I also think that it is likely that a port is not freed quickly enough... There are a couple of fixes in Apache 2.0.54 that may be relevant to this problem: Refer to the changelist for details. I therefore think that it might be worthwhile to try to reproduce the restart problem with 2.0.54. Although I am not a great fan of upgrading production systems to the latest and greatest version every time there is a release, but there have been so many fixes since 2.0.49 (and 2.0.50) that we currently use, that I have started testing 2.0.54. I have myself stopped using Apache restart and graceful for other reasons than "stuck" ports. We can do this because we have a load balancer that also provides failover. This allows us to restart one web server at the time. The reason we stopped using restart/graceful was because we ran into problems which I thought might be due to inadequate cleanup of shared memory segments and semaphores on Solaris when restart or graceful are invoked. I may be wrong about this, but since I adopted this policy and removed the restart and graceful from apachectl to force our administrators to use stop and start instead, we have not experienced any problems with orphaned shared memory blocks or orphaned semaphores. Specifically, the problem we had was that Apache (SSL) complained about the SSL shm session cache file existing when starting Apache. We then removed the file and tried again, and again we got the same message. And sure enough, the file had been recreated. The only solution was to either change SSL session cache to dbm, or reboot the server. -ascs -----Original Message----- From: Andres Monroy-Hernandez [mailto:andres@xxxxxxxxxxxxxxxx] Sent: Wednesday, June 08, 2005 10:04 PM To: users@xxxxxxxxxxxxxxxx Subject: Re: [users@httpd] Restart Errors (apache 2.0.50) My guess is that the graceful restart is not freeing the port after all the children exit. Is using apache restart out of the question? Perhaps your web applications don't let the children exit and the request to exit times out? -Andres brad bowman wrote: >I was wondering if anyone could give some insight as to the possible >cause of a restart error i've been getting intermittently over the last >8 months or so. > >my apache version is: >Apache/2.0.50 (Unix) > >My kernel version is: >Linux 2.4.20-8smp (redhat 9) > >Basically, if I restart the server using 'apachectl graceful', it will >restart without problems 99 percent of the time, but occasionally it >will fail to restart, giving the error message: >(98)Address already in use: make_sock: could not bind to address >0.0.0.0:80 no listening sockets available, shutting down Unable to open >logs > >>From my understanding, this error is normally the result of either an >apache server already running, or something else listening on that >port. There is nothing else running on the server that should ever pick >up on port 80, so typically i've tried to fix the problem by killing >all apache/httpd processes and then restarting apache, but the same >error persists. Since this is a production server, at this point > > >i've just hedged my bets and rebooted for a 2-3 minute downtime, at >which point operation resumes normally. > >Basically i'm looking for advice on what the issue could be, so I can >stop it from happening.... or at least a _FAST_ way to get apache back >up and running if/when it does happen, so I don't have to enter "oh >crap the server is down" mode. > >any input would be appreciated >thanks, >brad bowman > --------------------------------------------------------------------- The official User-To-User support forum of the Apache HTTP Server Project. See <URL:http://httpd.apache.org/userslist.html> for more info. To unsubscribe, e-mail: users-unsubscribe@xxxxxxxxxxxxxxxx " from the digest: users-digest-unsubscribe@xxxxxxxxxxxxxxxx For additional commands, e-mail: users-help@xxxxxxxxxxxxxxxx
|