I recently put together a large Raid 5 file server.
Computer is running Fedora Core 3. Raid 5 is software raid using mdadm.
Because the network here at my house is used by a bunch of semi-literate
(just enough to fuck things up) people, I run a chown script against the
root of the samba share every so often. This has been working fine, but
has brought some things to my attention. I am loosing files slowly.
The Raid itself is 5 WD 320gig SATA drives. Using a supermicro enclosure
and 2 crappy little PCI SATA controller cards. /dev/md1 is formatted as
an ext3 file system. I will be getting a nicer controller card in the
next few months (moving to a PCI-X architechture, but for now this is
all I have).
The errors I'm getting are:
code:
------------------------------------------------------------------------
Message 72:
From root@xxxxxxxxxxxxxxxxxxx <mailto:root@xxxxxxxxxxxxxxxxxxx> Fri Sep 2 10:30:32 2005
Date: Fri, 2 Sep 2005 10:30:25 -0400
From: root@xxxxxxxxxxxxxxxxxxx <mailto:root@xxxxxxxxxxxxxxxxxxx> (Cron Daemon)
To: root@xxxxxxxxxxxxxxxxxxx <mailto:root@xxxxxxxxxxxxxxxxxxx>
Subject: Cron <root@RIAA> chown -R nobody.nobody /fileserv
chown: cannot access `/fileserv/Audio/Jay's Shit/Rap Hip-Hop/Talib Kweli/Talib Kweli & Hi-tek - Reflection Eternal/010-reflection_eternal-name_of_the_game-ego.mp3':
Input/output error
------------------------------------------------------------------------
This file is now GONE. It will show up in an ls, but i can not access
said file at all. The only thing I can do with it is rm -f it or run an
fsck (on the md device) and it just disappears after the fsck is complete.
/proc/mdstat seems to show everything as ok
code:
------------------------------------------------------------------------
[root@RIAA ~]# cat /proc/mdstat
Personalities : [raid5]
md1 : active raid5 sde1[4] sdd1[1] sdc1[0] sdb1[3] sda1[2]
1250274304 blocks level 5, 64k chunk, algorithm 2 [5/5] [UUUUU]
unused devices: <none>
------------------------------------------------------------------------
Do I have a bad drive? Or is something else going on here? Need more
info? Just ask, I'll provide what I can.
-
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
