Sync Mount/Flush Block Device on NodeUnstage
Closed this issue · 5 comments
The NodeUnpublishVolume -> NodeUnstageVolume -> ControllerUnpublishVolume flow does not explicitly perform a filesystem sync in linux. The driver calls umount on the staged block device path. If the block device is not mounted elsewhere in the OS, this results in a successful sync and unmount. Files are correctly flushed to the disk.
If the mount is being used elsewhere in the system (via private mount in an alternate namespace), the removal of the driver's mount may not result in the block device being flushed and unmounted. The consequence is that ControllerUnpublishVolume may proceed with removing a block device from a VM while there could be pending block writes at the kernel layer (cached at the filesystme).
This is being done explicitly in windows (https://github.com/kubernetes-csi/csi-proxy/blob/ba9a0a80d071f1ad4650d84abaea043b08e6496e/pkg/os/volume/api.go#L141), but not in linux.
Two options:
- Explicitly call
syncon the underlying mount prior to unmounting in NodeUnstage - Flush the block device with
blockdev --flushduring NodeUnstage
Related to #1657. We may still need to sync/flush the block device after validating it is not being used as a filesystem. Keeping this open still.
The Kubernetes project currently lacks enough contributors to adequately respond to all issues.
This bot triages un-triaged issues according to the following rules:
- After 90d of inactivity,
lifecycle/staleis applied - After 30d of inactivity since
lifecycle/stalewas applied,lifecycle/rottenis applied - After 30d of inactivity since
lifecycle/rottenwas applied, the issue is closed
You can:
- Mark this issue as fresh with
/remove-lifecycle stale - Close this issue with
/close - Offer to help out with Issue Triage
Please send feedback to sig-contributor-experience at kubernetes/community.
/lifecycle stale
The Kubernetes project currently lacks enough active contributors to adequately respond to all issues.
This bot triages un-triaged issues according to the following rules:
- After 90d of inactivity,
lifecycle/staleis applied - After 30d of inactivity since
lifecycle/stalewas applied,lifecycle/rottenis applied - After 30d of inactivity since
lifecycle/rottenwas applied, the issue is closed
You can:
- Mark this issue as fresh with
/remove-lifecycle rotten - Close this issue with
/close - Offer to help out with Issue Triage
Please send feedback to sig-contributor-experience at kubernetes/community.
/lifecycle rotten
The Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs.
This bot triages issues according to the following rules:
- After 90d of inactivity,
lifecycle/staleis applied - After 30d of inactivity since
lifecycle/stalewas applied,lifecycle/rottenis applied - After 30d of inactivity since
lifecycle/rottenwas applied, the issue is closed
You can:
- Reopen this issue with
/reopen - Mark this issue as fresh with
/remove-lifecycle rotten - Offer to help out with Issue Triage
Please send feedback to sig-contributor-experience at kubernetes/community.
/close not-planned
@k8s-triage-robot: Closing this issue, marking it as "Not Planned".
In response to this:
The Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs.
This bot triages issues according to the following rules:
- After 90d of inactivity,
lifecycle/staleis applied- After 30d of inactivity since
lifecycle/stalewas applied,lifecycle/rottenis applied- After 30d of inactivity since
lifecycle/rottenwas applied, the issue is closedYou can:
- Reopen this issue with
/reopen- Mark this issue as fresh with
/remove-lifecycle rotten- Offer to help out with Issue Triage
Please send feedback to sig-contributor-experience at kubernetes/community.
/close not-planned
Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository.