Server monitoring

[Sarky78]

Hi all,

We are using a open source CMS for a small site for quite a large client and the site keeps crashing. As we didn't write the open source CMS we don't know the core code intimately and are having trouble finding the reason for the crashes.

Does anyone know of a tool that will monitor the server and allow us to see what resources are being used by what file, what pages are taking a long time to process etc. basically any bottle necks in the system.

Coming from a few years of ColdFusion development we had a tool called Seefusion (

http://www.seefusion.com/).

does anyone know of a tool that does something close to this? I've had a look on Google and the only ones i have found just ping a server

TIA

Tony

 

[jpadie]

can you tell us what the open source tool is?

and do the webserver logs not help with the debugging? what is the environment?

i don't know of any tool that will be helpful for an interpreted language like php (server resources will appear used by apache or php.cgi or the like), but there may be other ways we can help you find the root cause. and i'm not saying that there are not any tools, just that i don't know of any. a custom extension to php, perhaps.

 

[Sarky78]

Thanks for the quick response

The CMS is Drupal 6, the environment is Linux. the log files just show what has been requested etc, not what is running slowly etc etc.

 

[ingresman]

How does the crash manifest itself ? e.g. does the user get an error on the browser, does apache crash etc??

 

[jpadie]

can you not tell from the log file what the last request is before the crash? you should be able to focus your debugging efforts on the offending script then.

you have not said what version of php nor what linux kernel you are running off. are you using php as a (f)cgi, cli, or as a sapi? what webserver are you running off etc?

php in general will not murder your webserver. bad coding might: particularly never ending loops. this is unlikely to be in drupal itself, but in some template page or other user content. php sometimes falls over with some of the zend extensions (like the cache), so it's worth turning those off to see whether they are the culprits.

i'm not familiar enough with Drupal to give you any immediate help. but if you let us know what the offending script is, i'll take a look.

 

[Sarky78]

Apparently the pages just don't load and then eventually time out, then the server/service has to be restarted.

 

[ingresman]

Can you replicate this on a dev server at all ?
If you can (or even if you can't if your brave !) turn on as much error reporting in the php.ini as you can and see what happens.

 

[Sarky78]

Guys,

Thanks for all of your really quick responses, I've just found out that there are other sites crashing on the entire rack and the hardware guys seem to think it is a hardware fault causing it which is different, as when a service is restarted things are fine for a while. This one site is only up there for another two weeks so i think they are going to put code in place to restart the service at set intervals which sucks ass, as i'd like to find out the reason.

Thanks again, but i think we'll leave it at that

Tony