Committed Space estimations are too high when User or Application Backups are being submitted to Advanced Disk Disk Pools

Article: 100000994
Last Published: 2018-02-11
Ratings: 0 0
Product(s): NetBackup & Alta Data Protection

Problem

When performing backups to Advanced Disk volumes/disks, NetBackup maintains several counters that reflect the amount of space already in use, plus the amount that in-progress jobs are expected to use.  The space that is not yet used but expected to be used is referred to as 'committed'. 

When jobs are started the ' committed_space' counter is updated based on the estimated size of the data being backed up.  Backups submitted manually or scheduled which have Full or Incremental schedules can be estimated based on the size of previous backups run of the same type and policy for the client.   The estimate is a fixed percentage (120%) and used to adjust the 'committed_space' for the disk receiving the backup image.

If a previous backup is not present, or the job is a User or Application Backup, then a prior image size cannot be used to reliably determine an estimate.  In those instances, an estimiate must still be made and the default calculation is (100 - High Water Mark)% of the disk. 

As a result, jobs may remain in a queued state or use an unexpected storage unit.
 

Error Message

The job details and/or logs (rem and dps) may show the current configuration and commit calculations that prevent a job from using the expected storage unit.  In this case the high water mark is 85% resulting in an estimate of 19 TB for this job and other jobs already have estimates that result in a committed space total that exceeds the available space on the volume.

[select_from_stu_list] ATTEMPTING TO ALLOCATE FROM STORAGE UNIT - stu_t: name = my-ost, ...
[validate_stu] list_mode = CONSIDER_DRIVE_MEDIA_AVAILABILITY
[kbytes_to_commit] kbytes_needed = 0, user_job_kbytes_estimate = 0
[kbytes_to_commit] remainder = 15% , calculated kbytes_to_commit = 20868141542
[check_disk_space] disk pool/volume does not have sufficient best-fit space, name = my-lsu, avail_space = 142240799784960, committed_space = 170441765601280, kbytes_needed = 20868141542
V-143-1546 [validate_stu] Disk volume is Full or down @aaaan

 

Cause

The 'committed_space' value for the disk is an accumulation of all estimates of backups being performed to that volume concurrently. The 'committed_space' for each backup in progress will decrement as the backup proceeds and reach 0 when either the job completes or the current backup image exceeds the size of the estimate.  Note that the the committed space for the volume will still be greater than zero if other backups are running.  When all backups have completed the committed space will return to zero.

Note that If the high water mark is 90%, 10 concurrent User/Application jobs will commit the entire volume - even if otherwise empty - and additional jobs will not be able to use the volume.

The nbdevquery command can be used to review the total and committed space.  In this case, the committed space is 0 so either all backups have completed or those in progress have exceeded their estimate.

$ /usr/openv/netbackup/bin/admincmd/nbdevquery -listdv -D -stype AdvancedDisk -dp < diskpool_name>
 name                : </MOUNT-POINT>
 id                  : </MOUNT-POINT>
 diskpool            : <DISKPOOL_NAME::STORAGE_SERVER::AdvancedDisk>
 disk_media_id       : <DISK_ID_for_example_@aaaab>
 total_capacity      : 2147483648 
 total_phys_capacity : 2147483648
 free_space          : 2126877696
 free_phys_space     : 20605952
 potential_free_space: 0
 committed_space     : 0
 precommitted_space  : 0
 nbu_state           : 2
 sts_state           : 0
 flags               : 0x6
 num_read_mounts     : 0
 max_read_mounts     : 0
 num_write_mounts    : 1
 max_write_mounts    : 2147483647
 system_tag          : <Unknown>

 

Solution

At NetBackup 6.5.6 or 7.0.1 and above, it is possible for the administrator to define a better estimate for User and Application Backup jobs. Otherwise a 100 TB volume with a 95% high water mark would result in 5 TB being committed for each non-scheduled user backup. 

/usr/openv/netbackup/bin/admincmd/nbemmcmd -changesetting -USER_JOB_KBYTES_ESTIMATE < kbytes> -machinename < masterservername>

Alternate options:

Increase the High Water Mark (HWM) to 100%. This prevents the Committed Space from being allocated

    or

Configure EMM to prevent job throttling.  The committed space is still calculated, however the number of jobs that can run simultaneously is not restricted.

/usr/openv/netbackup/bin/admincmd/nbemmcmd -changesetting -DISABLE_DISK_STU_JOB_THROTTLING 1 -machinename < master_server>

For further information on Advanced Disk options and configurations,  refer to the Related Article linked below.
 

References

Etrack : 1938338 Etrack : 1938334

Was this content helpful?