Hi Phil, On Wed, 07 Jun 2006, Philip Hands wrote: > Hi, > > I notice (from nagios and hobbit reports) that costa's HTTP service is off > the air. Perhaps you could deal with that. This is the second time in a > couple of days (Wichert fixed it yesterday after I mentioned it on IRC)
Well it happens very frequently. It's very probably related to log rotation. /etc/logrotate.d/apache2 rotates 2 serie of files and after each rotation it does restart apache2. For apache 1.x the postrotate script is a simple reload (and not a restart), we should probably also use that if it's enough. But that's really a bug in apache2: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=301702 For the apache maintainers and the bug log: we are bitten by this bug on an almost daily basis. I don't know if costa's HTTPD configuration has something specific which triggers the bug regularly but it's very annoying and it should really be fixed. apache1 postrotate script seems much nicer (use of invoke-rc.d and use reload instead of restart). Why don't use the same script for apache2? Would using reload create problems? In the mean time, I'll use reload on costa and see how it works. > On both days, it went down somewhen near 08:00 UTC, which may help you > narrow down the cause. Yes thanks for the hint. Cheers, -- Raphaël Hertzog Premier livre français sur Debian GNU/Linux : http://www.ouaza.com/livre/admin-debian/
