HP NetRAID-4M Configuration and Upgrade Guide (Release 5)

Chapter 9 Issues and Problem Resolution
42
Problem: The PASSTHRU command s/g 3-4GB mapping routine added in an
earlier build was not checking for a ZERO s/g count and was thus processing a non-
existent s/g list resulting in a infinite loop in the kernel code.
Impact to customer: Serious. Controller can hang causing data unavailability.
Resolution: Pass-thru code was fixed to check for zero SG count.
11 Problem with double SCSI bus reset
Problem: A change was made for a double SCSI bus reset issue in the error isolation
process module. The change eliminated the double SCSI bus reset. However, the
scope of the SCSI bus reset that was eliminated was also used in a separate process
of the error isolation module. The removal of the SCSI bus reset from this process
causes the controller to panic during error isolation.
Impact to customer: The recovery routine is now more robust.
Resolution: The code for the separate process of the error isolation module was
fixed to handle this new change, so it will no longer hang.
12 OS-Managed container cannot be deleted
Problem: The main problem is due to the fact that after the failure, there is a copy
of the original container without a drive letter or file system assigned. This makes
the container still OS-Managed, not allowing a user to delete the container. The only
way to delete the container is to take the disks out of the diskset, which causes data
loss on these drives.
Impact to customer: Moderate. To completely delete the container, in this case,
take down the desksets first and rebuild the container from a current backup.
Resolution: The fix involved removing the collision between the nodes trying to do
a domain validation on the same enclosure.
13 RAID-5 or RAID-50 hang
Problem:When more than one disk is lost in RAID-5 or RAID-50, the failure of
RAID-5or RAID-50 hangs the operating system. In a configuration where the
operating system is residing on the NetRAID-4M, the operating system is not
accessible even after powering off the HP Server.
Impact to customer: Serious. The NetRAID-4M is not accessible thereby causing
the data to be unavailable.