-
Notifications
You must be signed in to change notification settings - Fork 3.4k
Closed
Copy link
Labels
area/CI-improvementTopic or proposal to improve the Continuous Integration workflowTopic or proposal to improve the Continuous Integration workflowarea/datapathImpacts bpf/ or low-level forwarding details, including map management and monitor messages.Impacts bpf/ or low-level forwarding details, including map management and monitor messages.area/encryptionImpacts encryption support such as IPSec, WireGuard, or kTLS.Impacts encryption support such as IPSec, WireGuard, or kTLS.feature/ipsecRelates to Cilium's IPsec featureRelates to Cilium's IPsec feature
Description
Problem description
Our IPSec tests use the migrate-svc app to detect whether or not there have been packet drops when doing upgrades or key rotations. This app includes a 500ms sleep, included to avoid saturating the system. The wait time was arbitrarily picked, no investigation was done as to how much to wait.
It would be worth reducing the time or even completely removing the sleep, because we currently miss catching some drops that are visible when using the more intensive httpperf instead.
Implementation suggestion
Whoever is doing this work, should check that the system doesn't get overloaded. And to make this easier it would make sense to make the sleep amount a parameter to the migrate-svc app, and then iterate from the testing side.
Metadata
Metadata
Assignees
Labels
area/CI-improvementTopic or proposal to improve the Continuous Integration workflowTopic or proposal to improve the Continuous Integration workflowarea/datapathImpacts bpf/ or low-level forwarding details, including map management and monitor messages.Impacts bpf/ or low-level forwarding details, including map management and monitor messages.area/encryptionImpacts encryption support such as IPSec, WireGuard, or kTLS.Impacts encryption support such as IPSec, WireGuard, or kTLS.feature/ipsecRelates to Cilium's IPsec featureRelates to Cilium's IPsec feature