Red Hat Bugzilla – Bug 802726
[glusterfs-3.3.30qa27] - 'rm -rf .glusterfs' succeeds on nfs mount
Last modified: 2016-05-31 21:55:51 EDT
Description of problem:
rm -rf .glusterfs succeeded on nfs mount with zero exit status. But anyway .glusterfs was not removed from the back-end.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Create and start a single node distribute volume.
2. Mount via fuse and create some data.
3. Now run stat .glusterfs, rm -rf .glusterfs, cd .glusterfs etc. All of them will fail with No data available error. (currently)
4. Now umount the volume and mount again from nfs with nolock option.
5. Run rm -rf .glusterfs from the nfs mountpoint. The command succeeds.
The rm -rf .glusterfs succeeds with zero exit status. But .glusterfs is not removed from the back-end.
Ideally behavior 'rm -rf .glusterfs' should be consistent in fuse and nfs volume. Even it can't be same, there should be some error must be thrown to the user.
After above operations I checked in back-end and .glusterfs had a gfid set on it.
*** This bug has been marked as a duplicate of bug 802478 ***
CHANGE: http://review.gluster.com/2936 (storage/posix: don't allow rmdir()/lookup() on HIDDEN_DIRECTORY) merged in master by Vijay Bellur (email@example.com)