Deleting Savesets in NetWorker 7.5x

I have observed a new behaviour in NetWorker 7.5.x that has been actively discussed with EMC, due to the potential for impact on customer sites.

In versions of NetWorker 7.4.x and below, it was always possible to use nsrmm to delete savesets that exist on disk backup units, then run a space reclamation (either via nsrstage's clean volume operation, or nsrim -X) to have NetWorker erase the space on the disk backup unit.

Under v7.5.x it would appear that this no longer works. If a saveset is deleted manually using nsrmm, NetWorker cannot reclaim its space either using the nsrstage clean volume operation or nsrim -X. This results in needing to manually review long saveset lists for each disk backup unit and purging files that are on disk that are not in the media database — something that is time consuming and potentially dangerous. Thus, this should not be attempted without support assistance.

In the interim if you have upgraded to 7.5.x and you need to manually reclaim space from disk backup units, we recommend you change your procedure to set the saveset browse/retention time to "now", then run nsrim —X.

  • Old procedure:
    # nsrmm -dy -S ssid
    # nsrim -X
    # nsrmm -dy -S ssid
    # nsrstage -C -V volume
  • New procedure:
    # nsrmm —w "+1 minute" -e "+1 minute" -S ssid/cloneid
    (wait 1 minute)
    # nsrim -X

If you have upgraded to NetWorker 7.5.x and issued nsrmm delete commands, you will need to compare and analyse disk backup contents against media database details (via "ssid(60)" to get the long saveset ID — the filename of the saveset).

As a result of an escalation with EMC NetWorker engineering over this issue, a number of patches to rectify this behaviour for key operating systems, and have been advised that a maintenance bundle for 7.5.1 has been released.