Slow I/O with megaraid and u160 scsi jbod

From: Phil Dier
Date: Mon Aug 29 2005 - 13:15:45 EST


Hi,

I've had luck with this patch[1] (it at least eliminated the oopses
I was getting), but now I'm having a different sort of problem with
my setup[2] (the 2.6.13 release exhibits this behaviour as well). I
have 2 u160 scsi jbods connected to this machine[3]. One is connected
to an Adaptec card, and the other is connected to the Fusion MPT card
(megaraid). All of the disks connected to the Adaptec card work fine,
but when doing I/O on disks 2 and 3 on the megaraid card, it stalls
considerably. When trying to sync a RAID1 device using the 4 148GB
disks, the sync speed never goes above 2KB/s. When they finally get
synched, IO stalls constantly. Watching with iostat confirms this. I've
formatted the disks in question with a single JFS partition and they
still exhibit this behaviour when used by themselves. I have verified
that this behaviour is not present up until at least 2.6.12.3. Let me
know what info I can collect that would be helpful.. Thanks.



[1] http://www.ussg.iu.edu/hypermail/linux/kernel/0508.1/1952.html
[2] http://www.icglink.com/debug-2.6.13-rc6.html
[3] Diagram:

+---------+
| Adaptec |
+---------+
|
+-------+-------+-------+-------+-------+
| id: 0 | id: 1 | id: 2 | id: 3 | id: 4 |
| 73GB | 73GB | 148GB | 148GB | 73GB |
+-------+-------+-------+-------+-------+

+----------+
| megaraid |
+----------+
|
+-------+-------+-------+-------+-------+
| id: 0 | id: 1 | id: 2 | id: 3 | id: 4 |
| 73GB | 73GB | 148GB | 148GB | 73GB |
+-------+-------+-------+-------+-------+

--

Phil Dier (ICGLink.com -- 615 370-1530 x733)

/* vim:set noai nocindent ts=8 sw=8: */
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/