-
Bug
-
Resolution: Unresolved
-
Normal
-
None
-
4.19.z
-
Quality / Stability / Reliability
-
False
-
-
None
-
Important
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
Description of problem: After OCP cluster update from 4.18.26 to 4.19.16 pods behind EgressService fail to reach external resources:
oc rsh -n rds-egress-ns rds-egress-deploy-5d778bc996-fvgkb ~ $ curl --connect-timeout 5 -Lv http://10.46.187.195:9090/clientip * Trying 10.46.187.195:9090... * Connection timed out after 5000 milliseconds * Closing connection 0 curl: (28) Connection timed out after 5000 milliseconds ~ $ curl --connect-timeout 5 -Lv http://[2620:52:0:2ebb::c3]:9090/clientip * Trying 2620:52:0:2ebb::c3:9090... * Connection timed out after 5001 milliseconds * Closing connection 0 curl: (28) Connection timed out after 5001 milliseconds
Service:
oc get svc -n rds-egress-ns NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE egress-svc-1 LoadBalancer 172.30.213.181 192.200.10.1,fd12:2222:3333:4444::1 9090:32545/TCP 3h57m
oc get egressservice -n rds-egress-ns NAME ASSIGNED HOST egress-svc-1 openshift-worker-0
Ingress access works via LoadBalancer's IPv4 address but fails over IPv6:
curl -Lv --connect-timeout 3 http://192.200.10.1:9090/clientip * Trying 192.200.10.1... * TCP_NODELAY set * Connected to 192.200.10.1 (192.200.10.1) port 9090 (#0) > GET /clientip HTTP/1.1 > Host: 192.200.10.1:9090 > User-Agent: curl/7.61.1 > Accept: */* > < HTTP/1.1 200 OK < Date: Tue, 14 Oct 2025 12:36:29 GMT < Content-Length: 16 < Content-Type: text/plain; charset=utf-8 < * Connection #0 to host 192.200.10.1 left intact 100.64.0.8:56420
curl -Lv --connect-timeout 3 http://[fd12:2222:3333:4444::1]:9090/clientip * Trying fd12:2222:3333:4444::1... * TCP_NODELAY set * Connection timed out after 3001 milliseconds * Closing connection 0 curl: (28) Connection timed out after 3001 milliseconds
External IPv4 address is reachable from all the cluster nodes.
External IPv6 address is reachable only from control-plane nodes.
Version-Release number of selected component (if applicable):
metallb-operator.v4.19.0-202509281724
OCP 4.19.16
How reproducible:
So far ran the upgrade once
Steps to Reproduce:
1. Deploy and configure baremetal dual stack cluster per RDS Core specification with OCP 4.18
2. Update the cluster to 4.19
3. Update day 2 operators (SRIOV, NMState, MetalLB, etc)
4. Check EgressService functionality
Actual results:
EgressService functionality is partially working
Expected results:
EgressService functions properly after upgrade
