Network problems

From: Seth Rothenberg (SROTHENB@emerginghealthit.com)
Date: Wed May 10 2006 - 11:21:12 EDT


Greetings,
I'm wondering if anyone has had an experience anything like this.

We are running two Sun Fireserver 4800's...on Solaris 8.
Each server has 2 trunks to a Nortel Switch (using MLT on Nortel).
These were configured for 100/Full on both the switch and the server.

Sunday night, about 1 am, we got called because the GUI for our
application wasn't working.
I could not connect to the server, but logging into another server
first, was able to log in. *
The server could not ping defaultrouter or anything else on the
network...
though apparently previously-existing TCP socket connections were
sustained.
(I just checked, logs say about 20-30 messages were received and
acknowledged
each minute throughout the problem.)

* note, the "other server" is on the same subnet, and both servers are
multi-homed,
but I believe I logged in at that time across the public network.

We ended up rebooting (Maybe I could have just shut and restarted
the
network :-)

When the server came up, we had another problem!
"anar not set with speed selection" repeating numerous times in
/var/adm/messages.

My colleague recalled that he had seen this on one of our other
servers....
He requested our network guys to switch to "AutoNegotiate" -

Whereas the switch and the server were BOTH set for 100/FULL
and it was resolved by changing the SWITCH side to autonegotiate.
The server remains at 100/Full.

Interesting thing, it is nice that Sun servers reply to autonegiotiate

requests even when they are hardcoded.

We also had a second problem, the second server in this cluster (VCS)
did NOT have network problems (though we had done the above change
on that switch previously)....but it had another problem...
ps would not work. It hung every time.
Rebooting that (after we saw that the other was recovering)
solved that problem.

Both of these issues are perplexing, and we really can't afford to be
at risk on
these servers. I'd appreciate any suggestions.

Thanks
Seth

[demime 1.01b removed an attachment of type application/octet-stream which had a name of Header]
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers



This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 23:39:50 EDT