Description of problem: This is to ascertain what should be the right way of going about, when a user has backup tool running along with snapshots. Say a user has snapshots getting created at time stamps t1, t2, t3 .. At timestamp t4, a backup is taken using glusterfind api. This results in all the changes to get synced until t4. The status file $SESSION_DIR/status is updated with t4. IF the user at this point has a requirement to go back in time, he/she restores file(s) at timestamp t2. Now the changed file(s) would have the timestamp t2, which will NOT get recorded as 'changed files' in the next run of glusterfind pre command (as t2 < t4). Possible ways to resolve: 1) We could ensure that 'snapshot restore' restores/updates the status file present in $SESSION_DIR, with the timestamp of the restored snapshot (in the above e.g., t2) 2) We could have clear warnings displayed to let the user know that the backup destination is not going to stay up-to-date, when he/she tries to run snapshot restore, unless a --full backup is done 3) We could update our documentation stating that caution would have to be exercised in case a user is using both backup as well as snap restore. Version-Release number of selected component (if applicable): Gluster 3.7 upstream nightly glusterfs-3.7dev-0.852.git3feaf16.el6.x86_64
This is not a bug, but needs documentation update. 1. following a snap restore, take a full backup to ensure the backup catalog/db stays in step the the volume state 2. expire backup copies taken between the original date of the snap and the restore date Moving this Bug to MODIFIED state.
This bug is getting closed because a release has been made available that should address the reported issue. In case the problem is still not fixed with glusterfs-3.7.0, please open a new bug report.