07-06-2012 11:16 AM - edited 07-06-2012 12:25 PM
A very simple question. What command do I run if one of my two internal boot disks fail?
These two disks are mirrored together via a HP SmartArray Raid controller. It is raid type 1/0.
When I look for the disk via ioscan, it is ONE disk.
When I look via command: saconfig /dev/ciss0, it is TWO disks.
Now I bet a saconfig or sautil command of some format might be used to cut off the dead disk, or maybe that's automatic.
Maybe there is a command to hook in the NEW disk that has replaced the dead, removed disk.
It would be nice to know these two magic phrases BEFORE just a disaster occurs. Oh sure.... I bet I can just pull out a disk and pretend I have a disaster to figure it out. But if I do NOT figure it out.....it would be a REAL DISASTER.
So what is the magic phrase? open-sesame? saconfig -stuff -wrkin -now?
I just checked the man page on saconfig again. And there is no command listed to unhook a dead disk and to hook in a replaced disk. Perhaps it is automatic? If so, you would think the man page might say something like that right?
Solved! Go to Solution.
07-06-2012 01:10 PM
With SmartArray hardware RAID controllers, the usual procedure is that the controller automatically cuts off the failed disk and lights up a LED to indicate which disk to replace. As soon as the failed disk is removed and a new disk inserted in its place, the controller will automatically start the recovery process. However, some (pre-SAS?) Integrity server models apparently don't have the circuitry to detect disk hot-swaps, so you'll have to explicitly tell the controller to have another look at the disks:
sautil /dev/ciss0 scan
As the controller detects that the failed disk has been replaced, you should get a drive status change message in syslog saying that the logical drive has transitioned from INTERIM RECOVERY to READY FOR RECOVERY. If auto-rebuild is enabled (by default it is), it will then immediately transition to RECOVERING.
If the auto-rebuild has been disabled or does not start for some reason, you can start it manually with:
sautil /dev/ciss0 start_recovery
You should have a look at the Knowledge Base document emr_na-c00858118-1 (for the earliest Integrity models with a SmartArray) and this URL (for the SmartArray SAS controllers):