well right now im thinking about moving users public-secure outside of -home-* so they can have o-rwx and making

joe-f: I would say yes but if you happen to open exactly the same stuff you already have open, then browsers could do something tricky
joe-f: but.. yes, most likely atleast HEAD requests are made to server host to check whether content has changed
joe-f: ..or not HEAD but GET with check whether content has changed

i didn't know REFERER had anything to do with the changing of content.. hm

it doesn't
anyway, what is your question/problem ?

i just want to know if a user opening a link in a new window, from my site, will still be marked as a REFERER on the site the user is opening
not completely apache related i know, but kinda..

joe-f: if user opens a link, the request has referer *unless* the user, browser, firewall or something strips it off. you can't trust referers.

alright, so having it opened in a new window/tab doesnt matter
makes sense
thx

1) Apache doesn't run as root (only the parent process responsible for openling logs and binding to ports 1024) and 2) The UID Apache runs as has to have traverse access to any directories (and their parents) it must read content from.

6 2007] [error] [client 192.168.103.69] File does not exist:

and?

my url was http://www.mysite.com/incoming

hello. quick question... will mod_deflate work with apache as a proxy server? i.e. will it deflate proxy requests?

for some reason it's getting rewritten and i can't figure out why

I doubt it

Roobarb:

what directory is your DocumentRoot ?

Is there a way to limit the amount of memory an httpd child process is allowed to use?

well, right now i'm thinking about moving user's public/secure outside of /home/* so they can have o-rwx and making symlinks so they can easily get to those directories. the directories are setgid apache so any new files or directories will be group owned by apache. is this bad or can you think of anything simpler?

RLimitMem

the goal is to keep user's out of each others' home and web directories

I thought that was only for processes that httpd spawned (I may have misunderstood the docs)

how can I activate php5 hosting with apache2? I can with apache 1 but I dont see the way with apache2

oh I see. I don't know whether you can control how mych ram Apache itself uses (short of ulimit)
php

is a scripting 'language' that is enabled by loading mod_php and by adding 'AddHandler application/x-httpd-php .php' or you can look at http://wiki.apache.org/httpd/PHPDownload

DocumentRoot is DocumentRoot /var/ee/www, http://apache.pastebin.ca/606435

Line 69

what is wrong ?

Questions of the form "what's wrong with ..." can't be answered unless we know what you're trying to do.

nothing, its Aliasing /incoming to "/home/ftp/pub/incoming"

that should be OK

Hi i have apache2.0.54 running on debian sarge with PHP5 (SUPHP/PHP5-CGI), i am having problems with large file uploads, php webhosting is configured correctly but when uploading a 50 odd meg file the client stops sending data near the end with but the browser just sits there still saying sending, server side apache is showing that its returning a 400 http status code for the post request? any ideas, please.

you're trying to use a VirtualHost and main config to serve content on the same ip I think

anyone have an opinion of Ruby on Rails - good? bad?

in my apache2.conf I dont have any refference to PHP, any

then add it

the last one?

the last one is the [L] one, by definition

Roobarb, what do i need to change?

load the php module, and put the AddHandler line next to it
that depends on what you are trying to achieve
you have a VirtualHost listening on Port 80, as well as a main config. I think the VHost is being selected first

In fact, I dont have any LoadModule line apache1 was more easy

Roobarb, that should be fine, either way

take a look at http://www.linux-corner.info/httpd-simple.conf for a proper VHost example
-- gone

you are using debian/ubuntu, I assume?

thumbs, debian
I just did it with apache1, just uncommented a line, but in 2 i cant find the line

default layout

Pop along to http://wiki.apache.org/httpd/DistrosDefaultLayout for an idea on where the main distros place Apache files.

read that factoid, please

I know where my apache2.conf is
but inside apache2.conf I cant find anything about php

Roobarb-Work, i just got rid of VirtualHost, but same behavior.

Roobarb-Work:

the trick is, you DON'T want to edit apache2.conf
debian does NOT use that file.

going to ask google groups.

php

is a scripting 'language' that is enabled by loading mod_php and by adding 'AddHandler application/x-httpd-php .php' or you can look at http://wiki.apache.org/httpd/PHPDownload

debian php

debian php is here http://wiki.apache.org/httpd/ScratchPad/DebianPHP

http://wiki.apache.org/httpd/DebianPHP

i'll see
apache1 was so easy :'(

no, debian is a mess
debian--

wine++

if you used a sane distro/package, you would simply edit httpd.conf
but no, debian decided to create unmanageable structures that require digging to simply find a config file

Bonsoir / Good evening

hey anapi, how are you doing?

fine, and you ? : )
)
hi thumbs, lamp

quite fine thanks, friday, free weekend, few glasses of wine

)

I need help

yes, you do

Thank you

i need help too :o

well, at last it runs

Can anyone in here tell me how to get flash to run through apache
google is at a loss for me..

http://www.newgrounds.com/audio/view.php?id=204148&sub=50029

flash does not run on apache

that sucks....why not..

now mysql :'(

flash is invoked from html

flash runs in your browser

flash run in the browser

tell your html to load flash content.
see #web and #html

run the file from browser it plays correctly.....access it through the server it;s there just not playing

apache does not run flash
flash is embedded in html content.
so embed it, and it'll work

I got an idea
send it with the proper mimetype

What can you explain (apache + suphp + php4) that after few hours, connections are never closed ? http://apache.pastebin.ca/606464

hang on
matt you're prolly having trouble with MIME type
matt ran away :o

he was clueless

yea
:|
and he didn't stay around
i hate that
you help someone and they just bail
anyone have any experience with Ruby on Rails?

a bit

Good experience? Bad experience?

fantastic. I would be doing it now if I had anything to do with it.

any reason why you're still using php4?

thumbs, not me, a customer... !!!!

