netapp + oracle bug = disaster

Oracle

13146182

ORA-1499 ORA-8102 ORA-600 [kdsgrp1] Bitmap Index / Table mismatch

Netapp http://mysupport.netapp.com/NOW/cgi-bin/bol?Type=Detail&Display=872000

872000 

Under certain circumstances, described below, a giveback could result in the
 root aggregate on the partner node not being brought back to the same point 
 in time as the other aggregates in the HA pair.  Depending on the operations
 that were happening while the node was taken over, this could result in
 stale data being served.
 
 This scenario happens only when all of the following criteria are met:
 
 * The wafl.group_cp option is enabled ('on' or 'default' if MetroCluster is
   configured)
 
 * The node is running Data ONTAP 8.2.x (and the Data ONTAP version does not
   include a fix for this issue).
 
 * The node is performing a giveback.
 
 * For user data to be affected, it must reside in the partner's root
   aggregate, and write operations must have been initiated in a specific
   window of time relative to the giveback.