Simulating a failed disk

[EML]

We have new server that is ready for production in a week's time. It's currently running on HP-UX 11i with VA7110. My question is how do I simulate a failed disk when system is running? Can I just pull out 1 disk, and that would be same as a failed disk in OS point of view?

[Steven E. Protter]

If the disk is a hot swappable disk you can safely do this:

pvdisplay -v /dev/dsk/c#t#d#

make sure whats on it is what you expect. If its raid 5 striping, I'm not sure how relavent this is.

Then pull the disk out of the array.

I wouldn't do this except during a maintenance window.

So long as the OS is mirrored, you should be okay.

lvlnboot -v
should give you details so you want to pull a boot disk, which one to pull.

SEP

[Steven E. Protter]

If the disk is a hot swappable disk you can safely do this:

pvdisplay -v /dev/dsk/c#t#d#

make sure whats on it is what you expect. If its raid 5 striping, I'm not sure how relavent this is.

Then pull the disk out of the array.

I wouldn't do this except during a maintenance window.

So long as the OS is mirrored, you should be okay.

lvlnboot -v
should give you details so you want to pull a boot disk, which one to pull.

SEP

[Helen French]

Since you got hot-swappable disks, you can do this to simulate a failed disk. If the disk you are pulling out is in hardware RAID (1 or 5) the disk array controller will take care about the missing disk and will reconfigure the LUN. If it's on software mirror, the OS will report the missing disk and you can make use of the mirror disk in that case.

If the pulled disk is not part of any type of redundancy, the file system (of that disk) will hang and may result in system hangs and issues if it's an important file system.

[Leif Halvarsson_2]

Hi,
I assume you mean a disk in the VA7110. The OS actually don't know anything about the physical disks in the array. The server only know "LUNs", logical disks, the array controller takes care about the physical disks. If you pull out a disk and then put it back the controller must rebuild the disk again, there may be a performance decrease when this process is running but you should not loose any data.

[Rajesh G. Ghone]

Hello,
Whenever any disk is failed in VA the red LED will start flashing in the disk if it is not so & still you want to confirm it you can give this command
#armdsp -a va sr. no. or the alias
NOTE:If you dont know the VA sr. no. & alias name you can isue this command
#armdsp -i
this command will give the VA sr. no. & alias name if it is set.
Once you five armdsp -a command you can able to see all the disk status from that you can find out the faulty disk.

If you dont want to issue this command you can still run launcher command through Xterm & you can see the status of Disk & Array.

Hope this will help you.

Regards,
Rajesh G