really? is it yer first framework?
I just discovered cakephp
so I'm on the fence
of learning either ruby or something i already know but not as popular

it was my first, but it hasn't been my last. Ruby in general is an infinitely better language to work with

cool

rails is still my favorite thing to work with. CherryPy isn't bad. Django I've only touched on a little but around here it seems popular

yeah php is nice n' easy but there's a massive trade off with what you can do with it

in terms of performance you may find this interesting: http://wiki.rubyonrails.org/rails/pages/Framework+Performance

nice thx - this is absolutely something i've been loking for

think its quite dated so not sure how true it is now, but if gives you a general idea

thx

people talk about speed as rails' weakpoint. it actually did better than I would have thought.
bpeople talk about speed as rails' weakpoint. it actually did better than I would have thought./b
in fact it did damn well
and on mongrel too

tbh i've never really got into ruby, ive always been more of a php hosting person. but then i dont like frameworks really either

i'm jsut cautious about technologies being popular just because their fads - you know what i mean? Like CORBA during the 90's
don't want to learn something that's gonna be traded up with the next fad

or java

java is great for engineering next generation solutions to enable maximization of developer income by means of enhanced buzzword useZthe future. assuming they can make the connector between tomcat/apache a little more solidQvery good at dumping meaningless exceptions giving no clue about the real problemba funny little story at http://steve-yegge.blogspot.com/2006/03/execution-in-kingdom-of-nouns.html

and if its about one mongrel instance, you can easily set up two!

fajita feels strongly about this

huh?

whats with her random letter bug?

i mean shit - java web hosting got blown away by flash in the client applet market

...if you missed this due fajitas outcome, "and if its about one mongrel instance, you can easily set up two!

shows it doesn't matter how powerful or feature ridden a tech has

i've always considered java to be too bloated, though in terms of managing the code java is quite nice for larger projects
never really considered it a contender for client side though

that was the java dream at the beginning
then they had to switch to server side - then to mobile

java is physically painful for me to program in. Its pure anti-logic.

java makes me wanna poo

well considering your a ruby person that makes sense

C# makes my head hurt. i think its perls fault.

java is kinda the opposite to functional style programming

i like my delphi - it's easy n' fast

huh, what language is more logical than java?
java being such a dump ass language

depends on your logic

well, ruby naturally. C certainly is, although its relating exponentially fewer concepts so it has less of a fight there.
python

Programming the way Guido indented it

this bot is a wit

yay! it works... after much delibration I give to you... my modified autoindex (please don't laugh) http://nightkhaos.homelinux.org/files/

30% shinier than the previous autoindex

ok. about ruby I have only basics on ruby on rails. about c, it's little bit on different level. but yeah, I have always laughed on java's "platform independence" and performance is a joke in first hand

well at least you didn't laugh.

you don't know that I don't HAVE to type lol every time I fall out of my chair with tears streaming down my face
its the internet. its not a big truck.
but it looks fine.

java and performance dont belong in the same sentence, anywhere!

yep

java is designed to have a good ICON score. ICON being Isolatable Components with Obfuscated Names (essentially a metric of how good a flowchart you can draw of a particular system)

yeah I really need to stratch up on my HTML and CSS and predify it more... having a new icon thing is an improvement through.

indeed

how does this relate to apache?

autoindex.

autoindex is See mod_autoindex

about as much as ruby/delphi/java/php etch

but at least I am talking about a direct function of apache rather than something that can be used in conjuction with apache.

sorry, didnt mean it like that, what i meant to say was more relevant than ....

hi. i have a page that will display in FF on linux, and IE on windows, but FF on windows gives a internal service error
this (unfortunately) isn't my apache, so i can't refer to the log

step one?

whatever the problem, step one is to look in the error log (and any other logs that may apply, such as suexec, mod_rewrite, or mod_security).

any suggestions on how to track things down?
essentially, i need to be able to point the admin in the right direction and take the heat off my back...
yes. im not a dumbass thanks though

internel service error? how exactly are you attempting to access the page?

presumably it's doing some sort of user-agent thing and blowing it

what type of page is it? html or a server side parsed page(php/perl)?

yeah. thats what im thinking also. which is going to be damn near impossible to detect on the client side of things

I didn't think so but what exactly does the error log tell?

php i image

you can set your user-agent string

2 thing i said was this isnt my apache install

experiment with curl/wget, for example

GREAT idea

go get the User Agent switcher extension for FF, it'll make that easier

livehttp headers in ff is also quite usefull to see what is being sent and recieved

2 thing? "ok"

2nd

oh, another thing could be content negotiation

grabbing it now, ty

maybe your ff/windows install is asking for a language which it has trouble with.
or a mime-type, even

its pretty consistent on the different machines in my office, so maybe, but more unlikely than the UA detection

Would that affect php phasing?

just an idea, trying to run through the possibilities

depends on what the php script is doing

I didn't notice your comments about the log before I wrote

livehttpheaders will help too, so you can see exactly what each browser is sending

in any event, looking at the headers sent by the different browsers should give you a better idea

its ok. sorry i was short

I wouldn't think mime type would unless there is an embedded object.

internal server error could be the php process crashing out

undoubtedly
well, actually, maybe not... it could be in the apache config, too, if they're using .htaccess
but probably it's in the php

aye. and if that is the problem, why does PHP phase in the other broswers throughout the building?

the php script may have some code that is dependant on the user agent

both http (or https?) ?

if that is so it wouldn't have phased in Linux FF.

why?

phased?

my bet is that its triggering some IE compatibility code by mistake that is miswritten

the UA contains the OS too in most cases

hmm true, but if you forget to add the FF-UA for windows, why do you remember it for Linux?

its difficult to say as the php script could be doing practically anything

im checking the UA now. brb

incorrect regex in rewritecond falls through to an invalid rewriterule that generates a 500 for example

well one would have thought you wouldnt bother checking the OS in the UA, just he main part like Gecko/ or Firefox/

in english?

but you never know

one would have thought, indeed
but people do the strangest things

incorrect rewriterule seldomly is dependent on browser

just need to look at the daily wtf to see that

and even more seldom generates 500s

is this a website I am nieve about?

you can generate a 500 with a disallowed directive in an .htaccess file. but it takes some work to make that conditional

http://dailywtf.com

bingo. that is it

sorry thedailywtf.com

now to pass the buck
thanks a ton everyone

where's the buck, where's the buck

tell them to stop trying to micromanage the user agent
it's bound to end in tears

with rewriterules being interpreted first, it would be tricky to hide bad syntax in a RewriteCond looking at request headers

impossible

it's not impossible, it's just infinitely improbable.

...as somewhat hockey fan

the directives are parsed first

Is there something (maybe a plugin) that allows to add some HTML stuff to a served (either a static file, or content generated by PHP) text/html and text/xhtml page ?
i would need that to add some advertissement on free hosting

mod_include

mod_include is http://httpd.apache.org/docs/mod/mod_include.html or http://httpd.apache.org/docs-2.0/mod/mod_include.html

mod_filter

mod_filter is http://httpd.apache.org/docs/2.2/mod/mod_filter.html

comedy option, mod_ext_filter

thank you very much

I am hooked.

lol, you can waste hours on there, its such a laugh

where?

thedailywtf.com / worsethanfailure.com

heh

hmm, anyone here use jmeter with mod_deflate?

hola

hola, como estas

errr, good, if i understood you correctly. You?

qué usted dice siempre
well, I don't speak a word of spanish except ordering a beer

haha
that's a little more than I speak then
it's been a quiet day in here today, hasn't it

not as delightfull as with simonko19 but not too busy
karma mcu

mcu has neutral karma

Is there any easy way to check which modules are running under a given apache (either statically compiled in or dynamically loaded)?

httpd -M

httpd -M lists both statically and dynamically loaded modules for apache version 2.2 and higher

check the status page?
pctony++

hehe, ta

or -l for the staticly compiled, previous versions

or check "httpd -l" and LoadModule lines

interesting. I guess I have apache 2.0.

you guess?
httpd -V

httpd -v returns Apache's version number. httpd -V returns compile settings, too.

upgrade?

Old Apaches are harder to support than newer Apaches. Please upgrade. (2.0-2.2 http://httpd.apache.org/docs/2.2/upgrading.html)

but the only way a module gets loaded if it's NOT listed under ./httpd -l is with a LoadModule directive, right?
I doubt it'll be a simple upgrade path. Solaris doesn't have apt-get...

pkg-get
(blastwave, the open source repository for solaris)

whoa.

I know kung-fu...

with solaris, rather compile from the source
packages from sun are not that good

Reading up on blastwave.

rather read httpd.apache.org

all in good time

hmm, I am trying to get the deflate module installed, and it doesn't seem to be working. I am using "--enable-deflate", any hints?

proxypass?

proxypass is http://httpd.apache.org/docs/2.0/mod/mod_proxy.html#proxypass

beer?
lol

nevermind, I got it

pound signs in you eyes?

any references in config.log to deflate or zlib? Do you have zlib headers installed?

no, I just didn't have my config var in the right place.

je ne t'aime plus mon amour
karma lamp?

lamp has karma of 27

seen Daveman?

oh my

i like this one:
fajita, windows

windows broken, please use door

manu chao is replysee http://www.manuchao.net/

OK, lamp.

manu chao?

see http://www.manuchao.net/

nice

fajita, button

See that big red button over there? Right. That One. Never ever press it. Seriously, I'm not joking around.

any Amazon S3 and EC2 users around?

html is such a jackass format

yeah good thing it never caught on.

rofl

grr, mod_perl!

Roobarb-Work: there is here?

hello, how to make https server?

ssl

ssl is *Secure Sockets Layer. Ask me about mod_ssl or ssl vhosts see also http://httpd.apache.org/docs/2.2/ssl/

ssl howto

ssl howto is http://httpd.apache.org/docs-2.0/ssl/

is there a way within the virtualhost to permanently redirect a www to a non-www?

www

www is deprecated (http://no-www.org/)(not deprecated (http://www.yes-www.org/).possibly deprecated (http://ww.maybe-www.org/)

lol

some solutions written up there that should cover your needs

Proceed with the installation of the 3 packages? (0 MB) (Y/n) y
installing libapr_memcache0-0.7.0-8mdv2007.1.i586.rpm apache-mod_ssl-2.2.4-6mdv2007.1.i586.rpm libdistcache1-1.5.1-6mdv2007.1.i586.rpm from /var/cache/urpmi/rpms

mixed amongst the rhetoric should be a tip or two

Actually, no-www is where I got my .htaccess rewrite rule. It removes the www, but also redirects to http from https :|
I got it from here: http://no-www.org/faq.php#how_class_b

so change it to redirect to https
I bet you can guess how to do it.

Well, I really want it to STAY on https if it's https, or STAY on http if it's http

use absolute urls
or relative urls

relative urls is only useful for relative things

but don't use a fully qualified uri

I'm sorry, what?

sorry, is that a fundamental feature

I'm pretty new to this

rewriterule blah /foo/bar

absolute url

rewriterule blah foo

rewriteguide

http://httpd.apache.org/docs-2.0/misc/rewriteguide.html http://httpd.apache.org/docs/misc/rewriteguide.html http://www.engelschall.com/pw/apache/rewriteguide/ http://httpd.apache.org/docs/2.2/rewrite/rewrite_guide.html http://rewrite.drbacchus.com/ http://httpd.apache.org/docs/1.3/misc/rewriteguide.html

relative url

fully qualified uri

but how do you remove the www if you don't use the full uri?

that's the rub
:P
rewritecond

rewritecond is http://httpd.apache.org/docs/2.0/mod/mod_rewrite.html#rewritecond

there's a list of env vars
that you can use in the rewriterule

hi, Roobarb-Work are there here??
hi, Roobarb-Work are u there here??

didn't we have a redirectmatch solution for non-www to www? coulda sworn we did
it's in her brane somewheres

i'm having trouble with CGIs on windows

she-bang line

[ http://pastie.caboo.se/76758 ]

usually the problem i see with windows and cgi's

well, the she-bang line is '#!/usr/bin/env ruby'

apache still needs the right she-bang line for cgi's
and does windows have a /usr/bin/env?
:P

the ruby docs say to use that she-bang line...
do you think they are in error?

that's for *nix

they say to use it for windows

most docs are written in the *nix context
well that would just be plain wrong

well, actually, a tutorial says that

if you're running ruby under mod_cgi

yes, you're probably right
no, just plain cgi

yeah, that's mod_cgi

but the script works when i call it from the dos prompt
oh....

dos prompt != apache running it
apache uses the she-bang line

and dos prompt does not?

nope

ZOMG!11!!!
okay, so i need to turn the shebang line into '#!/where/ruby/is

something like that

did you read the logs i posted? [ http://pastie.caboo.se/76758 ]

#!c:/path/to/ruby

oh, i see

don't need to... not until you put in the right she-bang line

oh, okay.
the logs say, 'system can not find the path...'

there ya go
that == bad she-bang

that's the only reason i asked you to look -- to confirm my suspicion that you were right
thank you

np
happy cgi'ing

winows is stupid like that

I solved my problem. It redirects my .com, .net, and .org all to the .com ...removes the www. ... and keeps the user on http or https (whichever they were on to begin with): http://paste-bin.com/12062

nice job

thanks...I asked for help in here so many times, I hoped someone else could use it.
http://aaron.xavisys.com/redirecting-all-serveraliases-to-a-preferred-domain-with-mod_rewrite/
Nasty long title, but it's accurate

Would this be the correct place to ask questions about mod_rewrite?

i'd ask anyway

That can be dangerous on Freenode :x

people ask about mod_rewrite here all the time

I'm just wondering if there's any way you can re-write everything after the URL, ie: http://www.mysite.com/rewrite-this I've tried: RewriteRule (.*) blah.php?page=$1, but for some reason it's storing "blah.php" in the page variable
I've also tried RewriteRule ^/(.*)$ .., but that doesn't even work aswell
s/aswell/at all

RewriteRule ^(.*)/$ /blah.php?page=$1 [L]
RewriteRule ^(.*)$ /blah.php?page=$1 [L]

tried that aswell :/
I thought it might've been that it was looping back and that's why it was getting "blah.php", so I added the [L] and got the same results
it seems it was X_x
added some checks with RewriteCond checking if the filename existed
and that's what it was doing, redoing the checks, even with [L] present
weird

So add these conditions:
RewriteCond %{REQUEST_FILENAME} !-f
RewriteCond %{REQUEST_FILENAME} !-d

That's what I did xD

cool

It's weird that it redid the checks with [L] specified though
well, thanks for the help

query_string

query_string is a reserved var

qsa

qsa is query string append or a flag for RewriteRule or see the RewriteRule Documentation http://rewrite.drbacchus.com/rewritewiki/Flags_2fQSA

query

The query is the part of the url starting with a question mark. RewriteRule and *Match directives only see the location, not the query. or check http://wiki.apache.org/httpd/RewriteQueryString for mod_rewrite related examples

read that
or both

I should dig out the mod_seo i wrote once
i still think you should do that crap in the script, rather than tying yourself into rewrite knots

is Keepalive better switched off on a dynaic website ?

dynamic? how is it dynamic? like php / cgi? or dynamic dns?

yeah PHP

yeah PHP is nice n' easy but there's a massive trade off with what you can do with it

doesnt matter

I read in a few places that the standard setting for apache 2.x isnt very efficient for serving PHP pages
like KeepAlive at 15 seconds will just kill a faorly busy box

benchmark it yourself. otherwise its premature optimization
and you could be making it worse

Ive benchmarked and found that Ive just got loads of clients hanging around
wiaitng for keepalive to time out

anyone got a good tool to diagnose apache bench results over time?

hey anybody know how I can change the directory to something other than /var/www? I've checked the apache2.conf file and can't find anything.

documentroot

DocumentRoot specifies where apache should look for content. See: http://httpd.apache.org/docs/2.2/mod/core.html#documentroot or http://httpd.apache.org/docs/2.0/mod/core.html#documentroot or default layout.

in deb/ubuntu, look in /etc/apache2/sites-enabled/000-default

k thx

i have a question about using a + or -, I have this-"Options MultiViews -Indexes Includes FollowSymLinks" , is FollowLinks in "-" or in a default "+" , anyone ?

it's enabled
it's not a default + or anything
+ means inherit previous options
- means disable this option in the current scope
so if you had /var/www/html with Options FollowSymlinks and you did /var/www/html/blah with Options +ExceCGI, that means that /var/www/html/blah will have FollowSymlinks and ExecCGI allowed
now let's say /var/www/html/ had Options FollowSymlinks Indexes and /var/www/html/blah had Options -Indexes ExecCGI... that means that /var/www/html/blah will not have Indexes set and will only allow ExecCGI

in short, + means "add this option", - means "remove this option", and leaving out the sign means "i was too lazy to type a character and i'll regret it in the morning"

eh?

it seemed like an easier explanation

leaving it out on all options means these listed options are explicitely allowed in this scope
leaving + out that is
and only these options

leaving it out on all options is ok, as long as you just have one options directive

afair, + is an inheritence operator

yes, so is -

yeah
technically

but it gets confusing if you have a mixture of +/- and unmarked ones

yes
i think the docs tell you not to do that
heh
although 2.2 is supposed to do something sane
and doesn't discourage it

for some definition of sane

yeah
which ironically isn't explained i don't think

it actually would have been way better to have had separate directives

I personally don't use +/-
every location/directory explicitely has their own options

that's the other possibility. list them all.

sometimes determining which options get inherited from where or what actually gets inherited is a pita

can be almost impossible if you have a complex config

indeed

is there a tool to run against the apache server, to spill out current settings for a url or local filesystem path?

hrm...
not that i can think of
just your eyes
and your brain

it's tricky because every module is responsible for it's own configs

yeah

they generally don't release the details of their config info

but a tool like that could be useful...

yes, i agree
there's a debugging tool in the oreilly modules book, sadly out of date
which tracks merge operations, sort of
but it doesn't insert itself into modules' private data

(and not Options +smthing) , means that the feature is inheritably available

you start with a bag of options

it means only that option is explicitely available
to that scope

then apache runs through all the applicable sections, directories, files, locations.

does anybody run a dynamic PHP site with apache ?

