mirroring S.O.

[Javier Ortiz Guajardo]

i have mirrorimg lvols from the operating system and one of the disk mirror fails with a read-write error and the system were hangs and i had to force the reboot, the question is, is correct that the system were hangs?.
why the system do not continue working with the principal disk.
thanks for your help and pardon my english.

jortiz@gan.com.mx

[Anu Mathew · Accepted solution]

Hello,

Well, it is kind of strange.
However I remember our DBAs complaining of errors as one of the drives with mirror copies of a raw logical volume went nuts. But filesystems from the other PV were standing tall..!!

As yours looks like OS mirrors, are you sure that you had lvlnboot'ed the important (root/boot/swap/dump) LVs while mirroring them?

Before rebooting, have you tried to lvreduce the mirrors from the failed PV?

Also, are you sure that the failed PV had no un-mirrored logical volumes?

Thanks

Anu Mathew

[A. Clay Stephenson]

I have had dozens of system disks to fail and have never hung the system. In fact, I have never had to shutdown to replace a system disk because I only use hot-plug disks. I suspect that you missed mirroring at least one of the LVOL's - maybe you missed a swap LVOL?

[S.K. Chan]

If the mirroring steps are done correctly, your system should not hang unless the quorum requirement is not met. For a volume group to be running all the time it must have half or more disks available. For example if you volume group (that is mirrored) have 2 disks and one of them failed, the volume group will still stay activate. Another opposite example is if you have 5 disks in a volume group and 3 of them failed, then your volume group will hang. Can you provide detail information of your LVM configuration ?

[James R. Ferguson]

Hi:

A quorum issue as SK Chan suggests seems like a good possibility. See if your AUTO lif contains the low quorum string deposited by 'mkboot'. Do this:

# lifcp /dev/rdsk/c0t6d0:AUTO - |xd -tc

...you want to see this (with the '-lq':

h p u x - l q \n \0 \0 \0 \0 \0 \0 \0

Regards!

...JRF...

[A. Clay Stephenson]

Lack of quorum would explain why a reboot wouldn't work but it would not explain why a running system might hang. It is possible that a badly defective disk might hang everything on a bus including some non-system but still critical disks. Also, if your system disks (both mirrors) were on the same SCSI bus that might hang the entire system.