SUMMARY: SS20, strange crashes. SOLVED?!

From: Grant Schoep (grant@storm.com)
Date: Tue Dec 08 1998 - 12:57:34 CST


Ok, this is more of a Summary part2. Way back a few months ago. I had a
SS20 that was being used as a file server, and it kept on doing bad bad
crashes, or, it would just stop talking to the network, and tring to shut
down and bring back up the le0 wouldn't work. I could always log into the
machine, but couldn't anything in the network, and no machine could see it.
I had alot of help from this group, many saying that they had heard of
SS20's doing this sorta thing, but no one ever knew really what cause it. I
think I finally figured it out.
        The SS20, being used as a file server, has a great deal of network
traffic. It only has a 10meg ethernet card in it(the built in one). Well,
it was on a 10/100 switch stack, with a great deal of 100 meg machines
accessing it. I set all the 100 meg machines back down to 10 megs(via the
switch stack) and presto, the SS20 didn't crash for at all. The day I
turned the 100meg ability for the other machiens back on, it crashed again.
        What believe is happening, is that the ethernet port on the SS20, some
home gets overloaded and just locks up. And seemingly nothing but a
powerdown will unfreeze it. I think the problem took a while to manifest it
self originally, becuase we didn't have that many 100 meg machines on the
switch, once we added a large number of machines, the SS20's port just
couldn't handle that many people talking to it. The Utilization of teh SS20
port with all capable machines running at 100megs was as around 80-90
percent average. With all these 100 meg machines set to 10megs, it was only
about 50percent average.
        So I think I figure out the problem. Cross my fingers, I figure I will
just send this message out, and 10 seconds later the machine will crash
again. Then back to the drawing board.
                        -grant

---------------------------------------------------------------------------
Grant Schoep, grant@storm.com
System/Network Administrator
L3 Communications Telemetry & Instrumentation
San Jose,CA (408)271-0800, Ext. 135



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:12:53 CDT