I just got an Exadata Customer Advisory e-mail from Oracle, identifying “an important issue that needs your immediate attention” and pointing to MOS note 1265396.1.
Note 1265396.1 talks of an issue with disk controllers in Exadata storage servers declaring false disk failures. In some “rare cases”, multiple such false failures occur at the same some, potentially causing data loss.
The issue applies to Sun (V2) database machines only, not the HP V1 or database machines or with the newer X2-2/X2-8 database servers.
The only fix is to upgrade to the newly-released version 184.108.40.206.1 of the storage server software. A combination of an upgrade and manually forcing disks back online at the disk controller and ASM levels can “bring back” the data in case of loss, assuming the disks haven’t been physically removed of course.
The 220.127.116.11.1 patch will run on any 18.104.22.168+ databases, though either 22.214.171.124 or 126.96.36.199 plus bundle patch 6 is recommended. If the database patches are in place, the storage server patch can be applied in a rolling fashion without downtime.