Received: (qmail 74875 invoked by uid 501); 15 Feb 2001 18:04:13 -0000 Message-Id: <20010215180413.74862.qmail@apache.org> Date: 15 Feb 2001 18:04:13 -0000 From: Lenny Miceli Reply-To: miceli@buffalo.edu To: submit@bugz.apache.org Subject: Running into problems at approx 232 virtual hosts. X-Send-Pr-Version: 3.110 >Number: 7251 >Category: os-solaris >Synopsis: Running into problems at approx 232 virtual hosts. >Confidential: no >Severity: critical >Priority: medium >Responsible: apache >State: closed >Quarter: >Keywords: >Date-Required: >Class: sw-bug >Submitter-Id: apache >Arrival-Date: Thu Feb 15 10:10:03 PST 2001 >Closed-Date: >Last-Modified: Thu Feb 15 20:40:15 PST 2001 >Originator: miceli@buffalo.edu >Release: 1.3.14 >Organization: >Environment: SunOS 5.6 Generic_105181-23 sun4u sparc SUNW,Ultra-4. I use sun's CC compiler. >Description: I've been running apache for years then approx 6 months ago when running apache v1.3.6 I started seeing problems when I tried to add more virtual hosts. It seems like the magic number back then was approx 235-237. Once I added one more VH the server would start but approx 50% of the accesses would get a server denied error and in the error log I would see the following messages: [The date] [error] [client 128.205.251.12] (2)No such file or directory: file permissions deny server access: /the_path_to_the_file But the permissions on the files were fine because you could come right back to the same url and the server would serve the page without a problem. The files are .html, .jpg, etc it didn't matter. Sometimes the server would serve them and sometimes it wouldn't. So I figured I'd upgrade to v1.3.12 but the problem is still there, then I tried v1.3.14 but the problem still exists. I didn't upgrade to v1.3.17 since this is a major server on campus and I have to schedule upgrades like this since we are serving so many VH's. Anyways, I'm currently running 232 VH's. As soon as I add one more I get the problems I describe above. But, I've also noticed when I make a change to the httpd.conf file and then issue a kill -USR1 the above problem starts happening but to a lessor degree. Instead of %50 of the accesses having problems is approx 3-5%. So what I do is now kill the apache server off totally and restart the server and the problem goes away. I now have time to work on this problem again. I've searched your bug reports and everything else I know about and can't find any mention of anything like this, except of the issue of too many error log definitions on solaris because they use stdout and you can only have approx 255 file handles open when using that. So a long time ago I removed separate error logs for all the VH's and I have them all log to the same file. In all the VH 's definitions I don't specify the same file, I just specify the error file in the main server section so they pick up the same file by default. All the other hosts do have a different access log but that should be OK to my knowledge. I have unlimited shell limits on the main apache process so that should pass those to all the children it creates. I tried to replicate this problem by starting a test server on the same machine with all these VH's but I can't replicate the problem. It seems like it only happens on a very busy server which this host is. It's our main webserver. So I was wondering if you can give me any help in debugging why this problem is happening since we need to add more VH's to this server and we currently can't. >How-To-Repeat: I can only repeat it by how I said above. >Fix: Nope. I ran out of idea's so that's why I'm coming to you guys. >Release-Note: >Audit-Trail: State-Changed-From-To: open-closed State-Changed-By: slive State-Changed-When: Thu Feb 15 11:26:49 PST 2001 State-Changed-Why: Please see: http://httpd.apache.org/docs/misc/FAQ.html#fdlim and also consider using mod_vhost_alias. Thanks for using Apache! From: Lenny Miceli To: slive@apache.org Cc: apbugs@Apache.Org Subject: Re: os-solaris/7251: Running into problems at approx 232 virtual hosts. Date: Thu, 15 Feb 2001 14:49:11 -0500 On Thu, Feb 15, 2001 at 07:26:53PM -0000, slive@apache.org wrote: > > Synopsis: Running into problems at approx 232 virtual hosts. > > State-Changed-From-To: open-closed > State-Changed-By: slive > State-Changed-When: Thu Feb 15 11:26:49 PST 2001 > State-Changed-Why: > > Please see: > http://httpd.apache.org/docs/misc/FAQ.html#fdlim > and also consider using mod_vhost_alias. We can't run mod_vhost_alias since all of our virtual hosts are very distinct per department so every Virtual host definition is not similar enough to use this functionality. I thought I would be able to have upto 1024 file descriptors, since solaris allows this max and I have the following in the script that starts the apache server: unlimit cputime unlimit filesize unlimit datasize limit stacksize 2090000 limit coredumpsize 20480 unlimit vmemoryuse limit descriptors 1024 Also, your answer doesn't explain why I can't use kill -USR1 anymore to have the server reread the config file. Thanks for you time in this matter, Lenny Comment-Added-By: marc Comment-Added-When: Thu Feb 15 20:40:15 PST 2001 Comment-Added: The FAQ entry links to http://httpd.apache.org/docs/misc/descriptors.html which has a more detailed description. Compiling using a 64-bit ABI may get around this particular problem of yours if your system supports it. This is a Solaris limitation. >Unformatted: [In order for any reply to be added to the PR database, you need] [to include in the Cc line and make sure the] [subject line starts with the report component and number, with ] [or without any 'Re:' prefixes (such as "general/1098:" or ] ["Re: general/1098:"). If the subject doesn't match this ] [pattern, your message will be misfiled and ignored. The ] ["apbugs" address is not added to the Cc line of messages from ] [the database automatically because of the potential for mail ] [loops. If you do not include this Cc, your reply may be ig- ] [nored unless you are responding to an explicit request from a ] [developer. Reply only with text; DO NOT SEND ATTACHMENTS! ]