Unsolved
This post is more than 5 years old
3 Posts
0
2088
parallerly backups jobs not happening with the same storage node client
Hi,
We have 8.2 EBS networker backup software install in the Solaris 10 server & also we have one of Storage node which is also installed on Solaris 10 server.
Tape Library :- SL48 Sun with two Tape drives.
The issue, if we run tow backups JOB at a time for the same storage node client the parallerly backups JOBs not happening and second one JOB is going in waiting mode where as second tape device is idle, the second JOB should be run on idle tape device.
where as if we run two group which has same client, it's running successfully.
Can some one help me plz.
sunadmin2
3 Posts
0
April 14th, 2016 05:00
Thanks for your reply Bingo,
There is no any media alerts.
Storage node is same but media pools and group are different.
Library has two tape device.
If we ran one job and its select tape device "0" and device "1" is idle so we ran another JOBs for the same storage node but Group and pool is different and second one job is not going on idle tape device and going in waiting mode.
bingo.1
2.4K Posts
0
April 14th, 2016 05:00
Obviously your pool filters by "savegroup" (automatisc backups). So it does not not apply to manual backups (save).
Consequently the selection criteria is different.
I assume one of the jobs is not going to the same pool and the other pool has the wrong device selected.
Is there any media alert at all?
gautamgp
226 Posts
0
April 14th, 2016 21:00
There are multiple things why your second device is not in use.
1. Please see if you have "Auto Media Mangement" checked in the Tape Library properties
2. Make sure you have not selected the Tape Devices in any of the Pool's Selection Criteria
3. You have enough tapes to be labeled or reused
StuartWhitby
45 Posts
0
April 18th, 2016 04:00
Here's my understanding of your initial statement:
- If you run 2 backups from the same group simultaneously on the client, they both go to the same tape drive where you expect them to direct to one backup to each tape drive
- If you run 2 groups containing the same client, the backups go to different tape drives
You then follow this up to state that the media pools are different. If this is the case then your groups or jobs are set up in the "single group" instance to back up to one pool, whereas 2 groups are selecting 2 pools.
Here's the general logic that NetWorker steps through in order to select which tape drive it's going to send a new backup stream to (apologies if I miss a step here or get something out of order, but it'll be fairly close)
- Confirm target storage node
- Confirm target media pool
- Confirm availability of a drive with mounted, writable media for that pool on that storage node
- Confirm device target sessions
- If incoming saveset breaches target sessions
- Confirm availability of any other mounted, writable media in another drive
- Confirm availability of an idle device
- Mount/label available media in that drive
- else check incoming saveset against device max sessions and defer
So in the first case, if you have 2 sessions going to the same tape, then those sessions are set up to go to the same pool and the device target sessions value has not been reached (default 4).
In the second case, from what I can see the sessions are going to separate pools. If the pools are the same, then the target sessions per device value has been exceeded and extra sessions will spill over onto the second tape drive.
Note - if you're using 2 pools on a system which has 2 tape drives available, then I'd recommend that you limit these anyway to one drive per pool to avoid having one pool grab both drives and leave the backups for the other pool unable to be serviced until one of those tape drives fills up entirely - at which point it's down to chance which pool gets its tape mounted first as far as I remember (last I saw, it's really down to which request manages to get the jukebox arm first to load a volume rather than ordered selection criteria)
sunadmin2
3 Posts
0
April 26th, 2016 21:00
Thanks for your suggestion
Now issue has been resolved by changing value
Issue
************************************************
Not able to run two concurrent groups within the environment even though one device appeared to be available for use.
Client, device, savegroup parallelism required adjusting to ensure drives were available to more than one group when client savesets configured exceeded 4.
************************************************
Resolution
The environment has a jukebox which has 2 physical drives both of which are shared between two storage nodes with a total of 4 devices configured.
Device Target sessions were set to 4 with Max set to 32 for the issued group.