P, and bare metal (maistra#670) * add config example scripts for IBM Systems Z and P * update multi-arch bookinfo deployment README, remove src * Update README.md * these are provided in the IBM repo * so README.md passes mdlinter * so README.md passes mdlinter * so README.md passes mdlinter * Update README.md * Move federation examples to samples/ directory * Rename template YAMLs to .yaml.template This makes the linter happy Co-authored-by: cfillekes <cfilleke@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com> * chore: removes obsolute TODO * chore: simplifies bool return expression * chore: removes redundant kubeClient check if initialization fails this func will not be reached anyway * chore(pkg): moves kube ctrl under servicemesh pkg folder * OSSM-2338: Remove env ISTIO_META_ROUTER_MODE from federation test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-2338: Remove "routerMode: sni-dnat" from federation samples Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * fix: adds operator.go customizations to kube.go clearly with cherry-pick we lost information about the file rename and thus the changes we made specificually for testing federation got lost * fix(test/operator): checks if east-west gw needs to be deployed * fix(federation): uses Unwrap to get instance of Federation registry * fix(tests): sets istiod-less remote flag to false that was the behaviour for maistra-2.3 and we need istiod to be present in order to have federation working * chore: gets registry just before it is needed * chore: explains why istiodlessremotes is needed to be set to false * chore: removes redundant import aliases * chore: removes name collisions * chore: removes redundant type conversion * fix: disables staticcheck linter for cluster req tests. * fix(tests): reverts timeout to original (but in minutes) * fix(tests): removes extra logging * chore: removes unnecessary logging * fix: uses existing CRD file references in charts * chore: removes multicluster label * chore: uses built-in namespace.NewOrFail instead of our impl * chore: introduces defaultTimeout const for federation tests * fix(lint): fixes go imports * fix(lint): removes unused variable * fix: naively wait 5s hoping that kind network will show up Signed-off-by: rcernich <rcernich@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Co-authored-by: Jacek Ewertowski <jewertow@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: maistra-bot <57098434+maistra-bot@users.noreply.github.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <xuyuanlin_00@hotmail.com> Co-authored-by: bmangoen <bmangoen@redhat.com> Co-authored-by: cfillekes <cfilleke@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com>
maistra#699) * [federation] Introduces federation deployment (maistra#585) * [federation] Initial federation implementation * MAISTRA-2194 Add server/client code for Federation Service Discovery v1 * MAISTRA-2195 Implement /watch endpoint * MAISTRA-2293 add CRD and controller for federating meshes * MAISTRA-2294 create CRD for federation ServiceExport (maistra#324) * MAISTRA-2294 update example VirtualService resources for ratings and mongodb (maistra#333) * [federation] MAISTRA-2295 create CRD for federation ServiceImport (maistra#336) Signed-off-by: rcernich <rcernich@redhat.com> * [misc] Use objects and clients from maistra/api repo - Remove local objects and clients - Update Makefile * [federation] MAISTRA-2309 create CRD for FederationStatus (maistra#348) Signed-off-by: rcernich <rcernich@redhat.com> * [federation] Federation fixes and improvements MAISTRA-2423 update federation api to v1 Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2424 minor updates to federation api Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2427 configure locality info on imported services Signed-off-by: rcernich <rcernich@redhat.com> Cherry-pick multi-root support (maistra#387) * Update go-control-plane to v0.9.9 * Support multiple roots Squashed commit, contains: - MAISTRA-2325 Distribute trust bundles over SDS - MAISTRA-2390 Push trust bundle updates through xDS (maistra#357) MAISTRA-2425 move spec.security.certificateChain to ConfigMap reference; add ability to specify ports for service and discovery (maistra#392) Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2426 move FederationStatus into MeshFederation (maistra#393) Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2513 federation API refinements Signed-off-by: rcernich <rcernich@redhat.com> [federation] MAISTRA-2237 Encrypt service discovery traffic (maistra#411) MAISTRA-2610 Prefix federation discovery endpoints with /v1/ (maistra#422) MAISTRA-2297 Support updates of federation resources (maistra#417) MAISTRA-2375: Do not create automatic routes for Federation Gateways Remove a redundant call `setHostname()` is already being called within `NameForService()` see https://github.com/maistra/istio/blob/21ee900cf8825711f70d88dc97afcf6862ed2626/pkg/servicemesh/federation/common/namemapping.go lines 83, 120, 129 Remove techPreview.meshConfig from PoC example It's set by default now. MAISTRA-2611 Fix deletion of service exports to federated mesh (maistra#421) Fix test MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil (maistra#437) * MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil * Fix test MAISTRA-2682 Fix watch mechanism in federation (maistra#439) Previously, no events were read from the watch response, because the read started with an endless loop that waited for data to be available in the decoder's buffer. This never happened, because the buffer is only written to when you call decoder.Decode(); this function was never called because the code waited for the buffer to have data. MAISTRA-2683 Properly close incoming watch connections when shutting down (maistra#440) Log actual error returned by pollServices() (maistra#441) Previously, instead of the actual error, only the following error message was logged: "expected condition not met". MAISTRA-2439: Prevent federation from exporting services that are not visible to the federation gateway (maistra#432) By taking into consideration the service annotation `networking.istio.io/exportTo`. This annotation restricts where this service is visible: https://istio.io/latest/docs/reference/config/annotations/ If a service is not reachable from the federation gateway namespace due to this annotation, it should not be exported. MAISTRA-2617: Do not watch all namespaces in Extensions controller (maistra#425) When using MemberRoll, we should rely on it to provide the list of namespaces to watch. If not using it, defaults to command line arguments. This fixes an istiod startup error as seen in the logs: ``` github.com/maistra/xns-informer/pkg/informers/informer.go:204: Failed to watch *v1.ServiceMeshExtension: failed to list *v1.ServiceMeshExtension: servicemeshextensions.maistra.io is forbidden: User "system:serviceaccount:i1:istiod-service-account-basic" cannot list resource "servicemeshextensions" in API group "maistra.io" at the cluster scope ``` * Remove package export and extensions Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix creating discovery.Controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix calling nil ResourceManager Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Remove panicing from AppendNetworkGatewayHandler Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [misc] OSSM-774 Fix flaky TestStatusManager (maistra#456) This adds a little sleep to our unit tests for the StatusManager, because without it, we're running into the issue that we're updating a ServiceMeshPeer's status very quickly, and in some cases it might be that the last change has not been propagated when we're generating the patch for the next status change, which can lead to failures. This can happen in the real world, but you would need to change a ServiceMeshPeer's status within a few milliseconds, I doubt that it affects users. It would also be fixed with the next status update. For those reasons, I'm only fixing it in the test, with a Sleep() call. * Refactor manager_test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-1150 Fix flaky TestStatusManager unit test (maistra#478) Co-authored-by: Marko Lukša <marko.luksa@gmail.com> * OSSM-1252 Fix federation status updates (maistra#512) * Copy federation privileges from base to istio-discovery * Remove unnecessary ServiceMeshExtensions CRD Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add model.NetworkGatewaysHandler to federation controller to implement AppendNetworkGatewayHandler Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [federation] MAISTRA-2640 Add federation integration test (maistra#447) * Fix building federation test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add package gogo from maistra-2.2 to temporarily fix TbdsGenerator Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Disable configuring remote cluster in federation deployment Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [federation] OSSM-1128 Fix federation (maistra#480) * Fix SecretCacheClient Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Send initial XDS request for trust bundle from proxy Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Disable using EndpointSlices to fix error on getting federation-egress endpoints Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Remove unused serviceMeshExtensionController Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix lint errors Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Update maistra CRDs Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * fedration_cp_version_update (maistra#521) * OSSM-1529 Improve federation example script (maistra#522) * OSSM-1529 Improve federation example install.sh Previously, the script would fall back to using nodeports when the load balancer IP wasn't set. This meant that if the provision of the load balancer took too long, the SMCPs would be misconfigured and you had to run the install script again. With this change, the script now waits for the load balancer IP to appear. It never falls back to using node ports, because they never really worked (the nodes' hostnames typically aren't FQDN and the node ports are typically protected by firewalls). If the user wants to expose the federation ingresses in a different way, they can now set the environment variables MESH1_ADDRESS, MESH1_DISCOVERY_PORT, and MESH2_SERVICE_PORT (likewise for MESH2) and run the script. * Update Federation example README * Better "Waiting for load balancer" message * OSSM-1211 Fix federation locality failover issues (maistra#561) Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Signed-off-by: rcernich <rcernich@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: maistra-bot <57098434+maistra-bot@users.noreply.github.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <xuyuanlin_00@hotmail.com> * fix: removes deprecated gogo protobuf conversion * fix: goimport format * fix(lint): removes unused funcs * fix(lint): removes deprecated io/ioutil * fix(lint): disables staticcheck for federation tests it requires at least two clusters to make sense * fix(lint): use anypb.UnmarshalTo instead of ptypes * fix: no need to exclude grpcgen_test.go it seems to be fixed in v1.39 see: grpc/grpc-go#4476 * chore(backoff): aligns backoff dependency with v4 used by upstream * chore: reverts removed blank line - irrelevant for merge * chore(revive): adds explanation why json:inline is skipped from linting * OSSM-1962: Use EndpointSlices instead of Endpoints in federation controller (maistra#614) Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-2049: Fix handling ServiceAccounts in federation controller (maistra#627) * Fix collecting empty or repeated ServiceAccounts in federation controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Collect ServiceAccounts in sorted order Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-1093 Shorten exported resource name (maistra#653) * Shorten exported resource name * Fix import createResourceName + unit tests * Rearrange unit tests + renaming function * Gen and lint * Rearrange unit tests + renaming function Gen and lint * Fix minor changes * Error message RFC 1123 * Reorganize structs in TestStatusManager Instead of having three arrays (events, expectedStatuses, and assertions), we now only have a single array, where each entry is a triplet containing the event, the expected status and the assertion. This allows you to see the event and its expected effects together and not have to scroll up and down, matching the indexes of the three arrays. * OSSM-2193 Fix flaky TestStatusManager See comment in https://issues.redhat.com/browse/OSSM-2193 to understand why this change fixes the test. * fix: runs make gen * chore: explains why staticcheck linter is disabled for federation_test * OSSM-728 Configuration scripts for Federation on Z and P, and bare metal (maistra#670) * add config example scripts for IBM Systems Z and P * update multi-arch bookinfo deployment README, remove src * Update README.md * these are provided in the IBM repo * so README.md passes mdlinter * so README.md passes mdlinter * so README.md passes mdlinter * Update README.md * Move federation examples to samples/ directory * Rename template YAMLs to .yaml.template This makes the linter happy Co-authored-by: cfillekes <cfilleke@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com> * chore: removes obsolute TODO * chore: simplifies bool return expression * chore: removes redundant kubeClient check if initialization fails this func will not be reached anyway * chore(pkg): moves kube ctrl under servicemesh pkg folder * OSSM-2338: Remove env ISTIO_META_ROUTER_MODE from federation test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-2338: Remove "routerMode: sni-dnat" from federation samples Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * fix: adds operator.go customizations to kube.go clearly with cherry-pick we lost information about the file rename and thus the changes we made specificually for testing federation got lost * fix(test/operator): checks if east-west gw needs to be deployed * fix(federation): uses Unwrap to get instance of Federation registry * fix(tests): sets istiod-less remote flag to false that was the behaviour for maistra-2.3 and we need istiod to be present in order to have federation working * chore: gets registry just before it is needed * chore: explains why istiodlessremotes is needed to be set to false * chore: removes redundant import aliases * chore: removes name collisions * chore: removes redundant type conversion * fix: disables staticcheck linter for cluster req tests. * fix(tests): reverts timeout to original (but in minutes) * fix(tests): removes extra logging * chore: removes unnecessary logging * fix: uses existing CRD file references in charts * chore: removes multicluster label * chore: uses built-in namespace.NewOrFail instead of our impl * chore: introduces defaultTimeout const for federation tests * fix(lint): fixes go imports * fix(lint): removes unused variable * fix: naively wait 5s hoping that kind network will show up Signed-off-by: rcernich <rcernich@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Co-authored-by: Jacek Ewertowski <jewertow@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: maistra-bot <57098434+maistra-bot@users.noreply.github.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <xuyuanlin_00@hotmail.com> Co-authored-by: bmangoen <bmangoen@redhat.com> Co-authored-by: cfillekes <cfilleke@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com>
