EMC released a technical advisory on May 31st about some interop issues with VAAI (vSphere API for Array Integration) and VNX arrays using RecoverPoint Appliances that may cause the storage processors in the array to reboot. This could be a major issue if you were using Array based replication using RPA (RecoverPoint Appliance) – possibly for SRM.

The issue as noted by EMC states that the single or dual storage processors reboots continuously. The issue was noted to occur with VNX Operating Environment Release 32 P204.

EMC details the issue as below,

Storage Processor reboots due to incorrect access of a pageable address at high Interrupt Request Level (IRQL).

Part of the timeout mechanism dedicated for Compare and Sweep (CAS) and Punch commands, the RPSplitter may the check timeout of a command containing an IRP that has already finished. This means that it has a master_bh field which is NULL, and that will cause that error and panic.

The fix will be to check if the command has already finished, and if so, ignore it.

EMC says they don’t have a permanent fix to this problem however this can be avoided by disabling VAAI on hosts in the described environment. EMC has also recommended a Non-Disruptive Flare upgrade to R32 P204.

To disable VAAI on ESXi hosts – http://kb.vmware.com/kb/1033665

Also ensure that after the Flare upgrade – VAAI should remain disabled until EMC hotfix is available.

You know these constant reboots remind me of good ol’ Spirit that went into reboots until the patch was applied. Spirit is the Mars rover! And if you did not know that – no no that’s fine, go back to Facebook – boooo 🙂

Comment to discuss 🙂