EMC Technical Advisory (or “ETA”) are notices to customers of important fixes or updates. Some are more important than others.
If you go to http://support.emc.com, you can get all the ETAs for your platforms. Shortly, this portal is getting a ton of updates, so customers can see a ton of info for all their EMC solutions in a nice place, with some useful analytics. EMC customers, partners – you can learn more about this at EMC World.
Every once in a while, I see an ETA I think could use extra eyeballs – this is one.
You can get all the info at the ETA link above, but the core issue is that you can have VMDK data inconsistency on the Target of an XCOPY session, while the Symmetrix VMAX storage array does not exhibit internal symptoms indicating data inconsistency.
The symptom is that a VM may not be able to boot after some time following an XCOPY operation.
The Root Cause: This is due to an Enginuity issue. During a WRITE SAME/UNMAP operation for an XCOPY Target volume, the optimization function and an extent overlap can result in data inconsistency.
This problem was introduced with Fix 62776, released in 5876.159.102. This fix was written to optimize the WRITE SAME command in vMotion when data is copied from the Source to Target and the extent is overlapped. Refer also to solution EMC317607. The problem occurs in this optimization for WRITE SAME/UNMAP, when it is issued to an XCOPY Target volume. (TimeFinder/Clone is not affected since it does not perform this optimization.) The optimization skips the background copy on the tracks for which the WRITE SAME/UNMAP command was issued. The problem is that the extents sent to stop the background copy use member-LBA instead of host-LBA. Only Meta devices are affected by this problem.
The fix: Enginuity Fix 66930 addresses the issue and is available via Enginuity Pack at 5876.159.102 and 5876.163.105. Contact your local EMC Customer Service Representative or the EMC Support Center for further information on this issue.