Inigo Tejedor Arrondo wrote:
scene: home webserver (athlon 1,2Mz via chipset, lan realtek) dmesg at
the end of email.
OpenBSD 3.9 (GENERIC) #617: Thu Mar 2 02:26:48 MST 2006
The server is conected to a 3com superstack hub, and witouth
keyboard/screen, managed with ssh.
Packet Filter is disable on this machine.
Uptimes are about 8/15 days, because network stops working. Sudenly the
httpd, sshd or ping disappear. rl0 and the switch port are as always,
with the green led on (ok)
can other devices on the network talk to each other when things "stop
working"?
If not, be suspicious of your switch. I've seen switches "crash" not
infrequently, and sometimes, simply turning one machine off and back on
will reset them...which makes troubleshooting somewhat confusing.
If that is really a hub and not a switch, I've seen a not uncommon
problem with 10/100 hubs: they lose the "bridge" between the 10 and
100Mbps sides...so the 100Mbps devices talk to each other, and the
10Mbps devices talk, but not 10-to-100. However, so few people have
10Mbps-only devices anymore, it is not so common an issue now. :)
I have try to put a kvm port, and it does not respond either when this
incidence happens. (Machine is on, fans are turning, the screen shows
the login, but keyboard or network doesn'y work.
careful on what conclusions you draw here.
You plug a PS/2 keyboard in after boot, you may lock the keyboard driver.
You plug a PS/2 mouse in after boot on, you WILL lock the keyboard
driver. If you are going to "hot-plug" the keyboards, use USB, and skip
the mouse.
So, you could have problem X (lost network) and then introduce problem Y
(locked keyboard). These may be completely unrelated.
If it is a KVM device you are plugging in, leave it in, with ONLY the
keyboard, not the mouse, attached. Some KVM switches switch the mouse
for OpenBSD just fine, many do not. Adding a KVM just adds one more
complication. Just Plain Mouse and Keyboard always attached is much
better for diagnosis.
/var/log/* and /var/www/logs stop record events, some hours or minutes
before I notice that 4it has gone away!. The only thing that I see
abnormal, is that they appear two files just at the time of logs stops
to work:
/var/backups/device.backup and /var/backups/device.current
Those files are made by the /etc/daily process... What are you
expecting to see in the logs AFTER that that you are not seeing?
The room temperature is +- 22 Celsius (garage). PSU is ok and protected.
I presume by "protected" you mean a UPS of some kind. Don't be too
sure: I've seen UPSs cause as many problems as they ever "prevented".
What can I search, read, enable/disable, test or do ?
Greetings.
# dmesg
OpenBSD 3.9 (GENERIC) #617: Thu Mar 2 02:26:48 MST 2006
[EMAIL PROTECTED]:/usr/src/sys/arch/i386/compile/GENERIC
cpu0: AMD Athlon(tm) Processor ("AuthenticAMD" 686-class, 256KB L2
cache) 1.26 GHz
As I recall, that's a hot running processor, make sure your heat sink is
properly mounted. OpenBSD runs so cool normally that you can get away
with all kinds of stuff...until you actually do some hard processing.
AMD processors of that vintage stalled when they overheat, as I recall.
IF the entire machine is really hanging, it could be any part of
hardware. If it is strictly the network, could be NIC, wire,
hub/switch, etc.
If you really think the daily process might have something to do with
the hangs, write a simple script to run date(1) then sleep for sixty
seconds. If it dies during the /etc/daily process (1:30, local time),
you might have a disk system issue. If it dies at 6:30am, it's your
coffee maker. :)
I've actually had pretty good results with rl(4) cards (this message is
coming to you through one, in fact), better than I have with 3c905
cards, but that isn't to say they don't break (I think I have yet to
have a 21143 card break on me, but I've seen bad cards of just about
every other type).
Nick.
