Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 21 Sep 1998 14:22:10 -0600
From:      "Justin T. Gibbs" <gibbs@plutotech.com>
To:        Robert Crowe <bob@boogie.ipinc.com>
Cc:        "Justin T. Gibbs" <gibbs@plutotech.com>, current@FreeBSD.ORG, bob@carlsbad.ipinc.com
Subject:   Re: Problems with DPT/CAM and RAIDed drives under 3.0-CURRENT 
Message-ID:  <199809212028.OAA05481@pluto.plutotech.com>
In-Reply-To: Your message of "Sun, 20 Sep 1998 21:57:28 PDT." <199809210457.VAA26664@boogie.ipinc.com> 

next in thread | previous in thread | raw e-mail | index | archive | help
>
>Not sure if this is a known problem or not, but under CURRENT (synced
>on 9/20), the kernel fails to boot from a raided disk.  If I break the
>raid array (or even just yank one of the disks and let the controller
>think the array is degraded) the kernel boots just fine.  When the
>array is built, I get the following lines right after probing npx0:
>
>(probe0:0:0:0): CCB 0xf875a07c - timed out

How quickly after "npx0"?  The current timeouts are 60s long, so it seems
surprising to me that the DPT could take that long to get going.  The
timeouts were bumped at around the same time as dpt rev 1.15 was committed.

>With dpt_scsi 1.14, this error would print about 5 times then panic
>with a trap type 12 at
>
>_camq_insert + 0xf  movl %ecx,0(%eax,%edx,4)

Yes.  This is the "ccb done twice" problem I fixed in the error recovery
code with 1.15.

>With dpt_scsi 1.15 it appears to just hang after the print (or maybe I 
>didn't wait long enough)

Hard to say.  I may be confused about how the DPT handles command abort
requests.  Either that or it takes "forever" for mirrored pairs to
respond to simple commands like "inquiry" soon after the DPT is booted.
Did you have a background mirror reconstruction task going on?

I have a RAID 5 array building on a DPT PM3224 right now and if I can't
reproduce the problem with that configuration, I'll try some mirrored
configurations.

--
Justin



To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199809212028.OAA05481>