maistra#699) * [federation] Introduces federation deployment (maistra#585) * [federation] Initial federation implementation * MAISTRA-2194 Add server/client code for Federation Service Discovery v1 * MAISTRA-2195 Implement /watch endpoint * MAISTRA-2293 add CRD and controller for federating meshes * MAISTRA-2294 create CRD for federation ServiceExport (maistra#324) * MAISTRA-2294 update example VirtualService resources for ratings and mongodb (maistra#333) * [federation] MAISTRA-2295 create CRD for federation ServiceImport (maistra#336) Signed-off-by: rcernich <rcernich@redhat.com> * [misc] Use objects and clients from maistra/api repo - Remove local objects and clients - Update Makefile * [federation] MAISTRA-2309 create CRD for FederationStatus (maistra#348) Signed-off-by: rcernich <rcernich@redhat.com> * [federation] Federation fixes and improvements MAISTRA-2423 update federation api to v1 Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2424 minor updates to federation api Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2427 configure locality info on imported services Signed-off-by: rcernich <rcernich@redhat.com> Cherry-pick multi-root support (maistra#387) * Update go-control-plane to v0.9.9 * Support multiple roots Squashed commit, contains: - MAISTRA-2325 Distribute trust bundles over SDS - MAISTRA-2390 Push trust bundle updates through xDS (maistra#357) MAISTRA-2425 move spec.security.certificateChain to ConfigMap reference; add ability to specify ports for service and discovery (maistra#392) Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2426 move FederationStatus into MeshFederation (maistra#393) Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2513 federation API refinements Signed-off-by: rcernich <rcernich@redhat.com> [federation] MAISTRA-2237 Encrypt service discovery traffic (maistra#411) MAISTRA-2610 Prefix federation discovery endpoints with /v1/ (maistra#422) MAISTRA-2297 Support updates of federation resources (maistra#417) MAISTRA-2375: Do not create automatic routes for Federation Gateways Remove a redundant call `setHostname()` is already being called within `NameForService()` see https://github.com/maistra/istio/blob/21ee900cf8825711f70d88dc97afcf6862ed2626/pkg/servicemesh/federation/common/namemapping.go lines 83, 120, 129 Remove techPreview.meshConfig from PoC example It's set by default now. MAISTRA-2611 Fix deletion of service exports to federated mesh (maistra#421) Fix test MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil (maistra#437) * MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil * Fix test MAISTRA-2682 Fix watch mechanism in federation (maistra#439) Previously, no events were read from the watch response, because the read started with an endless loop that waited for data to be available in the decoder's buffer. This never happened, because the buffer is only written to when you call decoder.Decode(); this function was never called because the code waited for the buffer to have data. MAISTRA-2683 Properly close incoming watch connections when shutting down (maistra#440) Log actual error returned by pollServices() (maistra#441) Previously, instead of the actual error, only the following error message was logged: "expected condition not met". MAISTRA-2439: Prevent federation from exporting services that are not visible to the federation gateway (maistra#432) By taking into consideration the service annotation `networking.istio.io/exportTo`. This annotation restricts where this service is visible: https://istio.io/latest/docs/reference/config/annotations/ If a service is not reachable from the federation gateway namespace due to this annotation, it should not be exported. MAISTRA-2617: Do not watch all namespaces in Extensions controller (maistra#425) When using MemberRoll, we should rely on it to provide the list of namespaces to watch. If not using it, defaults to command line arguments. This fixes an istiod startup error as seen in the logs: ``` github.com/maistra/xns-informer/pkg/informers/informer.go:204: Failed to watch *v1.ServiceMeshExtension: failed to list *v1.ServiceMeshExtension: servicemeshextensions.maistra.io is forbidden: User "system:serviceaccount:i1:istiod-service-account-basic" cannot list resource "servicemeshextensions" in API group "maistra.io" at the cluster scope ``` * Remove package export and extensions Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix creating discovery.Controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix calling nil ResourceManager Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Remove panicing from AppendNetworkGatewayHandler Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [misc] OSSM-774 Fix flaky TestStatusManager (maistra#456) This adds a little sleep to our unit tests for the StatusManager, because without it, we're running into the issue that we're updating a ServiceMeshPeer's status very quickly, and in some cases it might be that the last change has not been propagated when we're generating the patch for the next status change, which can lead to failures. This can happen in the real world, but you would need to change a ServiceMeshPeer's status within a few milliseconds, I doubt that it affects users. It would also be fixed with the next status update. For those reasons, I'm only fixing it in the test, with a Sleep() call. * Refactor manager_test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-1150 Fix flaky TestStatusManager unit test (maistra#478) Co-authored-by: Marko Lukša <marko.luksa@gmail.com> * OSSM-1252 Fix federation status updates (maistra#512) * Copy federation privileges from base to istio-discovery * Remove unnecessary ServiceMeshExtensions CRD Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add model.NetworkGatewaysHandler to federation controller to implement AppendNetworkGatewayHandler Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [federation] MAISTRA-2640 Add federation integration test (maistra#447) * Fix building federation test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add package gogo from maistra-2.2 to temporarily fix TbdsGenerator Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Disable configuring remote cluster in federation deployment Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [federation] OSSM-1128 Fix federation (maistra#480) * Fix SecretCacheClient Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Send initial XDS request for trust bundle from proxy Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Disable using EndpointSlices to fix error on getting federation-egress endpoints Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Remove unused serviceMeshExtensionController Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix lint errors Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Update maistra CRDs Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * fedration_cp_version_update (maistra#521) * OSSM-1529 Improve federation example script (maistra#522) * OSSM-1529 Improve federation example install.sh Previously, the script would fall back to using nodeports when the load balancer IP wasn't set. This meant that if the provision of the load balancer took too long, the SMCPs would be misconfigured and you had to run the install script again. With this change, the script now waits for the load balancer IP to appear. It never falls back to using node ports, because they never really worked (the nodes' hostnames typically aren't FQDN and the node ports are typically protected by firewalls). If the user wants to expose the federation ingresses in a different way, they can now set the environment variables MESH1_ADDRESS, MESH1_DISCOVERY_PORT, and MESH2_SERVICE_PORT (likewise for MESH2) and run the script. * Update Federation example README * Better "Waiting for load balancer" message * OSSM-1211 Fix federation locality failover issues (maistra#561) Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Signed-off-by: rcernich <rcernich@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: maistra-bot <57098434+maistra-bot@users.noreply.github.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <xuyuanlin_00@hotmail.com> * fix: removes deprecated gogo protobuf conversion * fix: goimport format * fix(lint): removes unused funcs * fix(lint): removes deprecated io/ioutil * fix(lint): disables staticcheck for federation tests it requires at least two clusters to make sense * fix(lint): use anypb.UnmarshalTo instead of ptypes * fix: no need to exclude grpcgen_test.go it seems to be fixed in v1.39 see: grpc/grpc-go#4476 * chore(backoff): aligns backoff dependency with v4 used by upstream * chore: reverts removed blank line - irrelevant for merge * chore(revive): adds explanation why json:inline is skipped from linting * OSSM-1962: Use EndpointSlices instead of Endpoints in federation controller (maistra#614) Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-2049: Fix handling ServiceAccounts in federation controller (maistra#627) * Fix collecting empty or repeated ServiceAccounts in federation controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Collect ServiceAccounts in sorted order Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-1093 Shorten exported resource name (maistra#653) * Shorten exported resource name * Fix import createResourceName + unit tests * Rearrange unit tests + renaming function * Gen and lint * Rearrange unit tests + renaming function Gen and lint * Fix minor changes * Error message RFC 1123 * Reorganize structs in TestStatusManager Instead of having three arrays (events, expectedStatuses, and assertions), we now only have a single array, where each entry is a triplet containing the event, the expected status and the assertion. This allows you to see the event and its expected effects together and not have to scroll up and down, matching the indexes of the three arrays. * OSSM-2193 Fix flaky TestStatusManager See comment in https://issues.redhat.com/browse/OSSM-2193 to understand why this change fixes the test. * fix: runs make gen * chore: explains why staticcheck linter is disabled for federation_test * OSSM-728 Configuration scripts for Federation on Z and P, and bare metal (maistra#670) * add config example scripts for IBM Systems Z and P * update multi-arch bookinfo deployment README, remove src * Update README.md * these are provided in the IBM repo * so README.md passes mdlinter * so README.md passes mdlinter * so README.md passes mdlinter * Update README.md * Move federation examples to samples/ directory * Rename template YAMLs to .yaml.template This makes the linter happy Co-authored-by: cfillekes <cfilleke@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com> * chore: removes obsolute TODO * chore: simplifies bool return expression * chore: removes redundant kubeClient check if initialization fails this func will not be reached anyway * chore(pkg): moves kube ctrl under servicemesh pkg folder * OSSM-2338: Remove env ISTIO_META_ROUTER_MODE from federation test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-2338: Remove "routerMode: sni-dnat" from federation samples Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * fix: adds operator.go customizations to kube.go clearly with cherry-pick we lost information about the file rename and thus the changes we made specificually for testing federation got lost * fix(test/operator): checks if east-west gw needs to be deployed * fix(federation): uses Unwrap to get instance of Federation registry * fix(tests): sets istiod-less remote flag to false that was the behaviour for maistra-2.3 and we need istiod to be present in order to have federation working * chore: gets registry just before it is needed * chore: explains why istiodlessremotes is needed to be set to false * chore: removes redundant import aliases * chore: removes name collisions * chore: removes redundant type conversion * fix: disables staticcheck linter for cluster req tests. * fix(tests): reverts timeout to original (but in minutes) * fix(tests): removes extra logging * chore: removes unnecessary logging * fix: uses existing CRD file references in charts * chore: removes multicluster label * chore: uses built-in namespace.NewOrFail instead of our impl * chore: introduces defaultTimeout const for federation tests * fix(lint): fixes go imports * fix(lint): removes unused variable * fix: naively wait 5s hoping that kind network will show up Signed-off-by: rcernich <rcernich@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Co-authored-by: Jacek Ewertowski <jewertow@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: maistra-bot <57098434+maistra-bot@users.noreply.github.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <xuyuanlin_00@hotmail.com> Co-authored-by: bmangoen <bmangoen@redhat.com> Co-authored-by: cfillekes <cfilleke@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com>
maistra#699) * [federation] Introduces federation deployment (maistra#585) * [federation] Initial federation implementation * MAISTRA-2194 Add server/client code for Federation Service Discovery v1 * MAISTRA-2195 Implement /watch endpoint * MAISTRA-2293 add CRD and controller for federating meshes * MAISTRA-2294 create CRD for federation ServiceExport (maistra#324) * MAISTRA-2294 update example VirtualService resources for ratings and mongodb (maistra#333) * [federation] MAISTRA-2295 create CRD for federation ServiceImport (maistra#336) Signed-off-by: rcernich <rcernich@redhat.com> * [misc] Use objects and clients from maistra/api repo - Remove local objects and clients - Update Makefile * [federation] MAISTRA-2309 create CRD for FederationStatus (maistra#348) Signed-off-by: rcernich <rcernich@redhat.com> * [federation] Federation fixes and improvements MAISTRA-2423 update federation api to v1 Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2424 minor updates to federation api Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2427 configure locality info on imported services Signed-off-by: rcernich <rcernich@redhat.com> Cherry-pick multi-root support (maistra#387) * Update go-control-plane to v0.9.9 * Support multiple roots Squashed commit, contains: - MAISTRA-2325 Distribute trust bundles over SDS - MAISTRA-2390 Push trust bundle updates through xDS (maistra#357) MAISTRA-2425 move spec.security.certificateChain to ConfigMap reference; add ability to specify ports for service and discovery (maistra#392) Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2426 move FederationStatus into MeshFederation (maistra#393) Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2513 federation API refinements Signed-off-by: rcernich <rcernich@redhat.com> [federation] MAISTRA-2237 Encrypt service discovery traffic (maistra#411) MAISTRA-2610 Prefix federation discovery endpoints with /v1/ (maistra#422) MAISTRA-2297 Support updates of federation resources (maistra#417) MAISTRA-2375: Do not create automatic routes for Federation Gateways Remove a redundant call `setHostname()` is already being called within `NameForService()` see https://github.com/maistra/istio/blob/21ee900cf8825711f70d88dc97afcf6862ed2626/pkg/servicemesh/federation/common/namemapping.go lines 83, 120, 129 Remove techPreview.meshConfig from PoC example It's set by default now. MAISTRA-2611 Fix deletion of service exports to federated mesh (maistra#421) Fix test MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil (maistra#437) * MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil * Fix test MAISTRA-2682 Fix watch mechanism in federation (maistra#439) Previously, no events were read from the watch response, because the read started with an endless loop that waited for data to be available in the decoder's buffer. This never happened, because the buffer is only written to when you call decoder.Decode(); this function was never called because the code waited for the buffer to have data. MAISTRA-2683 Properly close incoming watch connections when shutting down (maistra#440) Log actual error returned by pollServices() (maistra#441) Previously, instead of the actual error, only the following error message was logged: "expected condition not met". MAISTRA-2439: Prevent federation from exporting services that are not visible to the federation gateway (maistra#432) By taking into consideration the service annotation `networking.istio.io/exportTo`. This annotation restricts where this service is visible: https://istio.io/latest/docs/reference/config/annotations/ If a service is not reachable from the federation gateway namespace due to this annotation, it should not be exported. MAISTRA-2617: Do not watch all namespaces in Extensions controller (maistra#425) When using MemberRoll, we should rely on it to provide the list of namespaces to watch. If not using it, defaults to command line arguments. This fixes an istiod startup error as seen in the logs: ``` github.com/maistra/xns-informer/pkg/informers/informer.go:204: Failed to watch *v1.ServiceMeshExtension: failed to list *v1.ServiceMeshExtension: servicemeshextensions.maistra.io is forbidden: User "system:serviceaccount:i1:istiod-service-account-basic" cannot list resource "servicemeshextensions" in API group "maistra.io" at the cluster scope ``` * Remove package export and extensions Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix creating discovery.Controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix calling nil ResourceManager Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Remove panicing from AppendNetworkGatewayHandler Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [misc] OSSM-774 Fix flaky TestStatusManager (maistra#456) This adds a little sleep to our unit tests for the StatusManager, because without it, we're running into the issue that we're updating a ServiceMeshPeer's status very quickly, and in some cases it might be that the last change has not been propagated when we're generating the patch for the next status change, which can lead to failures. This can happen in the real world, but you would need to change a ServiceMeshPeer's status within a few milliseconds, I doubt that it affects users. It would also be fixed with the next status update. For those reasons, I'm only fixing it in the test, with a Sleep() call. * Refactor manager_test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-1150 Fix flaky TestStatusManager unit test (maistra#478) Co-authored-by: Marko Lukša <marko.luksa@gmail.com> * OSSM-1252 Fix federation status updates (maistra#512) * Copy federation privileges from base to istio-discovery * Remove unnecessary ServiceMeshExtensions CRD Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add model.NetworkGatewaysHandler to federation controller to implement AppendNetworkGatewayHandler Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [federation] MAISTRA-2640 Add federation integration test (maistra#447) * Fix building federation test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add package gogo from maistra-2.2 to temporarily fix TbdsGenerator Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Disable configuring remote cluster in federation deployment Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [federation] OSSM-1128 Fix federation (maistra#480) * Fix SecretCacheClient Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Send initial XDS request for trust bundle from proxy Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Disable using EndpointSlices to fix error on getting federation-egress endpoints Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Remove unused serviceMeshExtensionController Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix lint errors Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Update maistra CRDs Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * fedration_cp_version_update (maistra#521) * OSSM-1529 Improve federation example script (maistra#522) * OSSM-1529 Improve federation example install.sh Previously, the script would fall back to using nodeports when the load balancer IP wasn't set. This meant that if the provision of the load balancer took too long, the SMCPs would be misconfigured and you had to run the install script again. With this change, the script now waits for the load balancer IP to appear. It never falls back to using node ports, because they never really worked (the nodes' hostnames typically aren't FQDN and the node ports are typically protected by firewalls). If the user wants to expose the federation ingresses in a different way, they can now set the environment variables MESH1_ADDRESS, MESH1_DISCOVERY_PORT, and MESH2_SERVICE_PORT (likewise for MESH2) and run the script. * Update Federation example README * Better "Waiting for load balancer" message * OSSM-1211 Fix federation locality failover issues (maistra#561) Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Signed-off-by: rcernich <rcernich@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: maistra-bot <57098434+maistra-bot@users.noreply.github.