Partitions are not bootstrapped after purge operation #35650
Description
Description
I'm running my process tests with CPT on a local Camunda 8 Run (8.8.0-alpha6) configured as a remote runtime.
This works fine with a fresh Camunda 8 Run distribution. However, after I restart the distribution, my process tests fail when deleting the data between the test cases.
This is a major issue for running process tests with CPT on a remote runtime.
I can reproduce the behavior without CPT by purging the data of the Camunda 8 Run distribution via HTTP request and checking the partitions in the topology. After the restart, the broker doesn't have any partitions in the topology.
Workaround
When I delete the data directory in the camunda-zeebe-* directory, the partitions are bootstrapped successfully after the purge, and the process tests run again.
Steps to reproduce
- Check the topology (
GET http://localhost:8080/v2/topology) - Purge the data and remember the change ID (
POST http://localhost:9600/actuator/cluster/purge) - Check the topology and verify that the change ID is successfully applied (
GET http://localhost:8080/v2/topology) - Restart the distribution
- Check the topology (
GET http://localhost:8080/v2/topology) - Purge the data and remember the change ID (
POST http://localhost:9600/actuator/cluster/purge) - Check the topology and verify that the change ID is successfully applied (
GET http://localhost:8080/v2/topology)
Then, verify that the broker doesn't have any partitions in the topology.
Current behavior
First run
Topology before purge:
{
"brokers": [
{
"nodeId": 0,
"host": "127.0.1.1",
"port": 26501,
"partitions": [
{
"partitionId": 1,
"role": "leader",
"health": "healthy"
}
],
"version": "8.8.0-alpha6"
}
],
"clusterSize": 1,
"partitionsCount": 1,
"replicationFactor": 1,
"gatewayVersion": "8.8.0-alpha6",
"lastCompletedChangeId": "-1"
}Purge response:
{
"changeId": 2,
"currentTopology": [
{
"id": 0,
"state": "ACTIVE",
"version": 0,
"lastUpdatedAt": "0001-01-01T00:00:00.000+0000",
"partitions": [
{
"id": 1,
"state": "ACTIVE",
"priority": 1,
"config": {
"exporting": {
"exporters": [
{
"id": "elasticsearch",
"state": "ENABLED"
},
{
"id": "camundaExporter",
"state": "ENABLED"
}
]
}
}
}
]
}
],
"plannedChanges": [
{
"operation": "PARTITION_LEAVE",
"brokerId": 0,
"partitionId": 1,
"brokers": []
},
{
"operation": "DELETE_HISTORY",
"brokers": []
},
{
"operation": "PARTITION_BOOTSTRAP",
"brokerId": 0,
"partitionId": 1,
"priority": 1,
"brokers": []
}
],
"expectedTopology": [
{
"id": 0,
"state": "ACTIVE",
"version": 4,
"lastUpdatedAt": "2025-07-21T09:08:31.991+0000",
"partitions": [
{
"id": 1,
"state": "ACTIVE",
"priority": 1,
"config": {
"exporting": {
"exporters": [
{
"id": "elasticsearch",
"state": "ENABLED"
},
{
"id": "camundaExporter",
"state": "ENABLED"
}
]
}
}
}
]
}
]
}Topology after purge:
{
"brokers": [
{
"nodeId": 0,
"host": "127.0.1.1",
"port": 26501,
"partitions": [
{
"partitionId": 1,
"role": "leader",
"health": "healthy"
}
],
"version": "8.8.0-alpha6"
}
],
"clusterSize": 1,
"partitionsCount": 1,
"replicationFactor": 1,
"gatewayVersion": "8.8.0-alpha6",
"lastCompletedChangeId": "2"
}Second run
Topology before purge:
{
"brokers": [
{
"nodeId": 0,
"host": "127.0.1.1",
"port": 26501,
"partitions": [
{
"partitionId": 1,
"role": "leader",
"health": "healthy"
}
],
"version": "8.8.0-alpha6"
}
],
"clusterSize": 1,
"partitionsCount": 1,
"replicationFactor": 1,
"gatewayVersion": "8.8.0-alpha6",
"lastCompletedChangeId": "2"
}Purge response:
{
"changeId": 4,
"currentTopology": [
{
"id": 0,
"state": "ACTIVE",
"version": 4,
"lastUpdatedAt": "2025-07-21T09:08:32.707+0000",
"partitions": [
{
"id": 1,
"state": "ACTIVE",
"priority": 1,
"config": {
"exporting": {
"exporters": [
{
"id": "elasticsearch",
"state": "ENABLED"
},
{
"id": "camundaExporter",
"state": "ENABLED"
}
]
}
}
}
]
}
],
"plannedChanges": [
{
"operation": "PARTITION_LEAVE",
"brokerId": 0,
"partitionId": 1,
"brokers": []
},
{
"operation": "DELETE_HISTORY",
"brokers": []
},
{
"operation": "PARTITION_BOOTSTRAP",
"brokerId": 0,
"partitionId": 1,
"priority": 1,
"brokers": []
}
],
"expectedTopology": [
{
"id": 0,
"state": "ACTIVE",
"version": 8,
"lastUpdatedAt": "2025-07-21T09:10:54.843+0000",
"partitions": [
{
"id": 1,
"state": "ACTIVE",
"priority": 1,
"config": {
"exporting": {
"exporters": [
{
"id": "elasticsearch",
"state": "ENABLED"
},
{
"id": "camundaExporter",
"state": "ENABLED"
}
]
}
}
}
]
}
]
}Topology after purge:
{
"brokers": [
{
"nodeId": 0,
"host": "127.0.1.1",
"port": 26501,
"partitions": [],
"version": "8.8.0-alpha6"
}
],
"clusterSize": 1,
"partitionsCount": 1,
"replicationFactor": 1,
"gatewayVersion": "8.8.0-alpha6",
"lastCompletedChangeId": "4"
}Expected behavior
After a restart of the distribution, the broker bootstraps the partitions successfully after a purge operation.
Environment
SM
Version
camunda-process-test-spring:8.8.0-alpha6- Camunda 8 Run:
8.8.0-alpha6
Rootcause
The issue is related to the purge operation.
The persisted data seems to cause an issue with the purge.
Solution ideas
No response
Dev -> QA handover
- Resources:
- Versions to validate:
- Release version (in which version this feature will be released):
Links
Camunda 8 Run log file (second run): camunda.log