Mailing List Archive: Hi & Segmentation fault

Hello cherokee community!

I have recently started using cherokee and I am a big fan so far,
however my first venture into the mailing lists is a support request :(

I am primarily a web developer, with very limited sys admin skills so I
used the cherokee docs to set up PHP and cherokee.

Currently cherokee serves static files and one PHP file, this PHP file
then queries an external service, it is hammered by AJAX requests. I
have the results of this external service stored with memcache for
around 30 seconds. However at peak times it seems that the system
slowly grinds to a halt, and 504 errors are shown. The HTML/JavaScript
etc are still as fast as ever though, I'm assuming this is because of
cherokee's internal cache?

Here is a message in /var/mail/root:

Subject: Segfault in Cherokee

The Cherokee 'panic action' script, /usr/share/cherokee/cherokee-panic,
was called for pid 7649 ().

This means there was a problem with the program, such as a segfault.
However, the executable could not be found for process 7649.
It may have died unexpectedly, or you may not have permission to
debug the process.

Here are a few errors in the cherokee.error file:

Couldn't create FastCGI listen socket on port 127.0.0.1:47990
Cannot bind/listen socket - [98] Address already in use.
Couldn't create FastCGI listen socket on port 127.0.0.1:47990
Cannot bind/listen socket - [98] Address already in use.
Couldn't create FastCGI listen socket on port 127.0.0.1:47990
[09/11/2011 21:40:48.476] (error) fdpoll-epoll.c:140 - epoll_ctl: ep_fd
6, fd 351: 'No such file or directory'

I have tried googling around and not really found anything that helps.

I am running Debian 6 and Cherokee 1.0.8 with PHP 5.3.8 from dotdeb.

Would anyone be able to help out with this issue, is this a
misconfiguration on my part or just the VPS not being powerful enough
for the amount of traffic that it gets? (I disabled RRD stats as I
thought that this would be a factor so I cannot give any real figures to
hits etc, however I know that the bandwidth usage has been around 20Gb+
for 1 and a half weeks worth of usage)

Thanks in advance.

John.

Based on the FastCGI error, I assume you're using the normal PHP FastCGI.
Try using php-fpm instead, as it's a separate daemon designed to handle
spawning of the PHP FastCGI processes (and works quite well). It might be
included in the php5-cgi package, or it might be in a separate "php5-fpm"
package (I haven't used the Dotdeb packages in a while).

Once installed you should be able to do */etc/init.d/php5-fpm start* and
get Cherokee to use its port (which is port 9000 by default). Set the
"interpreter" in the Cherokee configuration to "false" as it's no longer
needed (php-fpm handles the spawning).

On Thu, Nov 10, 2011 at 7:18 AM, John <john@phpremium.com> wrote:

> Hello cherokee community!
>
> I have recently started using cherokee and I am a big fan so far, however
> my first venture into the mailing lists is a support request :(
>
> I am primarily a web developer, with very limited sys admin skills so I
> used the cherokee docs to set up PHP and cherokee.
>
> Currently cherokee serves static files and one PHP file, this PHP file
> then queries an external service, it is hammered by AJAX requests. I have
> the results of this external service stored with memcache for around 30
> seconds. However at peak times it seems that the system slowly grinds to a
> halt, and 504 errors are shown. The HTML/JavaScript etc are still as fast
> as ever though, I'm assuming this is because of cherokee's internal cache?
>
> Here is a message in /var/mail/root:
>
> Subject: Segfault in Cherokee
>
> The Cherokee 'panic action' script, /usr/share/cherokee/cherokee-panic,
> was called for pid 7649 ().
>
> This means there was a problem with the program, such as a segfault.
> However, the executable could not be found for process 7649.
> It may have died unexpectedly, or you may not have permission to
> debug the process.
>
> Here are a few errors in the cherokee.error file:
>
> Couldn't create FastCGI listen socket on port 127.0.0.1:47990
> Cannot bind/listen socket - [98] Address already in use.
> Couldn't create FastCGI listen socket on port 127.0.0.1:47990
> Cannot bind/listen socket - [98] Address already in use.
> Couldn't create FastCGI listen socket on port 127.0.0.1:47990
> [09/11/2011 21:40:48.476] (error) fdpoll-epoll.c:140 - epoll_ctl: ep_fd 6,
> fd 351: 'No such file or directory'
>
> I have tried googling around and not really found anything that helps.
>
> I am running Debian 6 and Cherokee 1.0.8 with PHP 5.3.8 from dotdeb.
>
> Would anyone be able to help out with this issue, is this a
> misconfiguration on my part or just the VPS not being powerful enough for
> the amount of traffic that it gets? (I disabled RRD stats as I thought that
> this would be a factor so I cannot give any real figures to hits etc,
> however I know that the bandwidth usage has been around 20Gb+ for 1 and a
> half weeks worth of usage)
>
> Thanks in advance.
>
> John.
>
> _______________________________________________
> Cherokee mailing list
> Cherokee@lists.octality.com
> http://lists.octality.com/listinfo/cherokee
>
>