One of my clients has a Wordpress site that uses CloudFlare. They are complaining that CF keeps giving a "Error 524: A timeout occurred" which indicates that the server did not issue an HTTP response within their 100 second timeout period. I have not personally seen their site down at any point. I tried a little curl magic to test the response time and found it to be well under a single second using this: Code: curl -vso /dev/null -w "Connect: %{time_connect} \n TTFB: %{time_starttransfer} \n Total time: %{time_total} \n" -H 'Host: www.someslowsite.com' http://123.45.67.89 That IP makes sure that the request is coming from my server and not using DNS to resolve back to CloudFlare. Of course, I could just be checking when everything is working perfectly. So I have two questions. Should I add the "%T" LogFormat modifier to the Apache logs to see the connection time? And how should I go about watching my server resources to check for spikes that may be causing slow responses or downtime? My gut is telling me this is a WP/DB access problem. They are using some plugins to do basic object caching. It looks like W3TC is one of them. I'm trying to narrow down a bottle neck to this specific site. I can see a huge number of plugins in the plugins folder so I'm betting that this problem is site specific.
Any errors that might be related to this in the error.log of the website? I would install munin and monit (see tutorials here at howtoforge).
I just installed munin and got it running. I also installed monit and it is running, but I can't view it's page. I did open port 2812 in CSF, but I'm still getting Chrome's cannot connect error. Even though it is running, `netstat -tulpn` doesn't show it listening. I'm going to have dive into the logs and try to figure out what's going on with it.
