Date: Sat, 23 Nov 1996 20:50:07 -0600 (CST) From: Joe Greco <jgreco@brasil.moneng.mei.com> To: andrew@pubnix.net Cc: isp@freebsd.org Subject: Re: News server disk Striping (more) Message-ID: <199611240250.UAA12755@brasil.moneng.mei.com> In-Reply-To: <Pine.BSF.3.91.961123211959.6459H-100000@guardian.fortress.org> from "Andrew Webster" at Nov 23, 96 09:21:16 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> I had posted this about a week ago, but no suggestions, I did notice > however that when it locks up a drive and SCSI controller both have their > activity lights lit. > > Bugs in the SCSI code under heavy load? > > The hardware consists of a pair of 2.1 ST32151W drives on an AHA-2940UW > controller. I have (very!) intermittently seen the same thing happen with ST-31055N's and NCR-810 controllers - the drive appears to "wedge". The machine in question is certainly under heavy load, and seems to develop stuck drive syndrome about once a month. Since the 31055N and 32151W's are cousins, and the NCR-810 and AHA-2940UW are totally different drivers (at least pre-2.2), this tends to confirm my suspicion that this may be some sort of obscure drive firmware bug. An additional data point may be that I have _only_ seen this on heterogeneous drive chains; i.e. I have not seen a 31055N that is on a chain with a 32550N and 15150N lock up. The lock ups only seem to happen on a homogeneous drive chain. (Sample set is three dozen drives, soon two dozen more). I think it is some low level SCSI drive bogon, probably not (strictly) a FreeBSDism. It may be possible that FreeBSD could "work around" this with some sort of timeout and recovery strategy that may not currently exist. ... JG
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199611240250.UAA12755>