e2e: corrupted mount not attempting to recover.

[iPraveenParihar]

          > > replacing, `k8s.io/mount-utils => k8s.io/mount-utils v0.29.3`

this works!, So, will add a TODO: update once fixed in mount-utils and visit later? @Madhu-1

@iPraveenParihar Thank you, lets get it merged and we will revisit mount-utils, lets check what is changed in mount-utils release notes.

Originally posted by @Madhu-1 in #4614 (comment)

[iPraveenParihar]

There is a change in the IsLikelyNotMountPoint method: kubernetes/mount-utils@v0.29.3...v0.30.2

From the method description, it is clear that IsLikelyNotMountPoint is not suitable for detecting bind mounts: https://github.com/kubernetes/mount-utils/blob/dfd453c575c09ba21aa9f45bf9a9ddb8724c5e0b/mount_linux.go#L498-L513

ceph-csi uses this method for fuse mount recovery to determine if the path is a valid MountPoint or not.

ceph-csi/internal/util/util.go

Lines 327 to 335 in 202f43c

	// IsMountPoint checks if the given path is mountpoint or not.
	func IsMountPoint(mounter mount.Interface, p string) (bool, error) {
	notMnt, err := mounter.IsLikelyNotMountPoint(p)
	if err != nil {
	return false, err
	}
	return !notMnt, nil
	}

we need alternative method to be used?
thoughts @nixpanic @Madhu-1 ?

[nixpanic]

Probably you can delete this function and call x.mounter.IsMountPoint(...) directly? I don't see why the indirection is beneficial for anything.

[iPraveenParihar]

Probably you can delete this function and call x.mounter.IsMountPoint(...) directly? I don't see why the indirection is beneficial for anything.

IsMountPoint(...) is an expensive operation as it enumerates mount points on the Node and then does the evaluate operation on each mount point.
https://github.com/kubernetes/mount-utils/blob/dfd453c575c09ba21aa9f45bf9a9ddb8724c5e0b/mount.go#L72-L79

may be because of that, they chose IsLikelyNotMountPoint()?

[Madhu-1]

Probably you can delete this function and call x.mounter.IsMountPoint(...) directly? I don't see why the indirection is beneficial for anything.

IsMountPoint(...) is an expensive operation as it enumerates mount points on the Node and then does the evaluate operation on each mount point. https://github.com/kubernetes/mount-utils/blob/dfd453c575c09ba21aa9f45bf9a9ddb8724c5e0b/mount.go#L72-L79

may be because of that, they chose IsLikelyNotMountPoint()?

@iPraveenParihar IsMountPoint is already used for RBD and IsLikelyNotMountPoint currently used only for cephfs. we can switch from IsLikelyNotMountPoint to IsMountPoint for cephfs as well.

[Madhu-1]

@nixpanic #3281 in this PR we switch to IsLikelyNotMountPoint for specific case for rbd, if we switch back to IsMountPoint wont we get into the same problem again, can you please check or we need to still use IsLikelyNotMountPoint for specific cases and switch to IsMointPoint in util function.

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed in a week if no further activity occurs. Thank you for your contributions.

[LinskId]

sorry, I might have missed it in your merged pr, but if indeed we identify a bind mount how do we recover from it?

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed in a week if no further activity occurs. Thank you for your contributions.