ceph-csi: [csi-cephfs] Plugin Pod dies
First of all, thanks for the csi-cephfs plugin implementations.
I have been using the csi-cephfs version 0.3.0 in our production Kubernetes 1.11.5-gke.5 for about a month. Most of the time the plugin works great, except when the plugin Pod was suddenly died and re-created by the DaemonSet. When this happened, I had to re-deploy the Pods and PVCs that use CephFS because the Pods got an error Transport endpoint is not connected (see #91). This situation happened 3-4 times so far.
Is there any way to mitigate this issue?
Has anyone considered to store the plugin’s state (including mountpaths) on the node’s volume? By this way, the plugin Pod may be able to restore the previous state when it is re-created.
About this issue
- Original URL
- State: closed
- Created 5 years ago
- Comments: 24 (10 by maintainers)
Commits related to this issue
- issue #91 issue #217 Goal we try to solve when csi exit unexpect, the pod use cephfs pv can not auto recovery because lost mount relation until pod be killed and reschedule to other node. i think th... — committed to ShyamsundarR/ceph-csi by deleted user 5 years ago
- issue #91 issue #217 Goal we try to solve when csi exit unexpect, the pod use cephfs pv can not auto recovery because lost mount relation until pod be killed and reschedule to other node. i think th... — committed to wilmardo/ceph-csi by deleted user 5 years ago
- Merge pull request #217 from red-hat-storage/sync_us--devel Syncing latest changes from devel for ceph-csi — committed to nixpanic/ceph-csi by openshift-merge-bot[bot] 7 months ago
The v1.0.0 image in quay should already be built with this fix. Every PR merge refreshes the image, and hence this should be updated with the fix.
In the near future we would get to a versioned system, for minor updates to the 1.0 branch.