Kimmo L writes:
Here is the list of processes.
That's an unfamiliar format to me. Is this a Linux system running systemd?
Not sure yet, which stores the most memory in the cache. For example at the moment i have 6GB ram - 250mb free and 3gb cached.
"Cache" in the context of Linux kernel operation normally means "data file contents we keep around for possible future use, but if you need memory we will evict some stuff and give you some memory". It's not memory committed to running processes. So a couple hundred MB of memory with lots of cache is to be expected after a Linux system has been running for a while. So if you have 6GB RAM and 3GB cached, that's consistent with a machine that's running Mailman, using about 2.5 GB of RAM and not much else is running except the kernel. This report does not indicate a machine on the edge of OOM-killing processes, and it's quite consistent with Mark's experience that the whole Mailman suite runs comfortably in 4 GB.
How much swap do you have, and how much is in use?
What other daemons are running on the host besides Mailman? I guess you must have an MTA (Postfix, Exim) running and filtering software, as well as a front-end webserver (you shouldn't be running gnuicorn directly on the web). Systemd, dbus, I guess. How much memory are they using?
Have you modified any web pages associated with Mailman, including any of the Django templates? Are there any pages with unusually long load times?
I checked, and neither of the errors that you report for the uwsgi configuration ("OSError: write error" and "django.request Internal Server Error") seems to be related to a memory problem, but rather to difficulties serving certain resources, typically non-existent ones, or a user leaning on the "refresh" function in their browser. (Information from checking a few StackOverflow articles.)
By the way, Mark is going to be mostly unavailable for about 4 weeks. I'll do what I can to help, but Mark is probably a lot more expert on this than I am, so bear with me.
Steve