I ran a 'yum update' on a CentOS based web server. Kernel version stayed the same at 2.6.21.7-2.fc8xen It upgraded a bunch of stuff, including - mysql from 5.0.45 to 5.0.91 apache from 2.2.3 to 2.2.15
I'm finding that my server response time is quite degraded. Web requests that would have completed in .5 seconds are taking over a second. My configuration files for both Apache and Mysql haven't changed.
I'm running on EC2, so I can run the old image and compare. Server processes appear to be the same.
Where should I look first to find out where the kink in the pipe is?
-
I would start my finding out if the problem is in Apache or if the problem is in MySQL. To do this with Apache you can look at Apache response times in ms with the %...D option in mod_log_config. For MySQL you can monitor queries times. The Slow Query Log is one way to do this for MySQL. Another way to see if it is MySQL from a more front-end standpoint is to see if Apache is slow serving out static content only. If one of these is taking longer than you would expect than you know where to go from there.
You can also of course check out system resource utilization with tools like top, vmstat, and iostat. Lastly, you might also have update your language modules (ie mod_php) and your problem could be there as well (or even network issues etc).
Laizer : This was helpful. After searching for a good couple hours, I found that it was an eaccelerator installation that was compiled for a later version of PHP than what I had. Once I isolated it to a server (And not network) issue, I should have started by looking at the startup logs of Apache and Mysql. Once I did that, it was pretty clear what was happening.From Kyle Brandt -
The problem could be also DNS (internal/lan) related.
What happen if you do a request from server itself (localhost)?
How apache (php or other...) does connect to mysql (localhost, ip, hostname...)?
From lg
0 comments:
Post a Comment