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <xuyuanlin_00@hotmail.com> * fix: removes deprecated gogo protobuf conversion * fix: goimport format * fix(lint): removes unused funcs * fix(lint): removes deprecated io/ioutil * fix(lint): disables staticcheck for federation tests it requires at least two clusters to make sense * fix(lint): use anypb.UnmarshalTo instead of ptypes * fix: no need to exclude grpcgen_test.go it seems to be fixed in v1.39 see: grpc/grpc-go#4476 * chore(backoff): aligns backoff dependency with v4 used by upstream * chore: reverts removed blank line - irrelevant for merge * chore(revive): adds explanation why json:inline is skipped from linting * OSSM-1962: Use EndpointSlices instead of Endpoints in federation controller (maistra#614) Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-2049: Fix handling ServiceAccounts in federation controller (maistra#627) * Fix collecting empty or repeated ServiceAccounts in federation controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Collect ServiceAccounts in sorted order Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-1093 Shorten exported resource name (maistra#653) * Shorten exported resource name * Fix import createResourceName + unit tests * Rearrange unit tests + renaming function * Gen and lint * Rearrange unit tests + renaming function Gen and lint * Fix minor changes * Error message RFC 1123 * Reorganize structs in TestStatusManager Instead of having three arrays (events, expectedStatuses, and assertions), we now only have a single array, where each entry is a triplet containing the event, the expected status and the assertion. This allows you to see the event and its expected effects together and not have to scroll up and down, matching the indexes of the three arrays. * OSSM-2193 Fix flaky TestStatusManager See comment in https://issues.redhat.com/browse/OSSM-2193 to understand why this change fixes the test. * fix: runs make gen * chore: explains why staticcheck linter is disabled for federation_test * OSSM-728 Configuration scripts for Federation on Z and P, and bare metal (maistra#670) * add config example scripts for IBM Systems Z and P * update multi-arch bookinfo deployment README, remove src * Update README.md * these are provided in the IBM repo * so README.md passes mdlinter * so README.md passes mdlinter * so README.md passes mdlinter * Update README.md * Move federation examples to samples/ directory * Rename template YAMLs to .yaml.template This makes the linter happy Co-authored-by: cfillekes <cfilleke@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com> * chore: removes obsolute TODO * chore: simplifies bool return expression * chore: removes redundant kubeClient check if initialization fails this func will not be reached anyway * chore(pkg): moves kube ctrl under servicemesh pkg folder * OSSM-2338: Remove env ISTIO_META_ROUTER_MODE from federation test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-2338: Remove "routerMode: sni-dnat" from federation samples Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * fix: adds operator.go customizations to kube.go clearly with cherry-pick we lost information about the file rename and thus the changes we made specificually for testing federation got lost * fix(test/operator): checks if east-west gw needs to be deployed * fix(federation): uses Unwrap to get instance of Federation registry * fix(tests): sets istiod-less remote flag to false that was the behaviour for maistra-2.3 and we need istiod to be present in order to have federation working * chore: gets registry just before it is needed * chore: explains why istiodlessremotes is needed to be set to false * chore: removes redundant import aliases * chore: removes name collisions * chore: removes redundant type conversion * fix: disables staticcheck linter for cluster req tests. * fix(tests): reverts timeout to original (but in minutes) * fix(tests): removes extra logging * chore: removes unnecessary logging * fix: uses existing CRD file references in charts * chore: removes multicluster label * chore: uses built-in namespace.NewOrFail instead of our impl * chore: introduces defaultTimeout const for federation tests * fix(lint): fixes go imports * fix(lint): removes unused variable * fix: naively wait 5s hoping that kind network will show up Signed-off-by: rcernich <rcernich@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Co-authored-by: Jacek Ewertowski <jewertow@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: maistra-bot <57098434+maistra-bot@users.noreply.github.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <xuyuanlin_00@hotmail.com> Co-authored-by: bmangoen <bmangoen@redhat.com> Co-authored-by: cfillekes <cfilleke@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com> OSSM-2376: Move kube controller to the federation package (maistra#718) Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> OSSM-2376: Don't start federation controllers until informers have synced (maistra#717) (maistra#720) * OSSM-2376: Don't start federation-discovery-controller until kube informer has synced Federation discovery controller fetches config map with remote CA root cert, so if the controller started before the informer has synced, it would fail to fetch the config map. Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Store ConfigMap informer in a field of the discovery.Controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Refactor ResourceManager and don't start federation controller until informers has synced Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Simplify Start and HasSynced functions in federation controllers Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Move kube controller to the federation package Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Federation example fixes (maistra#758) * Use default version in federation example SMCPs * Fix paths in federation example OSSM-3599 Federation egress-gateway gets wrong network gateway endpoints (maistra#775) * OSSM-3599 Federation egress-gateway gets wrong update of network gateway endpoints * Deprecate GatewayEndpoints on server side * Remove resyncNetworkGateways in unit tests * Fix lint * Deprecate NetworkGatewayEndpoints and fix tests Refactor federation tests (maistra#841) * Refactor federation tests Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add more test cases Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> --------- Signed-off-by: Jacek Ewertowski <jewertow@redhat.com>
* [federation] Introduces federation deployment (#585) * [federation] Initial federation implementation * MAISTRA-2194 Add server/client code for Federation Service Discovery v1 * MAISTRA-2195 Implement /watch endpoint * MAISTRA-2293 add CRD and controller for federating meshes * MAISTRA-2294 create CRD for federation ServiceExport (#324) * MAISTRA-2294 update example VirtualService resources for ratings and mongodb (#333) * [federation] MAISTRA-2295 create CRD for federation ServiceImport (#336) Signed-off-by: rcernich <rcernich@redhat.com> * [misc] Use objects and clients from maistra/api repo - Remove local objects and clients - Update Makefile * [federation] MAISTRA-2309 create CRD for FederationStatus (#348) Signed-off-by: rcernich <rcernich@redhat.com> * [federation] Federation fixes and improvements MAISTRA-2423 update federation api to v1 Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2424 minor updates to federation api Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2427 configure locality info on imported services Signed-off-by: rcernich <rcernich@redhat.com> Cherry-pick multi-root support (#387) * Update go-control-plane to v0.9.9 * Support multiple roots Squashed commit, contains: - MAISTRA-2325 Distribute trust bundles over SDS - MAISTRA-2390 Push trust bundle updates through xDS (#357) MAISTRA-2425 move spec.security.certificateChain to ConfigMap reference; add ability to specify ports for service and discovery (#392) Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2426 move FederationStatus into MeshFederation (#393) Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2513 federation API refinements Signed-off-by: rcernich <rcernich@redhat.com> [federation] MAISTRA-2237 Encrypt service discovery traffic (#411) MAISTRA-2610 Prefix federation discovery endpoints with /v1/ (#422) MAISTRA-2297 Support updates of federation resources (#417) MAISTRA-2375: Do not create automatic routes for Federation Gateways Remove a redundant call `setHostname()` is already being called within `NameForService()` see https://github.com/maistra/istio/blob/21ee900cf8825711f70d88dc97afcf6862ed2626/pkg/servicemesh/federation/common/namemapping.go lines 83, 120, 129 Remove techPreview.meshConfig from PoC example It's set by default now. MAISTRA-2611 Fix deletion of service exports to federated mesh (#421) Fix test MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil (#437) * MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil * Fix test MAISTRA-2682 Fix watch mechanism in federation (#439) Previously, no events were read from the watch response, because the read started with an endless loop that waited for data to be available in the decoder's buffer. This never happened, because the buffer is only written to when you call decoder.Decode(); this function was never called because the code waited for the buffer to have data. MAISTRA-2683 Properly close incoming watch connections when shutting down (#440) Log actual error returned by pollServices() (#441) Previously, instead of the actual error, only the following error message was logged: "expected condition not met". MAISTRA-2439: Prevent federation from exporting services that are not visible to the federation gateway (#432) By taking into consideration the service annotation `networking.istio.io/exportTo`. This annotation restricts where this service is visible: https://istio.io/latest/docs/reference/config/annotations/ If a service is not reachable from the federation gateway namespace due to this annotation, it should not be exported. MAISTRA-2617: Do not watch all namespaces in Extensions controller (#425) When using MemberRoll, we should rely on it to provide the list of namespaces to watch. If not using it, defaults to command line arguments. This fixes an istiod startup error as seen in the logs: ``` github.com/maistra/xns-informer/pkg/informers/informer.go:204: Failed to watch *v1.ServiceMeshExtension: failed to list *v1.ServiceMeshExtension: servicemeshextensions.maistra.io is forbidden: User "system:serviceaccount:i1:istiod-service-account-basic" cannot list resource "servicemeshextensions" in API group "maistra.io" at the cluster scope ``` * Remove package export and extensions Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix creating discovery.Controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix calling nil ResourceManager Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Remove panicing from AppendNetworkGatewayHandler Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [misc] OSSM-774 Fix flaky TestStatusManager (#456) This adds a little sleep to our unit tests for the StatusManager, because without it, we're running into the issue that we're updating a ServiceMeshPeer's status very quickly, and in some cases it might be that the last change has not been propagated when we're generating the patch for the next status change, which can lead to failures. This can happen in the real world, but you would need to change a ServiceMeshPeer's status within a few milliseconds, I doubt that it affects users. It would also be fixed with the next status update. For those reasons, I'm only fixing it in the test, with a Sleep() call. * Refactor manager_test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-1150 Fix flaky TestStatusManager unit test (#478) Co-authored-by: Marko Lukša <marko.luksa@gmail.com> * OSSM-1252 Fix federation status updates (#512) * Copy federation privileges from base to istio-discovery * Remove unnecessary ServiceMeshExtensions CRD Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add model.NetworkGatewaysHandler to federation controller to implement AppendNetworkGatewayHandler Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [federation] MAISTRA-2640 Add federation integration test (#447) * Fix building federation test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add package gogo from maistra-2.2 to temporarily fix TbdsGenerator Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Disable configuring remote cluster in federation deployment Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [federation] OSSM-1128 Fix federation (#480) * Fix SecretCacheClient Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Send initial XDS request for trust bundle from proxy Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Disable using EndpointSlices to fix error on getting federation-egress endpoints Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Remove unused serviceMeshExtensionController Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix lint errors Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Update maistra CRDs Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * fedration_cp_version_update (#521) * OSSM-1529 Improve federation example script (#522) * OSSM-1529 Improve federation example install.sh Previously, the script would fall back to using nodeports when the load balancer IP wasn't set. This meant that if the provision of the load balancer took too long, the SMCPs would be misconfigured and you had to run the install script again. With this change, the script now waits for the load balancer IP to appear. It never falls back to using node ports, because they never really worked (the nodes' hostnames typically aren't FQDN and the node ports are typically protected by firewalls). If the user wants to expose the federation ingresses in a different way, they can now set the environment variables MESH1_ADDRESS, MESH1_DISCOVERY_PORT, and MESH2_SERVICE_PORT (likewise for MESH2) and run the script. * Update Federation example README * Better "Waiting for load balancer" message * OSSM-1211 Fix federation locality failover issues (#561) Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Signed-off-by: rcernich <rcernich@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: maistra-bot <57098434+maistra-bot@users.noreply.github.