xgrr.de – the whole not the half

For years now I was using PHP as a Apache module and not thinking twice about it. Easy to install not much to maintain and working reasonably well. Server too slow? Buy a new server. Recently one of our servers got really slow again with varying timeouts while serving pages. Memory shortage. No new server available I needed to make the most of it and started investigating how to decrease the memory footprint of Apache.

FastCGI was the answer. With PHP as a module installed every Apache process can take up to the maximum amount of memory you defined in php.ini plus what the Apache process needs by himself even though this process is only serving static content. With the switch to FastCGI it was possible to use the threaded version of Apache (mpm_worker). Static contents are now served really fast without need for invoking PHP at all. PHP scripts are served by calling the FastCGI IPC.

As memory is still low Apache and PHP are both configured to “die” after quite a low number of requests served. This can sometimes lead to a small delay while the processes are restarted but ensures that the memory footprint is kept low.

How did I do it? I won’t go into details on how to install Apache, mod_fcgid and PHP because there are loads of howto’s out there. Please find the configuration I’m using below. I’m using Debian and Apache, mod_fcgid and PHP are installed out of the repository. If you have a custom compiled version or different distribution your paths can differ.

Depending on your available hardware it is feasible to tweak the settings for the mpm_worker module so more concurrent clients can be served and/or the amount of requests which is handled before re-creating the thread is higher. It is also very recommended to use eAccelerator in conjunction with this setup. See my post on the topic for more information.

/etc/apache2/apache2.conf (excerpt):

<IfModule mpm_worker_module>
	StartServers           4
	ServerLimit            4
	MaxClients           128
	MinSpareThreads        8
	MaxSpareThreads       16
	ThreadsPerChild       32
	MaxRequestsPerChild  500
</IfModule>

/etc/apache2/mods-available/fcgid.conf:

<IfModule mod_fcgid.c>
	AddHandler fcgid-script .fcgi .php
	FCGIWrapper /var/www/php-fcgi-starter .php

	IdleTimeout		3600
	BusyTimeout		 300
	ProcessLifeTime		7200
	IPCConnectTimeout	  10
	IPCCommTimeout		 360
	MaxProcessCount		  15
	MaxRequestsPerProcess	  -1
	PHP_Fix_Pathinfo_Enable    1
</IfModule>

/var/www/php-fcgi-starter:

#!/bin/sh
PHPRC=/etc/php5/cgi/
export PHPRC
export PHP_FCGI_MAX_REQUESTS=250
export PHP_FCGI_CHILDREN=1
exec /usr/lib/cgi-bin/php

In my series of optimizing one of our servers I added eAccelerator to the setup to improve the performance. One of the bottlenecks of the server is unfortunately memory. Nevertheless even with small memory settings and caching to disk eAccelerator is able to enhance the performance and user experience of our websites and eCommerce platforms by 2-2.5.

For some projects it might come in handy to have a pre-compiled binary shipped to the customer or run as a separate instance. I took a peek at Roadsend PHP which implemented their own engine and is able to compile PHP source into C binaries which can either run on the command line or with special compilation arguments as a FastCGI program.

But before using it productively and changing a lot of source to make it compatible with the way how Roadsend works I wanted to make sure that the promised performance bonus would be really there. I created a very crude script which does nothing more than to iterate an integer to a certain maximum and measures the time to do this.

NOTE: I know that a proper performance test looks differently and I’m aware that Roadsend PHP can put his design into the works when we deal with a lot of includes. I’ll make another test with including random files to compare if this is impacting the results.

<?php
echo “Simple iteration test to compare performance between interpreted PHP and Roadsend PHP<hr />”;

$max = 1000;
if($_GET["max"] != “” && is_numeric($_GET["max"])) {
$max = $_GET["max"];
}

$start = microtime();

for($i = 0; $i < $max; $i++) {
echo $i.”<br />”;
}

$stop = microtime();
$elapsed = $stop – $start;
echo “<hr />Script took $elapsed seconds to execute”;
?>

To the environment. I didn’t set up a clean room environment. I used my server at home with 2G of memory Athlon64 X2 and a RAID5. Nothing special but I think it resembles real-world situations perfectly (unless you setup a new box for every PHP site you build).

Apache (worker – threaded) is used for serving the requests. FastCGIs are served via mod_fcgid. PHP is also run via mod_fcgid which improved the performance of PHP dramatically so far.

Let’s get to it:

max = 1000
Roadsend PHP: 0.005817
Interpreded PHP: 0.001201

max = 10000
Roadsend PHP: 0.06846
Interpreded PHP: 0.012625

These results are not dramatically but significant. It seems that compiling PHP adds overhead to the execution.

