CODEWITHSUNDEEP
kubernetesredissigtermredis-serverkeydb
PERSIKIDA
PERSIKIDA

Reputation: 1

KeyDB in Kubernetes receives SIGTERM and restarts unexpectedly

I’m running KeyDB v6.3.2 in a Kubernetes sts, and after an extended uptime (39+ days), the pod receives SIGTERM and restarts, even though there was no manual shutdown or scaling event

Here is the relevant log before termination:

1:signal-handler (1740559193) Received SIGTERM scheduling shutdown... 1:22:S 26 Feb 2025 08:39:53.711 # User requested shutdown... 1:22:S 26 Feb 2025 08:39:53.711 * Saving the final RDB snapshot before exiting. 1:22:S 26 Feb 2025 08:39:54.027 * DB saved on disk 1:22:S 26 Feb 2025 08:39:54.027 * Removing the pid file. 1:22:S 26 Feb 2025 08:39:54.029 # KeyDB is now ready to exit, bye bye...

Pods lastState shows:

    lastState:
  terminated:
    containerID: containerd://8cb24568ab14bd0579427dd04505ad2fea2a0ea685ef63e4c64579b8f5f92888
    exitCode: 0
    finishedAt: "2025-02-26T08:39:54Z"
    reason: Completed
    startedAt: "2025-01-18T07:30:35Z"
restartCount: 2

Pod was running for over a month before this restart

What I’ve checked so far:

  1. SIGTERM was received (kubectl logs --previous confirms it)
  2. No manual shutdown or scaling event (no Helm or StatefulSet update)
  3. No eviction or OOMKill (kubectl describe pod/keydb shows no resource issues)
  4. No liveness probe failures (kubectl describe pod/keydb)
  5. PVC is stable (kubectl get pvc shows volumes are bound)

Deployment Details:

  1. KeyDB Version: v6.3.2
  2. Kubernetes Version: v1.21
  3. Helm Chart: keydb-0.48.0
  4. StatefulSet with PVC: Yes

Liveness Probe:

livenessProbe:
  exec:
    command:
    - sh
    - -c
    - /health/ping_liveness_local.sh 5
  failureThreshold: 5
  initialDelaySeconds: 20
  periodSeconds: 5
  timeoutSeconds: 6

Readiness Probe:

readinessProbe:
  exec:
    command:
    - sh
    - -c
    - /health/ping_readiness_local.sh 1
  failureThreshold: 5
  initialDelaySeconds: 20
  periodSeconds: 5
  timeoutSeconds: 2

Are there any keydb specific conditions where it could request its own shutdown? Any debugging steps to track the origin of the SIGTERM signal? I've tried to set loglevel to debug/verbose, but there are no results

Upvotes: 0

Views: 10

Answers (0)

Related Questions