SUMMARY: SunSwift problem on SPARCstation 20/Solaris 2.5

From: Marc S. Gibian (gibian@stars1.hanscom.af.mil)
Date: Wed Jul 16 1997 - 16:07:13 CDT


I asked about running a SunSwift SCSI on a SPARCstation 20/Solaris 2.5 system,
complicated by the fact that the system is heavily dependent on Solstice
DiskSuite and thus reported a DiskSuite error. The bottom line is that it
appears that the driver for the SunSwift SCSI port is NOT part of the core
install for Solaris 2.5 or 2.5.1, even though the documentation does not
indicate any software install necessary for the SCSI controller on the board. I
have not had time on my server to test this, but it makes a great deal of sense.

My thanks to:

Sydney Weinstein <syd@myxa.com>
Jim Harmon <jharmon@telecnnct.com>
Francis Liu <fxl@pulse.itd.uts.edu.au>
Glenn Satchell - Uniq Professional Services <Glenn.Satchell@Uniq.com.au>
robin.landis@imail.exim.gov
Michael <mcgeown@BBN.COM>

Marc S. Gibian
Telos Comsys phone: (617) 377-6350
PRISM/TFS email: gibian@stars1.hanscom.af.mil
                           or is it: gibian@hanscom.af.mil
                        well, maybe: gibianm@hanscom.af.mil
              and if all else fails: marc.gibian@acm.org

attached mail follows:


I have been chasing a problem on my server for a number of weeks now regarding
the SunSwift fast-wide-scsi/fast-ethernet card. After trying to work through
Sun, and having lost enough server time trying their suggestions, its time to
try this list...

My customer recently purchased a Sun 12 disk multipack and a SunSwift card to
provide the necessary additional fast-wide-scsi controller on their SPARCstation
20 server. The order arrived, but rather than the SunSwift card, I found a
single function fast-wide-scsi controller. I was hesitant to install this card
knowing I would have to swap it out for the SunSwift card when it arrived, but
was told it would be a month before that happened. I was also assured that
removing the single function card and adding the SunSwift card would be a simple
card swap. So, I went ahead and installed the single function card and moved the
team's critical data onto the 12pack. This configuration is running great. It is
providing us a real performance boost while finally providing the disk space
necessary to continue working on the team's projects.

Well, the SunSwift card arrived literally days after the 12pack on the single
function card went into production use, about three weeks early! I generally
don't complain when things arrive ahead of schedule, but...

I scheduled some downtime for my server and attempted the simple swap of the two
cards. Naturally, when I tried to boot the server with the SunSwift installed it
failed. I am only using the fast-wide-scsi portion of the card at this point, as
I can work on moving from the SPARCstation 20's internal le0 device to the
SunSwift's hme0 later without rushing. But, I must have a fast-wide-scsi
controller in my server, and that is supposed to be the SunSwift card.

Before providing the error message, I should note that I am running Solstice
Disksuite 4.0 with the latest version of its jumbo patch on this machine. The
12pack is configured as a single stripe, with each disk partitioned into a
single whole disk partition. A copy of the metadb is located on all 12 disks.

On booting the machine, be it a normal boot, a single user boot, a boot -r, I
get the following error:

Boot device:/iommu/sbus/espdma@f,400000/esp@f,800000/sd@3,0 File and args:
SunOS Release 5.5 Version Generic_103093-10 [UNIX(R) System V Release 4.0]
Copyright (c) 1983-1995 Sun Microsystems, Inc.
configuring network interfaces: le0.
Hostname: drizzle
WARNING: md: state database is stale
mount: /dev/md/dsk/d8 is not this fstype
/etc/rcS.d/S35SUNWmd.init: rm: not found
/sbin/swapadd: expr: not found
/sbin/swapadd: /usr/sbin/swap: not found

WARNING - /usr/sbin/fsck not found. Most likely the
mount of /usr failed or the /usr filesystem is badly
damaged. The system is being halted. Either reinstall
the system or boot with the -b option in an attempt
to recover.

syncing file systems... done
Program terminated
Type help for more information
ok

A probe-scsi shows all 12 of the disks on the controller. One things I find odd
is that the problem reported is a metadb error. While I do have 12 metadb
instances on the SunSwift controller in the 12pack, there are another 3 devices
with metadb instances serviced by other fast-narrow (scsi-2) controllers on this
system. And all the devices necessary to run the /usr filesystem are serviced by
these two controllers, so the /usr filesystem actually has no involvement with
the 12pack and the fast-wide controller other than the fact that I have 12
metadb replicas on it?

I have tried the SunSwift card in the same sbus slot as the single function
controller as well as the "other" top sbus slot in the machine. Both of the
lower sbus slots are already in use, one by a turbo GX+ graphics card, the other
by the second fast-narrow-scsi controller for this system. It still has enough
devices connected that I can not make due with just the internal scsi controller
and a fast-wide-scsi controller. One last sbus slot tidbit... I discovered when
swaping the single function card back in after my last failed test of the
SunSwift card that the single function card fails in the "other" top sbus slot.
Only when returned to the slot I initially had installed it in does IT work.

Finally, since the SunSwift card was purchased (well, since the paperwork to
purchace it was started), the purchase of a new Ultra-2 to replace my current
server has begun. Thus, since the Ultra-2 has an internal fast-ethernet device
of its own, I have no use for the SunSwift card's fast-ethernet capability. So,
it might make sense to just keep the single function fast-wide-scsi card. I
hesitate to do this, though, as I have seen people post to this list reporting
problems with the single function card while the SunSwift card is reported as
working terrific in Ultra-2 machines?

1. Does anyone have any explination as to why the SunSwift card is not working
when I swap it into my system as instructed by Sun?

2. Do I really need the SunSwift card at all? Can I just stick with the single
function card that is working right now in my SPARCstation 20 and be sure it
will work in my new Ultra-2 server when that hardware shows up?

Thank you very much for taking the time to read this message. I know it is very
long, but this has been a very time consuming and frustrating problem.
Particularly since it impacts the only server machine that my entire team relies
on. If the server is down, the team is down.

TIA,
Marc

Marc S. Gibian
Telos Comsys phone: (617) 377-6350
PRISM/TFS email: gibian@stars1.hanscom.af.mil
                           or is it: gibian@hanscom.af.mil
                        well, maybe: gibianm@hanscom.af.mil
              and if all else fails: marc.gibian@acm.org



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:11:59 CDT