Test fails at the end of execution when api server is temporary down

[thradec]

Describe the bug

The test workflow fails, when during its execution, the testkube-api-server is temporary un-available, for example when the node where it is scheduled is evicted. The error message in log is following:

failed to run post process function: failed to create internal storage for metrics: failed to connect with Cloud: connection error: desc = "transport: error while dialing: dial tcp 172.20.144.180:8089: connect: connection refused": failed to connect with Cloud: connection error: desc = "transport: error while dialing: dial tcp 172.20.144.180:8089: connect: connection refused"�

Expected behavior

The best would be if the api server can run in high availability configuration, but I understand that this can be hard to achive. It would help, if the testworflow-init be more resilient and implement some retry mechanism idealy with backoff.

Version / Cluster

• testkube version: 2.1.166 (standalone agent)
• kubernetes version: eks 1.33

[vsukhin]

thank you @thradec let us check it out

[vsukhin]

hey, @thradec if testkube-api-server has crashed, then it can restart the execution after reload. but if the node was evicted,then yes,you need to have high availability support in opace. We have more advanced scheduler in Control Panel for commercial edition @admacleod