Poor old zest.lemonia.org dropped off air, apparently with a dead CPU or motherboard. Its standby box,
juice.lemonia.org was brought up with 15 hour old data (the backup sync cron’d at 0300, the failure happened at ~1800), but not before attempting to stick one half of zest’s mirror into juice. The root fs at /dev/ad1s1a was fine, but the main data fs wasn’t:
[root@ /var]# fsck /dev/ad1s1d
** /dev/ad1s1d
CANNOT READ BLK: 204735648
CONTINUE? [yn] y
THE FOLLOWING DISK SECTORS COULD NOT BE READ: 204735648, 204735649, 204735650, 204735651,
/dev/ad1s1d: INCOMPLETE LABEL: type 4.2BSD fsize 0, frag 0, cpg 0, size 205012094
[root@ /var]# mount -o ro /mnt/ark
WARNING: /mnt/ark was not properly dismounted
[root@ /var]# cd /mnt/ark
[root@ /mnt/ark]# ls
.snap home old
[root@ /mnt/ark]# ls -l
panic: vrele: negative ref cnt
Uptime: 28m42s
Cannot dump. No dump device defined.
Automatic reboot in 15 seconds - press a key on the console to abort
Is this the BIOS-y raid storing its metadata where it shouldn’t? Chipset is a Promise PDC20276.
Of course, there could be a genuine problem with this disk, but I won’t know till I’ve found a replacement CPU / mobo.
One Response to “Onboard RAID”
Leave a Reply
Recent articles
- MySQL replication health
(Tuesday, 09. 30. 2008 – No Comments) - Firewalls
(Friday, 09. 5. 2008 – No Comments) - Using Single Sign-On To Integrate Ning With An External Site
(Wednesday, 08. 20. 2008 – No Comments) - High Availability Rails Cluster
(Tuesday, 08. 19. 2008 – No Comments)
August 25th, 2006 at 3:17 pm
It wasn’t. Both mirror components mounted fine elsewhere. It seemed juice’s age made it incapable of groking 120GB disks, despite the BIOS being reasonably happy with them.