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <xuyuanlin_00@hotmail.com> * fix: removes deprecated gogo protobuf conversion * fix: goimport format * fix(lint): removes unused funcs * fix(lint): removes deprecated io/ioutil * fix(lint): disables staticcheck for federation tests it requires at least two clusters to make sense * fix(lint): use anypb.UnmarshalTo instead of ptypes * fix: no need to exclude grpcgen_test.go it seems to be fixed in v1.39 see: grpc/grpc-go#4476 * chore(backoff): aligns backoff dependency with v4 used by upstream * chore: reverts removed blank line - irrelevant for merge * chore(revive): adds explanation why json:inline is skipped from linting * OSSM-1962: Use EndpointSlices instead of Endpoints in federation controller (#614) Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-2049: Fix handling ServiceAccounts in federation controller (#627) * Fix collecting empty or repeated ServiceAccounts in federation controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Collect ServiceAccounts in sorted order Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-1093 Shorten exported resource name (#653) * Shorten exported resource name * Fix import createResourceName + unit tests * Rearrange unit tests + renaming function * Gen and lint * Rearrange unit tests + renaming function Gen and lint * Fix minor changes * Error message RFC 1123 * Reorganize structs in TestStatusManager Instead of having three arrays (events, expectedStatuses, and assertions), we now only have a single array, where each entry is a triplet containing the event, the expected status and the assertion. This allows you to see the event and its expected effects together and not have to scroll up and down, matching the indexes of the three arrays. * OSSM-2193 Fix flaky TestStatusManager See comment in https://issues.redhat.com/browse/OSSM-2193 to understand why this change fixes the test. * fix: runs make gen * chore: explains why staticcheck linter is disabled for federation_test * OSSM-728 Configuration scripts for Federation on Z and P, and bare metal (#670) * add config example scripts for IBM Systems Z and P * update multi-arch bookinfo deployment README, remove src * Update README.md * these are provided in the IBM repo * so README.md passes mdlinter * so README.md passes mdlinter * so README.md passes mdlinter * Update README.md * Move federation examples to samples/ directory * Rename template YAMLs to .yaml.template This makes the linter happy Co-authored-by: cfillekes <cfilleke@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com> * chore: removes obsolute TODO * chore: simplifies bool return expression * chore: removes redundant kubeClient check if initialization fails this func will not be reached anyway * chore(pkg): moves kube ctrl under servicemesh pkg folder * OSSM-2338: Remove env ISTIO_META_ROUTER_MODE from federation test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-2338: Remove "routerMode: sni-dnat" from federation samples Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * fix: adds operator.go customizations to kube.go clearly with cherry-pick we lost information about the file rename and thus the changes we made specificually for testing federation got lost * fix(test/operator): checks if east-west gw needs to be deployed * fix(federation): uses Unwrap to get instance of Federation registry * fix(tests): sets istiod-less remote flag to false that was the behaviour for maistra-2.3 and we need istiod to be present in order to have federation working * chore: gets registry just before it is needed * chore: explains why istiodlessremotes is needed to be set to false * chore: removes redundant import aliases * chore: removes name collisions * chore: removes redundant type conversion * fix: disables staticcheck linter for cluster req tests. * fix(tests): reverts timeout to original (but in minutes) * fix(tests): removes extra logging * chore: removes unnecessary logging * fix: uses existing CRD file references in charts * chore: removes multicluster label * chore: uses built-in namespace.NewOrFail instead of our impl * chore: introduces defaultTimeout const for federation tests * fix(lint): fixes go imports * fix(lint): removes unused variable * fix: naively wait 5s hoping that kind network will show up Signed-off-by: rcernich <rcernich@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Co-authored-by: Jacek Ewertowski <jewertow@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: maistra-bot <57098434+maistra-bot@users.noreply.github.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <xuyuanlin_00@hotmail.com> Co-authored-by: bmangoen <bmangoen@redhat.com> Co-authored-by: cfillekes <cfilleke@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com> OSSM-2376: Move kube controller to the federation package (#718) Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> OSSM-2376: Don't start federation controllers until informers have synced (#717) (#720) * OSSM-2376: Don't start federation-discovery-controller until kube informer has synced Federation discovery controller fetches config map with remote CA root cert, so if the controller started before the informer has synced, it would fail to fetch the config map. Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Store ConfigMap informer in a field of the discovery.Controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Refactor ResourceManager and don't start federation controller until informers has synced Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Simplify Start and HasSynced functions in federation controllers Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Move kube controller to the federation package Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Federation example fixes (#758) * Use default version in federation example SMCPs * Fix paths in federation example OSSM-3599 Federation egress-gateway gets wrong network gateway endpoints (#775) * OSSM-3599 Federation egress-gateway gets wrong update of network gateway endpoints * Deprecate GatewayEndpoints on server side * Remove resyncNetworkGateways in unit tests * Fix lint * Deprecate NetworkGatewayEndpoints and fix tests Refactor federation tests (#841) * Refactor federation tests Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add more test cases Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> --------- Signed-off-by: Jacek Ewertowski <jewertow@redhat.com>
maistra#699) * [federation] Introduces federation deployment (maistra#585) * [federation] Initial federation implementation * MAISTRA-2194 Add server/client code for Federation Service Discovery v1 * MAISTRA-2195 Implement /watch endpoint * MAISTRA-2293 add CRD and controller for federating meshes * MAISTRA-2294 create CRD for federation ServiceExport (maistra#324) * MAISTRA-2294 update example VirtualService resources for ratings and mongodb (maistra#333) * [federation] MAISTRA-2295 create CRD for federation ServiceImport (maistra#336) Signed-off-by: rcernich <rcernich@redhat.com> * [misc] Use objects and clients from maistra/api repo - Remove local objects and clients - Update Makefile * [federation] MAISTRA-2309 create CRD for FederationStatus (maistra#348) Signed-off-by: rcernich <rcernich@redhat.com> * [federation] Federation fixes and improvements MAISTRA-2423 update federation api to v1 Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2424 minor updates to federation api Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2427 configure locality info on imported services Signed-off-by: rcernich <rcernich@redhat.com> Cherry-pick multi-root support (maistra#387) * Update go-control-plane to v0.9.9 * Support multiple roots Squashed commit, contains: - MAISTRA-2325 Distribute trust bundles over SDS - MAISTRA-2390 Push trust bundle updates through xDS (maistra#357) MAISTRA-2425 move spec.security.certificateChain to ConfigMap reference; add ability to specify ports for service and discovery (maistra#392) Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2426 move FederationStatus into MeshFederation (maistra#393) Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2513 federation API refinements Signed-off-by: rcernich <rcernich@redhat.com> [federation] MAISTRA-2237 Encrypt service discovery traffic (maistra#411) MAISTRA-2610 Prefix federation discovery endpoints with /v1/ (maistra#422) MAISTRA-2297 Support updates of federation resources (maistra#417) MAISTRA-2375: Do not create automatic routes for Federation Gateways Remove a redundant call `setHostname()` is already being called within `NameForService()` see https://github.com/maistra/istio/blob/21ee900cf8825711f70d88dc97afcf6862ed2626/pkg/servicemesh/federation/common/namemapping.go lines 83, 120, 129 Remove techPreview.meshConfig from PoC example It's set by default now. MAISTRA-2611 Fix deletion of service exports to federated mesh (maistra#421) Fix test MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil (maistra#437) * MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil * Fix test MAISTRA-2682 Fix watch mechanism in federation (maistra#439) Previously, no events were read from the watch response, because the read started with an endless loop that waited for data to be available in the decoder's buffer. This never happened, because the buffer is only written to when you call decoder.Decode(); this function was never called because the code waited for the buffer to have data. MAISTRA-2683 Properly close incoming watch connections when shutting down (maistra#440) Log actual error returned by pollServices() (maistra#441) Previously, instead of the actual error, only the following error message was logged: "expected condition not met". MAISTRA-2439: Prevent federation from exporting services that are not visible to the federation gateway (maistra#432) By taking into consideration the service annotation `networking.istio.io/exportTo`. This annotation restricts where this service is visible: https://istio.io/latest/docs/reference/config/annotations/ If a service is not reachable from the federation gateway namespace due to this annotation, it should not be exported. MAISTRA-2617: Do not watch all namespaces in Extensions controller (maistra#425) When using MemberRoll, we should rely on it to provide the list of namespaces to watch. If not using it, defaults to command line arguments. This fixes an istiod startup error as seen in the logs: ``` github.com/maistra/xns-informer/pkg/informers/informer.go:204: Failed to watch *v1.ServiceMeshExtension: failed to list *v1.ServiceMeshExtension: servicemeshextensions.maistra.io is forbidden: User "system:serviceaccount:i1:istiod-service-account-basic" cannot list resource "servicemeshextensions" in API group "maistra.io" at the cluster scope ``` * Remove package export and extensions Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix creating discovery.Controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix calling nil ResourceManager Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Remove panicing from AppendNetworkGatewayHandler Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [misc] OSSM-774 Fix flaky TestStatusManager (maistra#456) This adds a little sleep to our unit tests for the StatusManager, because without it, we're running into the issue that we're updating a ServiceMeshPeer's status very quickly, and in some cases it might be that the last change has not been propagated when we're generating the patch for the next status change, which can lead to failures. This can happen in the real world, but you would need to change a ServiceMeshPeer's status within a few milliseconds, I doubt that it affects users. It would also be fixed with the next status update. For those reasons, I'm only fixing it in the test, with a Sleep() call. * Refactor manager_test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-1150 Fix flaky TestStatusManager unit test (maistra#478) Co-authored-by: Marko Lukša <marko.luksa@gmail.com> * OSSM-1252 Fix federation status updates (maistra#512) * Copy federation privileges from base to istio-discovery * Remove unnecessary ServiceMeshExtensions CRD Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add model.NetworkGatewaysHandler to federation controller to implement AppendNetworkGatewayHandler Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [federation] MAISTRA-2640 Add federation integration test (maistra#447) * Fix building federation test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add package gogo from maistra-2.2 to temporarily fix TbdsGenerator Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Disable configuring remote cluster in federation deployment Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [federation] OSSM-1128 Fix federation (maistra#480) * Fix SecretCacheClient Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Send initial XDS request for trust bundle from proxy Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Disable using EndpointSlices to fix error on getting federation-egress endpoints Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Remove unused serviceMeshExtensionController Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix lint errors Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Update maistra CRDs Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * fedration_cp_version_update (maistra#521) * OSSM-1529 Improve federation example script (maistra#522) * OSSM-1529 Improve federation example install.sh Previously, the script would fall back to using nodeports when the load balancer IP wasn't set. This meant that if the provision of the load balancer took too long, the SMCPs would be misconfigured and you had to run the install script again. With this change, the script now waits for the load balancer IP to appear. It never falls back to using node ports, because they never really worked (the nodes' hostnames typically aren't FQDN and the node ports are typically protected by firewalls). If the user wants to expose the federation ingresses in a different way, they can now set the environment variables MESH1_ADDRESS, MESH1_DISCOVERY_PORT, and MESH2_SERVICE_PORT (likewise for MESH2) and run the script. * Update Federation example README * Better "Waiting for load balancer" message * OSSM-1211 Fix federation locality failover issues (maistra#561) Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Signed-off-by: rcernich <rcernich@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: maistra-bot <57098434+maistra-bot@users.noreply.github.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <xuyuanlin_00@hotmail.com> * fix: removes deprecated gogo protobuf conversion * fix: goimport format * fix(lint): removes unused funcs * fix(lint): removes deprecated io/ioutil * fix(lint): disables staticcheck for federation tests it requires at least two clusters to make sense * fix(lint): use anypb.UnmarshalTo instead of ptypes * fix: no need to exclude grpcgen_test.go it seems to be fixed in v1.39 see: grpc/grpc-go#4476 * chore(backoff): aligns backoff dependency with v4 used by upstream * chore: reverts removed blank line - irrelevant for merge * chore(revive): adds explanation why json:inline is skipped from linting * OSSM-1962: Use EndpointSlices instead of Endpoints in federation controller (maistra#614) Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-2049: Fix handling ServiceAccounts in federation controller (maistra#627) * Fix collecting empty or repeated ServiceAccounts in federation controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Collect ServiceAccounts in sorted order Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-1093 Shorten exported resource name (maistra#653) * Shorten exported resource name * Fix import createResourceName + unit tests * Rearrange unit tests + renaming function * Gen and lint * Rearrange unit tests + renaming function Gen and lint * Fix minor changes * Error message RFC 1123 * Reorganize structs in TestStatusManager Instead of having three arrays (events, expectedStatuses, and assertions), we now only have a single array, where each entry is a triplet containing the event, the expected status and the assertion. This allows you to see the event and its expected effects together and not have to scroll up and down, matching the indexes of the three arrays. * OSSM-2193 Fix flaky TestStatusManager See comment in https://issues.redhat.com/browse/OSSM-2193 to understand why this change fixes the test. * fix: runs make gen * chore: explains why staticcheck linter is disabled for federation_test * OSSM-728 Configuration scripts for Federation on Z and P, and bare metal (maistra#670) * add config example scripts for IBM Systems Z and P * update multi-arch bookinfo deployment README, remove src * Update README.md * these are provided in the IBM repo * so README.md passes mdlinter * so README.md passes mdlinter * so README.md passes mdlinter * Update README.md * Move federation examples to samples/ directory * Rename template YAMLs to .yaml.template This makes the linter happy Co-authored-by: cfillekes <cfilleke@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com> * chore: removes obsolute TODO * chore: simplifies bool return expression * chore: removes redundant kubeClient check if initialization fails this func will not be reached anyway * chore(pkg): moves kube ctrl under servicemesh pkg folder * OSSM-2338: Remove env ISTIO_META_ROUTER_MODE from federation test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-2338: Remove "routerMode: sni-dnat" from federation samples Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * fix: adds operator.go customizations to kube.go clearly with cherry-pick we lost information about the file rename and thus the changes we made specificually for testing federation got lost * fix(test/operator): checks if east-west gw needs to be deployed * fix(federation): uses Unwrap to get instance of Federation registry * fix(tests): sets istiod-less remote flag to false that was the behaviour for maistra-2.3 and we need istiod to be present in order to have federation working * chore: gets registry just before it is needed * chore: explains why istiodlessremotes is needed to be set to false * chore: removes redundant import aliases * chore: removes name collisions * chore: removes redundant type conversion * fix: disables staticcheck linter for cluster req tests. * fix(tests): reverts timeout to original (but in minutes) * fix(tests): removes extra logging * chore: removes unnecessary logging * fix: uses existing CRD file references in charts * chore: removes multicluster label * chore: uses built-in namespace.NewOrFail instead of our impl * chore: introduces defaultTimeout const for federation tests * fix(lint): fixes go imports * fix(lint): removes unused variable * fix: naively wait 5s hoping that kind network will show up Signed-off-by: rcernich <rcernich@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Co-authored-by: Jacek Ewertowski <jewertow@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: maistra-bot <57098434+maistra-bot@users.noreply.github.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <xuyuanlin_00@hotmail.com> Co-authored-by: bmangoen <bmangoen@redhat.com> Co-authored-by: cfillekes <cfilleke@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com> OSSM-2376: Move kube controller to the federation package (maistra#718) Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> OSSM-2376: Don't start federation controllers until informers have synced (maistra#717) (maistra#720) * OSSM-2376: Don't start federation-discovery-controller until kube informer has synced Federation discovery controller fetches config map with remote CA root cert, so if the controller started before the informer has synced, it would fail to fetch the config map. Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Store ConfigMap informer in a field of the discovery.Controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Refactor ResourceManager and don't start federation controller until informers has synced Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Simplify Start and HasSynced functions in federation controllers Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Move kube controller to the federation package Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Federation example fixes (maistra#758) * Use default version in federation example SMCPs * Fix paths in federation example OSSM-3599 Federation egress-gateway gets wrong network gateway endpoints (maistra#775) * OSSM-3599 Federation egress-gateway gets wrong update of network gateway endpoints * Deprecate GatewayEndpoints on server side * Remove resyncNetworkGateways in unit tests * Fix lint * Deprecate NetworkGatewayEndpoints and fix tests Refactor federation tests (maistra#841) * Refactor federation tests Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add more test cases Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> --------- Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Yann Liu <yannliu@redhat.com>
