I just saw something really odd with RT 3.8.6.
One of our system process generated 30K tickets in RT. From that point on,
Apache would eat up all the memory on the server and force us to reboot.
I ran an rt-crontool job to reject all those messages. This part has a huge
memory leak, so I had to watch the process and restart it and when it used
to much memory.
Where it got weird is once all the 30K messages were rejected, I restarted
the web server, and right at the first web page request, one of the apache
process took 3GB of RAM. Then another one did the same, and on… So I had
to reboot the server once more.
I managed to fix it by cleaning mason_data and session_data. I have no idea
why that would do anything, but that fixed it.
Maybe some weird memory leak.