I started stress-testing my server at 00:47, and it happily pumped out ~320kB/s
until 01:03. I killed the server processes at 01:20, where the debug log
stopped.
Jan 26 00:49:31 genji kernel: droping syn ack:20 max:20
Jan 26 00:50:02 genji last message repeated 4 times
Jan 26 00:51:05 genji last message repeated 20 times
Jan 26 00:51:10 genji last message repeated 2 times
Jan 26 00:51:13 genji kernel: BUG: syn_recv:socket exists
Jan 26 00:51:28 genji kernel: droping syn ack:20 max:20
Jan 26 00:55:59 genji last message repeated 222 times
Jan 26 00:57:00 genji last message repeated 238 times
Jan 26 00:58:00 genji last message repeated 255 times
Jan 26 00:58:58 genji last message repeated 252 times
etc.....
26 01:19:00 genji last message repeated 349 times
Jan 26 01:20:00 genji last message repeated 349 times
Jan 26 01:20:57 genji last message repeated 288 times
The kernel: BUG: syn_recv: socket exists line is from net/ipv4/tcp_ipv4.c.
I cannot find where the "dropping syn ack" message is coming from, or
indeed if it is even something to be concerned about. The behaviour is
repeatable, although the "BUG: syn_recv" message does not always appear.
Any thoughts on all of this? Here is a little more info:
kernel 2.1.21 patch2.
GCC 2.7.2.1
libg++ 2.7.2.1 <--- Compiled with -D-REENTRANT
libc 5.4.17
linuxthreads 0.5
3Com 3c509 ethernet
Dell OptiPlex GXMT 5100 (P100)
BTW - None of this happened while running under my previous kernel, 2.0.0
------------
Vance Huntley
Director of Technology (& Chat Guy)
WebGenesis, Inc., Ithaca, NY 14850
vance@webgenesis.com --- 607.255.8499
Check out "The Globe"! --- http://www.theglobe.com/
------------