I'm trying to use mod_rewrite but I've run into some trouble. I tried the rule "RewriteRule ^(.*)$ $1". I don't think this rule will change the URL. However, when this rule is in effect, the server shows a 404 instead of the desired page. Why?

vhosts etc.
in order

rici, and "+ smthing" means it is not to be available for anything else, correct?

no
there's just a bag of options
if the section says +foo, apache puts foo in the bag
if the section says -foo, apache takes foo out of the bag
if the section says options foo bar,
apache empties the bag and just puts foo and bar in it
after it gets through all the sections, whatever's in the bag are the options
so it's really simple, except for figuring out what order apache applies the sections

rici, what i meant about inheritably available, doesn't mean it getting set for any possible subdirectory context for example..

it is no more complicated than i described it

anyone know how to load apache at startup for ubuntu when installed from source?

secctions, it would be easy

rici, where's this in the doc?

you need to create your own init.d script

apache applies the matching ones in order shortest-to-longest
where it gets complicated is where you mix location and directory and use wildcard patterns

apache.org's source tarball won't provide that for you

sections

sections is http://httpd.apache.org/docs/2.2/sections.html

is where it's described

do you know a site that has a walkthrough for that?

i can give you my piece of shit one
oh but i'm not sure if it'll work for debian
since it's a redhat init.d script

thats ok
ill look online

i understand sections, but it's the + and - with Options

chkconfig blah blah /sbin/service style
rici really gave a good simple explanation of it
take a step back and let it sink in

megaspaz, yes I know.. I'd also like to know where in the doc this is..

sections

sections is http://httpd.apache.org/docs/2.2/sections.html

options

options is http://httpd.apache.org/docs/2.0/mod/core.html#options or http://httpd.apache.org/docs/1.3/mod/core.html#options or see 'Options Context' Don't mix 'relative' and 'absolute' values. See: http://www.onlamp.com/pub/a/apache/2005/09/08/apacheckbk.html

that'll describe the +/-

the sections one is really important

yep
merge both
so you don't need to look at mine then?

particularly "how the sections are merged"

i have this line in a conf file-"Options MultiViews -Indexes Includes FollowSymLinks" , why would -Indexes be there if it was not in the bag in the first place?

it wouldn't
well, there's no need for it

it came from somewhere else
but that's the sort of line you should never put in a config

but that depends on where it's at

that line is wrong wrong wrong

rici, so there is inheritance taking place..

don't try to figure out what it means. fix it.

again it depends

rici, no i'm not trying to figure out-- just understanding why this came bundled in a package like this..

because someone fsked up

if that line you pasted came from /var/www/html/blah but /var/www/html never had indexes, then -indexes is a useless line

IfModule mod_userdir.c
Directory /home/*/public_html
AllowOverride All

AllowOverride All is a really bad idea, because it implies AllowOverride Options, which lets people do stuff like Options FollowSymlinks and Options ExecCGI, even though they are explicitly disabled in the main config.

i pasted jsut 3 lines..
this is an insecure setup of course-- this isn't my setting..
fajita:test

what?

fajita:override

i'm not following you...

i think that -indexes in that particular line is useless

it's also mixed

since the first option is neither + nor -, apache's already emptied the bag by the time it hits it

yep

so it's not doing any more than confusing you and making more work for apache's parser
however, it's a bad bad idea

and rici :P

luckily i touchtype

heh
options

rici, what do you mean making more work for apache's praser? it's in the ifmod section and will be parsed once no?

Mixing Options with a + or - with those without is not valid syntax, and is likely to cause unexpected results.
from the docs

have you run a heavily traffic server with PHP ?

nope

shiznet ?

although you will find a certain cynicism about php in this channel, i think that if you look around,
you'll see an awful lot of sites using it

like php.net
:P

thousands of sites use it

so can I place an .htaccess in ~/public_html or do I have to edit the ifmod section ?

you mean that a lot of people in this channel dislike PHP ?

if you're using that config you pasted, you can do .htaccess since you have allowoverride all

php?

is a scripting 'language' that is enabled by loading mod_php and by adding 'AddHandler application/x-httpd-php .php' or you can look at http://wiki.apache.org/httpd/PHPDownload

blame php

PHP should be presumed to be at fault until conclusively proven otherwise. And even thereafter, if it's convenient.

FollowSymLinks .. and is not working..

doesn't work?

Look buddy, doesn't work is a strong statement. Does it sit on the couch all day? Does it want more money? Is it on IRC all the time? Be specific! Examples of what doesn't work (or the URL) tend to help too, or pastebin the config if that's the problem

i presume it may be my setting, or because the target is on another partition-- (though my id user name is mounted as owner on the target directory)
fajita, php

is a scripting 'language' that is enabled by loading mod_php and by adding 'AddHandler application/x-httpd-php .php' or you can look at http://wiki.apache.org/httpd/PHPDownload

followsymlinks can be tricky

I think theres nothing wrong with php

it's just a tool.

I'm trying to figure out how to use mod_rewrite. Maybe somebody can help me.

so is apache

indeed

From my understanding, "RewriteRule ^(.*)$ $1" will not change the URL
is that correct?

have you used httplite ?

yes, that's correct
you can also use - as the second argument if you want to do nothing
however, you should be aware that mod_rewrite rewrites urls into filepaths
not urls into urls

ok, so without the rule my page loads correctly

although it can get confusing with .htaccess

but when I add the rule in, I get a 404 for the same URL that previously worked

yes, the url doesn't work as a filepath

this code is located in httpd.conf if that makes a difference

read the bit in the rewrite docs about [PT]

ok, thank you
I don't understand why it doesn't work
the rewrite rule is supposed to generate a file path?

yes
that's the point of rewriting
to derive a filepath from a url

sections, mod_rewrite helps out by translating the prefix of the url to the appropriate filepath prefix

outside, in server context, it just expects the rewrite rule to produce a filepath

I guess I don't understand what a filepath is

the path to the file
relative to doc root

how can I find out what the doc root is?

DocumentRoot

not necessarily
you can rewrite to a filepath outside the docroot

can be outside of doc root?

sure, like alias can be
only relative filepaths are relative to docroot

Hmm, never tried that.

you can replace alias with a rewrite rule

sorry for the misinformation Rigdern

so it says that "/Users/rigdern/Sites/somesite" is not found
but I can cd to that path so the folders do exist

I've got a question about mod_rewrite
is it possible to do a rewrite rule specific to a server alias?

yes, put it inside a vhost

if it's an existing absolute path, why does apache give me a 404?

ok...

