G-Loaded Journal » Errors http://www.g-loaded.eu An open-source software and technology related journal Mon, 05 Dec 2011 19:55:24 +0000 en hourly 1 http://creativecommons.org/licenses/by-nc-sa/3.0/ Issues with the feeds are now resolved http://www.g-loaded.eu/2010/03/28/issues-with-the-feeds-are-now-resolved/ http://www.g-loaded.eu/2010/03/28/issues-with-the-feeds-are-now-resolved/#comments Sun, 28 Mar 2010 06:06:46 +0000 George Notaras http://www.g-loaded.eu/?p=1634 This is just a quick notice that during the last five days there was a problem with the website feeds. The web server returned a 500 Internal Server Error to almost all requests for /feed/ URLs. Also, there was a big increase of the server’s CPU load behind the scenes, which was caused by php-cgi processes. It turned out that a custom plugin (unpublished) I had written for WordPress in the past was the cause of all the trouble, but, before coming to that conclusion, the fact that the problems had started without me doing anything on the blog during the last weeks together with a series of other random observations led me do a thorough examination of the whole server just in case.

So, after spending two days in front of a Linux terminal, I came to the conclusion that the following code had been problematic from the beginning and it was like a time-bomb for which the time had come to cause issues to the web site and the server that hosted it.

function rsse_add_related_posts($PostBody) {
  if (!rsse_verify_feed()) { return $PostBody; }
  if (function_exists('yarpp_related')) {
    $PostBody .= yarpp_related(array('post'),array(),false,false,'rss');
  }
  return $PostBody;
}
 
add_filter('the_content', 'rsse_add_related_posts', 252);

This code used a function of another plugin to attach a ‘related posts‘ section at the end of the full content of each feed entry. As soon as I commented out the above filter and started attaching the related-posts section to the feed content through the YARPP plugin’s administration interface, everything was normal again. No more php-cgi processes timing-out or over-consuming the CPU cycles.

The following graph shows clearly the problematic period:

php-cgi processes consuming the CPU

php-cgi processes consuming the CPU

The httpd error_log included many errors like the following:

[...]
Allowed memory size of 33554432 bytes exhausted (tried to allocate 18893 bytes)
[...]
mod_fcgid: process ... exit(communication error), terminated by calling exit(), return code: 1
[...]
[warn] (104)Connection reset by peer: mod_fcgid: read data from fastcgi server error.
[...]
[warn] mod_fcgid: stderr: PHP Fatal error:  Allowed memory size of 33554432 bytes exhausted (tried to allocate 4864 bytes)
[...]
Premature end of script headers: index.php
[...]

Note that these errors might appear in several occasions.

In my case, finding that faulty piece of code was a huge relief, I can tell you.

Issues with the feeds are now resolved, unless otherwise expressly stated, is licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported License. Terms and conditions beyond the scope of this license may be available at www.g-loaded.eu.

Related Articles

]]> http://www.g-loaded.eu/2010/03/28/issues-with-the-feeds-are-now-resolved/feed/ 0 http://creativecommons.org/licenses/by-nc-sa/3.0/ Strange mod_dav_svn error http://www.g-loaded.eu/2009/10/05/strange-mod_dav_svn-error/ http://www.g-loaded.eu/2009/10/05/strange-mod_dav_svn-error/#comments Mon, 05 Oct 2009 20:54:10 +0000 George Notaras http://www.g-loaded.eu/?p=1213 mod_dav_svn lets you serve subversion repositories through the Apache HTTP Server. mod_dav.so is a requirement for mod_dav_svn.so. If while setting up mod_dav_svn you see the error “undefined symbol: dav_register_provider“, make sure you check the order that the aforementioned modules are loaded. mod_dav must precede mod_dav_svn:

LoadModule dav_module modules/mod_dav.so
LoadModule dav_svn_module modules/mod_dav_svn.so

This will resolve the issue.

Strange mod_dav_svn error, unless otherwise expressly stated, is licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported License. Terms and conditions beyond the scope of this license may be available at www.g-loaded.eu.

Related Articles

]]> http://www.g-loaded.eu/2009/10/05/strange-mod_dav_svn-error/feed/ 0 http://creativecommons.org/licenses/by-nc-sa/3.0/ When it comes to error messages… http://www.g-loaded.eu/2007/07/29/when-it-comes-to-error-messages/ http://www.g-loaded.eu/2007/07/29/when-it-comes-to-error-messages/#comments Sat, 28 Jul 2007 23:00:42 +0000 George Notaras http://www.g-loaded.eu/2007/07/29/when-it-comes-to-error-messages/ This post is both a tip about vsftpd‘s configuration and an example of what an error message should not look like.

If you use vsftpd’s user_config_dir directive in order to set a directory that will hold configuration files for per-user customized server options, be warned that, if one of those files contains a directive whose value is a path that does not exist – for example a mistyped custom local_root -, you will not be able to connect to the FTP service with any user account. Finding the cause of the problem is going to be very difficult as there is no sign about errors in the log files, even if you have set their detail to the maximum.

Moreover, if you use an FTP client like gftp, be prepared for error messages like the following:

Error: Could not read from socket: Success

Yeah, a successful error message at last!

When it comes to error messages…, unless otherwise expressly stated, is licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported License. Terms and conditions beyond the scope of this license may be available at www.g-loaded.eu.

