Thanks for the suggestion Jay, though in my situation a 'squid -k reconfigure' puts too much load on the system (as it kills all the auth children and forces them to reload, as well as squid and squidguard). I need a way to force only squidguard to reconfigure.
I'm looking at squidGuard -u, as I am hoping that when a database file is rebuilt all the running squidGuard processes will pick up the change. Thanks for the help, Matt Smith On Fri, 17 Sep 2004 05:49 pm, Jay Turner wrote: > I use Squid with SquidGuard and Windbind or LDAP authentication helpers.. > > I always just issue 'squid -k reconfigure' which forces Squid and > SquidGuard to reload their config files again (including all > domain/user/url lists in SquidGuard) it is very fast and results in no > noticable disruption of service to end users. > > This works fine for me and I have never had a problem doing so in over two > years of use. > > Regards > Jay > > > -----Original Message----- > > From: [EMAIL PROTECTED] > > [mailto:[EMAIL PROTECTED] Behalf Of Matthew Smith > > Sent: Friday, 17 September 2004 12:59 PM > > To: [EMAIL PROTECTED] > > Subject: SquidGuard and HUP > > > > > > Hello SquidGuard List, > > > > I'm new to the list and squidguard, so please excuse me if I ask > > some newbie > > questions... :) > > > > I am running squid with a number of squidguard subprocesses, and > > a number of > > auth processes. These auth processes take a large time to reload, > > and put the > > system under pressure when they do. > > > > Every now and then I want to make alterations to the > > squidguard.conf, or add > > or remove a user from a external user list. At the moment I am restarting > > squid to get squidguard to reload its configuration. I have since > > descovered > > that sending squid a HUP using 'kill -HUP' forces squid to reload > > it's config > > and it's redirector and auth children, but this means that my > > auth processes > > restart as well. > > > > I noticed on the squidguard website it mentions: "Currently a squidGuard > > process will reincarnate on a HUP signal directly to the > > squidGuard process > > -- if at all possible (i.e. execvp() and replace itself with a fresh > > squidGuard process with the same command line arguments). Thus if > > you suspect > > a squidGuard process has gone mad, you don't necessarily need to > > reconfigure > > Squid; a "kill -HUP that_squidGuard_pid" will probably do" > > > > However, a 'killall -HUP squidGuard' causes my squidguard > > processes to die, if > > I do it a few more times then squid itself will shutdown. > > > > I get the following in my squid cache.log: > > error execve: 2 > > 2004/09/17 14:39:40| WARNING: redirector #4 (FD 31) exited > > error execve: 2 > > 2004/09/17 14:39:40| WARNING: redirector #5 (FD 32) exited > > error execve: 2 > > 2004/09/17 14:41:16| WARNING: redirector #1 (FD 11) exited > > error execve: 2 > > 2004/09/17 14:41:17| WARNING: redirector #2 (FD 26) exited > > error execve: 2 > > 2004/09/17 14:41:17| WARNING: redirector #3 (FD 27) exited > > 2004/09/17 14:41:17| Too few redirector processes are running2004/09/17 > > 14:41:17| Starting new helpers > > 2004/09/17 14:41:17| helperOpenServers: Starting 5 'squidGuard' processes > > > > and a bit later (if I send another HUP): > > > > 2004/09/17 14:43:11| WARNING: redirector #1 (FD 7) exited > > error execve: 2 > > 2004/09/17 14:43:11| WARNING: redirector #2 (FD 8) exited > > error execve: 2 > > 2004/09/17 14:43:12| WARNING: redirector #3 (FD 9) exited > > 2004/09/17 14:43:12| Too few redirector processes are running2004/09/17 > > 14:43:12| storeDirWriteCleanLogs: Starting... > > 2004/09/17 14:43:12| WARNING: Closing open FD 6 > > 2004/09/17 14:43:12| WARNING: Closing open FD 20 > > 2004/09/17 14:43:12| Finished. Wrote 8582 entries. > > 2004/09/17 14:43:12| Took 0.0 seconds (2193202.1 entries/sec). > > FATAL: The redirector helpers are crashing too rapidly, need help! > > > > Squid Cache (Version 2.5.STABLE5): Terminated abnormally. > > > > In my squidguard.log, I get the following: > > > > 2004-09-17 14:49:52 [19638] got sigHUP reload config > > 2004-09-17 14:49:52 [19639] got sigHUP reload config > > 2004-09-17 14:49:53 [19640] got sigHUP reload config > > 2004-09-17 14:49:53 [19641] got sigHUP reload config > > 2004-09-17 14:49:53 [19642] got sigHUP reload config > > > > But the squidguard processes must be crashing, because squid only > > reloads them > > when a request comes through (and the redirector fails until they come > > online). > > > > From what I can gather, when the squidgard processes recieve a HUP, they > > attempt to reload using execvp(), but crash. Is there a better > > way that I can > > force a squidguard reload (eg with the -u command line option)? > > Or is there a > > fix I can apply to squidguard to stop it crashing on a HUP? > > > > I am using squid-2.5.STABLE5-1 and squidGuard-1.2.0 on a Redhat > > Enterprise > > Linux server. > > > > Thanks for any help you can give, > > > > Matthew Smith
