Purging Physical Disks to Repair Failed Pools

When one or more physical disks in a pool are failed or missing, the pool will have an offline or unavailable status and read/write operations to all virtual disks in the pool will fail.

The failed or missing physical disks can be purged in order to make the pool healthy and functional again. This operation is useful when physical disks fail and are not mirrored in disk pools.

Purging a physical disk removes the disk from the pool and from the configuration. During this process, recoverable data will be copied to other physical disks in the same pool and mirrored virtual disks will be resynchronized using a log recovery. Data on the failed disk can be fully recovered for a mirrored virtual disk. Single and dual virtual disks have no fault tolerance at the physical disk level and data that existed on the failed physical disk will be lost. After the purge disk operation is complete, lost data in a single or dual virtual disks will appear as zeroes. A failed pool will go online again after all failed disks have been purged.

When a physical disk is selected to be purged, a complete list of affected virtual disks is displayed in the wizard along with information pertaining to those virtual disks. The wizard will determine and report any actions needed to prepare the physical disk to be purged (see Important notes below). After all virtual disk preparations are complete, the amount of recoverable data for every affected virtual disk will be listed in the wizard.

This operation can be performed while mirrored virtual disks are served to hosts.

 Important:

o           Before beginning the purge operation, check for the cause of the disk failure to ensure that the failure is not a transient or repairable issue. Once it is determined that the data residing on the physical disk is lost, then proceed with the purge.  

o           If you are running a version of this software prior to 10.0 PSP3, contact DataCore Technical Support for assistance before running the purge.

o           Affected virtual disks must be prepared before the purge operation can be completed. Affected replications must be split, snapshots created from affected sources must be split or deleted, rollbacks created from affected sources must be deleted and affected virtual disks with Continuous Data Protection or Sequential Storage enabled, must be split and the affected source must be deleted. Affected mirrored virtual disks that have an unhealthy remote side will have to be split and unserved. The inactive storage source in a 3-copy virtual disk must be split and unserved if affected. Required actions for specific virtual disks will be outlined in the wizard.

o           Data can be recovered for affected mirrored virtual disks, ensure both servers owning storage sources in affected mirrored virtual disks are online. If a server that owns an affected virtual disk is not or cannot be powered on, the mirrored virtual disk will need to be split and unserved. Required actions for specific virtual disks will be outlined in the wizard.

o           Sufficient space must be available in the pool to accommodate the SAUs copied from the failed or missing disk. If insufficient space exists in the pool, the operation will be executed and the pool will go online, but mirror virtual disks will remain in a Redundancy Failed state until space is available.

o           The DataCore Server must be running for the Purge operation to be available.

o           This operation is only available for failed/offline physical disks. In the case of pool-mirrored physical disks, both disks should be in an offline state. The operation is not available when all disks in a pool have failed; in that case, the pool must be deleted.

o           A physical disk that has been renamed (and therefore has an alias) will remain displayed in the configuration even after it has been purged. To permanently remove the physical disk with an alias, simply remove the alias by using the Rename command or by editing the name in the details page. This can be done before or after the purge operation.

o           Purging disks retains the current formatting and pool identifier information written to the LUN. A previously purged disk added to another pool will be seen as a foreign pool disk instead of usable storage. In order to add a previously purged disk to another pool the disk should be formatted.

To purge a failed or missing disk:

1            Remove the alias of any physical disks to be removed. The alias can be removed from the details page or renaming from the DataCore Servers Panel or list.

2           In the Disk Pool Details page>Physical Disks tab, right-click on the failed or missing physical disk and select Purge Disk.

3           In Step 1 of 2 of the wizard, this page will list affected virtual disks and actions to be performed if the virtual disks require any preparation before proceeding with the Purge Disk operation. If virtual disks are listed, perform the actions for the virtual disks. Most operations required can be directly accessed by a context menu available from the list. As the preparations are completed, the virtual disks will be removed from the list. When no virtual disks are listed, click Next to continue with the operation.

4          In Step 2 of 2 in the wizard, this page will list all affected virtual disks and the amount of recoverable data for each. Click Purge to continue with the operation.

5           After the Purge Disk operation is complete, the failed physical disk will be removed from the configuration. The failed physical disk can be physically removed from the server when convenient.