For anyone who''s curious... Begin forwarded message:> From: "Jason A. Hoffman" <jason-xKtDo/uLHBtl57MIdRCFDg@public.gmane.org> > Date: June 10, 2005 4:19:44 AM GMT-03:00 > To: Ben Jackson <ben-p14LI7ZcAE/pVLaUnt/cCQC/G2K4zDHf@public.gmane.org> > Subject: Re: lighttpd caching - RESOLVED?? > > On Jun 9, 2005, at 7:03 PM, Ben Jackson wrote: > >> It looks like the fastcgi process that was started by lighttpd isn''t >> getting killed when I restart the server, which would explain why >> changing the name of the directory was clearing the cache: it was the >> only way to get around the name conflict which I assume was blocking >> the new process from starting. Is it possible for me to kill lighttpd >> and all the fcgi processes spawned by it in a shot? > > That was the next thing I was going to say. This wasn''t an > > What you need to do is not simply kill the process, you need to get rm > the pid and then rm all the sockets (in case they weren''t). The > startup script should do that and it is best to use it rather than > just killing the proc. > > - Jason >
How about when using Apache/1.3.33 (Debian GNU/Linux), mod_fastcgi 2.4.2-6, libfcgi 2.4.0-5? If I do /etc/init.d/apache stop then ps aux | grep dispatch there''s still 3 dispatch.fcgi processes pointing at my rails app - this happens in development or production environment? Any ideas how I can clear these out as part of the Apache restart process? I''ve just been doing a killall -9 dispatch.fcgi then clearing the entries in /var/lib/apache/fastcgi/dynamic Cheers Kevin> For anyone who''s curious... > > Begin forwarded message: > >> From: "Jason A. Hoffman" <jason-xKtDo/uLHBtl57MIdRCFDg@public.gmane.org> >> Date: June 10, 2005 4:19:44 AM GMT-03:00 >> To: Ben Jackson <ben-p14LI7ZcAE/pVLaUnt/cCQC/G2K4zDHf@public.gmane.org> >> Subject: Re: lighttpd caching - RESOLVED?? >> >> On Jun 9, 2005, at 7:03 PM, Ben Jackson wrote: >> >>> It looks like the fastcgi process that was started by lighttpd isn''t >>> getting killed when I restart the server, which would explain why >>> changing the name of the directory was clearing the cache: it was the >>> only way to get around the name conflict which I assume was blocking >>> the new process from starting. Is it possible for me to kill lighttpd >>> and all the fcgi processes spawned by it in a shot? >> >> That was the next thing I was going to say. This wasn''t an >> >> What you need to do is not simply kill the process, you need to get rm >> the pid and then rm all the sockets (in case they weren''t). The >> startup script should do that and it is best to use it rather than >> just killing the proc. >> >> - Jason >> > > _______________________________________________ > Rails mailing list > Rails-1W37MKcQCpIf0INCOvqR/iCwEArCW2h5@public.gmane.org > http://lists.rubyonrails.org/mailman/listinfo/rails > >
Ben Jackson
2005-Jun-10 21:38 UTC
Clearing lighttpd production cache [was: lighttpd caching - RESOLVED??]
Well, I''m almost there... Everything now works fine in development mode. However (as I just learned the hard way by ticking off my server admins) this is not the ideal way to do things. After switching to production mode, lighttpd seems to be serving files from the cache, and bugs from earlier versions of the app are showing up again. Anyone know how to clear the production cache? Restarting lighttpd does not seem to help. - Ben On Jun 10, 2005, at 11:16 AM, Ben Jackson wrote:> For anyone who''s curious... > > Begin forwarded message: > >> From: "Jason A. Hoffman" <jason-xKtDo/uLHBtl57MIdRCFDg@public.gmane.org> >> Date: June 10, 2005 4:19:44 AM GMT-03:00 >> To: Ben Jackson <ben-p14LI7ZcAE/pVLaUnt/cCQC/G2K4zDHf@public.gmane.org> >> Subject: Re: lighttpd caching - RESOLVED?? >> >> On Jun 9, 2005, at 7:03 PM, Ben Jackson wrote: >> >>> It looks like the fastcgi process that was started by lighttpd isn''t >>> getting killed when I restart the server, which would explain why >>> changing the name of the directory was clearing the cache: it was >>> the only way to get around the name conflict which I assume was >>> blocking the new process from starting. Is it possible for me to >>> kill lighttpd and all the fcgi processes spawned by it in a shot? >> >> That was the next thing I was going to say. This wasn''t an >> >> What you need to do is not simply kill the process, you need to get >> rm the pid and then rm all the sockets (in case they weren''t). The >> startup script should do that and it is best to use it rather than >> just killing the proc. >> >> - Jason >> > > _______________________________________________ > Rails mailing list > Rails-1W37MKcQCpIf0INCOvqR/iCwEArCW2h5@public.gmane.org > http://lists.rubyonrails.org/mailman/listinfo/rails >