-
Notifications
You must be signed in to change notification settings - Fork 387
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
e2e test "TestBatchCreatePods" is Flaky #4086
Comments
@tnqn Hi, the TestBatchCreatePods failure often blocks ipv6 CI pipeline. Do you have any idea about the root cause? |
Having checked agent logs, I do not see re-connection between agent and OVS. |
I doubt the socket is not unix domain socket, but TCP connections from Agent to API Server or to Controller. The same issue is seen on both IPv6 and dual-stack testbed recently, but not found on IPv4 only. |
Having looked through the code, I doubt it is related with the socket is not closed which is used to send Gratuitous IPv6 NDP packet for a new Pod. https://github.com/antrea-io/antrea/blob/main/pkg/agent/util/ndp/ndp.go#L45 . I would try to verify if it can be fixed by closing the socket. |
Nice catch. Glad to see TestBatchCreatePods still has some value. Do you know why it sometimes can succeed? The FD were garbage collected sometimes? |
I didn't investigate why it succeeds sometimes.. |
Describe the bug
Hi,
I hit this failure when running e2e test on IPv6-only or dual-stack testbed. The error shows the fds are changed after batch creating Pods. I hit the failure several times, and the failure looks flaky because it may pass after I re-run the tests. I hit the error with my code change to switch OpenFlow version from 1.3 to 1.5, and I don't think the OpenFlow version change is the reason leads to the errors. So I create this issue to track it.
Below is the output of the error:
To Reproduce
Expected
Actual behavior
Versions:
Antrea: main branch ( v1.7+)
Additional context
The text was updated successfully, but these errors were encountered: