I see that about every 15 minutes (or later), tasks are killed by the systems OOM-Killer (out-of-memory).

Is there something wrong with the current Tom Sawyer version, or is this specific server's memory size too low?

Remark: I already reduced the mariadb innodb-buffer pool to ~100 MB (because mariadb got killed now and then before this change). Now it is our lovely mailman3 which got killed as the program with the largest memory footprint...

Any idea?