How to start image cleanup fo HWM from cmd

I'm no expert on this facet but isnt the point of it to be between high and low water marks and left to its own devices? If the HWM is crossed then the software reacts , ie its self managing. If the usage is exceeding HWM and images arent deleted then it suggests the storage is incorrectly sized.

Hopefully a better authority can provide more insight.

Jim

Hello, 

You should not have to worry about High and Low water marks unless your running out of space. Make sure the drive you allocated, is only being used for the disk pool, and that you don't have to many images on there with high retentions.

Also you should see the image cleanup job running about every 12 hours. This runs the bpimage -cleanup -allclients command and unless there are to many images that have not expired yet, you should see more space. 

Hello:

You did not state what version of Netbackup you are currently using. The following technote was written for Netbackup 6.5, however the process still applies in 7.x. You will note that the documentation states:

"the fact that staged images are not cleaned up when the DSSU is below high water mark does not represent a problem by itself...The High Water Mark setting is there as a flag to begin clearing staged images and prevent a full disk condition."

http://www.symantec.com/docs/TECH66149

 

Hope this helps.

 

You say Advanced disk... 

Can we start at the beginning?

Are these Advanced Disk STU part of Storage Lifecycle Policies? What do SLPs look like? 
Fixed retention on disk or 'try to keep' retentions with duplication to tape?

If no SLPs, how are you expecting expiration/cleanup of disk images to happen?

OK, seems to be not enough info. NetBackup 7.6.0.2 RHEL 6.5 AdvancedDisk with 20 TB of Capacity all backups to pool with slp and capacity managed retention. SLP use STU groups (is a must) Our problem is that we have more than 40 thousand backups over weekend. if the pool (pool device) reaches HWL no backup is allowed to write to pool and we get status 129. If image cleanup start to delete under HWL everything is fine. But in the few minutes till image cleanup was startet we ge a lot of 129'ers (sometimes 10 to 50 per pool). Most of the backups are RMAN backups (user backups). So backups fail and there is no retry from NBU. My thinking was to start the command with is used with Image cleanup to clean HWM. There where a lot of dicussions that earlier NBU versions queue the backups till HWM ist under boundary but at the moment NBU works as designed.

In addition to mph's excellent post I would add some documentation for fine tuning capacity managed retention. Of particular note: "Smantec does not recommend allowing capacity-managed images and fixed-retention images to be written to the same volume in a disk storage unit. The volume may fill with fixed-retention images and not allow the space management logic to operate as expected." I don't know if you have mixed retention type images in your enviornment, but these guidelines could be helpful in avoiding the 129 errors.

http://www.symantec.com/docs/HOWTO85955

http://www.symantec.com/docs/HOWTO87044

 

Regards

I am currently exploring a way to have OpsCenter alerting me with certain "disk pools" getting to the limit of its usage. Still in the progress...seeing your issue I think of a possibility, not quite automated but maybe good enough to act before it fails:

1) Say your HWM of the pool is 90%
2) In OpsCenter, set an alert policy to have it triggered when your disk pool usage reaches a pre-limit, say 85% (this is the part I am still exploring of how the value can be captured by OpsCenter)

3) Set a script in master server to trigger the "manual relocation" which would start duplication (or image cleanup if you don't need 2nd copies) to start cleaning up space.

 

We use AD SLP and have error 129 when doing Oracle RMAN full backup. It doesn't happen for Unix or RMAN incremental backup which image size may be much smaller. All backup images are duplicated to tape twice before next backup. 2 weeks ago, I wrote a script to record the free space of a AD every 15 seconds. The AD is only for one Oracle RMAN database backup.

The size of the AD is 1TB (1099GB). HWM is 81% and LWM is 70%. Database full backup is about 50GB (after compression in AD). The AD should have 19% free or 208G when it hit HWM. RMAN full backup was aborted when its image reached 9.7G. The free size of AD was 207.360G. The Oracle message is

ORA-19511: Error received from media manager layer, error text:
   VxBSACreateObject: Failed with error:
   Server Status: Disk Storage unit is full

The 207G free size has been kept for 45 seconds and increased to 331.461G in the next 15 seconds. I know NBU won't allow new backup when HWM is over but the 207G is useless for the current backup. It doesn't allow the current backup even waiting for 1 minute.

Our version is 7.1 and I don't know if the behaviour is different in other versions. I wish NBU has one more water mark, say Maximum Water Mark (MWM), and allows the current backup continue or even accept new backup until the MWM. Start the cleanup at HWM before it hit the wall MWM. Maybe some NBU features cover it which I don't know.

I am thinking to schedule a job to lower the HWM, trigger a cleanup and then raise the HWM back before full backup so I can ensure that there has at least 100GB usable AD. I get the comand to change the HWM. Can someone tell me how to trigger the cleanup of an identified AD (e.g. OracleRMAN)? We only have the production environment but I don't want to test it there.

nbdevconfig -changedp -dp OracleRMAN -stype AdvancedDisk -hwm 80 -lwm 70

(nbdelete or bpimage -cleanup or ...???)

nbdevconfig -changedp -dp OracleRMAN -stype AdvancedDisk -hwm 90 -lwm 70

 

Thanks