| View previous topic :: View next topic |
| Author |
Message |
eXess
Apprentice
Joined: 17 Aug 2003
Posts: 243
Location: Belgium
|
 Posted: Sun Jan 23, 2005 1:08 pm Post subject: Strange, recurring squid issue |
 |
|
Hi all,
I've got this strange problem with my SQUID installation. Basically, it doesn't work. Apart from configuration files issues (I'm working on that) I noticed that I always get this error :
| Code: | # /etc/init.d/squid restart
* Stopping squid...
* Waiting ............................................................. * Failed. [ !! ]
[ ok ]
* Starting squid...
2005/01/23 14:00:32| Squid is already running! Process ID 11021 [ ok ] |
And it seems really strange. Killing process 11021 (pid changes everytime of course) allows to restart, but I still got a very long "Waiting" prompt (altough it worked this time)... Strange uh? I'd like to sort this out prior to strafing through the config files. Actually this problem arose during an emerge -u and was simultaneous to the malfunction of Squid. And my config files have not changed, so I guess there's something else...
_________________
. . . . . . . .
eXess |
|
| Back to top |
|
 |
eXess
Apprentice
Joined: 17 Aug 2003
Posts: 243
Location: Belgium
|
| Posted: Sun Jan 23, 2005 1:16 pm Post subject: |
|
|
[UPDATE]
After killing and restarting Squid, and checking through the config files, it seems that now it's working again! Strangest thing. So from now on, the only problem left is slow restart. This can be considered as non critical. However, if you ever have a clue on this... 
_________________
. . . . . . . .
eXess |
|
| Back to top |
|
|
adaptr
Watchman
Joined: 06 Oct 2002
Posts: 6730
Location: Rotterdam, Netherlands
|
| Posted: Sun Jan 23, 2005 1:19 pm Post subject: |
|
|
If the initscripts did not properly stop Squid at one time or another, there will still be a link to the initscript in /var/lib/init.d/started.
This is what causes the "squid already started" message.
Remove the link and try again.
The slow restart is definitely a configuration issue - which you admit you haven't actually figured out yet...
_________________
>>> emerge (3 of 7) mcse/70-293 to /
Essential tools: gentoolkit eix profuse screen |
|
| Back to top |
|
|
eXess
Apprentice
Joined: 17 Aug 2003
Posts: 243
Location: Belgium
|
| Posted: Sun Jan 23, 2005 1:35 pm Post subject: |
|
|
I guess you are perfectly right adaptr. However, Squid config is quite complicated in the end. I could use some help here  Has anyone any idea where the slow restart issue could come from ? Actually it's more accurate to say "slow stop" issue, cause starting causes no problem at all. And btw, I see that about everything is set to default in the config files... :s
_________________
. . . . . . . .
eXess |
|
| Back to top |
|
|
adaptr
Watchman
Joined: 06 Oct 2002
Posts: 6730
Location: Rotterdam, Netherlands
|
| Posted: Sun Jan 23, 2005 1:40 pm Post subject: |
|
|
Let's see.. it's been awhile since I configured it..
The most important config choices - away from the defaults - are your ACLs (obviously), network settings, memory settings (to keep it from running rampant) and any upstream caches.
Oh and your cache directories - spread them over multiple disks if you have any chance at all...
_________________
>>> emerge (3 of 7) mcse/70-293 to /
Essential tools: gentoolkit eix profuse screen |
|
| Back to top |
|
|
eXess
Apprentice
Joined: 17 Aug 2003
Posts: 243
Location: Belgium
|
| Posted: Sun Jan 23, 2005 7:29 pm Post subject: |
|
|
I'll try to be short as well as precise, because I imagine posting all my config files would be rather inefficient.
- Memory settings are all default
- ACLs: All default except one supplement: one ACL set on "external auth", with one user in the list. In the restrictions, it is last but one, set on "allom".
- The user referenced in my ACL was defined in "/etc/webmin/squid/users", using Webmin Squis module.
- The "Basic authentication program" was defined as "/etc/webmin/squid/squid-auth.pl /etc/webmin/squid/users", number=5, cache time=2h, realm name changed of course, but the rest is default.
- Cache directory and all cache settings are default (cache is on /var/cache/squid) but I'm not using it as a cache, if necessary I can turn that off completely (if possible)...
- As far as I can see, everything else is default.
Note that I'm looking at the config through Webmin. If necessary, I can have a look directly in the config files...
_________________
. . . . . . . .
eXess |
|
| Back to top |
|
|
andreac
n00b
Joined: 03 Aug 2004
Posts: 14
Location: Mergo
|
| Posted: Thu Jan 27, 2005 12:38 pm Post subject: Re: Strange, recurring squid issue |
|
|
| eXess wrote: | Hi all,
I've got this strange problem with my SQUID installation. Basically, it doesn't work. Apart from configuration files issues (I'm working on that) I noticed that I always get this error :
| Code: | # /etc/init.d/squid restart
* Stopping squid...
* Waiting ............................................................. * Failed. [ !! ]
[ ok ]
* Starting squid...