404

404 is Not Found

I've got a vhost with a server name and several aliases...
where in there would I put the rewrite rule?

oh, you just want it to apply to one of the aliases?
well, that's only slightly trickier
precede the rewriterule with a rewritecond matching on %{HTTP_HOST}

oh!
keen

that's the path you're generating with the rewriterule?

yes

I've never used mod_rewrite, so ... crossing fingers

dunno then.
try enabling rewrite logging

rewritelog

rewritelog is http://httpd.apache.org/docs-2.0/mod/mod_rewrite.html#rewritelog http://httpd.apache.org/docs/mod/mod_rewrite.html#rewritelog. Use this logging to help design and track rewrite; see also rewriteloglevel; and cannot be used from .htaccess

did you check your error log?
maybe it's refusing to follow a symlink

there aren't any symlinks in that path

do you maintain heavy traffic servers ?

not any more

define heavy?

arreyder 1 million page impressions a day

ok then I do, we get than in an hour

you get a million page impressions an hour ?

yes.

on one server ?

no, many servers fronted by a cluster of reverse proxies
most of the applications are not clustered however

ah ok

a few big ones are.

rici, ok, i found out, that using nether + or - makes a new bag explicitly..

Im doing my 1 million impressions a day on one server

congratulations

heh, thanks
what type of content do you serve ?

you name it, I serve it.
did you have some questions, tuning, architecture?
just no mod_rewrite. I already blew one of those tonight

yeah I did, and I guess you sound like the guy who will probably no the answers
heh, dont worry, its not mod_rewrite
Basically I server PHP pages, but each process is 10Mg each
also I have KeepAlive set to on, with a 10 second KeepAliveTime out

I like KeepAliveTimeout 1

the load on my server always hovers around 10

what kind of load?
io wait, cpu?

disk

disk is slower than ram

databse running on same server?
logging to same filesystem as content?

what I was guessing, is that since each http process uses 10Mg to 15meg...MaxClients 500.....KeepAliveTimeout at 10 secs....
I run outta ram and disk swap apache processes

the keepalived connections arent going to eat much ram
so not relavent.

same server yes....although last night I move the db onto a different server and my load is around 0.* to 3.* at peak times on each server

yeah that had to help.
you might consider logging remotely as well

at peak times on the same server I was getting load of like 20+
the logs get wrote to a different RAID array

chances are thought the db was quite a bit of your io wait

it could have been....but the now db server has a load of 0.x as well

still, even to a different array it will cause some wait

I never would have expected database and apache on same server would cause the load to be 10 times greater !

so how is it now? the web server?
what type of db reads? random?

webserver is between 0.* to 3.* at peak times
so it the database server

ram is about all you can do for random reads

db is random reads

I've never had much luck optimising for random reads situations
I just throw memory at them

how do you mean random reads exactly ?

not a lot of cache hits
send the drive heads scurring around
you can go with a higher number of smaller disks

ah right, the cache gets hit quite fairly

more disk heads looking for data == less seek time

yeah, we had like 6 disks in raid 10

more disks would help

What are the is the recommended ways of supporting different SSL hosts for name based virtual hosts? Ports or IPs for each SSL service?

ssl

ssl is *Secure Sockets Layer. Ask me about mod_ssl or ssl vhosts see also http://httpd.apache.org/docs/2.2/ssl/

ssl vhosts

When using SSL, each virtual host must have either its own IP address or its own port. Or both. or for details see http://httpd.apache.org/docs-2.0/ssl/ssl_faq.html#vhosts2 or http://www.onlamp.com/pub/a/apache/2005/02/17/apacheckbk.html or see SNI or consider wildcard certificates or http://wiki.cacert.org/wiki/VhostTaskForce or see ip based vhosts

IP's but if you have wildcards certs can get by with a single address for each wildcarded domain
ssl howto

ssl howto is http://httpd.apache.org/docs-2.0/ssl/

so theres not much benefits from tweaking apache config, in relation to KeepAliveTimeout etc ?

I have that implemented already for wildcards, but now my customers want their own SSL certs for thier own domains
already implmented for a single wildcard.

are you hitting max clients? if not, then no I do not think the keep alive can do much to help you

Ok, looks like I am going IP shopping, thanks.

yep, more IP's

yes, I used to hit max clients before I split the db onto a seperate server
I had max clients of 500, and each http process used 10 to 15 Mg
thats 6.5G

not using keep alives at all will keep you able to serve more requests, but will hurt the user experience a bit and cause a little more overhead for connection builds/closes
I really like keepalivetimeout 1
it's about as a good of a comprimise as there is

is that 1 second after all requests are servered to thet client ?

whats this app?

or 1 second including serving tim eand waiting afterwards

between the end of the response and the next request

really ?

it's almost always enough if they're streaming requests for images or whatever

so if someone clicks a link....say it takes 12 requests to css, images etc, taking 3 seconds to build the page
then it would wait for 1 seond in arreyder 's case

those requests will happen on one connection

its an online game

or two if your browser decides to split them

so whats the point in having keepalivetieout greater than 1 ?
I can't see the point

there was a theory at one time that people tend to click pretty often
and that it's worth holding open the connection for the next click

i set mine to 300.
but I'm also using the event mpm

also, in places where connectivity is lousy, 1 second might not be enough

1 second isn't enough for most of the world

for example, if you're connecting over a satellite link

metro US on broadband, 1 second is fine

browsers should pipeline. i don't know how many do

not enough

but is it not 1 second after the content has been server...ie 1 seond idle

the last byte leaves the server
bounces off a bird

its one second after apache sends the last byte

arrives at the client
the client responds
bounces off the bird again
and reaches the server
that's four ground to sat delays

the client responds, renders, runs 300kb of ajvascript

the birds hang out 40000 km about earth

and then about a second or five later, renders a bunch of DOM nodes.

the speed of light is 300000 km/sec

heh

so thats about 0.6 seconds just in satellite time
not counting the terrestrial links, routing delays, and various etc.

so its got 1 second to reach the client

2.8 :P

yeah, i'm approximating
a typical sat conn rtt from, say, la paz to london is 900ms on the average

you approximate, and I use the wrong word that sounds the same but is spelled differently

but pings are short

higher keepalive is nice, but the best possible situation is to get 100% cache hits for all the extra resoruces.
mod_never_expire

i'm not following you...

mod_buy_a_lot_of_ram

mod_never_expire is http://people.apache.org/~pquerna/modules/mod_never_expire.c

OK, chipig.

hmm? Just version every resource referenced by html.
and then put it all on akamai....
and bingo.

and bingo was his name-o.

