Solved

DDB Backups: Is the media agent that has a DDB partition associated with it supposed to back that partition up (and not another media agent)?

  • 14 February 2022
  • 7 replies
  • 1358 views

Userlevel 2
Badge +11

We have several DDB’s, all partitioned across several media agents. When the DDB backups run, I’m seeing most of the Media agents doing a backup for “themselves” meaning the client and Media agent are the same when the DDB backup runs.  But for one of them, I cannot seem to get the DDB backup copy to choose the primary/default media agent (in the copy → data paths settings), to do any of the DDB backups, it always chooses the alternate media path for both DDB backup partitions.

I have *not* yet chosen the “use preferred data path” setting (where it should only? use the primary media argent and not use any alternates) as I feel that it should choose the primary and it would auto choose the secondary media agent for the other partition if it needs to.

Also: I want the DDBBackups to be slit over 2 media agents because 1 media agent is very overpowered (lots of CPU/memory) relative to the other (older and few CPU’s).  The media agent the DDB backups is choosing is this underpowered media agent and it appears to not even be trying to use teh “default” one.

I have tried setting the “DAta path configuration” settings to “roud robin” , as well as “use alternate data paths” and the “when resources are busy” settings, and nothing seems to change, it always chooses the non default, “less powerful” media agent to run the DDB backups for both partitins, and they take a long time to run.

 

Anyway: For my question: is the media agent DDB backup supposed to back up the partition it has associated with it ( like “best practice” or “recommended to make it work this way” or “yes, the system should auto choose it unless there’s an issue”) and is there a way to determine why my DDB backup is always choosing the non primary/default media agent, as if the “Data path configuration” settings are having no effect?

 

icon

Best answer by tigger2 16 February 2022, 18:45

View original

If you have a question or comment, please create a topic

7 replies

Userlevel 2
Badge +11

Update: I set it to “preferred” and the DDB backups just kicked off at 4 PM (they are scheduled to run every 8 hours after midnight) and it *still* chose the alternate (non preferred) media agent for both DDB backups!

For the media agent (the low CPU powered one) doing its *own* DDB backup,  its taking about 1.5 hours ( mostly normal for the size. throughput we are getting for the other DDB backups (~300 GB/hr) for other media agents), but when it backs up the other one (the high power CPU media agent), the throughput drops by 1/2 and it takes about 4 hours.  This DDB partition is a little larger too.  I feel if i can just get the more powerful one to back its own DDB up, it will hopefully take an hour or less and be “more like the others”

Another note: poking around, I see a note in the properties of this DDB backup. Apparently it was “cloned” from the other DDB backup policy (the other one that is working correctly, with other media agents). This makes some sense as in the “Data paths” tab it has entries for *all* of our media agents that do backups/aux copies (I couldn't figure out why they were in there, so I recently disabled the ones that are not associated with the DDB but didn’t want to delete them. I figured that those additional media agents might be causing some of the “which alternate to pick” issue but I guess its not)

 

edit: I’ll open a ticket tomorrow for it.

Userlevel 7
Badge +23

Thanks for the solution sharing!  I wish I thought of checking there. Great post for posterity to check out, thanks to you!

I’ll mark your reply as the Best Answer.

Userlevel 7
Badge +23

@tigger2 , generally, a job will run and the CS will assign whatever resource is best able to handle the job the fastest.

For whatever reason, it’s not picking the main MA (not saying there is not a valid reason, just one we don’t know yet).

If you want it to ALWAYS pick that MA, then definitely set the preferred MA.

To find out why it isn’t picking the main MA, you’ll likely want to open a support case so they can go through the log files, trace the pre-job decisions, and get an answer (and I’m not 100% sure if the default logging will provide that).

If you do, share the case number here so I can track it.

Userlevel 2
Badge +11

Thanks!  I’ll try the “set preferred” option and see how much faster is does/does not go.  if its noticeably quicker, then I’ll open a ticket, if its slower, then I guess the CS is picking properly.

Userlevel 7
Badge +23

Please keep me posted :nerd:

Userlevel 7
Badge +23

Thanks for the (quite detailed) update!  Keep me posted on the case number.

Userlevel 2
Badge +11

Ticket: CVLT::0177082227

Result: Apparently there is a way to override the “Data paths” being used at the subclient level. The “DDB Backup” Subclient on only one of our media agents (The one that was never chosen for a DDB Backup) had a “override data paths” set (under subclient → properties → Storage Device → ‘Data Paths” button), and it was overriding to choose another media agent.  That override config was removed/unchecked and now the DDB Backups are being “properly/correctly/the default method” backed up by the media agent(s) responsible for their partition. 

 

Additionally: while looking at the behavior closely, it appears we had an extra DDB backup being run (on the media agent that was “doing all the DDB Backups”) because it was attached to 2 schedules.  i removed the ‘incorrect” schedule.

 

Additionally (just for info) we are almost migrated to use all V5 DDB’s, and I was able to confirm with support that we can likely just run the DDB Backups 1x per day.  Right now we have them configured for 3x per day, which apparently is an older recommendation when you’re on V4 of the DB’s. So very soon I’ll be able to reschedule them for 1x per day as well as move them to a time that’s not very active, hopefully speeding them up even more.