No pod logs and failing healthcheck with dashboards helm chart

OpenSearch Dashboards
1

Versions (relevant - OpenSearch/Dashboard/Server OS/Browser): Opensearch Dashboards Helm Chart 3.7.0

Describe the issue: When I install a vanilla helm chart (no extra configuration is needed for me to run into this issue, adding specific configuration doesn’t help either), I get a pod where the healthcheck always times out and nothing is logged to the console (not even a starting message).

Configuration:

Just used the default helm chart config

Relevant Logs or Screenshots:

No logs, to show, which is kinda the problem :frowning:

The startup probe fails with “connect: connection refused”

2

@mkstephenson Welcome to the forum!

Have you tried increasing startup grace period and cpu, see example below

# values.yaml
startupProbe:
  tcpSocket:
    port: 5601
  initialDelaySeconds: 30    # was 10, give Node.js time to load plugins
  periodSeconds: 10
  failureThreshold: 30       # was 20,  30×10 + 30 = 330s total window
  timeoutSeconds: 5
  successThreshold: 1

resources:
  requests:
    cpu: "500m"              # was 100m, critical for startup speed
    memory: "1Gi"
  limits:
    cpu: "1000m"
    memory: "1Gi"
3

Hi @Anthony ,

Thanks for the tip! Increasing the resources from the default worked wonders (I didn’t need to modify the startupProbe). Is this just an issue just with slower systems or would it make sense to open a PR to increase the resources generally in the dashboard helm chart?

4

@mkstephenson great to hear you got it working.

I would recommend to open a PR for this, as the current resource requests are indeed very low and potentially cause problems for other users new to this.