Message-ID: <5B4D0C5BA65ECA46969C1419122317E610088D@mchh161e> From: Kross Joachim ICM N PG U ID A 1 <Joachim.Kross@icn.siemens.de> Date: Thu, 4 Apr 2002 16:35:28 +0200 Subject: [suse-sparc] happy meal unhappy
Hi all,
when looking through the archives, I found that other people had trouble
with ethernet cards before, also with the happy meal interface. However, I
did not find an answer that would help me with my problem, which is the
following:
I recently installed SuSE Linux 7.3 with the default kernel (uname -a gives:
Linux ouessant 2.4.14 #1 Mon Nov 12 11:25:00 GMT 2001 sparc64 unknown
) on a Sun Ultra 1. At first, I was quite happy, the machine was doing its
job as a nameserver for a small development network (not much traffic going
to or from the machine, just a few telnet sessions, ntp, DNS queries, and
the like). However, after a few days, it stopped responding to connection
requests from the network, and existing connections froze, and eventually
died, e.g. the windows from X clients transmitted to another machine
disappeared. UDP didn't work any longer, either. However, interestingly, the
machine seemed to still respond to ARP queries, at least other machines had
entries for the machine in their ARP tables, which expired and were renewed,
e.g. when trying to ping the machine after it had stopped responding for
hours.
While the console was still accessible and everything else on the machine
(i.e. things that did not depend on the network) was working fine, only a
reboot of the machine would alleviate the situation, a restart of the
networking would not do.
In /var/log/messages, the following messages could be seen reappearing about
once a minute:
Apr 4 15:38:18 ouessant kernel: NETDEV WATCHDOG: eth0: transmit timed out
Apr 4 15:38:18 ouessant kernel: eth0: transmit timed out, resetting
Apr 4 15:38:18 ouessant kernel: eth0: Happy Status 03030000
TX[000003ff:00000301]
Apr 4 15:38:22 ouessant kernel: eth0: Link is up using internal transceiver
at 100Mb/s, Full Duplex.
The built-in ethernet card was detected automatically during installation,
here is what the driver says when booting the machine:
Apr 4 15:43:08 ouessant kernel: sunhme.c:v1.99 12/Sep/99 David S. Miller
(davem@redhat.com)
Apr 4 15:43:08 ouessant kernel: eth0: HAPPY MEAL (SBUS) 10/100baseT
Ethernet 08:00:20:89:0f:f5
The problem appears to depend to a certain degree on the network traffic
going to and from the machine. I.e. while it was first sitting rather idle
for a few days before stopping the first time, it seems that large amounts
of data transferred (e.g. downloads of large packages) seem to shorten the
time before the machine stops.
Unfortunately, I do not have another machine to see whether it might be a
hardware problem. However, the machine was running Solaris 8 fine before I
installed Linux, and the Boot-PROM tests (e.g. test net-all) do not report
any errors.
I am afraid I have lost a lot of time already trying to figure out by myself
what might be wrong, however I am no kernel hacker or even too good a
programmer to even try delving into the kernel sources, so I have come to my
wits end on this one. If I don't find a solution real soon, I fear I'll have
to put Solaris back onto the machine, as the DNS server, while not very
busy, is rather critical to our current project. As I would strongly prefer
to keep Linux, which I am using on Intel machines without trouble, I would
really appreciate any help and advice you could give. If you need any other
information on the machine for that, please let me know.
Thanks,
Joachim
-- To unsubscribe, e-mail: suse-sparc-unsubscribe@suse.com For additional commands, e-mail: suse-sparc-help@suse.com
This archive was generated by hypermail 2.1.0 : Thu Apr 04 2002 - 06:34:39 PST