I've been testing 2.4 on one of my webservers and it seems to exhibit an
extreme case of slowdowns every 2-3 days. It slows down to the point where
I can't really type anything into the telnet screen (remotely admin'd).
However, when I was able to get a few commands to the system (w, memstat,
ps, free), it showed absolutely nothing out of the ordinary. The system
runs like a breeze and then just _slows_ to a complete crawl. Needless to
say, all the services also slowed down. It's the second time it happened
since I started testing this new kernel.
A little background on the server: It's a webserver that runs Apache and
MySQL. It has ~896MB RAM, and runs with an Adaptec 2940U2W SCSI card and
several SCSI drives. The system experiences a load average of about 1-2
throughout the day with the new kernel. It used to experience load averages
of 8-14 with 2.2.17. I don't make very much use of shared memory, and from
what I can tell, the system never swaps to swap space. Below are the last
stats I was able to retrieve before power-cycling the system (nothing else I
could do to fix it. It was unresponsive to a shutdown -r now).
uptime:
4:53pm up 2 days, 15 min, 4 users, load average: 2.57, 1.87, 2.26
free:
total used free shared buffers cached
Mem: 899712 896924 2788 0 24476 331660
-/+ buffers/cache: 540788 358924
Swap: 1052248 0 1052248
Peering through the aftermath, syslog contains this:
Jan 25 02:59:54 omega kernel: reset_xmit_timer sk=f00c9680 1 when=0x3744,
caller
=c0218172
Jan 25 02:59:58 omega kernel: reset_xmit_timer sk=f00c9680 1 when=0x3087,
caller
...skipping...
Jan 25 16:53:27 omega named[182]: ns_req: sendto([141.133.112.25].53):
Resource
temporarily unavailable
[... manually shortened ... ]
Jan 25 16:53:27 omega named[182]: ns_req: sendto([65.24.0.167].46484):
Resource
temporarily unavailable
Jan 25 16:53:27 omega named[182]: ns_req: sendto([198.77.116.8].37441):
Resource
temporarily unavailable
Jan 25 16:53:27 omega named[182]: ns_req: sendto([64.59.128.213].2102):
Resource
temporarily unavailable
^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@
^@^@
^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@
^@^@
^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@
^@^@
^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@
^@^@
^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@
^@^@
^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@
^@^@
[... manually shortened ... ]
^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@
^@^@
^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@
^@^@
^@^@^@^@^@^@^@^@Jan 25 18:01:02 omega syslogd 1.3-3#33.1: restart.
Jan 25 18:01:02 omega kernel: klogd 1.3-3#33.1, log source = /proc/kmsg
started.
Jan 25 18:01:02 omega kernel: Inspecting /boot/System.map-2.4.0
Jan 25 18:01:03 omega kernel: Loaded 15215 symbols from
/boot/System.map-2.4.0.
Jan 25 18:01:03 omega kernel: Symbols match kernel version 2.4.0.
Jan 25 18:01:03 omega kernel: No module symbols loaded.
Jan 25 18:01:03 omega kernel: Linux version 2.4.0 (root@omega) (gcc version
2.95
.2 20000220 (Debian GNU/Linux)) #1 SMP Fri Jan 19 23:03:21 PST 2001
Does anyone have any ideas as to what is causing this? I will likely be
taking 2.4 out of testing and replacing it with good old 2.2.17 if I can't
resolve this problem.
-- Vibol Hou KhmerConnection, http://khmer.cc "Connecting Cambodian Minds, Art, and Culture"- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org Please read the FAQ at http://www.tux.org/lkml/