Subject: Re: RAIDFrame problem
To: NetBSD/alpha Discussion List <port-alpha@NetBSD.ORG>
From: Matt Dainty <matt@bodgit-n-scarper.com>
List: port-alpha
Date: 07/05/2003 19:39:52
--4Ckj6UjgE2iN1+kY
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

* Greg A. Woods <woods@weird.com> [2003-07-04 22:03:59]:
> [ On Friday, July 4, 2003 at 21:32:19 (+0100), Matt Dainty wrote: ]
> > Subject: RAIDFrame problem
> >
> > I recently converted my PC164LX loaded with NetBSD 1.6, to running on a
> > RAIDFrame RAID 1 setup using a pair of Seagate ST39102LC disks on an
> > Adaptec 2940UW.
>=20
> The 2940UW has an aic7880 chip, right?

Yup. I have got a 2940U2W too, but it appears SRM can't boot from that,
which is a PITA as I believe the disks are Ultra2 spec. Are there any
Ultra2 cards known to work in a PC164LX, preferably 64-bit PCI? (Am I
okay to use the 64-bit PCI slots?). I've seen some LSI cards based on
either 53C895 or 53C896, but I don't know if the SRM will boot from
them. My SRM is updated to 5.8-1 IIRC.

> > sd1(ahc0:0:5:0): SCB 6a - timed out while idle, SEQADDR =3D=3D 0xa
> > SCSIRATE =3D=3D 0x0
> > sd1(ahc0:0:5:0): Queuing a BDR SCB
> > sd1(ahc0:0:5:0): Bus Device Reset Message Sent
> > sd1(ahc0:0:5:0): no longer in timeout, status =3D 0
> > sd1: async, 8-bit transfers, tagged queueing
> > ahc0: Bus Device Reset on A:5. 10 SCBs aborted
>=20
> I was seeing spontaneous bus resets on my 2944UW (also an aic7880 chip,
> but with an HV-differential bus, not SE) when a couple of the Kingston
> DE100 hot-swap bays (with their hot-swap/repeater isolator boards) I
> have on that bus started to go bad.  Sometimes there would be a parity
> error, but IIRC sometimes it was just an instant bus reset.  (my host
> system is intel, but that shouldn't matter)
>=20
> I suspect the new AHC driver will deal with such problems a little more
> gracefully, but of course if youv'e got SCSI bus problems, of any kind
> (poor termination, noise, bad connectors, bad cables, poor power, etc.)
> or failing disks, then there's not much the driver can do but report the
> errors....

Just to rule as much out I've now replaced the cable with the shortest
one I have, problem with the disks being LVD is that there's no TERM
jumper on the disks, so I've got to use a terminated cable, and the only
ones I have are LVD ones about 1.5m long with about 5 connectors. A bit
of swapping between machines has now got me a shorter one about 75cm.
Obviously if I can get a proper Ultra2 card, that should give me a
little more flexibility with cabling.

Cheers

Matt
--=20
"The way to a man's belief is through confusion and absurdity."

--4Ckj6UjgE2iN1+kY
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (NetBSD)

iD8DBQE/Bxt4KP58eR+X2TMRAoMZAJ9sTChIfRx9xLWC8q4gYIDOYC7eXACghYNf
5KjeqbBJ+9VdZlw7rNuOeb0=
=A/n1
-----END PGP SIGNATURE-----

--4Ckj6UjgE2iN1+kY--