maistra#699) (maistra#844) * [federation] Introduces federation deployment (maistra#585) * [federation] Initial federation implementation * MAISTRA-2194 Add server/client code for Federation Service Discovery v1 * MAISTRA-2195 Implement /watch endpoint * MAISTRA-2293 add CRD and controller for federating meshes * MAISTRA-2294 create CRD for federation ServiceExport (maistra#324) * MAISTRA-2294 update example VirtualService resources for ratings and mongodb (maistra#333) * [federation] MAISTRA-2295 create CRD for federation ServiceImport (maistra#336) Signed-off-by: rcernich <rcernich@redhat.com> * [misc] Use objects and clients from maistra/api repo - Remove local objects and clients - Update Makefile * [federation] MAISTRA-2309 create CRD for FederationStatus (maistra#348) Signed-off-by: rcernich <rcernich@redhat.com> * [federation] Federation fixes and improvements MAISTRA-2423 update federation api to v1 Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2424 minor updates to federation api Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2427 configure locality info on imported services Signed-off-by: rcernich <rcernich@redhat.com> Cherry-pick multi-root support (maistra#387) * Update go-control-plane to v0.9.9 * Support multiple roots Squashed commit, contains: - MAISTRA-2325 Distribute trust bundles over SDS - MAISTRA-2390 Push trust bundle updates through xDS (maistra#357) MAISTRA-2425 move spec.security.certificateChain to ConfigMap reference; add ability to specify ports for service and discovery (maistra#392) Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2426 move FederationStatus into MeshFederation (maistra#393) Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2513 federation API refinements Signed-off-by: rcernich <rcernich@redhat.com> [federation] MAISTRA-2237 Encrypt service discovery traffic (maistra#411) MAISTRA-2610 Prefix federation discovery endpoints with /v1/ (maistra#422) MAISTRA-2297 Support updates of federation resources (maistra#417) MAISTRA-2375: Do not create automatic routes for Federation Gateways Remove a redundant call `setHostname()` is already being called within `NameForService()` see https://github.com/maistra/istio/blob/21ee900cf8825711f70d88dc97afcf6862ed2626/pkg/servicemesh/federation/common/namemapping.go lines 83, 120, 129 Remove techPreview.meshConfig from PoC example It's set by default now. MAISTRA-2611 Fix deletion of service exports to federated mesh (maistra#421) Fix test MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil (maistra#437) * MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil * Fix test MAISTRA-2682 Fix watch mechanism in federation (maistra#439) Previously, no events were read from the watch response, because the read started with an endless loop that waited for data to be available in the decoder's buffer. This never happened, because the buffer is only written to when you call decoder.Decode(); this function was never called because the code waited for the buffer to have data. MAISTRA-2683 Properly close incoming watch connections when shutting down (maistra#440) Log actual error returned by pollServices() (maistra#441) Previously, instead of the actual error, only the following error message was logged: "expected condition not met". MAISTRA-2439: Prevent federation from exporting services that are not visible to the federation gateway (maistra#432) By taking into consideration the service annotation `networking.istio.io/exportTo`. This annotation restricts where this service is visible: https://istio.io/latest/docs/reference/config/annotations/ If a service is not reachable from the federation gateway namespace due to this annotation, it should not be exported. MAISTRA-2617: Do not watch all namespaces in Extensions controller (maistra#425) When using MemberRoll, we should rely on it to provide the list of namespaces to watch. If not using it, defaults to command line arguments. This fixes an istiod startup error as seen in the logs: ``` github.com/maistra/xns-informer/pkg/informers/informer.go:204: Failed to watch *v1.ServiceMeshExtension: failed to list *v1.ServiceMeshExtension: servicemeshextensions.maistra.io is forbidden: User "system:serviceaccount:i1:istiod-service-account-basic" cannot list resource "servicemeshextensions" in API group "maistra.io" at the cluster scope ``` * Remove package export and extensions Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix creating discovery.Controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix calling nil ResourceManager Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Remove panicing from AppendNetworkGatewayHandler Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [misc] OSSM-774 Fix flaky TestStatusManager (maistra#456) This adds a little sleep to our unit tests for the StatusManager, because without it, we're running into the issue that we're updating a ServiceMeshPeer's status very quickly, and in some cases it might be that the last change has not been propagated when we're generating the patch for the next status change, which can lead to failures. This can happen in the real world, but you would need to change a ServiceMeshPeer's status within a few milliseconds, I doubt that it affects users. It would also be fixed with the next status update. For those reasons, I'm only fixing it in the test, with a Sleep() call. * Refactor manager_test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-1150 Fix flaky TestStatusManager unit test (maistra#478) Co-authored-by: Marko Lukša <marko.luksa@gmail.com> * OSSM-1252 Fix federation status updates (maistra#512) * Copy federation privileges from base to istio-discovery * Remove unnecessary ServiceMeshExtensions CRD Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add model.NetworkGatewaysHandler to federation controller to implement AppendNetworkGatewayHandler Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [federation] MAISTRA-2640 Add federation integration test (maistra#447) * Fix building federation test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add package gogo from maistra-2.2 to temporarily fix TbdsGenerator Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Disable configuring remote cluster in federation deployment Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [federation] OSSM-1128 Fix federation (maistra#480) * Fix SecretCacheClient Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Send initial XDS request for trust bundle from proxy Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Disable using EndpointSlices to fix error on getting federation-egress endpoints Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Remove unused serviceMeshExtensionController Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix lint errors Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Update maistra CRDs Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * fedration_cp_version_update (maistra#521) * OSSM-1529 Improve federation example script (maistra#522) * OSSM-1529 Improve federation example install.sh Previously, the script would fall back to using nodeports when the load balancer IP wasn't set. This meant that if the provision of the load balancer took too long, the SMCPs would be misconfigured and you had to run the install script again. With this change, the script now waits for the load balancer IP to appear. It never falls back to using node ports, because they never really worked (the nodes' hostnames typically aren't FQDN and the node ports are typically protected by firewalls). If the user wants to expose the federation ingresses in a different way, they can now set the environment variables MESH1_ADDRESS, MESH1_DISCOVERY_PORT, and MESH2_SERVICE_PORT (likewise for MESH2) and run the script. * Update Federation example README * Better "Waiting for load balancer" message * OSSM-1211 Fix federation locality failover issues (maistra#561) Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Signed-off-by: rcernich <rcernich@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: maistra-bot <57098434+maistra-bot@users.noreply.github.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <xuyuanlin_00@hotmail.com> * fix: removes deprecated gogo protobuf conversion * fix: goimport format * fix(lint): removes unused funcs * fix(lint): removes deprecated io/ioutil * fix(lint): disables staticcheck for federation tests it requires at least two clusters to make sense * fix(lint): use anypb.UnmarshalTo instead of ptypes * fix: no need to exclude grpcgen_test.go it seems to be fixed in v1.39 see: grpc/grpc-go#4476 * chore(backoff): aligns backoff dependency with v4 used by upstream * chore: reverts removed blank line - irrelevant for merge * chore(revive): adds explanation why json:inline is skipped from linting * OSSM-1962: Use EndpointSlices instead of Endpoints in federation controller (maistra#614) Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-2049: Fix handling ServiceAccounts in federation controller (maistra#627) * Fix collecting empty or repeated ServiceAccounts in federation controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Collect ServiceAccounts in sorted order Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-1093 Shorten exported resource name (maistra#653) * Shorten exported resource name * Fix import createResourceName + unit tests * Rearrange unit tests + renaming function * Gen and lint * Rearrange unit tests + renaming function Gen and lint * Fix minor changes * Error message RFC 1123 * Reorganize structs in TestStatusManager Instead of having three arrays (events, expectedStatuses, and assertions), we now only have a single array, where each entry is a triplet containing the event, the expected status and the assertion. This allows you to see the event and its expected effects together and not have to scroll up and down, matching the indexes of the three arrays. * OSSM-2193 Fix flaky TestStatusManager See comment in https://issues.redhat.com/browse/OSSM-2193 to understand why this change fixes the test. * fix: runs make gen * chore: explains why staticcheck linter is disabled for federation_test * OSSM-728 Configuration scripts for Federation on Z and P, and bare metal (maistra#670) * add config example scripts for IBM Systems Z and P * update multi-arch bookinfo deployment README, remove src * Update README.md * these are provided in the IBM repo * so README.md passes mdlinter * so README.md passes mdlinter * so README.md passes mdlinter * Update README.md * Move federation examples to samples/ directory * Rename template YAMLs to .yaml.template This makes the linter happy Co-authored-by: cfillekes <cfilleke@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com> * chore: removes obsolute TODO * chore: simplifies bool return expression * chore: removes redundant kubeClient check if initialization fails this func will not be reached anyway * chore(pkg): moves kube ctrl under servicemesh pkg folder * OSSM-2338: Remove env ISTIO_META_ROUTER_MODE from federation test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-2338: Remove "routerMode: sni-dnat" from federation samples Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * fix: adds operator.go customizations to kube.go clearly with cherry-pick we lost information about the file rename and thus the changes we made specificually for testing federation got lost * fix(test/operator): checks if east-west gw needs to be deployed * fix(federation): uses Unwrap to get instance of Federation registry * fix(tests): sets istiod-less remote flag to false that was the behaviour for maistra-2.3 and we need istiod to be present in order to have federation working * chore: gets registry just before it is needed * chore: explains why istiodlessremotes is needed to be set to false * chore: removes redundant import aliases * chore: removes name collisions * chore: removes redundant type conversion * fix: disables staticcheck linter for cluster req tests. * fix(tests): reverts timeout to original (but in minutes) * fix(tests): removes extra logging * chore: removes unnecessary logging * fix: uses existing CRD file references in charts * chore: removes multicluster label * chore: uses built-in namespace.NewOrFail instead of our impl * chore: introduces defaultTimeout const for federation tests * fix(lint): fixes go imports * fix(lint): removes unused variable * fix: naively wait 5s hoping that kind network will show up Signed-off-by: rcernich <rcernich@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Co-authored-by: Jacek Ewertowski <jewertow@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: maistra-bot <57098434+maistra-bot@users.noreply.github.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <xuyuanlin_00@hotmail.com> Co-authored-by: bmangoen <bmangoen@redhat.com> Co-authored-by: cfillekes <cfilleke@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com> OSSM-2376: Move kube controller to the federation package (maistra#718) Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> OSSM-2376: Don't start federation controllers until informers have synced (maistra#717) (maistra#720) * OSSM-2376: Don't start federation-discovery-controller until kube informer has synced Federation discovery controller fetches config map with remote CA root cert, so if the controller started before the informer has synced, it would fail to fetch the config map. Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Store ConfigMap informer in a field of the discovery.Controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Refactor ResourceManager and don't start federation controller until informers has synced Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Simplify Start and HasSynced functions in federation controllers Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Move kube controller to the federation package Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Federation example fixes (maistra#758) * Use default version in federation example SMCPs * Fix paths in federation example OSSM-3599 Federation egress-gateway gets wrong network gateway endpoints (maistra#775) * OSSM-3599 Federation egress-gateway gets wrong update of network gateway endpoints * Deprecate GatewayEndpoints on server side * Remove resyncNetworkGateways in unit tests * Fix lint * Deprecate NetworkGatewayEndpoints and fix tests Refactor federation tests (maistra#841) * Refactor federation tests Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add more test cases Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> --------- Co-authored-by: Bartosz Majsak <bartosz.majsak@gmail.com> Co-authored-by: Jacek Ewertowski <jewertow@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <yuanlin.xu@redhat.com> Co-authored-by: Brian Mangoenpawiro <bmangoen@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com> Signed-off-by: Yann Liu <yannliu@redhat.com>
… (maistra#844) * [federation] Introduces federation deployment (maistra#585) * [federation] Initial federation implementation * MAISTRA-2194 Add server/client code for Federation Service Discovery v1 * MAISTRA-2195 Implement /watch endpoint * MAISTRA-2293 add CRD and controller for federating meshes * MAISTRA-2294 create CRD for federation ServiceExport (maistra#324) * MAISTRA-2294 update example VirtualService resources for ratings and mongodb (maistra#333) * [federation] MAISTRA-2295 create CRD for federation ServiceImport (maistra#336) Signed-off-by: rcernich <rcernich@redhat.com> * [misc] Use objects and clients from maistra/api repo - Remove local objects and clients - Update Makefile * [federation] MAISTRA-2309 create CRD for FederationStatus (maistra#348) Signed-off-by: rcernich <rcernich@redhat.com> * [federation] Federation fixes and improvements MAISTRA-2423 update federation api to v1 Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2424 minor updates to federation api Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2427 configure locality info on imported services Signed-off-by: rcernich <rcernich@redhat.com> Cherry-pick multi-root support (maistra#387) * Update go-control-plane to v0.9.9 * Support multiple roots Squashed commit, contains: - MAISTRA-2325 Distribute trust bundles over SDS - MAISTRA-2390 Push trust bundle updates through xDS (maistra#357) MAISTRA-2425 move spec.security.certificateChain to ConfigMap reference; add ability to specify ports for service and discovery (maistra#392) Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2426 move FederationStatus into MeshFederation (maistra#393) Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2513 federation API refinements Signed-off-by: rcernich <rcernich@redhat.com> [federation] MAISTRA-2237 Encrypt service discovery traffic (maistra#411) MAISTRA-2610 Prefix federation discovery endpoints with /v1/ (maistra#422) MAISTRA-2297 Support updates of federation resources (maistra#417) MAISTRA-2375: Do not create automatic routes for Federation Gateways Remove a redundant call `setHostname()` is already being called within `NameForService()` see https://github.com/maistra/istio/blob/21ee900cf8825711f70d88dc97afcf6862ed2626/pkg/servicemesh/federation/common/namemapping.go lines 83, 120, 129 Remove techPreview.meshConfig from PoC example It's set by default now. MAISTRA-2611 Fix deletion of service exports to federated mesh (maistra#421) Fix test MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil (maistra#437) * MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil * Fix test MAISTRA-2682 Fix watch mechanism in federation (maistra#439) Previously, no events were read from the watch response, because the read started with an endless loop that waited for data to be available in the decoder's buffer. This never happened, because the buffer is only written to when you call decoder.Decode(); this function was never called because the code waited for the buffer to have data. MAISTRA-2683 Properly close incoming watch connections when shutting down (maistra#440) Log actual error returned by pollServices() (maistra#441) Previously, instead of the actual error, only the following error message was logged: "expected condition not met". MAISTRA-2439: Prevent federation from exporting services that are not visible to the federation gateway (maistra#432) By taking into consideration the service annotation `networking.istio.io/exportTo`. This annotation restricts where this service is visible: https://istio.io/latest/docs/reference/config/annotations/ If a service is not reachable from the federation gateway namespace due to this annotation, it should not be exported. MAISTRA-2617: Do not watch all namespaces in Extensions controller (maistra#425) When using MemberRoll, we should rely on it to provide the list of namespaces to watch. If not using it, defaults to command line arguments. This fixes an istiod startup error as seen in the logs: ``` github.com/maistra/xns-informer/pkg/informers/informer.go:204: Failed to watch *v1.ServiceMeshExtension: failed to list *v1.ServiceMeshExtension: servicemeshextensions.maistra.io is forbidden: User "system:serviceaccount:i1:istiod-service-account-basic" cannot list resource "servicemeshextensions" in API group "maistra.io" at the cluster scope ``` * Remove package export and extensions Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix creating discovery.Controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix calling nil ResourceManager Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Remove panicing from AppendNetworkGatewayHandler Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [misc] OSSM-774 Fix flaky TestStatusManager (maistra#456) This adds a little sleep to our unit tests for the StatusManager, because without it, we're running into the issue that we're updating a ServiceMeshPeer's status very quickly, and in some cases it might be that the last change has not been propagated when we're generating the patch for the next status change, which can lead to failures. This can happen in the real world, but you would need to change a ServiceMeshPeer's status within a few milliseconds, I doubt that it affects users. It would also be fixed with the next status update. For those reasons, I'm only fixing it in the test, with a Sleep() call. * Refactor manager_test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-1150 Fix flaky TestStatusManager unit test (maistra#478) Co-authored-by: Marko Lukša <marko.luksa@gmail.com> * OSSM-1252 Fix federation status updates (maistra#512) * Copy federation privileges from base to istio-discovery * Remove unnecessary ServiceMeshExtensions CRD Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add model.NetworkGatewaysHandler to federation controller to implement AppendNetworkGatewayHandler Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [federation] MAISTRA-2640 Add federation integration test (maistra#447) * Fix building federation test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add package gogo from maistra-2.2 to temporarily fix TbdsGenerator Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Disable configuring remote cluster in federation deployment Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [federation] OSSM-1128 Fix federation (maistra#480) * Fix SecretCacheClient Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Send initial XDS request for trust bundle from proxy Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Disable using EndpointSlices to fix error on getting federation-egress endpoints Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Remove unused serviceMeshExtensionController Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix lint errors Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Update maistra CRDs Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * fedration_cp_version_update (maistra#521) * OSSM-1529 Improve federation example script (maistra#522) * OSSM-1529 Improve federation example install.sh Previously, the script would fall back to using nodeports when the load balancer IP wasn't set. This meant that if the provision of the load balancer took too long, the SMCPs would be misconfigured and you had to run the install script again. With this change, the script now waits for the load balancer IP to appear. It never falls back to using node ports, because they never really worked (the nodes' hostnames typically aren't FQDN and the node ports are typically protected by firewalls). If the user wants to expose the federation ingresses in a different way, they can now set the environment variables MESH1_ADDRESS, MESH1_DISCOVERY_PORT, and MESH2_SERVICE_PORT (likewise for MESH2) and run the script. * Update Federation example README * Better "Waiting for load balancer" message * OSSM-1211 Fix federation locality failover issues (maistra#561) Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Signed-off-by: rcernich <rcernich@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: maistra-bot <57098434+maistra-bot@users.noreply.github.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <xuyuanlin_00@hotmail.com> * fix: removes deprecated gogo protobuf conversion * fix: goimport format * fix(lint): removes unused funcs * fix(lint): removes deprecated io/ioutil * fix(lint): disables staticcheck for federation tests it requires at least two clusters to make sense * fix(lint): use anypb.UnmarshalTo instead of ptypes * fix: no need to exclude grpcgen_test.go it seems to be fixed in v1.39 see: grpc/grpc-go#4476 * chore(backoff): aligns backoff dependency with v4 used by upstream * chore: reverts removed blank line - irrelevant for merge * chore(revive): adds explanation why json:inline is skipped from linting * OSSM-1962: Use EndpointSlices instead of Endpoints in federation controller (maistra#614) Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-2049: Fix handling ServiceAccounts in federation controller (maistra#627) * Fix collecting empty or repeated ServiceAccounts in federation controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Collect ServiceAccounts in sorted order Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-1093 Shorten exported resource name (maistra#653) * Shorten exported resource name * Fix import createResourceName + unit tests * Rearrange unit tests + renaming function * Gen and lint * Rearrange unit tests + renaming function Gen and lint * Fix minor changes * Error message RFC 1123 * Reorganize structs in TestStatusManager Instead of having three arrays (events, expectedStatuses, and assertions), we now only have a single array, where each entry is a triplet containing the event, the expected status and the assertion. This allows you to see the event and its expected effects together and not have to scroll up and down, matching the indexes of the three arrays. * OSSM-2193 Fix flaky TestStatusManager See comment in https://issues.redhat.com/browse/OSSM-2193 to understand why this change fixes the test. * fix: runs make gen * chore: explains why staticcheck linter is disabled for federation_test * OSSM-728 Configuration scripts for Federation on Z and P, and bare metal (maistra#670) * add config example scripts for IBM Systems Z and P * update multi-arch bookinfo deployment README, remove src * Update README.md * these are provided in the IBM repo * so README.md passes mdlinter * so README.md passes mdlinter * so README.md passes mdlinter * Update README.md * Move federation examples to samples/ directory * Rename template YAMLs to .yaml.template This makes the linter happy Co-authored-by: cfillekes <cfilleke@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com> * chore: removes obsolute TODO * chore: simplifies bool return expression * chore: removes redundant kubeClient check if initialization fails this func will not be reached anyway * chore(pkg): moves kube ctrl under servicemesh pkg folder * OSSM-2338: Remove env ISTIO_META_ROUTER_MODE from federation test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-2338: Remove "routerMode: sni-dnat" from federation samples Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * fix: adds operator.go customizations to kube.go clearly with cherry-pick we lost information about the file rename and thus the changes we made specificually for testing federation got lost * fix(test/operator): checks if east-west gw needs to be deployed * fix(federation): uses Unwrap to get instance of Federation registry * fix(tests): sets istiod-less remote flag to false that was the behaviour for maistra-2.3 and we need istiod to be present in order to have federation working * chore: gets registry just before it is needed * chore: explains why istiodlessremotes is needed to be set to false * chore: removes redundant import aliases * chore: removes name collisions * chore: removes redundant type conversion * fix: disables staticcheck linter for cluster req tests. * fix(tests): reverts timeout to original (but in minutes) * fix(tests): removes extra logging * chore: removes unnecessary logging * fix: uses existing CRD file references in charts * chore: removes multicluster label * chore: uses built-in namespace.NewOrFail instead of our impl * chore: introduces defaultTimeout const for federation tests * fix(lint): fixes go imports * fix(lint): removes unused variable * fix: naively wait 5s hoping that kind network will show up Signed-off-by: rcernich <rcernich@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Co-authored-by: Jacek Ewertowski <jewertow@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: maistra-bot <57098434+maistra-bot@users.noreply.github.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <xuyuanlin_00@hotmail.com> Co-authored-by: bmangoen <bmangoen@redhat.com> Co-authored-by: cfillekes <cfilleke@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com> OSSM-2376: Move kube controller to the federation package (maistra#718) Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> OSSM-2376: Don't start federation controllers until informers have synced (maistra#717) (maistra#720) * OSSM-2376: Don't start federation-discovery-controller until kube informer has synced Federation discovery controller fetches config map with remote CA root cert, so if the controller started before the informer has synced, it would fail to fetch the config map. Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Store ConfigMap informer in a field of the discovery.Controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Refactor ResourceManager and don't start federation controller until informers has synced Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Simplify Start and HasSynced functions in federation controllers Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Move kube controller to the federation package Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Federation example fixes (maistra#758) * Use default version in federation example SMCPs * Fix paths in federation example OSSM-3599 Federation egress-gateway gets wrong network gateway endpoints (maistra#775) * OSSM-3599 Federation egress-gateway gets wrong update of network gateway endpoints * Deprecate GatewayEndpoints on server side * Remove resyncNetworkGateways in unit tests * Fix lint * Deprecate NetworkGatewayEndpoints and fix tests Refactor federation tests (maistra#841) * Refactor federation tests Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add more test cases Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Reimplement InstancesByPort method `InstancesByPort` is used by the federation server, which was removed in the upstream istio/istio#46329. We reimplement it to support the federation server. --------- Co-authored-by: Bartosz Majsak <bartosz.majsak@gmail.com> Co-authored-by: Jacek Ewertowski <jewertow@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <yuanlin.xu@redhat.com> Co-authored-by: Brian Mangoenpawiro <bmangoen@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com> Signed-off-by: Yann Liu <yannliu@redhat.com>
