Hello,
I am running a fresh install of 4.1 on a dedicated redhat linux box. The server is running a plesk interface. The mt system is running 3 small blogs with little traffic. I've had the mt system running for about a month and yesterday it brought the server down. I had to disable mt.cgi to keep the server from crawling. My server admin said that perl is not running under fastCGI. Are there any other variables that might cause this memory problem?

Thanks,
Jonas

We're having the same problem. Started having it with 4.0 and then 4.1 and now 4.2. The performance improvements in 4.2 haven't helped a bit.

Have had to kill the mt.cgi process more than once. For us the problem seems centered around comments and even just the comments pages in the dashboard.

Something on that page (in a blog) will set off mt.cgi in such a way that it starts eating server memory up to 95%+ and stalling every other app running in the server. It killed off our mail system yesterday and always bogs down our forums.

We've been trying to adjust MySQL variables to see if that will help but no luck so far. Have turned off the LaunchBackgroundTasks also but that seems to have had zero effect.

We're experiencing this on both our production server and our dedicated test server.

Will post here if we figure anything out. Hope you'll do the same.

Can you post your mt-check.

Found a solution: probably the solution.

Install or reinstall ImageMagick on your server. Even if you don't use it. Be absolutely sure to run the Perl make at the end of the process. Go here for a super how-to by Anu Gupta: Movable Type and Image::Magick - digital<?>query

After weeks of bashing our brains out on this, we stumbled upon (yet another) someone else on a (weirdly) this very forum with a different way of describing exactly the same problem. Fortunately for them and us, user Yntouse posted the solution with backup from the aforelinked Anu.gupta. It was so exactly our problem that I immediately checked our test server's Apache error log for messages about "Deep recursion on subroutine "Image::Magick...." (see the second post in the thread).

We had a current ImageMagick on that server and working for other systems just fine but I went ahead and installed a slightly newer one -- being very sure to run the Perlmagick install too.

It's a lenghty compile but once it was done the MT problem was gone. Poof. Gone.

I checked the Apache error logs on our production server but (NOTE:) I did NOT find the same error message. That was a little frustrating, I must say, but I went ahead and reinstalled Image::Magick and PerlMagick and again, like magic, the problem is GONE. Fantastic.

Many parts of MT were being affected by this. Ironically, you can't run the MT system check from the dashboard while you're experiencing this problem as your MT.cgi system will dive into the crushing ImageMagick memory eating loop of death. You can, however, safely run mt-check.cgi directly from your web browser as: http://www.your-domain-here.com/your-path-...gi/mt-check.cgi. Run that and look to see if MT thinks that you don't have ImageMagick installed. If that's the case then this is probably the fix for you!

Good luck!

(Will post this on my blog at ThisllWork.com too.)