Segfault httpd error 6 and system crash

Portal Home > Knowledgebase > Articles Database > Segfault httpd error 6 and system crash

Posted by sh4ka, 03-21-2012, 02:52 PM
Hello, My server (CentOS 5.x, running php 5.3.8 and httpd 2.2.3) recently crashed and it seems was a lack of memory: Before this crash, I've seen a lot of segfaults: I had coredump generation activated at httpd.conf and I found this while reading the core dump file: Does the file seem to be corrupted in some how? Do you see any signs of the memory leak or the segfault cause on this? Thanks
Posted by mike86, 03-21-2012, 06:43 PM
Have you got any type of monitoring on the server? It seems that your memory was fully used and you swap was full (swap is disk used as memory, so is very slow). Once this all filled up, your apache crashed. You should install monitoring software to monitor your memory usage. I use Cacti. Maybe it's time you upgraded your RAM? How many visitors does your site get and what spec is your server?
Posted by sh4ka, 03-22-2012, 08:07 AM
Hi, thanks for your answer. I use MRTG and Newrelic for monitoring, I will install Cacti to see if it helps. The server is a Intel E5410 @ 2.33GHz, guess with HT as it shows 8 cores at /proc/cpuinfo. 8 GB RAM. The server only serves php-apache and a few images from CentOS 5.x, there is another server dedicated to mysql only and another one for all the static stuff. The website has more than 50k uniq visitors daily. Right now this is the ram usage: Any other suggestions please? Thanks
Posted by mike86, 03-22-2012, 08:18 AM
As long as you use some monitoring software. Your memory usage is extremely high, if is reaches full then it goes to swap space (slow and not good!). I suggest that you invest in more RAM or maybe try a more efficient webserver such as Litespeed or nginx, these should reduce RAM usage. The E5410 is a quad core processor, so if it says you have 8 cores, then you have 2 processors.
Posted by sh4ka, 03-22-2012, 08:22 AM
Thanks. Due to programming incompatibilities we can't use Nginx, already tried. We may try Litespeed this time. Any clue about how to get more info from the coredumps? I'm very interested about fixing the segfaults, which I think are causing the memory to get filled. That would be the root of the problem. Thanks a lot.
Posted by mike86, 03-22-2012, 12:19 PM
Hey, sorry I am not able to help you with that. Hopefully someone on here can shed some light on why that is happening.

Add to Favourites Print this Article