Im still lost as to why anyone would have it set more than 1 second

because the cost of setting up a tcp connection over any high latency connection is insane?

I mean after yourve sent your last byte, so what if it closes after 1 second

agore, if i'm in la paz and you're in europe, say, there's not a hope in hell that my browser can send requests less than a second apart
arreyder doesn't have that problem, i believe

yeah, my users are all very close

and if i lose that connection, it takes about 4-5 seconds to set up a new one, minimum
that really slows down page loading if the pages have a lot of graphics on them
i personally recommend 3
although chipig is more generous

sounds reasonable
he's just trying to show off that event mpm

well yeah, 300 sounds very high, must be mostly same users logged in all day

yes :-)

with that silly timeout

and the new patch that lets it work with mod_ssl!

ah yeah....I forgot he uses that

thats what let me turn it all on

hmm, soon I too will have the event mpm

that's cool.

looking forward to it

did they fix ssl?

oh, no.
it just reverts to the Worker MPM behavoirs, when mod_ssl is active

chipig, can you use php modules with it

poo

its not perfectly optimal, but its the only way out for 2.2
and it lets me mix https and http vhosts on a single mpm

what do you do for dns lookups?

which is what always killed the event mpm for me before in production
dns lookups?

I can't find the machine name "lookups?"

asynch?

httpd never does a dns lookup for me...

ah, i suppose not
you're not running any ldap stuff i guess

Im just gonna leave my timout on 10 cos to be honest Im not sure how it works, but thanks anyhow

best bet is split your disks
watch out for log files

but for the crawlers we use lots of djb's dnccache stuff.
although, i've heard good things about newer bind in a pure cache config

i've been looking through libcurl

curl--

it uses c-ares

c-ares--

yeah, it's got it's issues
i know, i hate c-ares
i was thinking of slugging djb's stuff in there

it works. It could be so much better, from an API perspective.

have you looked at the new multi api?
it's actually quite nice

i was following it a bit when it was being written

it does a callback to tell you about sockets it's interested in

as I said, it's not really going to help your problem much anyway, focus on reducing that io wait

and you call it with a socket and event flag

I really want one consistent event system that I can plug into a dozen different libraries

so it integrates perfectly into libevent
or kqueue

kqueue is chipig's pet APR patch or a poll() replacement for FreeBSD that has increaed performance with many sockets or a poll() replacement for {Free,Net,Open}BSD and OS X

yeah. kinda down the path i was going with apr_pollcb

yeah, it's a good general approach i think

if everyone used it, life would be good

well the server splitting sees to have dramitic effects

how much ram do you have, agore?

16Gb

i think in most of our programs, we have lots of apr_pollset() sets from apr_memcache, a couple libevent contexts from stuff, libcurl multi contexts, and sometimes a litle custom aio for our own services.
it sucks.

yeah

yes, if you have gigabit ethernet between them also look at using jumbo frames to keep some overhead down.

i'd like one simple system

9000mtu vs 1500

but if 500 processes are sitting around for 10 seconds each after finishing serving a client...and I have 1500 people hitting the site at once....1000 are waiting for 10 seconds

they're not waiting
they just have a socket open

but there not getting there pages served
and in that time more people are comming along

they aren't hanging on the socket
but yeah, it does use up resources
so lower it
10 seconds is useless

many images with this app?

people don't click that fast, and click streams don't wait that long
3-5 is good

ok
maybe 10 small images per page

it's the count, not the size, really. these days.

i was going to suggest serving the images from another machine

ah ok

or even another apache

or squid

you can set up a stripped down apache to just serve image files

yeah, ive heard that one before...we only have a small amount of images

yeah, or squid.

right

but our experiments with squid and apache 2.2 don't show much difference

yep, my reverse proxy clusters is a lean mean image serving machine.

it would be nice if you could tell apache's cache to expire a resource
that's one thing that squid has

didnt they add a cache manager to 2.2.x?

does it do the expiry protocol?

keep everything static in memory!

i used to think the same way, but really, versioning URLs/images has made stuff so much easier

god, i've got to keep up with the changes better

I havent tried it yet

i don't think thats in there, is it?

lemmie look

brian from CNN talked about it at apacheCon US, the one he wrote for CNN

image changes, name changes?

that used sqlite and other neat stuff
yes
we actually use SVN revision numbers in the URL

yeah, versioning is probably better

/images/r324324/foo.gif

makes the url's ugly

http://httpd.apache.org/docs/2.2/programs/htcacheclean.html
doesnt do much.

and then an AliasMatch /images/r(\d)+/(.*) /foo/bar/images/$2
yeah, it can't purge a specific URL
and none added PURGE support
it can manage the total size of the diskc ache

brian from CNN ?

that sounds nice, I'll remember that when I play with content stuff next time

the hardest part is geting your content generators to do it
but once they do, it makes lots of things easier
like you can use mod_never_expires

brian akins

and export the images to a CDN

actually plone has a unique resource id for every image
that's translated into a real filename when the page is output
i should talk to them about not doing the translation... hmmm...

mm. can't find brian's slides from ACUS 2006.
anyways, they did lots of things to mod_cache to make it better.

yeah

but most never found their way upstream, because CNN.corp is... paranoid about IP

so I enabled logging for mod_rewrite and I think the problem has to do with the difference between URL's and filepaths

perhaps you're right, versioning might be better than purge

I have 2 rewrite rules that generate the same filepath

never heard of Brian Akins

however, in the one that works, mod rewrite says that it is ignoring the rewrite because it equals the path that was passed to the rule
and I also noticed when apache gives me a 404 for the rewrite rule, it's a filepath
but when I get a legitimate 404, apache prints a URL

would you like to show us your rules or do we guess?

