-
Notifications
You must be signed in to change notification settings - Fork 1.4k
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
Cannot connect to Pod IP when windows hns service restarted #5164
Comments
@kais271 Thanks for reporting this issue. Normally it is the CNI plugin who creates/deletes the endpoint and calico/felix service implements network policies onto it. What CNI you are using? Calico? calico/felix can't recreate missing endpoints as those are managed by other CNI plugins. You would need to find another way for CNI plugins to get endpoint created again. Restart kubelet may do the job but I'm not 100 sure. |
@song-jiang Hi sir, I using calico 3.20.2. Through further understanding, the hnsendpoint created by docker. When hns service restarted, felix log will record 'Not found endpoint'. |
If you are using Calico with Calico CNI plugin, the hnsendpoint is created by Calico CNI plugin. The sequence of actions are
You may check kubelet logs to see the details. |
Thank you. We refer to this doc to deploy the calico. |
Add some kubelet logs create pod
|
@kais271 Sorry I missed your reply. It seems Calico CNI does not support HNS restart, that is not something we are testing. Could you share the reason why HNS service need to be restarted? |
@song-jiang this sounds like a bug to me - we need to be resilient to these sorts of error conditions (unless I'm misunderstanding something subtle here) |
Expected Behavior
Connections can be automatically restored.
Current Behavior
When hns service restarted:
Possible Solution
N/A
Steps to Reproduce (for bugs)
1.Have k8s windows node and a running pod
2.hnsdiag.exe list endpoints
3.on windows node restart hns service
4.hnsdiag.exe list endpoints
Context
Your Environment
The text was updated successfully, but these errors were encountered: