What are some reasons that will prevent OnCommand Unified Manager 5.X from purging backups/snapshots
- Last Updated:
- OnCommand Unified Manager 5.X ( DFM )
The following are the two main causes that can prevent OnCommand Unified Manager 5.x ( DFM ) from purging snapshots (backups):
- Free space within the DFM installation path drops below 10%. When this occurs, monitoring, perf data collection, and backup retention stops. However, the backups continue until the containing volumes or aggregates fill up. The free space can be checked in the "
dfm diag" output.
Installation Directory C:/Program Files/NetApp/DataFabricManager/DFM
45.3 GB free (83.1%)
- DFM will only purge the snapshots registered against a dataset. These are typically either created by DFM during a protection job or are individually registered against a dataset by a SnapManager product using an API call. If the snapshots that are not getting purged are not created by DFM, or are created by DFM, but during a time window that had been removed from the database (for example, by reverting), then it can be explained that DFM is no longer aware of these snapshots, although it created them.
Note: If reverting to a previous backup did occur, then DFM is no longer aware of any snapshots or any other changes that occurred during the time between the start of the pre-upgrade backup and restore. DFM will therefore not purge any snapshots created during this time window based on any retention rules. The snapshots can be manually deleted as required, by using the storage system CLI, OnCommand System Manager, or DFM.
There are other reasons why snapshot retention might not be working, such as the following:
- Controller communication issues (purging is performed at the start of the backup jobs through API calls to the hosting controller)
- Incorrect retention settings (purging Daily but only creating Weekly style backups)
The reasons above are the most likely causes of unpurged backups. If the cause for the issue is not one of the above, then the following data needs to be collected and attached to a support case:
- A recent DFMDC
- The output of the ' dfpm job detail JOBID ' command for a few DFM backup jobs where purging has not occurred as expected.
- The output of the ' dfpm policy node get POLICY_NAME_OR_ID ' command for the protection policy of the affected dataset for which the jobs were run for the retention settings.
- The output of the ' dfm host diag CONTROLLER_NAME_OR_IP ' command to a few controllers within the dataset that are hosting the aging snapshots.
- The names of a few volumes or aging snapshots on these controllers along with new AutoSupports displaying these snapshots.
- The controller ‘auditlog’ logs covering the time frame of the DFM backup jobs collected in item 2 above; these are located in the controller’s /etc/log directory.