Gandi, CMSMS1.11.9, and intermittent errors

[JohnnyNM]

I'm using Gandi.net to host on a Simple Hosting Service.
PHP Version 5.4.4-14+deb7u3
Apache/2.4.6 (Unix)
MySQL 5.5

I get the following error, and I'm really stumped! It's intermittent, sometimes and sometimes not related to the pages I'm working on and frankly unpredictable! Does anyone here have any ideas?

[Thu Oct 03 01:52:24.579452 2013] [:error] [pid 2961:tid 3368393058048] (104)Connection reset by peer: [client 151.226.44.23:36544] FastCGI: failed to read from backend server

There's no .htaccess files in the root so I can't have messed anything up in that.
The only thing different between installations on this server and others I've tried is it doesn't allow the use of the following php commands:

pcntl_alarm, pcntl_fork, pcntl_waitpid, pcntl_wait, pcntl_wifexited, pcntl_wifstopped, pcntl_wifsignaled, pcntl_wexitstatus, pcntl_wtermsig, pcntl_wstopsig, pcntl_signal, pcntl_signal_dispatch, pcntl_get_last_error, pcntl_strerror, pcntl_sigprocmask, pcntl_sigwaitinfo, pcntl_sigtimedwait, pcntl_exec, pcntl_getpriority, pcntl_setpriority,

Thanks in advance, any advice as to where to start would be great, it's really driving me crazy!
John.


----------------------------------------------
Cms Version: 1.11.9

Installed Modules:
CMSMailer: 5.2.2
CMSPrinting: 1.0.5
FileManager: 1.4.3
MenuManager: 1.8.6
MicroTiny: 1.2.5
ModuleManager: 1.5.5
News: 2.13
Search: 1.7.11
ThemeManager: 1.1.8
Gallery: 1.6.1
FormBuilder: 0.7.4

Config Information:
php_memory_limit:
process_whole_template:
max_upload_size: 64000000
url_rewriting: none
page_extension:
query_var: page
image_manipulation_prog: GD
auto_alias_content: true
locale:
default_encoding: utf-8
admin_encoding: utf-8
set_names: true

Php Information:
phpversion: 5.4.4-14+deb7u3
md5_function: On (True)
gd_version: 2
tempnam_function: On (True)
magic_quotes_runtime: Off (False)
E_STRICT: 0
E_DEPRECATED: 0
memory_limit: 128M
max_execution_time: 120
output_buffering: 8192
safe_mode: Off (False)
file_uploads: On (True)
post_max_size: 64M
upload_max_filesize: 64M
session_save_path: /srv/data/var/php/www (0770)
session_use_cookies: On (True)
xml_function: On (True)
xmlreader_class: On (True)

Server Information:
Server Api: fpm-fcgi
Server Db Type: MySQL (mysqli)
Server Db Version: 5.5.31
Server Db Grants: Could not find a suitable "GRANT ALL" permission. This may mean you could have problems installing or removing modules. Or even adding and deleting items, including pages
Server Time Diff: No filesystem time difference found

----------------------------------------------

[calguy1000]

a: the CMSMS core doesn't use any of those pcntl functions, so you should be safe.
b: Check your error logs, and clear your cache
not being able to reliably reproduce a problem is troublesome, and 99% of the time has been server related. like not being able to write session files, or disk space problems. Both of which, should show up in your error logs (if error logging is properly configured).

[JohnnyNM]

a: the CMSMS core doesn't use any of those pcntl functions, so you should be safe.
b: Check your error logs, and clear your cache
not being able to reliably reproduce a problem is troublesome, and 99% of the time has been server related. like not being able to write session files, or disk space problems. Both of which, should show up in your error logs (if error logging is properly configured).

Ah, pants, I was hoping it was the CMS. At least then I'd have something I could control to work on! Good to know the pcntl stuff isn't connected at least.

The only two server logs I have access to are the Apache, which is full of repeats of these (as well as my many, many fastcgi errors):


[Thu Oct 03 01:18:30.388598 2013] [mpm_event:notice] [pid 156:tid 3368643209024] AH00493: SIGUSR1 received. Doing graceful restart
[Thu Oct 03 01:18:30.435862 2013] [auth_digest:notice] [pid 156:tid 3368643209024] AH01757: generating secret for digest authentication ...
[Thu Oct 03 01:18:30.437389 2013] [mpm_event:notice] [pid 156:tid 3368643209024] AH00489: Apache/2.4.6 (Unix) configured -- resuming normal operations
[Thu Oct 03 01:18:30.437400 2013] [core:notice] [pid 156:tid 3368643209024] AH00094: Command line: '/usr/sbin/apache2 -d /srv/data/.config/apache -f apache2.conf'


...and the php log which I don't understand (and admittedly just found now! :s)...

...
[03-Oct-2013 01:50:42] WARNING: [pool www] child 3054 exited on signal 11 (SIGSEGV) after 2.030309 seconds from start
[03-Oct-2013 01:52:16] WARNING: [pool www] child 3056 exited on signal 11 (SIGSEGV) after 93.430695 seconds from start
[03-Oct-2013 01:52:23] WARNING: [pool www] child 3077 exited on signal 11 (SIGSEGV) after 6.954347 seconds from start
[03-Oct-2013 01:52:24] WARNING: [pool www] child 3079 exited on signal 11 (SIGSEGV) after 1.509487 seconds from start
[03-Oct-2013 02:22:37] WARNING: [pool www] child 3285 exited on signal 11 (SIGSEGV) after 93.065688 seconds from start


Does this make any sense to you?
Cheers again.
John.

[calguy1000]

So ... php is crashing, and apache is restarting frequently. points to a server problem. or a problem with another app. Now, this does not mean it is not CMSMS. However, since CMSMS crashes randomly I don't think so. if it were a CMSMS problem then the error would be more consistent.

I believe that some other application running on the server is causing PHP to crash, and you're experiencing the fallout.

Sounds like you need to contact the host.

[JohnnyNM]

So ... php is crashing, and apache is restarting frequently. points to a server problem. or a problem with another app. Now, this does not mean it is not CMSMS. However, since CMSMS crashes randomly I don't think so. if it were a CMSMS problem then the error would be more consistent.

I believe that some other application running on the server is causing PHP to crash, and you're experiencing the fallout.

Sounds like you need to contact the host.

Thanks for the feedback.

I walked back through the apache log, not that much of it made sense, but that restart sequence/script is seemingly running at 18m30s past the hour, every hour. Do you know if that's standard, or should the apache log be completely empty if it's working correctly?

I'll get on to the hosters again. It'll only take them eight hours to reply, but hey, what can you do!

Cheers,
John.