… (maistra#844) * [federation] Introduces federation deployment (maistra#585) * [federation] Initial federation implementation * MAISTRA-2194 Add server/client code for Federation Service Discovery v1 * MAISTRA-2195 Implement /watch endpoint * MAISTRA-2293 add CRD and controller for federating meshes * MAISTRA-2294 create CRD for federation ServiceExport (maistra#324) * MAISTRA-2294 update example VirtualService resources for ratings and mongodb (maistra#333) * [federation] MAISTRA-2295 create CRD for federation ServiceImport (maistra#336) Signed-off-by: rcernich <rcernich@redhat.com> * [misc] Use objects and clients from maistra/api repo - Remove local objects and clients - Update Makefile * [federation] MAISTRA-2309 create CRD for FederationStatus (maistra#348) Signed-off-by: rcernich <rcernich@redhat.com> * [federation] Federation fixes and improvements MAISTRA-2423 update federation api to v1 Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2424 minor updates to federation api Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2427 configure locality info on imported services Signed-off-by: rcernich <rcernich@redhat.com> Cherry-pick multi-root support (maistra#387) * Update go-control-plane to v0.9.9 * Support multiple roots Squashed commit, contains: - MAISTRA-2325 Distribute trust bundles over SDS - MAISTRA-2390 Push trust bundle updates through xDS (maistra#357) MAISTRA-2425 move spec.security.certificateChain to ConfigMap reference; add ability to specify ports for service and discovery (maistra#392) Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2426 move FederationStatus into MeshFederation (maistra#393) Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2513 federation API refinements Signed-off-by: rcernich <rcernich@redhat.com> [federation] MAISTRA-2237 Encrypt service discovery traffic (maistra#411) MAISTRA-2610 Prefix federation discovery endpoints with /v1/ (maistra#422) MAISTRA-2297 Support updates of federation resources (maistra#417) MAISTRA-2375: Do not create automatic routes for Federation Gateways Remove a redundant call `setHostname()` is already being called within `NameForService()` see https://github.com/maistra/istio/blob/21ee900cf8825711f70d88dc97afcf6862ed2626/pkg/servicemesh/federation/common/namemapping.go lines 83, 120, 129 Remove techPreview.meshConfig from PoC example It's set by default now. MAISTRA-2611 Fix deletion of service exports to federated mesh (maistra#421) Fix test MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil (maistra#437) * MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil * Fix test MAISTRA-2682 Fix watch mechanism in federation (maistra#439) Previously, no events were read from the watch response, because the read started with an endless loop that waited for data to be available in the decoder's buffer. This never happened, because the buffer is only written to when you call decoder.Decode(); this function was never called because the code waited for the buffer to have data. MAISTRA-2683 Properly close incoming watch connections when shutting down (maistra#440) Log actual error returned by pollServices() (maistra#441) Previously, instead of the actual error, only the following error message was logged: "expected condition not met". MAISTRA-2439: Prevent federation from exporting services that are not visible to the federation gateway (maistra#432) By taking into consideration the service annotation `networking.istio.io/exportTo`. This annotation restricts where this service is visible: https://istio.io/latest/docs/reference/config/annotations/ If a service is not reachable from the federation gateway namespace due to this annotation, it should not be exported. MAISTRA-2617: Do not watch all namespaces in Extensions controller (maistra#425) When using MemberRoll, we should rely on it to provide the list of namespaces to watch. If not using it, defaults to command line arguments. This fixes an istiod startup error as seen in the logs: ``` github.com/maistra/xns-informer/pkg/informers/informer.go:204: Failed to watch *v1.ServiceMeshExtension: failed to list *v1.ServiceMeshExtension: servicemeshextensions.maistra.io is forbidden: User "system:serviceaccount:i1:istiod-service-account-basic" cannot list resource "servicemeshextensions" in API group "maistra.io" at the cluster scope ``` * Remove package export and extensions Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix creating discovery.Controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix calling nil ResourceManager Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Remove panicing from AppendNetworkGatewayHandler Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [misc] OSSM-774 Fix flaky TestStatusManager (maistra#456) This adds a little sleep to our unit tests for the StatusManager, because without it, we're running into the issue that we're updating a ServiceMeshPeer's status very quickly, and in some cases it might be that the last change has not been propagated when we're generating the patch for the next status change, which can lead to failures. This can happen in the real world, but you would need to change a ServiceMeshPeer's status within a few milliseconds, I doubt that it affects users. It would also be fixed with the next status update. For those reasons, I'm only fixing it in the test, with a Sleep() call. * Refactor manager_test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-1150 Fix flaky TestStatusManager unit test (maistra#478) Co-authored-by: Marko Lukša <marko.luksa@gmail.com> * OSSM-1252 Fix federation status updates (maistra#512) * Copy federation privileges from base to istio-discovery * Remove unnecessary ServiceMeshExtensions CRD Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add model.NetworkGatewaysHandler to federation controller to implement AppendNetworkGatewayHandler Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [federation] MAISTRA-2640 Add federation integration test (maistra#447) * Fix building federation test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add package gogo from maistra-2.2 to temporarily fix TbdsGenerator Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Disable configuring remote cluster in federation deployment Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [federation] OSSM-1128 Fix federation (maistra#480) * Fix SecretCacheClient Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Send initial XDS request for trust bundle from proxy Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Disable using EndpointSlices to fix error on getting federation-egress endpoints Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Remove unused serviceMeshExtensionController Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix lint errors Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Update maistra CRDs Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * fedration_cp_version_update (maistra#521) * OSSM-1529 Improve federation example script (maistra#522) * OSSM-1529 Improve federation example install.sh Previously, the script would fall back to using nodeports when the load balancer IP wasn't set. This meant that if the provision of the load balancer took too long, the SMCPs would be misconfigured and you had to run the install script again. With this change, the script now waits for the load balancer IP to appear. It never falls back to using node ports, because they never really worked (the nodes' hostnames typically aren't FQDN and the node ports are typically protected by firewalls). If the user wants to expose the federation ingresses in a different way, they can now set the environment variables MESH1_ADDRESS, MESH1_DISCOVERY_PORT, and MESH2_SERVICE_PORT (likewise for MESH2) and run the script. * Update Federation example README * Better "Waiting for load balancer" message * OSSM-1211 Fix federation locality failover issues (maistra#561) Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Signed-off-by: rcernich <rcernich@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: maistra-bot <57098434+maistra-bot@users.noreply.github.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <xuyuanlin_00@hotmail.com> * fix: removes deprecated gogo protobuf conversion * fix: goimport format * fix(lint): removes unused funcs * fix(lint): removes deprecated io/ioutil * fix(lint): disables staticcheck for federation tests it requires at least two clusters to make sense * fix(lint): use anypb.UnmarshalTo instead of ptypes * fix: no need to exclude grpcgen_test.go it seems to be fixed in v1.39 see: grpc/grpc-go#4476 * chore(backoff): aligns backoff dependency with v4 used by upstream * chore: reverts removed blank line - irrelevant for merge * chore(revive): adds explanation why json:inline is skipped from linting * OSSM-1962: Use EndpointSlices instead of Endpoints in federation controller (maistra#614) Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-2049: Fix handling ServiceAccounts in federation controller (maistra#627) * Fix collecting empty or repeated ServiceAccounts in federation controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Collect ServiceAccounts in sorted order Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-1093 Shorten exported resource name (maistra#653) * Shorten exported resource name * Fix import createResourceName + unit tests * Rearrange unit tests + renaming function * Gen and lint * Rearrange unit tests + renaming function Gen and lint * Fix minor changes * Error message RFC 1123 * Reorganize structs in TestStatusManager Instead of having three arrays (events, expectedStatuses, and assertions), we now only have a single array, where each entry is a triplet containing the event, the expected status and the assertion. This allows you to see the event and its expected effects together and not have to scroll up and down, matching the indexes of the three arrays. * OSSM-2193 Fix flaky TestStatusManager See comment in https://issues.redhat.com/browse/OSSM-2193 to understand why this change fixes the test. * fix: runs make gen * chore: explains why staticcheck linter is disabled for federation_test * OSSM-728 Configuration scripts for Federation on Z and P, and bare metal (maistra#670) * add config example scripts for IBM Systems Z and P * update multi-arch bookinfo deployment README, remove src * Update README.md * these are provided in the IBM repo * so README.md passes mdlinter * so README.md passes mdlinter * so README.md passes mdlinter * Update README.md * Move federation examples to samples/ directory * Rename template YAMLs to .yaml.template This makes the linter happy Co-authored-by: cfillekes <cfilleke@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com> * chore: removes obsolute TODO * chore: simplifies bool return expression * chore: removes redundant kubeClient check if initialization fails this func will not be reached anyway * chore(pkg): moves kube ctrl under servicemesh pkg folder * OSSM-2338: Remove env ISTIO_META_ROUTER_MODE from federation test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-2338: Remove "routerMode: sni-dnat" from federation samples Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * fix: adds operator.go customizations to kube.go clearly with cherry-pick we lost information about the file rename and thus the changes we made specificually for testing federation got lost * fix(test/operator): checks if east-west gw needs to be deployed * fix(federation): uses Unwrap to get instance of Federation registry * fix(tests): sets istiod-less remote flag to false that was the behaviour for maistra-2.3 and we need istiod to be present in order to have federation working * chore: gets registry just before it is needed * chore: explains why istiodlessremotes is needed to be set to false * chore: removes redundant import aliases * chore: removes name collisions * chore: removes redundant type conversion * fix: disables staticcheck linter for cluster req tests. * fix(tests): reverts timeout to original (but in minutes) * fix(tests): removes extra logging * chore: removes unnecessary logging * fix: uses existing CRD file references in charts * chore: removes multicluster label * chore: uses built-in namespace.NewOrFail instead of our impl * chore: introduces defaultTimeout const for federation tests * fix(lint): fixes go imports * fix(lint): removes unused variable * fix: naively wait 5s hoping that kind network will show up Signed-off-by: rcernich <rcernich@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Co-authored-by: Jacek Ewertowski <jewertow@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: maistra-bot <57098434+maistra-bot@users.noreply.github.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <xuyuanlin_00@hotmail.com> Co-authored-by: bmangoen <bmangoen@redhat.com> Co-authored-by: cfillekes <cfilleke@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com> OSSM-2376: Move kube controller to the federation package (maistra#718) Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> OSSM-2376: Don't start federation controllers until informers have synced (maistra#717) (maistra#720) * OSSM-2376: Don't start federation-discovery-controller until kube informer has synced Federation discovery controller fetches config map with remote CA root cert, so if the controller started before the informer has synced, it would fail to fetch the config map. Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Store ConfigMap informer in a field of the discovery.Controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Refactor ResourceManager and don't start federation controller until informers has synced Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Simplify Start and HasSynced functions in federation controllers Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Move kube controller to the federation package Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Federation example fixes (maistra#758) * Use default version in federation example SMCPs * Fix paths in federation example OSSM-3599 Federation egress-gateway gets wrong network gateway endpoints (maistra#775) * OSSM-3599 Federation egress-gateway gets wrong update of network gateway endpoints * Deprecate GatewayEndpoints on server side * Remove resyncNetworkGateways in unit tests * Fix lint * Deprecate NetworkGatewayEndpoints and fix tests Refactor federation tests (maistra#841) * Refactor federation tests Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add more test cases Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Reimplement InstancesByPort method `InstancesByPort` is used by the federation server, which was removed in the upstream istio/istio#46329. We reimplement it to support the federation server. --------- Co-authored-by: Bartosz Majsak <bartosz.majsak@gmail.com> Co-authored-by: Jacek Ewertowski <jewertow@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <yuanlin.xu@redhat.com> Co-authored-by: Brian Mangoenpawiro <bmangoen@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com> Signed-off-by: Yann Liu <yannliu@redhat.com>
* [federation] Introduces federation deployment (#585) * [federation] Initial federation implementation * MAISTRA-2194 Add server/client code for Federation Service Discovery v1 * MAISTRA-2195 Implement /watch endpoint * MAISTRA-2293 add CRD and controller for federating meshes * MAISTRA-2294 create CRD for federation ServiceExport (#324) * MAISTRA-2294 update example VirtualService resources for ratings and mongodb (#333) * [federation] MAISTRA-2295 create CRD for federation ServiceImport (#336) Signed-off-by: rcernich <rcernich@redhat.com> * [misc] Use objects and clients from maistra/api repo - Remove local objects and clients - Update Makefile * [federation] MAISTRA-2309 create CRD for FederationStatus (#348) Signed-off-by: rcernich <rcernich@redhat.com> * [federation] Federation fixes and improvements MAISTRA-2423 update federation api to v1 Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2424 minor updates to federation api Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2427 configure locality info on imported services Signed-off-by: rcernich <rcernich@redhat.com> Cherry-pick multi-root support (#387) * Update go-control-plane to v0.9.9 * Support multiple roots Squashed commit, contains: - MAISTRA-2325 Distribute trust bundles over SDS - MAISTRA-2390 Push trust bundle updates through xDS (#357) MAISTRA-2425 move spec.security.certificateChain to ConfigMap reference; add ability to specify ports for service and discovery (#392) Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2426 move FederationStatus into MeshFederation (#393) Signed-off-by: rcernich <rcernich@redhat.com> MAISTRA-2513 federation API refinements Signed-off-by: rcernich <rcernich@redhat.com> [federation] MAISTRA-2237 Encrypt service discovery traffic (#411) MAISTRA-2610 Prefix federation discovery endpoints with /v1/ (#422) MAISTRA-2297 Support updates of federation resources (#417) MAISTRA-2375: Do not create automatic routes for Federation Gateways Remove a redundant call `setHostname()` is already being called within `NameForService()` see https://github.com/maistra/istio/blob/21ee900cf8825711f70d88dc97afcf6862ed2626/pkg/servicemesh/federation/common/namemapping.go lines 83, 120, 129 Remove techPreview.meshConfig from PoC example It's set by default now. MAISTRA-2611 Fix deletion of service exports to federated mesh (#421) Fix test MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil (#437) * MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil * Fix test MAISTRA-2682 Fix watch mechanism in federation (#439) Previously, no events were read from the watch response, because the read started with an endless loop that waited for data to be available in the decoder's buffer. This never happened, because the buffer is only written to when you call decoder.Decode(); this function was never called because the code waited for the buffer to have data. MAISTRA-2683 Properly close incoming watch connections when shutting down (#440) Log actual error returned by pollServices() (#441) Previously, instead of the actual error, only the following error message was logged: "expected condition not met". MAISTRA-2439: Prevent federation from exporting services that are not visible to the federation gateway (#432) By taking into consideration the service annotation `networking.istio.io/exportTo`. This annotation restricts where this service is visible: https://istio.io/latest/docs/reference/config/annotations/ If a service is not reachable from the federation gateway namespace due to this annotation, it should not be exported. MAISTRA-2617: Do not watch all namespaces in Extensions controller (#425) When using MemberRoll, we should rely on it to provide the list of namespaces to watch. If not using it, defaults to command line arguments. This fixes an istiod startup error as seen in the logs: ``` github.com/maistra/xns-informer/pkg/informers/informer.go:204: Failed to watch *v1.ServiceMeshExtension: failed to list *v1.ServiceMeshExtension: servicemeshextensions.maistra.io is forbidden: User "system:serviceaccount:i1:istiod-service-account-basic" cannot list resource "servicemeshextensions" in API group "maistra.io" at the cluster scope ``` * Remove package export and extensions Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix creating discovery.Controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix calling nil ResourceManager Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Remove panicing from AppendNetworkGatewayHandler Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [misc] OSSM-774 Fix flaky TestStatusManager (#456) This adds a little sleep to our unit tests for the StatusManager, because without it, we're running into the issue that we're updating a ServiceMeshPeer's status very quickly, and in some cases it might be that the last change has not been propagated when we're generating