SUMMARY:Shared SCSI-Problems

From: S.Hergeth <S.Hergeth_at_informatik.fh-muenchen.de>
Date: Wed, 16 Sep 1998 07:16:22 +0200

-- 
Stefan Hergeth, Fachhochschule Muenchen, Fachbereich
Informatik/Mathematik 
Lothstrasse 34 , D-80323 Muenchen  
Tel.:089/1265-1628 Fax:089/1265-1643 
email: S.Hergeth_at_informatik.fh-muenchen.de


Thanks to Steve Walker and Chris Ruhnke

Steve Walker wrote : (Answer to Q: 2)


> DEC have told us that to do any form of dual-scsi attachment TruCluster
> has to be installed as it
> patches the kernel drivers. Not to do so causes problems with bus-resets
> and gives disk corruption.
>
>
> Steve Walker

TCR still not installed, but I will install it this week

Chris Runke wrote :
 
> Have you readdressed the KZPSA in one of the Alphas to something other
> than "7". By default the SCSI adapters are programmed for address 7.
> At the console prompt (">>>") enter "show" and look for the
> "kzpsa<unit>_host_id = <address>" lines. They should be "7" on one
> system and (I suggest) "6" on the other.
>
> --CHRis

This was not the problem, because I've done it with the adapter cnfgdiag.exe




My original question :
=======================

Hello DU-Managers,

we want to setup a TruCluster Available Server Environment with two
servers.

server #1 : AlphaServer 2000 4/200,Firmware 5.2 (latest),DU V4.0B with
patchkit #8, KSPSA-BB Adapter (Firmware A11) SCSI-ID 6

server #2 : AlphaServer 800 5/333, Firmware 5.2 (latest), DU V4.0B with
patchkit #8, KZPSA-BB (Firmware A11) SCSI-ID 7

The resistor-packs on both KZPSA are removed. The servers are connected
to a Storage-Works Shelf BA356 via DWZZB-AA signal converter (terminator
resistor packs removed on differential-side) and Y-cables. The SCSI-Bus
is terminated on both sides with appropriate terminators.

We have following problems :

1.) After powering up the systems, sometimes the following
    messages apears on the screen

waiting for pkc.0.6.0.8.0 to poll
CAM command EXECUTE_SCSI_INFO timed out
SIMPORT set device state command timed out

Maybe this problem is solved, because after replacing a SCSI-cable
this message dissapears and both servers can see the disk in the
storagebox. Has everybody seen this before ?

2.) SCSI-Bus-Reset OK ?

Although the KZPSA-Adapters are configured via cnfgdiag.exe to not
performing a SCSI-Bus reset, both servers performing a reset on the
shared bus, when booted. Will this be a problem for a running ASE ?
(Note : TruCluster - Software is not yet installed; see Question 3)

3.) Differences between TCR1.4 and TCR1.4a

In the moment I don't have access to TCR1.4a (Associated Product
CD #2 for V4.0B contains only TCR1.4; newer CD from DEC-Campus seems
to be lost, but PAK CD available; (when it was released ?)). Searching
for problems with
TCR1.4 in this mailing list shows only a problem with rpc.rquotad.
This problem is fixed in PatchKit #8 (Patch-ID OSF410-400214).
It is possible to use TCR1.4 with patch #8, or will I have a lot
of other problems. I want to install it now and perform an upgrade
to TCR1.4a as soon I have access to the CD's

TIA
   
Stefan Hergeth
-- 
Stefan Hergeth, Fachhochschule Muenchen, Fachbereich
Informatik/Mathematik 
Lothstrasse 34 , D-80323 Muenchen  
Tel.:089/1265-1628 Fax:089/1265-1643 
email: S.Hergeth_at_informatik.fh-muenchen.de
Received on Wed Sep 16 1998 - 05:17:20 NZST

This archive was generated by hypermail 2.4.0 : Wed Nov 08 2023 - 11:53:38 NZDT