SLP: suspending and restarting cancelled jobs

[X2]

Hi,

We are in the process of moving our policies to a new disk based storage unit and all of policies will use SLPs. The backup to disk and tape run as per policy schedule and the offsite copy part of the SLP is restricted to certain hours during the day.

Earlier in the week I had to restart NetBackup on one of the media servers. I did the normal process that we had in place and also suspended the SLPs. However, the SLP operations which were already ongoing, did not cancel when I shut down NetBackup. I got a whole list of SLP operations in progress. The tasks which didn't stop on NetBackup shutdown were 1 bpcd, 1 bpduplicate and bpdm (several processes).

Here is what I thought to get NetBackup to really shutdown (script with following workflow).

1. Suspend SLPs
2. List incomplete images for SLPs
3. Get backup IDs from above step and then cancel them
4. Shutdown NetBackup
5. ..... do schedule maintenance....
6. Startup NetBackup
7. Use backup IDs from step 2, and restart/reissue those operations only

Does this sound like a viable workaround? Is there a better way out? Any suggestions would be appreciated.

Thanks.

[StefanosM]

you can not manually restart a canceled SLP. Instead you can manually duplicate the backup IDs from step 2.

[Amol_Nair]

instead of canceling the backupids in SLP why not just suspend them and then reactivate them. So instead of cancel use the nbstlutil inactive command to suspend them and then reactivate them by using the nbstlutil active command. Another easier method would be to simple suspend the SLP itself.. This will ensure that the backups run successfully and all the secondary operations are suspended till your media server's maintenance is completed.. You can suspend the SLP directly from the GUI by right clicking on the SLP

[X2]

Suspending SLP only suspends actions which have not started. Ongoing actions continue. Hence, the plan of cancelling.

[X2]

Ah, I realise that a single operation of the SLP which was cancelled cannot be restared by itself. Manual duplications can be possible. Is there any other method to cut this pie?

[Amol_Nair · Accepted solution]

Suspend the SLP and then stop the Netbackup services on the media server for your maintenance work.. This will lead to the active SLP jobs failing but they would not ve retried as the SLP would be suspended then. Once the media server's maintenance is completed activate the SLP and they should process normally.. The suspend operation will work for all the secondary operations in the SLP other than active duplication jobs in th activity monitor.. If yoy would like to manually stop the jobs then just filter the active SLP jobs in the activity monitor and cancel the jobs from the GUI i.e. the activity monitor after yoy suspend the SLP.. After maintenance once the SLP is activated the jobs would start running normally

[Genericus]

My steps are as follows:

1. deactivate SLP

2. deactivate policies

3. cancel all active jobs - they will retry when you restart NetBackup

I have scripts that do steps 1 & 2, or you can use manual commands at the GUI or command line.

You do not need to list out SLP not complete, NetBackup does that for you with "nbstlutil report"

[X2 · Accepted solution]

The steps posted by Genericus do not take care of ongoing SLP operations, hence should not be market as solution. (DONE).

Suggestion by Anmol is more relevant to the situation I described.

A couple of days ago I did the following during maintenance:

1. Suspended all SLPs/policies about 2hours before the downtime*

2. Got a list of ongoing SLP operations

3. Shutdown NetBackup/maintenance

4. Restarted NetBackup

5 Reactivated policies and SLPs.

6. Later verified that images which were in flight (incomplete operations) before shutdown and after suspension were all in the catalog and had their duplicated copies too.

* - can exclude the SLPs in advance for the whole day if maintenance duration is longer.

[sdo]

I think we're mixing up the term "cancel" here. If you say "cancel" an SLP to me, this makes me ask a question... Do you actually mean cancel (abort) the SLP job in activity monitor (which will auto restart later), or do you mean cancel the SLP from the CLI which will cause the SLP to never be re-attempted? Genericus' advice is the way I would do it... because the SLP jobs will auto restart when NetBackup is restarted. There should be no need to list SLPs or list incomplete images... all you would be achieving is to second guess the software... there should be no need to test Symantec's software for them. SLP processing is pretty solid these days... trust it to do what it should do. It is exactly because we sometimes have to bounce/reboot that SLP processing was developed the way it has been. To help us, and reduce our required effort... hence Genericus' short procedure.