Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 17 Apr 1999 20:34:45 +1000 (EST)
From:      Andrew MacIntyre <andymac@bullseye.apana.org.au>
To:        jonathan michaels <jon@caamora.com.au>
Cc:        freebsd-questions@FreeBSD.ORG
Subject:   Re: tcp/ip failing on freebsd network
Message-ID:  <Pine.OS2.3.95.990417195754.126A-100000@CENTRAL>
In-Reply-To: <19990417130904.A3459@caamora.com.au>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, 17 Apr 1999, jonathan michaels wrote:

> my small network consists off 4 intel computers, 2xi486dx33, a 
> p5-133 and a p6-180.

{...}

> all worked very well till about two weeks ago with the p6 
> started to drop teh netork conection while reading mail with 
> mutt over a ssh -C -v hostname -l username connection tot eh 
> mail host. i chacked teh logs, at both ends. after i killed off 
> teh session, by loging in as root n another terminal session 
> and issuing a kill on teh relevant processes.
> 
> this wnet on for a week on and off. finally i started to reboot 
> using teh -verbose flag, all i could find that was different 
> was that teh nic was now trying to use irq 9 as its irq. from 
> memory i originally told teh computers bios to use irq 10, i 
> think.

{...}

Are you able to associate any external events, particularly power
disturbances, with the time you started to see the problems?

Several things occur to me:

1.  the Flash BIOS chip is playing silly buggers..  IIRC most PCI systems
record PCI plug'n'play info in a part of the Flash BIOS chip.  If the
system is correctly configured and behaving properly, this info only gets
updated when there is actually some device change detected.  If this gets
zapped, behaviour becomes undefined.  Not sure how to force a rebuild of
this info (assuming the flash chip itself is not faulty) short of
removing a card, rebooting, shutting down, reinstalling card and
rebooting again.

2.  your CMOS configuration has been subtly changed or reset.  The PCI
BIOes I've played with (Award mainly, a couple of AMIs) do allow you some
control over whether IRQs are reserved for Legacy(ISA) or PCI/PnP devices.
As far as possible, only set IRQs to the Legacy setting when you know they
are actually being used.  You could try setting the config back to
defaults and then work from there.  I have encountered one motherboard
that allocated the USB port an IRQ even when USB was completely disabled,
so be warned that some odd things happen to make life difficult.

3.  its not a direct observation on your circumstances, but I have seen
a similar OK-dying-dead sequence in PCI hardware.  It involved an Adaptec
2940AU SCSI controller that was purchased to drive a CD-R drive at work.
It was installed and ran fine for several weeks.  Then, increasingly
frequently, it would report an initialisation error on booting (Win95 in
this case...) leaving the CD-R unusable.  Over the space of about 4 weeks,
it went from "no problems" to "hard failure".  The supplier claimed that
the card was perfectly OK, but very grudgingly eventually agreed to swap
the card.  In the mean time, as we needed the CD-R, I dropped an Adaptec
1542 ISA controller into the system which has worked perfectly since.
I've not had leisure to try installing the swapped 2940AU.

--
Andrew I MacIntyre                     "These thoughts are mine alone..."
E-mail: andrew.macintyre@aba.gov.au    (work) | Snail: PO Box 370
        andymac@bullseye.apana.org.au  (play) |        Belconnen  ACT  2616
Fido:   Andrew MacIntyre, 3:620/243.18        |        Australia



To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.OS2.3.95.990417195754.126A-100000>