Lighttpd

I don't see a problem.

I have the very same problem,

lighttpd-1.4.20 (ssl) - a light and fast webserver on freebsd 7
92.115.105.143 handle-req 99890 .....md.com /takeupload.php (/takeupload.php) /home.........md/takeupload.php
89.28.47.48 handle-req 99286 .....md.com /details.php?id= (/details.php?id=480083) /home.........md/details.php
86.106.247.97 handle-req 92583 www......md.com /search.php?search_str= (/search.php?search_str=dedication) /home.........md/search.php
212.56.209.26 handle-req 92583 .....md.com /download.php?returnto=... (/index.php) /home.........md/index.php
92.114.235.252 handle-req 92102 .....md.com /details.php?id=487435&viewcomm=4766665 (/details.php?id=487435&viewcomm=4766665) /home.........md/details.php
89.40.113.97 handle-req 92100 www......md.com /download.php (/download.php/489015/v...) /home.........md/download.php
92.115.77.149 handle-req 92104 .....md.com /search.php?search_str=2008&page=1 (/search.php?search_str=2008&page=1) /home.........md/search.php
...............

It's a server behind a lighttpd(1.4.20 freebsd 6.1) proxy (proxy module). The proxy server don't have such a problem. Maybe it's because of a php scripts I don't know (I have max-execution time 30 in php so it shouldn't be it).

The problem is that they never go away, and eat the cpu, and I have no other choice than to restart lighttpd every some minutes.
Another problem is when the workers is down(or maybe when it's not fast enoguth to process everything) lighttpd proxy for some reason start eat all the cpu.

If somebody can help me do a patch to not allow handle-req to wait so long, for example if it's more than 30 second, then die. I have though server.max-write-idle=30
server.max-read-idle=3 but it doesnt seem to help, and I don't find a param ModFastCGI to put some max-execution time on processes..

Is this still valid for the current version?

This is still an issue in 1.4.24. Using fastcgi with PHP 5.3.1, I get ~3 dozen handle-req building up every day (from about ~1mil total requests per day). Server crashes almost every day as well, at peak time.

Related to bug #575 ?

I can give you guys any more information you need.

php-cgi processes are created via spawn-fcgi.

fastcgi.server = (
    ".php" => ((
        "socket" => "/var/www/sockets/sock.nobody",
        "idle-timeout" => 300
    ))
)

What do you mean with "server crashes"? segfault? backtrace? What kind of requests are stuck in "handle-req" ? POST/GET/...?

@samkline, @adrenaline, @Anonymous, while I know this is old, if you're still seeing this issue, I'd like to dig into it. Would you be willing to test some patches?

Submitted pull request https://github.com/lighttpd/lighttpd1.4/pull/53

For those seeing the issue reported in this ticket, please test the patch and report back if the patch resolves the issue. Thank you.

I believe this issues has since been fixed in historical commits, potentially this one:

commit d00e1e79b94e0f4da35292d2293595dac02993c7
Author: Stefan Bühler <stbuehler@web.de>
Date:   Sat Feb 7 13:32:54 2015 +0000

    [connections] fix bug in connection state handling

      if a request was finished (con->file_finished = 1) and the state
      machine was triggered, but the write queue was empty, it didn't
      actually finish the request.

    From: Stefan Bühler <stbuehler@web.de>

    git-svn-id: svn://svn.lighttpd.net/lighttpd/branches/lighttpd-1.4.x@2973 152afb58-edef-0310-8abb-c4023f1b3aa9

Other commits related to closing connections: svn r2636 (476c5d48) and svn r2645 (20c4cd55)

See #1829 for further discussion, and please re-open this ticket if you are still seeing this issue.