On Mon, 2009-09-28 at 06:25 -0400, Phil Pinkerton wrote: > > Jonathan Zuckerman wrote: > > On Fri, Sep 25, 2009 at 10:54 AM, Phil Pinkerton <pcpinkerton@xxxxxxxxx> wrote: > > > >> On Fri, Sep 25, 2009 at 12:45 PM, Jonathan Zuckerman <j.zuckerman@xxxxxxxxx> > >> wrote: > >> > >>> On Fri, Sep 25, 2009 at 6:15 AM, Phil Pinkerton <pcpinkerton@xxxxxxxxx> > >>> wrote: > >>> > >>>> Anyone experience performance issues with regards to Apache cache ? > >>>> > >>>> Â From time to time we are experience access issues ( users accessing > >>>> Subversion ) that seem to be due to an over loaded Apache cache. We > >>>> suspect > >>>> the cache because when we refresh (stop,start) the Apache server the > >>>> problems goes away for a time. This happens so regularly we have set > >>>> crontab > >>>> to refresh every morning during our maintenance period. And this has > >>>> reduced > >>>> the number of problems. > >>>> > >>>> This is Apache 2.2.11 running on a Solaris Sparc server. Providing HTTPD > >>>> services for or Subversion users, > >>>> > >>>> Any thoughts on managing Apache Cache ? > >>>> > >>>> -- > >>>> " The fundamental principle here is that the justification for a > >>>> physical > >>>> concept lies exclusively in its clear and unambiguous relation to the > >>>> facts > >>>> that it can be experienced" Â AE > >>>> > >>>> Please Feed and Educate the Children... it's the least any of us can do. > >>>> > >>>> > >>>> > >>> Is there anything at all in the error log to suggest it's the cache? I > >>> think that bouncing the web server affects a lot of other stuff > >>> besides the cache.. > >>> > >>> > >> Nothing in the logs tells us anything about why we get the "Other" error > >> messages, but if it's not a cache problem (possibly related to LDAP), why > >> would a recycle fix it? > >> > >>> --------------------------------------------------------------------- > >>> 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 > >>> Â " Â from the digest: users-digest-unsubscribe@xxxxxxxxxxxxxxxx > >>> For additional commands, e-mail: users-help@xxxxxxxxxxxxxxxx > >>> > >>> > >> > >> -- > >> " The fundamental principle here is that the justification for a physical > >> concept lies exclusively in its clear and unambiguous relation to the facts > >> that it can be experienced" Â AE > >> > >> Please Feed and Educate the Children... it's the least any of us can do. > >> > >> > >> > > > > When you restart the webserver every thread completes and stops > > handling requests, every process ends, _everything_ about the > > webserver stops and then starts again, this potentially affects more > > than just the files used by mod_cache. > > Have you looked at the diskspace used by mod_cache to see if there are > > any patterns to it (size, usage, whatever) when the webserver stops > > responding? How about monitoring the health of the box to see if > > anything else could be the issue (CPU pegged, RAM, disk..)? Anything > > at all in apache error logs. > > I'm sorry that I don't have more suggestions; mod_cache might indeed > > be the problem but from a troubleshooting standpoint it seems to me > > that you're narrowing down the possible causes just a bit too > > recklessly. > > > > > Not sure you need to be so abusive. When this issue arises 100's of > developers cannot access Subversion, several automated scripts fail to > checkout from Subversion and it could literally halt production. We > quickly implemented this "band-aid" to keep production running and since > the refresh takes only about 3 seconds and works to clear the issue. I > would think letting production crash while taking detail RCA would be > much worse than reckless. We don't have the Apache expertise and we did > very well to at least find a short term solution. I asked this group for > help not criticism. > > The unnecessary "reckless" comment aside your questions are valid and > maybe somewhat useful. > > Thanks for the assistance. > > Phil How frequently does it occur if the cache is disabled? Cheers Tom --------------------------------------------------------------------- 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
|