We have a community member running Mattermost 5.3.1 in Kubernetes and receiving “Please check connection, Mattermost unreachable. If issue persists, ask administrator to check WebSocket port.” errors.
Chrome Developer Console shows websocket calls all have status „Finished“.
There‘s no proxy involved, but k8s ingress and service. config.json unchanged, regarding all WebSocket related entries (WebsocketURL „“, WebsocketSecurePort 443, WebsocketPort 80, ListenAdress „:8000“, SiteUrl set).
Does anyone have ideas on how to resolve the issue?
Hi @jasonblais, the community member in question would be me and I’d be happy to answer further questions here. For the record: I suspected a company proxy first, but another team member not using that proxy seems to have the same WebSocket problems accessing our Mattermost instance.
in the service side we are using ClusterIP instead of NodePort.
the ingress looks ok.
and we did not define in the config.json the WebsocketURL can you try to remove?
Hello @cpanato, thanks for the suggestions. WebsocketURL in the config.json is already empty. (I tried setting it explicitly to the same value as SiteURL earlier in my configuration attempts though.)
Could you detail your ClusterIP type service configuration please?
apiVersion: v1
kind: Service
metadata:
name: mattermost-app-service
labels:
app: <LABEL HERE IF YOU ARE USING THAT>
spec:
selector:
app: <LABEL HERE IF YOU ARE USING THAT>
type: ClusterIP
ports:
- port: 8065
targetPort: 8065
protocol: TCP
name: app
Hi @cpanato, thanks. Can you also tell me what ports you exposed in the container? First I did not explicitly specify container ports, but I also tried with: