Re: [SQU] comm_select: 0 FDs ready; - syslog explodes!

Henrik,

Thanks you for replying. Here's the details that I posted to the LRP
mailing list.
Also, I eliminated the time-hack.
As you can see, I'm fairly flustered at this point...

LRP 2.2.16-1 is stable, right? RIGHT!
squid 2.3.STABLE4 is stable, right? RIGHT!

ldd squid shows:

 /lib/libNoVersion.so.1 => /lib/libNoVersion.so.1 (0x40015000)
 libcrypt.so.1 => /lib/libcrypt.so.1 (0x4001b000)
 libpthread.so.0 => /lib/libpthread.so.0 (0x40048000)
 libm.so.6 => /lib/libm.so.6 (0x4005b000)
 libresolv.so.2 => /lib/libresolv.so.2 (0x40079000)
 libnsl.so.1 => /lib/libnsl.so.1 (0x40088000)
 libc.so.6 => /lib/libc.so.6 (0x4009e000)
 /lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x40000000)

Why does squid freeze on me?? Then ssh2 stops working.
Everything else works fine throughout - dhclient, dhcpd,
thttpd,dnscache, tinydns, smb-2. I'm perplexed. The LRP logs and the
squid logs reveal
nothing to me. The only way to get squid to start again is to reboot LRP -
not
acceptable!

Squid only freezes once or twice a day, otherwise squid works great -
ugh! Using "--enable-async-io" on squid, and "no-cache deny all" in
squid.conf. Not using IDE because I'm caching no data.

If this becomes stable, then I plan on using a redirector to filter and
then do some caching.

Help! I don't know how to shoot this bug. I thought it was a squid
bug, but then ssh2 was affected too. (Problem began before I used ssh2)

By the way, squid cache.log in debug mode reveals comm_poll: 0 FDs
ready, but the squid guys say this is ok.

Any ideas at all would be welcomed...or stable idiot images...THANKS.

I've tried init 1, then going back to run level 2 - but squid still
hangs when started, and ssh2 still does not work. Perhaps if I could find
a way to "reset
LRP" without doing a reboot, I could begin to find the problem. As you can
see, I'm
grasping at straws at this point.

P.S. When squid hangs and I stop it, I still see periodic IP traffic.
IPCHAINS -log reveals that IO is still being attempted from the sites I was
connected
to before the hang. I also see a little ssh IO even though ssh was killed.

Please copy dbaldwin@optonline.net in your reply, thank you.

Henrik Nordstrom wrote:

> Doug wrote:
> >
> > All's well, until -k debug. Then, the syslog explodes with
> > megabytes/second of the following:
> >
> > comm_select: 0 FDs ready at ....
> > comm_select: time out: ....
>
> Quite normal. "squid -k debug" enables full debugging output from Squid,
> and will easily choke syslog it sent there.
>
> > ...and, without the time-hack squid hangs after about
> > 12 hours and can not be restarted.
>
> If a process hangs and cannot get killed then you have a kernel problem.
>
> Squid is reported to have a couple of issues with the --enable-time-hack
> option. One is 100% CPU usage under certain conditions, other is loss of
> proper time measurements making response time analysis go completely
> wild. Probably other issues as well.
>
> --
> Henrik Nordstrom
> Squid hacker
>
> --
> To unsubscribe, see http://www.squid-cache.org/mailing-lists.html

--
To unsubscribe, see http://www.squid-cache.org/mailing-lists.html