Networker

Re: [Networker] SCSI errors (What cards supported for Linux?)

2003-07-22 19:24:57
Subject: Re: [Networker] SCSI errors (What cards supported for Linux?)
From: Matts Nilsson <matts_nilsson AT YAHOO DOT COM>
To: NETWORKER AT LISTMAIL.TEMPLE DOT EDU
Date: Wed, 23 Jul 2003 01:24:50 +0200
 Hi!
I just remembered I case where TSM took a couple of
drives off line every now and then. The reason seems
to have been too long SCSI cables.

The config was a HP/Compaq DL380 with HP/Compaq HVD
SCSI cards and an HP/Compaq ESL9326. The two drives
were sitting on the same SCSI bus as the robotics,
connected to the server via a 12ft cable. Re-wiring
the drives directly to the server, leaving the
robotics on its own bus, shortened the cable length by
~10ft.

With this solution the total cable length would be
~18ft. I think the standard says something like 10
meters (~30ft) for differential SCSI, but keep in mind
that all circuitry on the drives, SCSI cards etc eats
off that distance. A SCSI card is probably the
equivalence of 1-2ft and so on.

HTH!

// Matts

 --- Matthew Temple <mht AT RESEARCH.DFCI.HARVARD DOT EDU>
skrev: > Sharon, et. al.,
>
>         What SCSI card do you have?  We have a
> surely underpowered
> Adaptec AHA2940U2.   But the throughput to the tape
> drives has never been
> the issue, so it's hard for me to imagine it's the
> cards.   How short
> does the scsi cable have to be?
>
>                                         mht
>
>
> On Tue, 22 Jul 2003, O'Brien, Sharon L wrote:
>
> >  we get similar problems on our windows 2000/san
> connected STK L180
> > using DDS. Not all the time, but most of the time.
> We also get a lot of
> > problems with tape unloads causing active tape
> drives to get timeout
> > errors. We never did get any resolution.
> >
> > -----Original Message-----
> > From: Matthew Temple
> [mailto:mht AT research.dfci.harvard DOT edu]
> > Sent: Tuesday, July 22, 2003 9:40 AM
> > To: NETWORKER AT LISTMAIL.TEMPLE DOT EDU
> > Subject: Re: [Networker] SCSI errors (What cards
> supported for Linux?)
> >
> > With reference to the conversation below, we're
> still getting
> > strange SCSI related errors on our machine.
> Strangely, the
> > problems only seem to occur when an tape is being
> labeled
> > by auto-media-management.   I can label normally
> when  no
> > backup is in progress.   If a backup is in
> progress, the tape
> > will (can) go into a permanent verify state, or
> stranger,
> > seems to be trying to eject nad load at the same
> time.   The robot
> > arm of the Qualstar library "clunks" for a very
> long time.   (I can
> > eject the tape through the Qualstar Console -- it
> ejects, gets put back
> > into its slot, then Networker remounts the tape
> and can't eject
> > it!   Is there a list of supported SCSI cards for
> Linux anywhere?
> >
> >
> > On Thu, 5 Jun 2003, Frank Altpeter wrote:
> >
> > >
> > > Hello!
> > >
> > > Matthew Temple wrote on 2003-06-04 16:01:03
> -0400:
> > > >
> > > > Frank,
> > > >
> > > >     I missed the beginning of this thread.
> Did it ever work?
> > > Oh, yes, it works, but not all the time...  It
> happens sometimes that
> > > the mentioned errors occur, but not permanently.
> Last night for
> > > example the backup processes were running fine
> without any error.
> > >
> > >
> > > >     I've had similar problems on two extended
> occasions with a
> > > > Qualstar library and three sony AIT-2 drives.
>  The card is an
> > adaptec
> > > > (I'm not near the machine) 2940U2.   The first
> time occurred after a
> > move.
> > > > The tape in the library, would go into a hung
> state when trying to
> > > > verify label.
> > > ACK, same was here.
> > >
> > > >   first it was all three drives.   Then I
> messed around with
> > > > termination and changed the scsi cable and the
> problem reverted to
> > > > just one drive. -- SCSI resets and timeouts.
> Then I switched
> > > > SCSI cards (same card, just a different one.)
>  I took a DDS tape
> > drive
> > > > off the internal channel.   Same result.
> Then I reattached the
> > > > internal tape drive.   Everything worked.   I
> don't know why.
> > > > It was all so ... SCSI.   SCSI cables have
> always been mysterious to
> > me --
> > > > we've had RAID stacks that worked when their
> cable was curved to the
> >
> > > > left but now when curved to the right.
> > >
> > > You don't want to know how often i changed the
> cabeling on my system
> > > the last weeks :-))
> > >
> > > >     1. Keep the cables as short as possible.
> > > >     2. Is everything REALLY terminated
> correctly (how about
> > > >             the jumpers on the devices?
> TERM-POWER?)
> > > >     3. Try different cable positions.
> > > >     4. The following is MOSTLY true -- It's
> probably NOT your card.
> > > >                     (except when it is 8-)   )
> > > Yep, and i think that i used to work according
> to these rules... we
> > > checked cables and termination more than a dozen
> times these days, but
> >
> > > it's hard to check for me when it comes to
> hardware - i'm more the
> > > software guy ...
> > >
> > > With kind regards,
> > >
> > >       Frank Altpeter
> > >
> >
> > --
> >
>
=============================================================
> > Matthew Temple                Tel:    617/632-2597
> > Director, Research Computing  Fax:    617/582-7820
> > Dana-Farber Cancer Institute
> mht AT research.dfci.harvard DOT edu
> > 44 Binney Street,  ML105
> http://research.dfci.harvard.edu
> > Boston, MA 02115              Choice is the
> Choice!
> >
> > --
> > Note: To sign off this list, send a "signoff
> networker" command via
> > email to listserv AT listmail.temple DOT edu or visit the
> list's Web site at
> > http://listmail.temple.edu/archives/networker.html
> where you can also
> > view and post messages to the list.
> >
>
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
> >
> > --
> > Note: To sign off this list, send a "signoff
> networker" command via email
> > to listserv AT listmail.temple DOT edu or visit the
> list's Web site at
> > http://listmail.temple.edu/archives/networker.html
> where you can
> > also view and post messages to the list.
> >
>
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
> >
>
> --
>
=============================================================
> Matthew Temple                Tel:    617/632-2597
> Director, Research Computing  Fax:    617/582-7820
> Dana-Farber Cancer Institute
> mht AT research.dfci.harvard DOT edu
> 44 Binney Street,  ML105
> http://research.dfci.harvard.edu
> Boston, MA 02115              Choice is the Choice!
>
> --
> Note: To sign off this list, send a "signoff
> networker" command via email
> to listserv AT listmail.temple DOT edu or visit the list's
> Web site at
> http://listmail.temple.edu/archives/networker.html
> where you can
> also view and post messages to the list.
>
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=

_____________________________________________________
Gå före i kön och få din sajt värderad på nolltid med Yahoo! Express
Se mer på: http://se.docs.yahoo.com/info/express/help/index.html

--
Note: To sign off this list, send a "signoff networker" command via email
to listserv AT listmail.temple DOT edu or visit the list's Web site at
http://listmail.temple.edu/archives/networker.html where you can
also view and post messages to the list.
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=