[2892] in linux-scsi channel archive
Disk Subsystem Flakiness
daemon@ATHENA.MIT.EDU (Troy Wu)
Sun Nov 30 17:36:36 1997
Date: Sun, 30 Nov 1997 17:26:03 -0500 (EST)
From: Troy Wu <troy@sixdegrees.com>
To: Linux RAID Group <linux-raid@vger.rutgers.edu>
cc: Linux SCSI Group <linux-scsi@vger.rutgers.edu>,
Mehlam Shakir <mehlam@sixdegrees.com>
I posted a message a week ago about a mail server with a SCSI/RAID
disk subsystem hanging.
I've come closer to the problem with each crash (grunt), and the
lastest incident where the machine hung (which I've seen twice
already) was on a ckraid after the machine is hard-rebooted. After a
while (duration varies on each recovery attempt), ckraid hangs.
I'm using:
BusLogic SCSI Driver Version 2.0.10 of 11 August 1997
BusLogic Model BT-958 PCI Wide Ultra SCSI Host Adapter
with
RAID-1
made with the
raidtools-0.42 RAID utilities
and the
raid145-0.36.3-2.0.30 patch
against the
linux-2.0.31-pre10 kernel
I'm running the same RAID configuration on our test machine; the big
difference being that the test machine has 2 IDE hard drives, and the
production machine (the hanging one), has 2 SCSI hard drives (and one
SCSI CD-ROM). The 2 IDE drives run on separate controllers (IDE-0 and
IDE-1). The 2 SCSI drives are on one SCSI controller/host adapter.
The test machine (IDE) never fails. The raidtools, kernel, and
patches, are identical.
Has anyone else observed this behavior? If so, does anyone have a
clue whether it's the SCSI driver or the RAID system? Again, I'm
still doing my homework with this, but any continued support is
greatly appreciated.
--troy
