Messages in this thread | | | Subject | Re: 'Socket destroy delayed'/raw socket | Date | Sat, 30 Mar 1996 18:29:55 +0100 | From | Axel Kohlmeyer <> |
| |
> > > I've noticed strange behavior out of raw sockets for a long time -- i.e. they > > never go away -- but it's never bothered me so... > > Tell me more. I need to know these sort of things... > Alan > ok, here are my observations regarding this "socket destroy delayed" issue:
I usally get some of those during the shutdown procedure. I didn't pay a lot attention to it until i realized that they were related to the xntp daemon. after a few hours uptime some ntp packets get stuck in the send queue. Example:
[72|18:09] akohlmey@rincewind:/tmp> uname -a Linux rincewind 1.3.81 #1 Sat Mar 30 13:32:18 GMT 1996 i486 [73|18:09] akohlmey@rincewind:/tmp> uptime 6:09pm up 2:13, 3 users, load average: 1.55, 1.38, 1.63 [74|18:09] akohlmey@rincewind:/tmp> netstat -tu Active Internet connections Proto Recv-Q Send-Q Local Address Foreign Address (State) User udp 0 0 localhost:734 localhost:amd ESTABLISHED udp 0 0 localhost:ntp *:* udp 0 504 rincewind.chemie.u:ntp *:* ^^^ after killing the xntpd the socket destroy delayed messages start to flood my syslog. Socket destroy delayed (r=0 w=504) Socket destroy delayed (r=0 w=504) ... and so on. ^^^
6 messages per minute. and no way to stop this except rebooting.
apart from the usual boot messages was only
double lock on device queue! eth0: Tx timed out, cable problem? TSR=0xc0, ISR=0x0, t=38. double lock on device queue!
in the dmesg output.
FWIW: i'm using kerneld. but only for cdrom, msdosfs, isofs, sound, tape and floppy.
Alan, if you want me to try something out, please let me know.
Cheers, Axel.
========================================================================= Axel Kohlmeyer email: axel.kohlmeyer@chemie.uni-ulm.de Abteilung fuer Theoretische Chemie Universitaet Ulm D-89069 Ulm/Donau =========================================================================
| |