When I was playing around with Funambol last week I was brutally stopped by an Exception thrown somewhere in the code of the Funambol DS Server.

java.lang.NullPointerException
at com.funambol.server.session.SyncSessionHandler.processInitSyncMapMessage(SyncSessionHandler.java:905)
at com.funambol.server.session.SyncSessionHandler.processMessage(SyncSessionHandler.java:521)
at com.funambol.server.engine.SyncAdapter.processInputMessage(SyncAdapter.java:533)
at com.funambol.server.engine.SyncAdapter.processXMLMessage(SyncAdapter.java:254)
at com.funambol.transport.http.server.LocalSyncHolder.processXMLMessage(LocalSyncHolder.java:97)
at com.funambol.transport.http.server.Sync4jServlet.doPost(Sync4jServlet.java:399)
at javax.servlet.http.HttpServlet.service(HttpServlet.java:637)
at javax.servlet.http.HttpServlet.service(HttpServlet.java:717)
at org.apache.catalina.core.ApplicationFilterChain.internalDoFilter(ApplicationFilterChain.java:290)
at org.apache.catalina.core.ApplicationFilterChain.doFilter(ApplicationFilterChain.java:206)
at com.funambol.transport.http.server.LogContextFilter.doFilter(LogContextFilter.java:132)
at org.apache.catalina.core.ApplicationFilterChain.internalDoFilter(ApplicationFilterChain.java:235)
at org.apache.catalina.core.ApplicationFilterChain.doFilter(ApplicationFilterChain.java:206)
at com.funambol.transport.http.server.SyncResponseTimeFilter.doFilter(SyncResponseTimeFilter.java:159)
at org.apache.catalina.core.ApplicationFilterChain.internalDoFilter(ApplicationFilterChain.java:235)
at org.apache.catalina.core.ApplicationFilterChain.doFilter(ApplicationFilterChain.java:206)
at org.apache.catalina.core.StandardWrapperValve.invoke(StandardWrapperValve.java:233)
at org.apache.catalina.core.StandardContextValve.invoke(StandardContextValve.java:191)
at org.apache.catalina.core.StandardHostValve.invoke(StandardHostValve.java:128)
at org.apache.catalina.valves.ErrorReportValve.invoke(ErrorReportValve.java:102)
at org.apache.catalina.core.StandardEngineValve.invoke(StandardEngineValve.java:109)
at org.apache.catalina.connector.CoyoteAdapter.service(CoyoteAdapter.java:293)
at org.apache.jk.server.JkCoyoteHandler.invoke(JkCoyoteHandler.java:190)
at org.apache.jk.common.HandlerRequest.invoke(HandlerRequest.java:291)
at org.apache.jk.common.ChannelSocket.invoke(ChannelSocket.java:769)
at org.apache.jk.common.ChannelSocket.processConnection(ChannelSocket.java:698)
at org.apache.jk.common.ChannelSocket$SocketConnection.runIt(ChannelSocket.java:891)
at org.apache.tomcat.util.threads.ThreadPool$ControlRunnable.run(ThreadPool.java:690)
at java.lang.Thread.run(Unknown Source)

I decided to make the effort and have a look at the code. That was actually what was taking me so long to fix the problem at hand. Today I got a fix on the actual problem. The offending code is in the handling of the session. When I configured my Funambol I supplied the URI of the service I want to create which is at this point in time not the real URI. What happens during the authentication handshake is that the session is bound to this specific URI. This means that if the client requests a specific URI the server overwrites this URI with the one set in Funambol.xml and the session is only valid for this URI.

During sync it is checked if the client is authenticated which is not the case or at least not for the URI specified in the Funambol.xml. In this case a clause in the code is reached which provokes the NullPointerException stated above.

If you have this problem, make sure that the URI in your Funambol.xml file is either empty (then the URI provided by the client is taken) or that your clients use the URI which is specified.

As an early adopter I am using Thunderbird 3 for quite a while now. The Betas have been really stable and other than a few minor glitches I had no problems.

I just saw that RC 2 is out there after RC 1 was released in the last week. Be careful with the release. My copy was crashing every time I tried to send an email. I couldn’t figure out the exact problem yet but not being able to send messages is quite inconvenient in my opinion.

I came across this nice picture. Just another reason to love Firefox!

I really find it astonishing what impact music can have. Last winter during the time I was concocting my BSc thesis I was very intensely listening to Trouble Over Tokyo. As a matter of fact it was playing almost all the time, during jogging writing and in the car.

This summer I didn’t listen to it at all. Probably because I couldn’t stand it anymore due to too much listening during winter. Now it is winter again though and suddenly I have the urge to play it again.

And then when I finally find the time to tune in I have this strong sense of déjà vu. You almost can remember the emotional status and feelings you had before some place else and in a different time.

This is not always preferred… For instance I can’t listen to System of A Down – Toxicity anymore. During the time I listened to it very intensely I was splitting up with my then girlfriend. Still every time I hear the song I can remember the exact moment back then.

This perfect example of Austrian-German relations found today in Bludenz, Austria.

Just found this passing by. Enjoy!

In professional areas virtualization of servers and workstations enables us to evaluate operating systems and software packages in an isolated environment and without great costs. But this is still the starting point of virtualization. As our current internet topology and software applications ask for integration – the Cloud is a good keyword here – we have to be not only able to virtualize physical PC hardware but also provide network environments which are abstracted from our current internet.

This would enable network engineers, system administrators and of course scientists to easily develop new network infrastructures, test distributed deployments or invent new protocols on top of the network stack. The impact of a commercial virtualization technology for networks would have a huge impact of the way we currently build networks.

With the development of NGN (Next Generation Networks) this idea is pushed by a few selected but until now it did not surface to the broad public.

Update: I made a bit of research on the topic. Nothing deep really but I still found some interesting material:

• Network Virtualization: a strategy for de-ossifying the internet
• Cisco gives virtual servers virtual networking

I also found a lot of sites talking about network virtualization in connection with classic computer virtualization. This is only a part of what is meant. Network virtualization should be an end to end implementation with virtual nodes on the end points itself in the form of a software or firmware on the NIC and also on the nodes in between. It has not necessarily be linked to virtualized servers although this would of course make a lot of sense.