Related Articles

]]> http://www.g-loaded.eu/2007/07/29/when-it-comes-to-error-messages/feed/ 0 http://creativecommons.org/licenses/by-nc-sa/3.0/ Application Testing: Zero Free Space http://www.g-loaded.eu/2007/02/20/application-testing-zero-free-space/ http://www.g-loaded.eu/2007/02/20/application-testing-zero-free-space/#comments Tue, 20 Feb 2007 05:41:54 +0000 George Notaras http://www.g-loaded.eu/2007/02/20/application-testing-zero-free-space/ what went wrong when I run out of free space".]]> I should clarify from the beginning that this is not a technical article containing any kind of debugging information – I don’t have the knowledge anyway – or any other type of proof that indicates that some commonly used applications have bugs. This is just about some observations, or better an unofficial user report, of “what went wrong when I ran out of free space“.

I had written a script which recorded information to a log file the other day. I had forgotten to turn it off and the result was that my home directory became full of data, leaving zero free space. This came at a moment while I was using firefox (v2.0.0.1) and gftp (v2.0.18) on a GNOME (v2.16.3) Fedora 6 desktop. Of course, as soon as I realised that my home was full of data, I deleted that huge log file. In the next moments, I noticed that the things I outline below had gone wrong during this incident. I have not tried to reproduce these errors, so I haven’t filed any bugs. Feel free to do so, if you are able to reproduce them.

First, I do not remember being warned about running out of free space. GNOME is supposed to do this, but, in my case, either it did not raise that warning message or I was too focused on my work so I didn’t notice it. I’d bet that the latter is what actually happened.

Second, I have setup several profiles in firefox and I have adjusted its launcher in the GNOME menu to start the profile manager each time I want to use the browser. Firefox keeps the profile information in the ~/.mozilla/firefox/profiles.ini file. Under the “no free space” conditions, something had caused the file to be re-created as a zero-size file, losing all of my profile information. Fortunately, this was the only file that was affected. My profile directories were in tact, so it was rather easy to recover from this situation by setting my profiles again with the profile manager.

The third observation has to do with the FTP client, gftp. As I have mentioned earlier I was also working with this client when I run out of free space. It is strange to me how some things happen sometimes, but at that moment I was adding some ftp sites to the gftp bookmarks. Well, it seems that if you happen to save your bookmarks or settings in gftp while there is no free space in the partition that holds gftp’s config files, you are LOST! All the files in ~/.gftp/ are replaced with zero-size files and all your settings and bookmarks are lost, unless you have a recent backup of them. Well, I didn’t have one and that’s why I was angry (against myself). This was fixed – but not my bookmarks :-( – by completely removing the ~/.gftp/ directory and getting it recreated when launching gftp.

After this incident, I kept wondering if there were more configuration files in my user account’s home that were affected. I guess it is impossible to know in advance. Time will show.

An application which acted absolutely professionally and actually was the one that made me realize that there was something wrong with the free space is Evolution. It displayed proper warning messages about not being able to write the newly fetched email and news messages to the disk and, generally, the problematic situation had absolutely no effect on it. I take my hat off to Evolution (v2.8.3).

This incident made me think if software developers ever test how their software would behave if there is no free space available, especially if its usage involves reading/writing/appending from/to files. I would bet that other programs might have problems under such conditions too, but I guess these situations are rare. On the other hand, if it happens and the software is not designed to work well, then your personal data could be at risk. At this point, I am rather happy I was not running my password manager at that critical period of time! ;-)

Anyway, as I have already written earlier, I blame noone. Unfortunately, I do not have the required free time or will to reproduce those errors in a test environment, but I guess I will do this at some later time. All the above might just be wrong conclusions and something else might was the cause of the problem. I really do not know, but I felt like writing about that experience.

Application Testing: Zero Free Space, unless otherwise expressly stated, is licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported License. Terms and conditions beyond the scope of this license may be available at www.g-loaded.eu.

Related Articles

]]> http://www.g-loaded.eu/2007/02/20/application-testing-zero-free-space/feed/ 0 http://creativecommons.org/licenses/by-nc-sa/3.0/ TrueCrypt on 2.6.18 kernels http://www.g-loaded.eu/2006/11/05/truecrypt-on-2618-kernels/ http://www.g-loaded.eu/2006/11/05/truecrypt-on-2618-kernels/#comments Sun, 05 Nov 2006 03:52:40 +0000 George Notaras http://www.g-loaded.eu/2006/11/05/truecrypt-on-2618-kernels/ It seems that there is a problem compiling the Linux version of Truecrypt v4.2a on the 2.6.18 series of kernels. A patch has been posted in the Truecrypt forums (you need to be logged-in to read), but, although this patch resolves the compilation error, there have been reports that the program has problems unmounting the encrypted volumes and releasing the loop devices. So, truecrypt is a no-go for Fedora 6 at the moment, at least until a new version that works flawlessly is released. An alternative way to create encrypted file containers pretty much like Truecrypt is by using dm-crypt and LUKS. More information about this method, can be found in another article of mine.

TrueCrypt on 2.6.18 kernels, unless otherwise expressly stated, is licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported License. Terms and conditions beyond the scope of this license may be available at www.g-loaded.eu.

Related Articles

]]> http://www.g-loaded.eu/2006/11/05/truecrypt-on-2618-kernels/feed/ 1 http://creativecommons.org/licenses/by-nc-sa/3.0/