[Bugs] [Bug 1529285] New: snapshot creation not working if thin provisioned brick LV is mounted via autofs

[bugzilla at redhat.com]

https://bugzilla.redhat.com/show_bug.cgi?id=1529285

            Bug ID: 1529285
           Summary: snapshot creation not working if thin provisioned
                    brick LV is mounted via autofs
           Product: GlusterFS
           Version: 3.10
         Component: snapshot
          Assignee: bugs at gluster.org
          Reporter: arnd at gronenberg.com
                CC: bugs at gluster.org



Description of problem:
Trying to create snapshot of thin provisioned gluster brick LV which is mounted
via autofs does not work

Version-Release number of selected component (if applicable):
glusterfs 3.10

How reproducible:
Happens for all thin provisioned gluster brick LVs which are mounted via autofs

Steps to Reproduce:
1. Create thin pool
2. Create thin provisioned LV for brick and format it with xfs
3. Create autofs setup
3.1. /etc/auto.master:
     +dir:/etc/auto.master.d
3.2. /etc/auto.master.d/glusterd.autofs:
     /- /etc/auto.glusterd --timeout=0
3.3. /etc/auto.glusterd:
     /data/glusterfs/test/brick1
-fstype=xfs,defaults,rw,noatime,nodiratime,largeio,inode64,swalloc,allocsize=64M
       :/dev/vg00/glusterfs_test_1
4. Start autofs
   systemctl start autofs
5. Create gluster volume
   gluster volume create test transport tcp
localhost:/data/glusterfs/test/brick1/brick
6. Try to create snapshot
   gluster snapshot create snapshot_test test no-timestamp

Actual results:
Error messsage:
"Snapshot is supported only for thin provisioned LV. Ensure that all bricks of
test are thinly provisioned LV."
In log:
[glusterd-snapshot.c:2004:glusterd_is_thinp_brick] 0-management: Failed to get
pool name for device /etc/auto.glusterd

Expected results:
Snapshot successfully created

Additional info:

Cause is in glusterd-utils.c function glusterd_get_mnt_entry_info:
getmntent_r reads through mount points, for autofs mounted fs there are two
mountpoints, first with fstype autofs and second with read fstype
Output of findmnt /data/glusterfs/test/brick1:
TARGET                      SOURCE                            FSTYPE OPTIONS
/data/glusterfs/test/brick1 /etc/auto.glusterd                autofs rw,...
/data/glusterfs/test/brick1 /dev/mapper/vg00-glusterfs_test_1 xfs    rw,...

Solution would be to check for fstype=autofs in while loop and skip to next
mount entry

-- 
You are receiving this mail because:
You are on the CC list for the bug.
You are the assignee for the bug.