Last night after posting the previous post, I noticed that the httpd processes till consumed resources, but not near as much as before. This leads me to believe that the RPC ping process that is controlled via PHP scripting is somehow spawning the child httpd processes that are causing the slow downs on the server.
As I have said, I am not too worried about it at this point, because I will be delivering the new server sometime in the next two weeks, so I won’t have to worry about it after that.
I have removed all the URLs in my ping list and will now post this and check the CPU utilization during and after the post to see where the httpd process ends up. It seems odd that posting would cause this, but nothing else on the server causes this, including other Wordpress installations. The only difference that I know of outside of a few plugins is the ping list contents.
I just updated the previous post by simply changing the title from Mblogging to Moblogging. Needless to say I ended up with about 30 httpd processes, three of them going defunct and two of them eating up 98-99% of CPU availability. Hmmm…..
I am going to try something, as when I post from this blog (I host several other blogs and they seem to be working just fine) I have a rather lengthy list of ping services that I send pings out to. I am wondering if maybe this list has some bad entries in it that is causing the traffic to bounce back and forth or something otherwise causing it to start trying to load several processes to handle it, thus bogging the server down by utilizing the CPU.
Moblogging –
my attempt at writing a post from a windows mobile based phone. the wordpress interface does not scale to the tiny screen well….but the theme I have right now does make the posts fairly readable.
One of the things I have noticed on my aging server, which is going to be replace as soon as I can get my new one delivered to the data center where I do the co-location, is when I make a post from within WordPress, the HTTPD process seems to spawn anywhere from 5-8 children. These each then eat up 4-5% cpu process and 7.5-15 MB memory.
I am not sure why this is, but it then makes the server slow to a crawl on serving web pages, as the HTTPD process is completely maxed and running many children. I can understand that during the post it may need to do this, but after the post is done, and the page is reloaded the processes should go back down, but they are not doing that. And it is only when posting through WordPress that I find it does this. All of the various forum software I have ran on the server has never done this, nor has anything else that I have tried on here. And it seems to have only started after I did the upgrade to Wordpress 2.2.3 at the beginning of the year.
Case in point – I just made a post, and I now have close to 23 instances of HTTPD running on the server and it is not loading any pages (I am restarting the httpd daemon before hitting publish on this post). So I will be glad to get the new server online and get past this, since I have not had the time to find out why it is doing that.
