SUMMARY: Problems with Qlogic isp-adapter, or?

From: Per Boussard T/ED <Per.Boussard_at_era-t.ericsson.se>
Date: Wed, 01 Jul 1998 09:39:39 +0200 (MET DST)

It seems the cable length could be the problem. The card has one external
SCSI interface and two internal with different style cables. One of the
internal cables was a really long (DEC-provided, terminated) cable. When I
removed that one, it works like charm. I'm still a bit puzzled by this. I
would have thought that three short cables would be a smaller problem than
one long cable (sum-of-lengths being equal). None of the cables involved
had a length in excess of 1m. I also noticed that it's real easy to jerk
the card out of its slot without really intending to, which makes it work
not at all :-).

I will know better than to trust my intuition on cable lengths now.

Thanks to --
alan_at_nabeth.cxo.dec.com

(Original question below)

//Per

----
Per Boussard, KI/ERA/T/ED          Office: +46 8 404 55 11
UNIX System Administrator          Fax: +46 8 757 55 50
Ericsson Radio Systems AB          Home: +46 8 570 349 67
S-164 80 STOCKHOLM, SWEDEN         Email: Per.Boussard_at_era-t.ericsson.se
On Wed, 24 Jun 1998, Per Boussard, ERA/T/ED wrote:
> I have a problem with a Digital Workstation 500au. It goes like this -- 
> 
> cam_logger: bus 2 target 2 lun 0 
> ss_perform_timeout
> timeout on disconnected request
> cam_logger: CAM_ERROR packet
> cam_logger: bus 2 target 2 lun 0 
> ss_perform_timeout
> timeout on disconnected request
> cam_logger: CAM_ERROR packet
> cam_logger: bus 2 
> isp_mailboxcomplete
> Timeout on mailbox command completion - scheduling chip reinit
> cam_logger: CAM_ERROR packet
> cam_logger: bus 2 
> isp_reinit
> Beginning Adapter/Chip reinitialization
> cam_logger: CAM_ERROR packet
> 
> etc.. I found a bug-report which may be related; it's (Patch ID: 
> OSF370-043). This one says that the system may fail to boot like 1 out of
> 20 tries. My system fails much more often than that for some disks on the
> SCSI chain (ironically, the only time I've seen it work was when the
> service-guy from digital was here). The problem is that I can't install
> the operating-system from CDROM because of the above errors. According the
> the description of the PatchID above, the patch should be installed for
> any system using a Qlogic chip. My boot-message shows that I have ... 
> 
> MESSAGE                                 Alpha boot: available memory from 
> == cut ==                                         _0x1160000 to 0x7ffe000 
>                                         isp0 at pci1000 slot 10 
>                                         isp0: QLOGIC ISP1040B/V2 
>                                         isp0: Firmware revision 5.27
> (loaded 
>                                          _by console) 
>                                         isp0: Fast RAM timing enabled. 
>                                         scsi2 at isp0 slot 0 
> 
> etc. I wonder if the Fast RAM timing is bad for me. How can I change that?
> saying isp_edit to the console-software is amusing --
> >>>help isp_edit
> NAME    
>          isp_edit
> FUNCTION
>           Edit ISP1020/ISP1040 NVRAM parameters.
> SYNOPSIS
>          isp_edit [-sd] [-offset] [-id] [-allids] [-value] 
>           [-bit] [-set] -[clear] [<isp_controller_name>] 
> 
> >>>isp_edit
> pkc0.7.0.1010.0
> old NVRAM
> 00000000   49 53 50 20 06 7b 01 00 01 f9 08 ab fa 00 00 01
> 00000010   1d 00 00 00 00 00 01 00 01 00 00 00 fd 10 0c 18
> 00000020   00 00 fd 10 0c 18 00 00 fd 10 0c 18 00 00 fd 10
> 00000030   0c 18 00 00 fd 10 0c 18 00 00 fd 10 0c 18 00 00
> 00000040   fd 10 0c 18 00 00 fd 10 0c 18 00 00 fd 10 0c 18
> 00000050   00 00 fd 10 0c 18 00 00 fd 10 0c 18 00 00 fd 10
> 00000060   0c 18 00 00 fd 10 0c 18 00 00 fd 10 0c 18 00 00
> 00000070   fd 10 0c 18 00 00 fd 10 0c 18 00 00 00 00 09 92
> 
> Now, which byte should I mess with?
> 
> Seriously. I'm in a pinch here. Except for getting a new host-adapter I
> can't say I have many ideas. I'm also sure somebody out there must have
> seen and solved the same problem.
> 
> Oh, when the system fails to boot, it would stop just after showing the
> size of the text-segment of the kernel, you know the line saying 
> text: some-interesting-number
> with the message "failed to send Read to dkc300.x.y.z.whatever.". I can
> boot from the cdrom since it's not on that scsi-chain.
> 
> In short -- help!
> 
> Thanks for your time
> //Per
> -----
> Per Boussard, KI/ERA/T/ED          Office: +46 8 404 55 11
> UNIX System Administrator          Fax: +46 8 757 55 50
> Ericsson Radio Systems AB          Home: +46 8 570 349 67
> S-164 80 STOCKHOLM, SWEDEN         Email: Per.Boussard_at_era-t.ericsson.se
> 
> 

Received on Wed Jul 01 1998 - 09:40:42 NZST