Re: Networking Status pre #2

Austin Donnelly (and1000@cam.ac.uk)
Wed, 15 May 96 01:16 BST


In article <m0uIjWp-0005FcC@lightning.swansea.linux.org.uk> you write:
>
>Can people mail the list (not just me directly) if they are seeing the
>following with 2.0 pre 2
>
[...]
>2. repeated socket destroy messages every 10 seconds

I've had this:

May 13 18:12:05 valour kernel: Socket destroy delayed (r=0 w=204)
May 14 18:49:59 valour kernel: Socket destroy delayed (r=236 w=0)
May 14 18:50:38 valour last message repeated 4 times
May 14 18:51:48 valour last message repeated 7 times
May 14 18:52:48 valour last message repeated 6 times
May 14 18:53:48 valour last message repeated 6 times
May 14 18:54:48 valour last message repeated 6 times
May 14 18:55:48 valour last message repeated 6 times
May 14 18:56:48 valour last message repeated 6 times
.....
<until reboot>

Note the (rough) 24H delay between the first isolated message and the
"1 every 10 seconds" problem.

The machine was under little load at the time - normal daemons running
(bind, apache, ftpd, etc), with a couple of interactive users (none of
which doing heavy network activity).

This is using 1.99.1.

I was running 1.3.98 for approx 1 week, and didn't have this problem.

Hardware is ISA 3c509, amd486DX66, stone-age PCI motherboard (with
interesting ideas about routeing ISA IRQs to PCI INTs).

>3. Stuck sockets (notably on big FTP or WWW servers)

I had 504 (bytes?) in the receive queue for port 1028 at the time the
"socket destroy delayed" problem happened. This had been the case for
quite a while before, and I was previously unworried about it. Here's
some sample output from netstat:

valour$ netstat -a
Active Internet connections (including servers)
Proto Recv-Q Send-Q Local Address Foreign Address (State) User
[...]
udp 0 0 *:talk *:*
udp 504 0 *:1028 *:*
udp 0 0 *:who *:*
raw 0 0 *:1 *:*
[...]

I haven't seen any of the other problems you describe.

I know this isn't much to go on, but I hope it helps,
Austin