fix(zfspv): fixing data loss in case of pod deletion #91
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
looks like a bug in ZFS as when you change the mountpoint property to none, ZFS automatically umounts the file system. When we delete the pod, we get the unmount request for the old pod and mount request for the new pod. Unmount is done by the driver by setting mountpoint to none and the driver assumes that unmount has done and proceeded to delete the mountpath, but here zfs has not unmounted the dataset ``` $ sudo zfs get all zfspv-pool/pvc-3fe69b0e-9f91-4c6e-8e5c-eb4218468765 | grep mount zfspv-pool/pvc-3fe69b0e-9f91-4c6e-8e5c-eb4218468765 mounted yes - zfspv-pool/pvc-3fe69b0e-9f91-4c6e-8e5c-eb4218468765 mountpoint none local zfspv-pool/pvc-3fe69b0e-9f91-4c6e-8e5c-eb4218468765 canmount on ``` here, the driver will assume that dataset has been unmouted and proceed to delete the mountpath and it will delete the data as part of cleaning up for the NodeUnPublish request. Shifting to use zfs umount instead of doing zfs set mountpoint=none for umounting the dataset. Also the driver is using os.RemoveAll which is very risky as it will clean child also, since the mountpoint is not supposed to have anything, just os.Remove is sufficient and it will fail if there is anything there. Signed-off-by: Pawan <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
cherry-pick :- #89
looks like a bug in ZFS as when you change the mountpoint property to none,
ZFS automatically umounts the file system. When we delete the pod, we get the
unmount request for the old pod and mount request for the new pod. Unmount
is done by the driver by setting mountpoint to none and the driver assumes that
unmount has done and proceeded to delete the mountpath, but here zfs has not unmounted
the dataset
here, the driver will assume that dataset has been unmouted and proceed to delete the
mountpath and it will delete the data as part of cleaning up for the NodeUnPublish request.
Shifting to use zfs umount instead of doing zfs set mountpoint=none for umounting the dataset.
Also the driver is using os.RemoveAll which is very risky as it will clean
child also, since the mountpoint is not supposed to have anything,
just os.Remove is sufficient and it will fail if there is anything there.
Signed-off-by: Pawan [email protected]