David Robinson ([EMAIL PROTECTED]) wrote on 8 June 1999 09:15:
>Raid1 root worked prefectly for me until I did a test:-)
>
>Have you tested your raid5?
Not explicitly removing one of the disks, because I can't do this with
a live system. I don't have special cases for this, and I'm not going
to just pull the plug. The only way is to power off, disconnect and
power on. But I can't fiddle too much with the machine, it's an
essential production server... Everything depends on it.
What happened many times some months ago is that the kernel crashed
due to quota problems, and it booted with root-raid and reconstructed
parity fine.
On another ocasion we had cable problems which produced LOTS of SCSI
bus resets. Once this caused a raid reject, it said the disk was out
of sync and left it out of the raid. It happened with the /var
partition only though. I didn't care to raidhotadd it back, because
shortly after I repartitioned. I now have only / and /home, and both
spread through all disks. So if any disk goes out of sync the root
partition will be affected.
