61 lines
2.6 KiB
Markdown
61 lines
2.6 KiB
Markdown
# im-server-test ArgoCD patch
|
|
|
|
The live StatefulSet is owned by the ArgoCD Application `im-server-test`.
|
|
Copy `statefulset-performance-patch.yaml` into that application's Git source
|
|
and reference it from the test overlay:
|
|
|
|
```yaml
|
|
patches:
|
|
- path: statefulset-performance-patch.yaml
|
|
```
|
|
|
|
Also include `backend-debug-metrics-rbac.yaml` as a resource in the test
|
|
overlay. It grants the existing `argocd/backend-debug` service account
|
|
namespaced read access to Pod metrics so the load-test collector can run
|
|
`kubectl top pods -n im-server-system`:
|
|
|
|
```yaml
|
|
resources:
|
|
- backend-debug-metrics-rbac.yaml
|
|
```
|
|
|
|
After ArgoCD sync, verify the effective configuration:
|
|
|
|
```bash
|
|
kubectl -n im-server-system get statefulset im-server -o jsonpath='{.spec.template.spec.containers[0].resources}'
|
|
kubectl -n im-server-system get statefulset im-server -o jsonpath='{.spec.template.spec.containers[0].livenessProbe.httpGet.path}'
|
|
kubectl -n im-server-system get statefulset im-server -o jsonpath='{.spec.template.spec.containers[0].env}'
|
|
kubectl -n im-server-system get statefulset im-server -o jsonpath='{.spec.template.spec.dnsConfig.options}'
|
|
kubectl auth can-i list pods.metrics.k8s.io \
|
|
--as=system:serviceaccount:argocd:backend-debug \
|
|
-n im-server-system
|
|
```
|
|
|
|
Expected values are requests `1 CPU / 512Mi`, limits `2 CPU / 1Gi`, and
|
|
liveness path `/livez`. The MySQL pool environment must be
|
|
`MYSQL_MAX_OPEN_CONNS=24`, `MYSQL_MAX_IDLE_CONNS=8`,
|
|
`MYSQL_REGISTRY_MAX_OPEN_CONNS=4`, and `MYSQL_REGISTRY_MAX_IDLE_CONNS=2`.
|
|
The WebSocket heartbeat environment must be `WS_PING_PERIOD_SECONDS=15` and
|
|
`WS_PONG_WAIT_SECONDS=90`.
|
|
The Pod DNS options must be `ndots:1`, `timeout:2`, and `attempts:2`. Keeping
|
|
`ClusterFirst` preserves short Kubernetes service-name resolution, while the
|
|
lower `ndots` value prevents fully qualified internal and external dependency
|
|
names from being expanded through every cluster search suffix first.
|
|
The application repository's deploy workflow waits for reconciliation and
|
|
fails if ArgoCD restores different values.
|
|
|
|
After the rollout, verify the generated resolver configuration on every Pod:
|
|
|
|
```bash
|
|
for pod in im-server-0 im-server-1 im-server-2; do
|
|
echo "--- ${pod}"
|
|
kubectl -n im-server-system exec "${pod}" -- cat /etc/resolv.conf
|
|
done
|
|
```
|
|
|
|
The output must retain the cluster DNS nameserver and search domains, with
|
|
`options ndots:1 timeout:2 attempts:2`. This mitigation reduces amplified DNS
|
|
traffic from the application Pods. If lookup timeouts remain under load,
|
|
CoreDNS capacity or NodeLocal DNSCache must be changed in the cluster
|
|
infrastructure repository by an administrator with `kube-system` access.
|