OCPBUGS-48790: Add a liveness probe to the extractor container

The extractor container relies on critctl to collect container information If the crio.service is restarted on the worker nodes when the CA bundle has been updated by the proxy/cluster, the container would not be able to connect anymore on the restarted CRI-O service. Adding a liveness probe that check that crictl works as expected ensure that the container will be killed and restarted with the correct TLS settings to connect to the CRI-O service. This fixes https://issues.redhat.com/browse/OCPBUGS-48790. Upstream PR is #1067. Signed-off-by: Jeff Mesnil <[email protected]>
openshift · Jan 23, 2025 · 6ac3354 · 6ac3354
1 parent 5c8b715
commit 6ac3354
Showing 1 changed file with 7 additions and 0 deletions.
diff --git a/manifests/10-insights-runtime-extractor.yaml b/manifests/10-insights-runtime-extractor.yaml
@@ -93,6 +93,13 @@ spec:
           env:
             - name: CONTAINER_RUNTIME_ENDPOINT
               value: unix:///crio.sock
+          livenessProbe:
+            exec:
+              command:
+                - crictl
+                - info
+              periodSeconds: 10
+              failureThreshold: 2
           resources: 
             requests:
               cpu: 10m