Apache HTTPD 2.0.28 Beta Released

This is a major 2.0 beta release for the Apache HTTPD Project. The last publicly available beta was released in April (and the ChangLog is nearly 500k!). See what the new features are going into the 2.0 release. Please remember to use a mirror.

Good Apache 2.x books?

[BroadbandBradley]

Apache 2.x is looking real good, I mean, is there anything Apache can't do? You Apache Developers are great people!
Thanks for 'A Patchy Server', a real showcase of what people can do when we work together as a community.

Re:Good Apache 2.x books?

[Anonymous Coward]

I asked the Apache team back in April "Can you help me quit my addiction to chewing tobacco" and they said "Boy, there are somethings we can't do."

So there you have it.

OpenBSD

[almeida]

I hope I can get this to work on OpenBSD. I was able to get one of the beta's to compile after a little work, but then it wouldn't actually do anything. The httpd daemon would start up, but it didn't actually listen for connections. Weird stuff.

front page news?

[Anonymous Coward]

this isnt newsworthy enough of the front page? WTF!? every time there's a minor version of mozilla released or minor release of a linux kernel it's front page news. but this a major release of apache - 2.0 beta. what - did the /. editors just think - nah no one's interested, just tuck it where no one will see it.

Re:front page news?

[DarkKnightRadick]

No, they just felt that it was only of real interest to us users who actually use Apache for various reasons. I'm currently running the 1.3.22 version for Win32 on my Win98 box and I must say, this is not of any interest to anyone I know, except a feee geek friends.

Re:New Apache

[GTRacer]

Well, I'll tell you why *I* was interested in this 2.0 beta: The ability to chain perl scripts and have the whole thing SSI'd.

I'd already modularized my pages as far as 1.3 would allow without using mod-perl but I had a few perl "utilities" that I couldn't get to work with my standard INCLUDEd navbar, header or footer scripts.

With 2.0, I can. In fact, I just did.

GTRacer
- Apache and Gimp on Win32...What's wrong with me?

forking

[[e-mail]]

i know, i know, rtfm, but does apache 2.x still serve pages by forking a new process (i know, it preforks, blah blah, still you end up with dozens of processes)? until apache is a multithreaded server like AOLServer [aolserver.com] (don't laugh, it's open source and very, very good) or even a non-blocking IO server like thttpd [acme.com] it is just unusable for truly scalable or database-centric stuff. i love all the mods for apache, and would love the apache API to evolve to a single process, advanced server instead of the 'patchy' mess on top of NCSA it started as.

for background on why things like non-blocking IO or multithreaded instead of forking is good, check out thttpd's section on non-blocking io [acme.com] or Philip Greenspun's own Introduction to AOLServer, part 1 [greenspun.com].

-sam

Re:forking

[the way]

i know, i know, rtfm, but does apache 2.x still serve pages by forking a new process (i know, it preforks, blah blah, still you end up with dozens of processes)?

Yes, it does. There are a range of choices when you compile, but the threaded MPM is the default. There is also a Win32-specific multi-threaded engine.

until apache is a multithreaded server like AOLServer [aolserver.com] (don't laugh, it's open source and very, very good) or even a non-blocking IO server like thttpd [acme.com] it is just unusable for truly scalable or database-centric stuff

Well, it's not an issue now with Apache 2, but the above statement is simply not true. I can tell you that from experience running a scalable database-centric web application. The trick is to put a http accelerator / reverse proxy up front. It's easy to do and works very nicely.

Re:forking

[the way]

Gee that came out wrong! I meant "Yes, it has now been changed to be multi-threaded", but it sounds like I meant the opposite!

To make it clear--Apache 2 is now multi-threaded by default on Win32 and Unix.

Re:forking

[Jerenk]

Yes, it does. There are a range of choices when you compile, but the threaded MPM is the default. There is also a Win32-specific multi-threaded engine.

The one correction I'd like to make is that we disabled the threaded MPM by default after 2.0.16 was released. And, we've actually removed the threaded MPM *after* we released 2.0.28. The threaded MPM has been superceded by the "worker" MPM (our names aren't the greatest). The "worker" MPM is essentially threaded (but has multiple processes to increase redunancy) - you'll see us call this a MPMT (multi-process/multi-thread) MPM.

There is a discussion on-list about whether we should enable threading MPM by default if the OS supports it. It's an open question, but for now (and what is in 2.0.28), we use prefork by default. A configure-time option allows the selection of which MPM it should compile with.

FWIW, special platforms (like Win32) have their own MPM that is specific to their OS. The "worker" MPM is targeted towards UNIX-like platforms (Linux, Solaris, etc.).

Filtering

[Graymalkin]

Yay for the addition of filtering, it's on of the strengths of ISAPI is the ability to set up filters on either incoming or outgoing data. To some people this seems like crap but if you're scripting in an OO language it's cool to have a really simple filter than can objectify POST and/or GET requests. Or you can jam the output of stuff into an SSI and then nobody has to worry about funky shit inside HTML.