services

kubectl get svc -n default NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE bureaublad-backend ClusterIP 10.43.186.230 80/TCP 3h3m bureaublad-frontend ClusterIP 10.43.108.163 80/TCP 3h3m bureaublad-redis-headless ClusterIP None 6379/TCP 3h3m bureaublad-redis-master ClusterIP 10.43.81.217 6379/TCP 3h3m collabora-online ClusterIP 10.43.242.247 9980/TCP 3h30m docs-minio ClusterIP 10.43.23.14 9000/TCP 3h5m docs-nginx ClusterIP 10.43.49.244 80/TCP,443/TCP 3h4m docs-postgresql ClusterIP 10.43.218.63 5432/TCP 3h5m docs-postgresql-hl ClusterIP None 5432/TCP 3h5m docs-redis-headless ClusterIP None 6379/TCP 3h5m docs-redis-master ClusterIP 10.43.67.8 6379/TCP 3h5m docs-static-nginx ClusterIP 10.43.25.75 80/TCP,443/TCP 3h5m element-postgresql ClusterIP 10.43.132.149 5432/TCP 3h30m element-postgresql-hl ClusterIP None 5432/TCP 3h30m element-redis-headless ClusterIP None 6379/TCP 3h30m element-redis-master ClusterIP 10.43.153.12 6379/TCP 3h30m element-web ClusterIP 10.43.86.254 8080/TCP 3h30m grist-minio ClusterIP 10.43.241.202 9000/TCP 12d grist-minio-console ClusterIP 10.43.145.71 9090/TCP 12d grist-postgresql ClusterIP 10.43.206.72 5432/TCP 12d grist-postgresql-hl ClusterIP None 5432/TCP 12d grist-redis-headless ClusterIP None 6379/TCP 12d grist-redis-master ClusterIP 10.43.46.27 6379/TCP 12d keycloak-keycloak ClusterIP 10.43.131.39 80/TCP 13d keycloak-keycloak-headless ClusterIP None 8080/TCP 13d keycloak-postgresql ClusterIP 10.43.159.226 5432/TCP 13d keycloak-postgresql-hl ClusterIP None 5432/TCP 13d kubernetes ClusterIP 10.43.0.1 443/TCP 13d livekit-livekit-server LoadBalancer 10.43.120.54 192.168.0.216 30001:30001/UDP,30002:30002/UDP,30003:30003/UDP,30004:30004/UDP,30005:30005/UDP,30006:30006/UDP,30007:30007/UDP,30008:30008/UDP,30009:30009/UDP,32669:32669/TCP 3h6m livekit-redis-headless ClusterIP None 6379/TCP 3h6m livekit-redis-master ClusterIP 10.43.16.191 6379/TCP 3h6m livekit-server ClusterIP 10.43.74.104 80/TCP 3h6m meet-backend ClusterIP 10.43.123.54 80/TCP 3h30m meet-frontend ClusterIP 10.43.230.94 80/TCP 3h30m meet-postgresql ClusterIP 10.43.42.238 5432/TCP 3h30m meet-postgresql-hl ClusterIP None 5432/TCP 3h30m meet-redis-headless ClusterIP None 6379/TCP 3h30m meet-redis-master ClusterIP 10.43.254.166 6379/TCP 3h30m meet-static-nginx ClusterIP 10.43.188.143 80/TCP,443/TCP 3h30m nextcloud ClusterIP 10.43.134.233 8080/TCP 3h6m nextcloud-minio ClusterIP 10.43.176.97 9000/TCP 3h29m nextcloud-minio-console ClusterIP 10.43.159.253 9090/TCP 3h29m nextcloud-postgresql ClusterIP 10.43.177.65 5432/TCP 3h29m nextcloud-postgresql-hl ClusterIP None 5432/TCP 3h29m nextcloud-redis-headless ClusterIP None 6379/TCP 3h29m nextcloud-redis-master ClusterIP 10.43.21.144 6379/TCP 3h29m synapse ClusterIP 10.43.222.1 8448/TCP 3h30m synapse-replication ClusterIP 10.43.141.53 9093/TCP 3h30m

Perfect — we now have the exact cause and the fix.

Root Cause

The nginx config contains:

upstream ... "docs-backend" ...

But there is no service named docs-backend in your cluster.

Looking at kubectl get svc, the correct backend for your “docs” service (Collabora Online) is:

→ collabora-online (ClusterIP 10.43.242.247, port 9980)

The Bitnami nginx chart / your “docs” Helm chart is hardcoded (or configured) to proxy to a non-existent service called docs-backend.

This is why nginx fails with [emerg] host not found in upstream "docs-backend".

Fix (Do this now)

You need to tell the nginx chart what the real upstream/backend is.

In your Helm values.yaml for the docs release, add or change this section:

nginx:
  serverBlock:
    upstream:
      backend: collabora-online   # ← Change this line
  # or depending on the exact chart structure, it might be:
  # backend:
  #   serviceName: collabora-online
  #   servicePort: 9980

If the chart uses a more specific key (common in Bitnami or custom “docs” charts), try one of these patterns:

# Option 1 (most common)
collabora:
  enabled: true
  service:
    name: collabora-online

# Option 2
nginx:
  proxy:
    upstream: collabora-online:9980

# Option 3
extraEnvVars:
  - name: NGINX_UPSTREAM_BACKEND
    value: collabora-online

After updating the values, upgrade the chart:

helm upgrade docs <chart-name> -n default -f values.yaml

(or whatever your release name and chart reference is — e.g. helm upgrade docs bitnami/nginx ... if it’s a plain nginx chart).

Quick & Dirty Temporary Workaround (if you want nginx up immediately)

You can patch the ConfigMap or the deployment to replace the upstream name, but it’s better to fix it properly via Helm so it survives restarts.

Next Steps After Fixing docs-nginx

Once docs-nginx becomes 1/1 Running, the Collabora integration for Nextcloud should start working (or at least show the next error).

Then we can tackle the remaining two crashing pods:

Nextcloud (still CrashLoopBackOff)
Synapse (OIDC config still broken)

For Nextcloud, please now run these two log commands and paste the output:

kubectl logs nextcloud-864d6c9854-9ts7h -n default --tail=150
kubectl logs nextcloud-864d6c9854-9ts7h -n default --previous --tail=100

Also run the cronjob log if you want:

kubectl logs nextcloud-cronjob-29611495-hpj6l -n default

For Synapse (quick reminder)

While we’re waiting, the fastest way to bring Matrix back is to set discover: true and a correct issuer in the OIDC block (remove the four manual endpoint lines), or temporarily set oidcProviders: [].

First, let’s get docs-nginx fixed — update the upstream to collabora-online and run the helm upgrade. Then paste the Nextcloud logs. We’re very close to a fully green cluster!