2013-04-22 07:44 AM - edited 2015-12-18 12:33 AM
Do I need to create as many snapdrive snap create commands as there are filesystem.
Current setup at my client enviornment:
13 databases runing on 13 mount poitns on 13 luns in a single volume.
I can get to quiesce all 13, but when snapshoting using :
NTAP_SNAPSHOT_CREATE_CMD01=snapdrive snap create -fs /mnt/db1 /mnt/db2 /mnt/db3.... -snapname %SNAME-%SNAP_TYPE_%SNAP_TIME
First snapshot goes through well, but it fails on second snapshot saying, cannot overwrite previous snapshot, as they have same time stamp.)
Can't I just create a sinlge snapshot (Ultimately snapshot is happening at the volume level) ? DO I need to use snapdrive command for snap create ?
2013-04-23 03:51 AM
This issue is now resolved.
Having to run 13 snapshot scripts were too messy and did not make sense with all the database on a single volume. snapshoting 13 together and mounting on another host was the trickiest part. We rather, re-designed the database storage storage container, instead of having each database on separate lun, we went for volume group creation (with one big lun), we created single VG and 13 LV, that meant just one snapshot was enough instead of 13. I don't know if this was the right strategy but then it worked.
2013-04-23 05:38 AM
Ashwin, when you do a volume restore from a snapshot, it restores all the 13 databases back to the snapshot when it was taken.
The recommendation is each database to be put on individual volume (or set of volumes like db,log,temp etc) so that volume restores from snapshot can occur independent of other databases.
2013-04-23 06:56 AM
thanks, appreciate your input.
It is little compliated, having as many volume as databases is not an option here for other reasons. Its an HPUX running Progress databases for various applicatons. They have their own backup system to take backup to tape. So, once they these databases are quiesced and snapshoted, they are mounted at specific time on another backup host as filesystems and back up to tape. Another requirement was to - snapshot all together, not in a separate script. I had done this with single database with single script from snapcreator framework, but to get all of them together was an issue. I had thought about the volume restore concern (restoring all together), but then snapshot backups works fine,. If the the point in time recovery is immediate, then flexclone works pretty well, they can choose a previous snapshot and create a volclone and mount that database at the 'original mount point' and they are up. Apart from this, they also do 'redo' log shipping to the remote standby database for any DR situation. As I said, it is little complicated, but I am open to discussion, ideas.
Any thoughts ?
2013-05-16 05:26 AM
I have a customer who wants to use the snapcreator framework to perform similar tasks with their progress db, as you can with the typically NetApp snapmanagers products i.e. snapshots, backup/restores & flexclones. From your post, it appears you're achieving all the above providing you use certain design layouts i.e. one db per volume.
Can I ask what version of Progress you are using? They are currently using v9 and I have found a document which assist with this on another communitity (http://media.netapp.com/documents/tr_3132.pdf) however they are looking to upgrade to version 11 and I wanted to know if this will still work in the same way or whether you had any useful comments/recommendats/documents?
Many thanks in advance,