Description of problem:
When the "downloads" pod scheduled on the worker node which is disabled IPv6, the pods does not work with the following log messages.
$ oc get pod -n openshift-console -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
pod/console-1111111111-aaaaa 1/1 Running 0 8d 10.129.0.7 node02.ocp4.example.com <none> <none>
pod/console-1111111111-bbbbb 1/1 Running 0 8d 10.128.0.12 node03.ocp4.example.com <none> <none>
pod/downloads-aaaaaaaaaa-xxxxx 0/1 CrashLoopBackOff 5 1m54s 10.129.2.18 node03.ocp4.example.com <none> <none>
pod/downloads-aaaaaaaaaa-yyyyy 0/1 CrashLoopBackOff 5 3m14s 10.129.2.68 node02.ocp4.example.com <none> <none>
$ oc logs pod/downloads-aaaaaaaaaa-xxxxx --container=download-server --timestamps
2020-06-11T00:47:48.113162338Z serving from /tmp/tmpC8Eoy8
2020-06-11T00:47:48.11326402Z Traceback (most recent call last):
2020-06-11T00:47:48.11326402Z File "/tmp/serve.py", line 59, in <module>
2020-06-11T00:47:48.11326402Z sock = socket.socket(socket.AF_INET6, socket.SOCK_STREAM)
2020-06-11T00:47:48.11326402Z File "/usr/lib64/python2.7/socket.py", line 187, in __init__
2020-06-11T00:47:48.11326402Z _sock = _realsocket(family, type, proto)
2020-06-11T00:47:48.11326402Z socket.error: [Errno 97] Address family not supported by protocol
Version-Release number of selected component (if applicable):
You can always reproduce this issue after the worker node is disabled IPv6 for running the "downloads" pods.
Steps to Reproduce:
"downloads" pod does not work when the node is disabled IPv6.
"downloads" pod work well regardless of the IPv6 enabled or disabled.
*** Bug 1795325 has been marked as a duplicate of this bug. ***
As I understand that we need following steps to verify this bug
1) Add RHEL worker nodes to the cluster(IPv4 cluster is ok)
2) Disable IPv6 on the RHEL worker node
3) Re-schedule downloads pods to the new added RHEL worker node to see if downloads pods are running
Is it correct? Can you also please give some guidance about how to disable IPv6 for a RHEL worker node?
> Is it correct? Can you also please give some guidance about how to disable IPv6 for a RHEL worker node?
You need not to add new worker. Just add "ipv6.disable=1" to kernel parameter when existing any worker node reboot, then you can disable IPv6 on the node.
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory (OpenShift Container Platform 4.6 GA Images), and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.