2005/01/23 14:00:32| Squid is already running! Process ID 11021 [ ok ] |
|
Hi eXess,
same stop problem here. I use squid in a production server from last year, but in recent updates, the init script is unable to stop the daemon.
So I must manually kill the process then do:
| Code: | | /etc/init.d/squid zap |
(wich reset to stopped state)
and then restart squid. 
_________________
Andrea |
|
| Back to top |
|
|
tinwood
n00b
Joined: 05 Oct 2002
Posts: 36
Location: Newcastle upon Tyne, UK
|
| Posted: Thu Jan 27, 2005 5:27 pm Post subject: |
|
|
Although it doesn't help, I likewise have the same stopping problem (not that I've worked out what is doing it).
Also, after today's update, http/1.1 doesn't appear to be working through the box .... sigh.  |
|
| Back to top |
|
|
DawgG
l33t
Joined: 17 Sep 2003
Posts: 874
|
| Posted: Wed Feb 09, 2005 3:03 pm Post subject: use SQUID-command directly |
|
|
i'm not having any problems with the squid-init-scripts, but if they don't work for u u can always issue commands to squid directly (shutdown, interrupt and kill may be the most interesting in ur case). often that's better than stopping and re-starting the whole process (eg -k reconfigure).
the squid command-line opts can be found here:
http://www.squid-cache.org/Doc/FAQ/FAQ-3.html#ss3.8
helped me a lot.
EDIT: now i got problems, too
too bad, now i got the same problems you all mentioned.
it must have SOMETHING to do with the gentoo-initscripts which i think are rather strange:
- the starting-part tries to change settings that are already defined in the config-file (FDs)
when i start it "by hand"
it just runs & serves after a couple of secs. with the init-script one cpu of the system was at 100% of load for some minutes (never-ever produced that much load before, and _I_ want to decide when or how to rebuild cachedirs or sth.) until i killed and restarted squid by hand.
-the stopping part somehow doesen't work right (gotta check EXACTLY how); it simply does not stop squid.
when i stop it with
| Code: |
squid -k shutdown (or kill)
|
it's down after 30 secs max.
i'll look into it some more (it's a "production" server) and maybe rewrite the initscripts. |
|
| Back to top |
|
|
tnt
Veteran
Joined: 27 Feb 2004
Posts: 1227
|
| Posted: Sun Apr 17, 2005 12:54 am Post subject: |
|
|
Same wait problem here.
Here's the stop part of the /etc/init.d/squid
| Code: | stop() {
ebegin "Stopping squid"
PID=`cat /var/run/squid.pid 2>/dev/null`
start-stop-daemon --stop --quiet --exec /usr/sbin/squid \
--pidfile /var/run/squid.pid
# Now we have to wait until squid has _really_ stopped.
sleep 2
if test -n "$PID" && kill -0 $PID 2>/dev/null
then
einfon "Waiting ."
cnt=0
while kill -0 $PID 2>/dev/null
do
cnt=`expr $cnt + 1`
if [ $cnt -gt 60 ]
then
# Waited 120 seconds now. Fail.
eend 1 "Failed."
break
fi
sleep 2
echo -n "."
done
echo -n "done."
eend 0
else
eend 0
fi
} |
Should we rewrite it something like (waiting for 40secs because default waiting for connection closing is 30secs):
| Code: | stop() {
ebegin "Stopping squid"
squid -k shutdown
# Now we have to wait until squid has _really_ stopped.
sleep 2
if test -n "$PID" && kill -0 $PID 2>/dev/null
then
einfon "Waiting ."
cnt=0
while kill -0 $PID 2>/dev/null
do
cnt=`expr $cnt + 1`
if [ $cnt -gt 20 ]
then
# Waited 40 seconds now. Force!
squid -k kill
eend 1 "Killed."
break
fi
sleep 2
echo -n "."
done
echo -n "done."
eend 0
else
eend 0
fi
} |
_________________
gentoo user |
|
| Back to top |
|
|
jamapii
l33t
Joined: 16 Sep 2004
Posts: 637
|
| Posted: Thu May 12, 2005 1:15 pm Post subject: unstoppable squid |
|
|
I have never seen any squid version+rc file from any distribution that works. At best, it works sometimes. I have used Suse 7.3 and Gentoo.
On Suse, I have modified the rc file to kill squid after some time. It worked finally. Usually I just do "killall squid", which can do collateral damage if someone calls their program squid.
The Gentoo rc file should be modified similarly.
I'm now trying it... looking at the rc file...
I wonder what's the reason. The PID may be wrong, or start-stop-daemon --stop doesn't work. Stopping squid fails only after it has been running for some time, so maybe it wants to do some stuff on exit that takes about forever.
I'm trying this modification to /etc/init.d/squid:
add this function before stop()
| Code: | force_kill() {
echo "trying harder"
killall squid
sleep 2
while kill -0 $PID 2>/dev/null; do
echo -n "."
cnt=`expr $cnt + 1`
[ $cnt -gt 65 ] && return 1
sleep 2
done
return 0
} |
make the "Failed" code conditional
| Code: | # Waited 120 seconds now. Fail.
force_kill || eend 1 "Failed."
break |
EDIT: fixed bugs |
|
| Back to top |
|
|
|
Networking & Security
