From Bugzilla Helper:
User-Agent: Mozilla/5.0 (compatible; Konqueror/3; Linux)
Description of problem:
Doing some experiments with the SAN here. After seeing that syncing 2 RAID 1's simultaneously is very slow, I wanted to speed things up a bit and did a raidstop /dev/md0
Result: kernel panic (see attachment)
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Make two RAID 1's (/dev/md) using sd[ac]1 and sd[bd]1
on 4 logical drives on a LSI storage (~73 GB each) connected via QLogic 2300 HBA (driven by qla2x00.o)
2. Do a mkraid -R /dev/md
3. Do a raidstop /dev/md1
Actual Results: Kernel panic
Expected Results: Obvious.....
Compaq DL 580, 4xXeon 700, 4 GB, Onboard Smart-2
2x Qlogic 2300 FC HBA
Created attachment 74684 [details]
Dump of the observation of /proc/mdstat and the kernel panic
qla2x00.o panics.. film at 11...at least it didn't eat your data this time.
Why aren't you using qla2300.o ?
Well, see #73356
I just did it again without watching /proc/mdstat, and it worked...
Ok, I think I have to apologize...
I just learned that the SAN guys made the four disks a 2x2 mirror,
although they explicitly told me they hadn't...
This explains some of the problems I had...
Nevertheless, I think the issue concerning "raidstop" while having a
"while true; do grep sync /proc/mdstat ; sleep 1 ; echo; done"
seems to be generic to me...
This has been fixed in the e.12 kernel errata that was released
at the beginning of this year.