CrashLoopBackOff on GKE cluster.

Expected Behavior

Actual Behavior

I'm using temporal on my GKE cluster. with following values. it works perfectly but i have an issue. whenever there is an gke update and node gets replaced. temporal crashes and goes in CrashLoopBackOff and cannot recover. complains about no usable database connection. I have to manual redeploy with helm.

server:
  config:
    namespaces:
      create: true
      namespace:
        - name: default
          retention: 1d
    persistence:
      default:
        driver: "sql"
        sql:
          driver: "postgres12"
          host: my-postgresql.microservices.svc.cluster.local
          port: 5432
          database: temporal
          user: postgres
          password: "mypass"
          maxConns: 20
          maxIdleConns: 20
          maxConnLifetime: "1h"
      visibility:
        driver: "sql"

        sql:
          driver: "postgres12"
          host: my-postgresql.microservices.svc.cluster.local
          port: 5432
          database: temporal_visibility
          user: postgres
          password: "mypass"
          maxConns: 20
          maxIdleConns: 20
          maxConnLifetime: "1h"
cassandra:
  enabled: false

mysql:
  enabled: false

postgresql:
  enabled: true

prometheus:
  enabled: false

grafana:
  enabled: false

elasticsearch:
  enabled: false

schema:
  createDatabase:
    enabled: true
  setup:
    enabled: true
  update:
    enabled: true

Specifications

• Version: temporal-0.65.0
• App-Version: 1.28.1
• Platform: GKE

No repository file or test is named. Start with the CrashLoopBackOff logs and the supplied Helm values, then investigate behavior after a GKE node replacement; done means the Temporal deployment recovers without a manual Helm redeploy.