a307607a99
Real data migrated, not a fresh start: 1.2GB Prometheus TSDB, Grafana's
existing dashboards/DB (its admin password was already changed from
default — confirmed real, in-use data), SonarQube's data/logs/extensions
and its own Postgres DB (confirmed: the actual `management-platform`
SonarQube project survived, matching the Jenkinsfile's own
-Dsonar.projectKey). Old docker-compose services stopped, not removed —
left as a rollback path.
Real problems hit and fixed along the way (see k8s/monitoring/README.md
for the full detail, worth reading before touching any of this again):
- k3s's bundled Traefik defaults to a LoadBalancer Service, which
immediately hijacked ports 80/443 via iptables DNAT (klipper-lb) out
from under this VM's existing unrelated nginx-served app ("nqks") the
moment k3s was installed. Patched Traefik to ClusterIP via a
HelmChartConfig; nginx now reverse-proxies the 3 new domains to
Traefik's ClusterIP instead of Traefik touching host ports at all.
- Caught my own mistake before it caused damage: "corrected" what I
assumed was a typo (grafanna.nav.ovh -> grafana.nav.ovh) — the two-N
version is what the user actually specified, and is what points at this
VM; plain grafana.nav.ovh is a different, unrelated, pre-existing live
server at 109.199.127.74. Renamed everything back before requesting any
cert or touching that domain.
- Adding hostPort to the SonarQube pod (for Jenkins' hardcoded
178.18.243.51:9000) broke pod-to-pod routing to that pod entirely —
Traefik couldn't reach it (hung/504) while direct host-to-pod and
host-to-ClusterIP both worked fine the whole time, which is what gave
it away. Replaced with a host-level socat proxy (systemd unit) to the
Service's ClusterIP instead, which touches nothing pod-level. Verified
Jenkins' exact address still works AND Traefik routing to the same pod
works simultaneously.
- SonarQube needed a DB migration after the version jump
(DB_MIGRATION_NEEDED), then OOM'd during post-migration rule
re-registration on the compose file's original 256MB web heap — bumped
SONAR_WEB_JAVAADDITIONALOPTS.
TLS: real Let's Encrypt certs via certbot (HTTP-01, nginx already owns
port 80 so no DNS-01/OVH-webhook replication needed on this second
cluster), auto-renewing. Access: HTTP Basic Auth in front of all three
domains for browser access; SonarQube also stays reachable unauthenticated
at 178.18.243.51:9000 specifically for Jenkins (which authenticates via
its own token, not BasicAuth — that path would break the scanner),
restricted to the main server's IP via ufw, not open to the internet.
Verified end-to-end: all three https://*.nav.ovh domains return 401
without credentials and load correctly with them; Jenkins' exact
SonarQube address (no auth) still returns 200; real data confirmed
present in Grafana, Prometheus, and SonarQube post-migration.
Monitoring stack on k8s (VM: 178.18.243.51)
Migrated from docker-compose (/root/monitoring, /root/CI-CD/sonarqube on
the VM — left docker compose stop'd, not removed, as a rollback path) into
the VM's own k3s cluster, with real data migrated (not a fresh start):
Prometheus TSDB, Grafana dashboards/DB, SonarQube data+DB (confirmed intact:
the real management-platform SonarQube project and Grafana's existing
admin password both survived).
Apply order
kubectl apply -f monitoring-pvcs.yaml
# migrate data into the PVCs via a loader pod per volume (see chat history —
# not scripted, was a one-time manual migration)
kubectl apply -f monitoring-apps.yaml
kubectl apply -f monitoring-ingress.yaml
Domains — READ THIS BEFORE TOUCHING DNS
grafanna.nav.ovh(note: two Ns —grafana.nav.ovhwithout the typo is a DIFFERENT, unrelated, pre-existing server at 109.199.127.74. Do not point anything at plaingrafana.nav.ovhor touch that record.)sonar.nav.ovhprom.nav.ovh
All three already existed as A records pointing at 178.18.243.51 before this
work; only nginx-vhost.conf + a Let's Encrypt cert (via certbot, HTTP-01,
auto-renews) make them actually resolve to something.
Why nginx is in front of Traefik, not Traefik directly on 80/443
This VM already runs an unrelated app ("nqks") on nginx, owning ports 80/443.
k3s's bundled Traefik defaults to a LoadBalancer Service, which on a
single-node k3s claims host 80/443 via iptables DNAT (klipper-lb) —
this silently hijacked nqks's traffic the first time (found and fixed
during this session). Traefik's Service is patched to ClusterIP-only via
a HelmChartConfig (kubectl get helmchartconfig -n kube-system traefik);
nginx reverse-proxies the 3 domains to Traefik's ClusterIP instead, and
handles TLS + BasicAuth itself.
Access control
- Browser access to all 3 domains: HTTP Basic Auth (
/etc/nginx/.htpasswd-monitoringon the VM, useradmin) — ask whoever ran this migration for the password, it's not in git. - SonarQube is also reachable directly at
178.18.243.51:9000, no auth — this is deliberate, not an oversight. Jenkins' SonarQube server config (main cluster,hudson.plugins.sonar.SonarGlobalConfiguration.xml) is hardcoded to that exact address and authenticates via its own token, not BasicAuth (which would break the scanner).sonarqube-jenkins-proxy.service(systemd, socat) forwards host:9000 -> the sonarqube Service's ClusterIP; ufw restricts port 9000 to the main server's IP only, not the open internet. Do not add BasicAuth in front of this specific path.- Fragility note: the socat unit hardcodes the Service's ClusterIP. If the
sonarqubeService in themonitoringnamespace is ever deleted and recreated (not just the pod — pod recreates keep the same ClusterIP), update theTCP:<ip>:9000target in/etc/systemd/system/sonarqube-jenkins-proxy.serviceon the VM to match.
- Fragility note: the socat unit hardcodes the Service's ClusterIP. If the
Gotchas hit building this (context for next time)
- Don't add a k8s Service of type LoadBalancer on this cluster without first patching it to ClusterIP/setting a HelmChartConfig — same hijack risk as Traefik's default.
- Don't add
hostPortto a pod that also needs normal in-cluster (Service/ Ingress) traffic to reach it — it breaks pod-to-pod routing to that pod (confirmed directly: Traefik -> sonarqube hung/504'd the whole time hostPort was present, while direct host-to-pod and host-to-ClusterIP both worked fine — the asymmetry was the tell). Use a host-level proxy (socat) or a NodePort in the 30000-32767 range instead. - SonarQube's DB was mid-upgrade (
DB_MIGRATION_NEEDED) after the version jump — needed onePOST /api/system/migrate_db(admin/admin was already changed on the real instance; used the actual admin session) — and the post-migration rule re-registration OOM'd on the compose file's original 256MB web heap; bumpedSONAR_WEB_JAVAADDITIONALOPTSto-Xms512m -Xmx1536m.