SCSI 'Cluster reserved disks' are marked offlined when an abort is triggered.

Issue

• SCSI disks are marked/considered offline under the following circumstances:

1) We have clustered disks.
2) A cluster node reboots when another system 'owns' the disks with a SCSI reservation.
3) While scanning for disks on reboot, a SCSI abort happens, usually from a dropped frame. This is why it is intermittent (it can only happen here).
4) The SCSI mid-layer is hard-coded to do a Test-Unit-Ready (TUR) when recovering from an abort.
5) TUR is hard-coded to interpret a 'SCSI Reservation Error' as a failure.