the patch for the next status change, which can lead to failures. This can happen in the real world, but you would need to change a ServiceMeshPeer's status within a few milliseconds, I doubt that it affects users. It would also be fixed with the next status update. For those reasons, I'm only fixing it in the test, with a Sleep() call. * Refactor manager_test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-1150 Fix flaky TestStatusManager unit test (#478) Co-authored-by: Marko Lukša <marko.luksa@gmail.com> * OSSM-1252 Fix federation status updates (#512) * Copy federation privileges from base to istio-discovery * Remove unnecessary ServiceMeshExtensions CRD Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add model.NetworkGatewaysHandler to federation controller to implement AppendNetworkGatewayHandler Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [federation] MAISTRA-2640 Add federation integration test (#447) * Fix building federation test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add package gogo from maistra-2.2 to temporarily fix TbdsGenerator Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Disable configuring remote cluster in federation deployment Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * [federation] OSSM-1128 Fix federation (#480) * Fix SecretCacheClient Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Send initial XDS request for trust bundle from proxy Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Disable using EndpointSlices to fix error on getting federation-egress endpoints Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Remove unused serviceMeshExtensionController Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Fix lint errors Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Update maistra CRDs Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * fedration_cp_version_update (#521) * OSSM-1529 Improve federation example script (#522) * OSSM-1529 Improve federation example install.sh Previously, the script would fall back to using nodeports when the load balancer IP wasn't set. This meant that if the provision of the load balancer took too long, the SMCPs would be misconfigured and you had to run the install script again. With this change, the script now waits for the load balancer IP to appear. It never falls back to using node ports, because they never really worked (the nodes' hostnames typically aren't FQDN and the node ports are typically protected by firewalls). If the user wants to expose the federation ingresses in a different way, they can now set the environment variables MESH1_ADDRESS, MESH1_DISCOVERY_PORT, and MESH2_SERVICE_PORT (likewise for MESH2) and run the script. * Update Federation example README * Better "Waiting for load balancer" message * OSSM-1211 Fix federation locality failover issues (#561) Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Signed-off-by: rcernich <rcernich@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: maistra-bot <57098434+maistra-bot@users.noreply.github.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <xuyuanlin_00@hotmail.com> * fix: removes deprecated gogo protobuf conversion * fix: goimport format * fix(lint): removes unused funcs * fix(lint): removes deprecated io/ioutil * fix(lint): disables staticcheck for federation tests it requires at least two clusters to make sense * fix(lint): use anypb.UnmarshalTo instead of ptypes * fix: no need to exclude grpcgen_test.go it seems to be fixed in v1.39 see: grpc/grpc-go#4476 * chore(backoff): aligns backoff dependency with v4 used by upstream * chore: reverts removed blank line - irrelevant for merge * chore(revive): adds explanation why json:inline is skipped from linting * OSSM-1962: Use EndpointSlices instead of Endpoints in federation controller (#614) Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-2049: Fix handling ServiceAccounts in federation controller (#627) * Fix collecting empty or repeated ServiceAccounts in federation controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Collect ServiceAccounts in sorted order Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-1093 Shorten exported resource name (#653) * Shorten exported resource name * Fix import createResourceName + unit tests * Rearrange unit tests + renaming function * Gen and lint * Rearrange unit tests + renaming function Gen and lint * Fix minor changes * Error message RFC 1123 * Reorganize structs in TestStatusManager Instead of having three arrays (events, expectedStatuses, and assertions), we now only have a single array, where each entry is a triplet containing the event, the expected status and the assertion. This allows you to see the event and its expected effects together and not have to scroll up and down, matching the indexes of the three arrays. * OSSM-2193 Fix flaky TestStatusManager See comment in https://issues.redhat.com/browse/OSSM-2193 to understand why this change fixes the test. * fix: runs make gen * chore: explains why staticcheck linter is disabled for federation_test * OSSM-728 Configuration scripts for Federation on Z and P, and bare metal (#670) * add config example scripts for IBM Systems Z and P * update multi-arch bookinfo deployment README, remove src * Update README.md * these are provided in the IBM repo * so README.md passes mdlinter * so README.md passes mdlinter * so README.md passes mdlinter * Update README.md * Move federation examples to samples/ directory * Rename template YAMLs to .yaml.template This makes the linter happy Co-authored-by: cfillekes <cfilleke@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com> * chore: removes obsolute TODO * chore: simplifies bool return expression * chore: removes redundant kubeClient check if initialization fails this func will not be reached anyway * chore(pkg): moves kube ctrl under servicemesh pkg folder * OSSM-2338: Remove env ISTIO_META_ROUTER_MODE from federation test Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * OSSM-2338: Remove "routerMode: sni-dnat" from federation samples Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * fix: adds operator.go customizations to kube.go clearly with cherry-pick we lost information about the file rename and thus the changes we made specificually for testing federation got lost * fix(test/operator): checks if east-west gw needs to be deployed * fix(federation): uses Unwrap to get instance of Federation registry * fix(tests): sets istiod-less remote flag to false that was the behaviour for maistra-2.3 and we need istiod to be present in order to have federation working * chore: gets registry just before it is needed * chore: explains why istiodlessremotes is needed to be set to false * chore: removes redundant import aliases * chore: removes name collisions * chore: removes redundant type conversion * fix: disables staticcheck linter for cluster req tests. * fix(tests): reverts timeout to original (but in minutes) * fix(tests): removes extra logging * chore: removes unnecessary logging * fix: uses existing CRD file references in charts * chore: removes multicluster label * chore: uses built-in namespace.NewOrFail instead of our impl * chore: introduces defaultTimeout const for federation tests * fix(lint): fixes go imports * fix(lint): removes unused variable * fix: naively wait 5s hoping that kind network will show up Signed-off-by: rcernich <rcernich@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Yuanlin <yuanlin.xu@redhat.com> Co-authored-by: Jacek Ewertowski <jewertow@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: maistra-bot <57098434+maistra-bot@users.noreply.github.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <xuyuanlin_00@hotmail.com> Co-authored-by: bmangoen <bmangoen@redhat.com> Co-authored-by: cfillekes <cfilleke@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com> OSSM-2376: Move kube controller to the federation package (#718) Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> OSSM-2376: Don't start federation controllers until informers have synced (#717) (#720) * OSSM-2376: Don't start federation-discovery-controller until kube informer has synced Federation discovery controller fetches config map with remote CA root cert, so if the controller started before the informer has synced, it would fail to fetch the config map. Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Store ConfigMap informer in a field of the discovery.Controller Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Refactor ResourceManager and don't start federation controller until informers has synced Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Simplify Start and HasSynced functions in federation controllers Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Move kube controller to the federation package Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Federation example fixes (#758) * Use default version in federation example SMCPs * Fix paths in federation example OSSM-3599 Federation egress-gateway gets wrong network gateway endpoints (#775) * OSSM-3599 Federation egress-gateway gets wrong update of network gateway endpoints * Deprecate GatewayEndpoints on server side * Remove resyncNetworkGateways in unit tests * Fix lint * Deprecate NetworkGatewayEndpoints and fix tests Refactor federation tests (#841) * Refactor federation tests Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> * Add more test cases Signed-off-by: Jacek Ewertowski <jewertow@redhat.com> Reimplement InstancesByPort method `InstancesByPort` is used by the federation server, which was removed in the upstream istio/istio#46329. We reimplement it to support the federation server. --------- Co-authored-by: Bartosz Majsak <bartosz.majsak@gmail.com> Co-authored-by: Jacek Ewertowski <jewertow@redhat.com> Co-authored-by: Daniel Grimm <dgrimm@redhat.com> Co-authored-by: Rob Cernich <rcernich@redhat.com> Co-authored-by: Jonh Wendell <jonh.wendell@redhat.com> Co-authored-by: Marko Lukša <marko.luksa@gmail.com> Co-authored-by: Praneeth Bajjuri <pbajjuri@redhat.com> Co-authored-by: Yuanlin Xu <yuanlin.xu@redhat.com> Co-authored-by: Brian Mangoenpawiro <bmangoen@redhat.com> Co-authored-by: Cheryl Fillekes <cfillekes@ibm.com> Signed-off-by: Yann Liu <yannliu@redhat.com>
[federation] Initial federation implementation
MAISTRA-2194 Add server/client code for Federation Service Discovery v1
MAISTRA-2195 Implement /watch endpoint
MAISTRA-2293 add CRD and controller for federating meshes
MAISTRA-2294 create CRD for federation ServiceExport (MAISTRA-2294 create CRD for federation ServiceExport #324)
MAISTRA-2294 update example VirtualService resources for ratings and mongodb (MAISTRA-2294 update example VirtualService resources for ratings and … #333)
[federation] MAISTRA-2295 create CRD for federation ServiceImport (MAISTRA-2295 create CRD for federation ServiceImport #336)
Signed-off-by: rcernich rcernich@redhat.com
Signed-off-by: rcernich rcernich@redhat.com
MAISTRA-2423 update federation api to v1
Signed-off-by: rcernich rcernich@redhat.com
MAISTRA-2424 minor updates to federation api
Signed-off-by: rcernich rcernich@redhat.com
MAISTRA-2427 configure locality info on imported services
Signed-off-by: rcernich rcernich@redhat.com
Cherry-pick multi-root support (#387)
Update go-control-plane to v0.9.9
Support multiple roots
Squashed commit, contains:
MAISTRA-2425 move spec.security.certificateChain to ConfigMap reference; add ability to specify ports for service and discovery (#392)
Signed-off-by: rcernich rcernich@redhat.com
MAISTRA-2426 move FederationStatus into MeshFederation (#393)
Signed-off-by: rcernich rcernich@redhat.com
MAISTRA-2513 federation API refinements
Signed-off-by: rcernich rcernich@redhat.com
[federation] MAISTRA-2237 Encrypt service discovery traffic (#411)
MAISTRA-2610 Prefix federation discovery endpoints with /v1/ (#422)
MAISTRA-2297 Support updates of federation resources (#417)
MAISTRA-2375: Do not create automatic routes for Federation Gateways
Remove a redundant call
setHostname()
is already being called withinNameForService()
see
https://github.com/maistra/istio/blob/21ee900cf8825711f70d88dc97afcf6862ed2626/pkg/servicemesh/federation/common/namemapping.go
lines 83, 120, 129
Remove techPreview.meshConfig from PoC example
It's set by default now.
MAISTRA-2611 Fix deletion of service exports to federated mesh (#421)
Fix test
MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil (#437)
MAISTRA-2658 Ensure ImportedServiceSet.status.importedServices is never nil
Fix test
MAISTRA-2682 Fix watch mechanism in federation (#439)
Previously, no events were read from the watch response, because the read started with an endless loop that waited for data to be available in the decoder's buffer. This never happened, because the buffer is only written to when you call decoder.Decode(); this function was never called because the code waited for the buffer to have data.
MAISTRA-2683 Properly close incoming watch connections when shutting down (#440)
Log actual error returned by pollServices() (#441)
Previously, instead of the actual error, only the following error message was logged: "expected condition not met".
MAISTRA-2439: Prevent federation from exporting services that are not visible to the federation gateway (#432)
By taking into consideration the service annotation
networking.istio.io/exportTo
.This annotation restricts where this service is visible: https://istio.io/latest/docs/reference/config/annotations/
If a service is not reachable from the federation gateway namespace due
to this annotation, it should not be exported.
MAISTRA-2617: Do not watch all namespaces in Extensions controller (#425)
When using MemberRoll, we should rely on it to provide the list
of namespaces to watch. If not using it, defaults to command line
arguments.
This fixes an istiod startup error as seen in the logs:
Signed-off-by: Jacek Ewertowski jewertow@redhat.com
Signed-off-by: Jacek Ewertowski jewertow@redhat.com
Signed-off-by: Jacek Ewertowski jewertow@redhat.com
Signed-off-by: Jacek Ewertowski jewertow@redhat.com
This adds a little sleep to our unit tests for the StatusManager,
because without it, we're running into the issue that we're updating
a ServiceMeshPeer's status very quickly, and in some cases it might be
that the last change has not been propagated when we're generating
the patch for the next status change, which can lead to failures.
This can happen in the real world, but you would need to change a
ServiceMeshPeer's status within a few milliseconds, I doubt that it
affects users. It would also be fixed with the next status update.
For those reasons, I'm only fixing it in the test, with a Sleep()
call.
Signed-off-by: Jacek Ewertowski jewertow@redhat.com
Co-authored-by: Marko Lukša marko.luksa@gmail.com
OSSM-1252 Fix federation status updates (OSSM-1252 Fix federation status updates #512)
Copy federation privileges from base to istio-discovery
Remove unnecessary ServiceMeshExtensions CRD
Signed-off-by: Jacek Ewertowski jewertow@redhat.com
Signed-off-by: Jacek Ewertowski jewertow@redhat.com
[federation] MAISTRA-2640 Add federation integration test (MAISTRA-2640 Add federation integration test #447)
Fix building federation test
Signed-off-by: Jacek Ewertowski jewertow@redhat.com
Signed-off-by: Jacek Ewertowski jewertow@redhat.com
Signed-off-by: Jacek Ewertowski jewertow@redhat.com
[federation] OSSM-1128 Fix federation ([federation] OSSM-1128 Fix federation #480)
Fix SecretCacheClient
Signed-off-by: Jacek Ewertowski jewertow@redhat.com
Signed-off-by: Jacek Ewertowski jewertow@redhat.com
Signed-off-by: Jacek Ewertowski jewertow@redhat.com
Signed-off-by: Jacek Ewertowski jewertow@redhat.com
Signed-off-by: Jacek Ewertowski jewertow@redhat.com
Signed-off-by: Jacek Ewertowski jewertow@redhat.com
fedration_cp_version_update (Fedration_cp_version_update #521)
OSSM-1529 Improve federation example script (OSSM-1529 Improve federation example script #522)
OSSM-1529 Improve federation example install.sh
Previously, the script would fall back to using nodeports when the load balancer IP wasn't set. This meant that if the provision of the load balancer took too long, the SMCPs would be misconfigured and you had to run the install script again.
With this change, the script now waits for the load balancer IP to appear. It never falls back to using node ports, because they never really worked (the nodes' hostnames typically aren't FQDN and the node ports are typically protected by firewalls).
If the user wants to expose the federation ingresses in a different way, they can now set the environment variables MESH1_ADDRESS, MESH1_DISCOVERY_PORT, and MESH2_SERVICE_PORT (likewise for MESH2) and run the script.
Update Federation example README
Better "Waiting for load balancer" message
OSSM-1211 Fix federation locality failover issues (OSSM-1211 Fix federation locality failover issues #561)
Signed-off-by: Yuanlin yuanlin.xu@redhat.com
Signed-off-by: rcernich rcernich@redhat.com
Signed-off-by: Jacek Ewertowski jewertow@redhat.com
Signed-off-by: Yuanlin yuanlin.xu@redhat.com
Co-authored-by: Daniel Grimm dgrimm@redhat.com
Co-authored-by: Rob Cernich rcernich@redhat.com
Co-authored-by: Jonh Wendell jonh.wendell@redhat.com
Co-authored-by: maistra-bot 57098434+maistra-bot@users.noreply.github.com
Co-authored-by: Marko Lukša marko.luksa@gmail.com
Co-authored-by: Praneeth Bajjuri pbajjuri@redhat.com
Co-authored-by: Yuanlin Xu xuyuanlin_00@hotmail.com