After the uni04delta-ipv6 base job got more stable recently, we started to observe failures in the update variant of this job.

As part of the Run update role there is an attempt to download cirros image that fails in IPv6 environment.

Two jobs where it happened recently:

• https://sf.apps.int.gpc.ocp-hub.prod.psi.redhat.com/zuul/t/components-integration/build/f997dff7a8574fc4bc551b3538e7cb06
  • https://sf.apps.int.gpc.ocp-hub.prod.psi.redhat.com/logs/f99/components-integration/f997dff7a8574fc4bc551b3538e7cb06/logs/controller-0/ci-framework-data/logs/ansible-edpm-update.log

• https://sf.apps.int.gpc.ocp-hub.prod.psi.redhat.com/zuul/t/components-integration/build/24b28ceecfa741e49569080d727b599c
  • https://sf.apps.int.gpc.ocp-hub.prod.psi.redhat.com/logs/24b/components-integration/24b28ceecfa741e49569080d727b599c/logs/controller-0/ci-framework-data/logs/ansible-edpm-update.log

Previously it was observed in the integrity pipeline in OSPCIX-817 when we executed limited set of tests (before the base job got stable), but the issue was ignored at that time when we decided to change the job to more stable scenario. From this, we know the problem occurs in https://github.com/openstack-k8s-operators/ci-framework/blob/3cc19802840ddc4deef62eb6d15a89b500d595cc/roles/update/templates/workload_launch.sh.j2#L294C1-L295C1