I thought it would be more fun to guess first
RewriteRule ^(.*)$ $1
RewriteRule (.*) app/webroot/$1 [L]

hmm, what is the point of the first rule?

that's the only rule I've gotten to work

lol
like rewriting /stuff to /stuff

yea
so for the first one, I type in http://127.0.0.1/~rigdern/cake/app/webroot/index.php

ok. get rid of that. how about the second one, what is problem? does it loop?

no, I get a 404 for a file that exists
what I did was I only used 1 rule at at ime
I typed the above URL when the first rule was being applied
URL: /Users/rigdern/Sites/cake/app/webroot/index.php [IGNORING REWRITE]
because it's the same as the URL before the rewrite
and I see the page in my web browser
so now I comment out rule 1 and apply rule 2
I type in the URL http://127.0.0.1/~rigdern/cake/index.php
so rule 2 should redirect me to the same URL that I got to from rule 1

are you using that in .htaccess? in ~ridgern or what directory?
(public_html or whatever your UserDir is)

I using it in server config file
Directory /Users/rigdern/Sites/cake

/Users/rigdern/Sites/cake/app/webroot/index.php

but when the page loads in my web browser, I see a Not Found error
so both rules rewrite to the same path
but rule 2 gives a 404
and in the log, it says rule 1's output was ignored
so I think the problem has to do with the difference between filepaths and URL's
no ideas?

little bit strange, you could turn your rewritelog to 9 and see if it tells you anything more

it is on 9
RewriteRule ^app/(.*)$ $1
this one also gives a 404 error even though the file does exist
where in httpd.cnf should I be putting this code?

hello?

Hail, kikkomanchow!

I'm reading help.ubuntu about installing apache.. i dunno what this means : Search both the strings starting by "User" and "Group", and change the names by the current username and groupname you are using. Then you'll need to restart Apache. (look at the next chapter concerning apache commands)
its refering to apache2.conf
can anyone explain?

user

user is http://httpd.apache.org/docs/2.0/mod/mpm_common.html#user or /http://httpd.apache.org/docs/mod/core.html#user

make_sock: could not bind to address 0.0.0.0:8080
im getting that when i start apache

did it ever occur to you, it's already running?
either that, or you have duplicated configs

hrmm i stopped it and start it again

netstat

netstat prints network connections, routing tables, interface statistics, masquerade connections, and multicast memberships. Use netstat -an grep -w 80 to check if port 80 is in use. (On linux, you can use netstat -pan to see which program uses the port, or -ltpn.) (For Windows use -ano to show the state, PID, and connection in IP not name format, or -abn).

duplicated configs?
i'm very new to this.. how to i troubleshoot

what distro are you using?
or platform

'What Operating System and apache version are you using?'

Ubuntu Fiesty Fawn?

the dpkg bot is debian specific. ask the ubuntu folks what's the ubuntu way of the same things.
get the factoid?

Apache2 core dumps on solaris.
Now how would I go about debugging this?

gdb?

gdb is the GNU debugger

do you have core dumps?

evening all

ay'

anyone know what these 'internal dummy connections' are?

I ran with -e debug and it seems to barf over fastcgi.
or whatever comes after that.

I've only seen them last I tested something.

I wouldn't know what to do with a core file.

i get a whole bunch of connections from localhost (::1), and they don't close, so Apache just spawns new processes all the time

internal dummy connection

internal dummy connection is http://mail-archives.apache.org/mod_mbox/httpd-users/200604.mbox/%3C4434308F.9030003@googlemail.com%3E or http://wiki.apache.org/httpd/Logs/InternalDummyConnection

imagine that

cheers

if gdb is installed, check the man page. I think it's $ gdb apache core
been ages since I gdb'd a core file

I can't find a core file, actually.
I did a full find.

ah, ulimit commonly prevents them.

removing fastchi seems to have done something.
that's primising.

ulimit -c

hi
I've got mod_auth_pam installed but I can't seem to login
s/login/authenticate/
user 'chris' - not authenticated: Module is unknown

mod_auth_pam

mod_auth_pam is a module that requires access to one's system password file. http://pam.sourceforge.net/mod_auth_pam/ recommends the use of mod_authnz_external as mod_auth_pam is no longer developed/supported

I'm starting it with -D AUTH_PAM

must be RH based?

I've got a shadow group and apache is a memeber

OK, unlord.

bot

Danger Will Robinson!

thanks
I really thought I had this working earlier
I wonder if my /etc/pam.d/apache is borked

Hmm, since when is 'Include' a syntax error?
or, probably, the error messages chain, nvm.

hmm, okay now it authenticates but svn is broken
ahh, apparently I need to move to subversion 1.4
http://forums.gentoo.org/viewtopic-t-567180.html
https://bugs.gentoo.org/show_bug.cgi?id=182453
okay, that should fix it
and it did

how do I check which config files are read?
For some reason, ServerName foo.bar.com in a VirtualHost block is used when I go to just 'bar.com'.
Now how could that happen?

you are speaking in riddles.

Okay, I have a directory with a bunch of scripts. They are all included using a directory Include directive.
All these scripts contain scriptly 1 VirtualHost definition per file.
One of the VHosts has a directive ServerName foo.bar.com
another has ServerName bar.com
Yet when I go to bar.com, I get the stuff I Defined in foo.bar.com

httpd -S

httpd -S tells you what your virtualhost configuration is.

and when I remark the whole foo.bar.com virtual host config file out.. I *DO* get bar.com
odd. they are all listed twice.

this one vhost per file is a nice idea -- but you end up messing stuff up.

It's the default behaviour of debian.
I kind of liked it there so I am (trying to) copy the scheme to solaris here.

into_a_file and pastebin that file.

okay.
one moment.

I'm suspecting one very, very, very simple thing to be happening.
faq1

exactly. For example, NameVirtualHost *:80 must be used with VirtualHost *:80

namevirtualhost

namevirtualhost is http://httpd.apache.org/docs-2.0/mod/core.html#namevirtualhost or http://httpd.apache.org/docs/mod/core.html#namevirtualhost or You must specify only one NameVirtualHost directive, hopefully with an IP address or a *. It says on which interfaces should Apache listen to name based vhosts requests, nothing more.

vhost examples

Look at these for examples http://wiki.apache.org/httpd/Recipes/ExampleVhosts :: http://httpd.apache.org/docs/2.2/vhosts/examples.html :: http://httpd.apache.org/docs-2.0/vhosts/examples.html :: http://httpd.apache.org/docs/vhosts/examples.html