Comment 12 for bug 1817936

Here is some finding , readiness or liveness deployment of nova-api-osapi or xxxx-api is using the tcp and port to check pod readiness and liveness
For example barbican-api readiness
readinessProbe:
          failureThreshold: 10
          periodSeconds: 10
          successThreshold: 1
          tcpSocket:
            port: 9311

‘curl --connect-timeout 10 http://barbican-api.openstack.svc.cluster.local:9311 result’

Host Name readinessProbe. periodSeconds Error times in 50000 times Comments
R22 10 29 only applied barbican , no cinder, glance , compute-toolkit were applied.
R22 10 28
R22 300 0
R22 300 0
R22 300 0
Edgetest901 10 8 Apply default helm chart
Edgetest901 300 2
Edgetest901 300 1

it seems change period of readiness will improve the result.

And I changed nova-api-osapi deploy readiness and liveness from tcp to exec command.
Changed to
        readinessProbe:
          exec:
            command:
            - cat
            - /tmp/nova-api.sh
          failureThreshold: 3
          periodSeconds: 10
          successThreshold: 1
          timeoutSeconds: 1

The result is very positive , there is 0 error on 2*50000 times.
will check how to fix this issue officially