abdf99ad03
The healthcheck used to liveness and readiness probes is considered intrusive and can produce false positives[0]. It's just returning that the pod is healthy if it's fully booted and whether the pod is waiting for a peer to appear to sync tables, it cannot succeed and eventually enters into a pod restart deadlock. The command is also deprecated and will be removed in future version. Updating the probes based on current recommenation from community[1]. Because of that, we change the liveness to check the rabbit connection to check if the rabbit is connected and running and also changed the readiness to ping the pod after boot to see if it's ready. Ref: [0] https://www.rabbitmq.com/monitoring.html#deprecations [1] https://www.rabbitmq.com/monitoring.html#health-checks Test plan: PASS: Close rabbitmq connection inside pod and verify if Liveness fail and rebooted the pod. PASS: Readiness check pod state and verify if can ping it to check if the pod is health after get up. PASS: Build on Debian Story: 2009892 Task: 44674 Signed-off-by: Arthur Luz de Avila <arthur.luzdeavila@windriver.com> Change-Id: Ib98df7e1135b7b5cee50d63a9ab06eb4a961cd4b |
||
|---|---|---|
| .. | ||
| centos | ||
| debian | ||
| stx-openstack-helm | ||