Re: [users@httpd] vhosts conf file efficiency

2012-02-14 Thread Noel Butler
On Tue, 2012-02-14 at 12:18 -0500, Sean Conner wrote: > It was thus said that the Great Tom Evans once stated: > > On Tue, Feb 14, 2012 at 4:32 PM, Sean Conner wrote: > > > And really, how often is Apache restarted? On a graceful restart, it can > > > still serve requests. > > > > > > > It's c

Re: [users@httpd] vhosts conf file efficiency

2012-02-14 Thread Tom Evans
On Tue, Feb 14, 2012 at 5:18 PM, Sean Conner wrote: > It was thus said that the Great Tom Evans once stated: >> On Tue, Feb 14, 2012 at 4:32 PM, Sean Conner wrote: >> >  And really, how often is Apache restarted?  On a graceful restart, it can >> > still serve requests. >> > >> >> It's clear you

Re: [users@httpd] vhosts conf file efficiency

2012-02-14 Thread Sean Conner
It was thus said that the Great Tom Evans once stated: > On Tue, Feb 14, 2012 at 4:32 PM, Sean Conner wrote: > >  And really, how often is Apache restarted?  On a graceful restart, it can > > still serve requests. > > > > It's clear you have a strong opinion on this. I prefer my mission > critica

Re: [users@httpd] vhosts conf file efficiency

2012-02-14 Thread Tom Evans
On Tue, Feb 14, 2012 at 4:32 PM, Sean Conner wrote: >  And really, how often is Apache restarted?  On a graceful restart, it can > still serve requests. > It's clear you have a strong opinion on this. I prefer my mission critical systems not take longer than necessary to start up, whilst you don'

Re: [users@httpd] vhosts conf file efficiency

2012-02-14 Thread Sean Conner
It was thus said that the Great Noel Butler once stated: > On Mon, 2012-02-13 at 22:10 -0500, Sean Conner wrote: > > > > > > So its going to open, read and close 2000 files, rather than open, read > > > and close one file, that may or may not be noticeable at > > > startup/reloads, if I was a bett

Re: [users@httpd] vhosts conf file efficiency

2012-02-14 Thread Sean Conner
It was thus said that the Great Tom Evans once stated: > On Tue, Feb 14, 2012 at 3:10 AM, Sean Conner wrote: > >  Nope.  I just ran a program [1] that opened and read 25,018 files in 2.5 > > seconds [2].  I'd bet unnoticeable. > > > > So that is an extra 5 seconds on a graceful restart - apache w

Re: [users@httpd] vhosts conf file efficiency

2012-02-14 Thread Noel Butler
On Mon, 2012-02-13 at 22:10 -0500, Sean Conner wrote: > > > > So its going to open, read and close 2000 files, rather than open, read > > and close one file, that may or may not be noticeable at > > startup/reloads, if I was a betting man, I'd say noticeable. > > Nope. I just ran a program [

Re: [users@httpd] vhosts conf file efficiency

2012-02-14 Thread Tom Evans
On Tue, Feb 14, 2012 at 3:10 AM, Sean Conner wrote: >  Nope.  I just ran a program [1] that opened and read 25,018 files in 2.5 > seconds [2].  I'd bet unnoticeable. > So that is an extra 5 seconds on a graceful restart - apache will read the config once, throw it away, and then read it again, so

Re: [users@httpd] vhosts conf file efficiency

2012-02-13 Thread Sean Conner
It was thus said that the Great Noel Butler once stated: > On Tue, 2012-02-14 at 12:21 +1000, Nick Edwards wrote: > > > On 2/12/12, Steve Swift wrote: > > > I don't think it would make a significant difference if you had a single > > > file with 2000 vhosts, or 2000 files with one vhost each. > >

Re: [users@httpd] vhosts conf file efficiency

2012-02-13 Thread Noel Butler
On Tue, 2012-02-14 at 12:21 +1000, Nick Edwards wrote: > On 2/12/12, Steve Swift wrote: > > I don't think it would make a significant difference if you had a single > > file with 2000 vhosts, or 2000 files with one vhost each. > > > > I think you're right, I added half a dozen test domains and c

Re: [users@httpd] vhosts conf file efficiency

2012-02-13 Thread Nick Edwards
On 2/13/12, Noel Butler wrote: > On Sun, 2012-02-12 at 13:14 +, Steve Swift wrote: > >> >> >> >> One caveat: Don't ever rely on a a wildcard INCLUDE for the order of >> your vhosts. This probably only matters for the first vhost, which is >> the default for any request which doesn't match any

Re: [users@httpd] vhosts conf file efficiency

2012-02-13 Thread Nick Edwards
On 2/12/12, Steve Swift wrote: > I don't think it would make a significant difference if you had a single > file with 2000 vhosts, or 2000 files with one vhost each. > I think you're right, I added half a dozen test domains and checked lsof, no apparent extra file handles. Just if I have to rebu

Re: [users@httpd] vhosts conf file efficiency

2012-02-12 Thread Noel Butler
On Sun, 2012-02-12 at 13:14 +, Steve Swift wrote: > > > > One caveat: Don't ever rely on a a wildcard INCLUDE for the order of > your vhosts. This probably only matters for the first vhost, which is > the default for any request which doesn't match any vhost. > > It absolutely matters,

Re: [users@httpd] vhosts conf file efficiency

2012-02-12 Thread Steve Swift
I don't think it would make a significant difference if you had a single file with 2000 vhosts, or 2000 files with one vhost each. Having them all in one file would make life easier if you needed to make a global change. One caveat: Don't ever rely on a a wildcard INCLUDE for the order of your vh

[users@httpd] vhosts conf file efficiency

2012-02-11 Thread Nick Edwards
Hi, Is it more efficient for all virtualhost blocks to be in a single file, eg httpd-vhosts.conf, or for each of them to be in their own "one file per domain" via an include hosts.d/*.conf We are talking a minimum of 2000 hosts per machine. Memory I assume would be the same since it needs to know