Flashes (Alerts)
Abstract
Deletion of a deduplicated volume on V8.1.3.0 or V8.1.3.1 may cause a rare race condition that could result in a data reduction pool going offline, followed by multiple node warmstarts and loss of some blocks of data on remaining volumes.
Content
Problem description
Deletion of a deduplicated volume may cause a rare race condition that could result in a data reduction pool going offline, followed by multiple node warmstarts and loss of some blocks data on remaining volumes.
Affected releases
IBM Spectrum Virtualize software V8.1.3.0 and V8.1.3.1, supporting Storwize V5000, V7000, FlashSystem V9000 and SAN Volume Controller and only when using deduplication in Data Reduction Pools.
Fix
This issue is resolved in V8.1.3.2.
Customers currently using deduplication with V8.1.3.0 or V8.1.3.1 are urged not to perform any of the following activities until the system has been upgraded to V8.1.3.2 or higher:
- Delete any deduplicated volumes or volume copies
- This includes allowing management software (such as Flash Copy Manager) to delete any deduplicated volumes
- Change a volume from deduplicated to non-deduplicated using the GUI
- Disable consistency protection for deduplicated volumes in Metro Mirror or Global Mirror relationships
- Use volume mirroring with the automatic delete option (where the source is a deduplicated volume)
If any of the above activities have already been performed on a system running V8.1.3.0 or V8.1.3.1, please contact IBM Support for guidance on avoiding any impacts to system availability.
Was this topic helpful?
Document Information
Modified date:
28 March 2023
UID
ibm10730411