Database container restarting every 30 seconds

[flexxyfluxx]

What happened?

I have the invite-only option enabled on my instance. As instructed, after generating the config (and setting the invite-only option), I launch stoat using docker compose up.

Problems arise when I then try to create an invite using the provided instructions to enter mongosh, and insert an invite ID: After some varying period of time, mongosh just dies; ostensibly from a SIGKILL.

After monitoring logs, I have found that this coincides quite neatly with the database container restarting every 30 seconds, which presumably is what's killing mongosh (in case it matters, the exit code is always 139).

This seems like incorrect behavior, because it is mentioned nowhere. It is unclear as of now what is causing the restart. I'll keep monitoring logs; will update if any other interesting patterns arise, or at the very least with a log dump.

Just to be certain though: Am I missing anything? Am I making a false assumption here? Am I doing something wrong? (I'm quite open to this. I am ..not any sort of seasoned veteran.)

[flexxyfluxx]

I suspect I've found the issue: https://www.mongodb.com/docs/manual/release-notes/8.0/

tl;dr: mongodb uses TCMalloc, which doesn't work on the very latest Linux kernel versions, and segfaults. I happen to be using the newest kernel on all my machines. Thus, mongodb crashes, and in this particular configuration, is auto-restarted.. over and over again.

I'll have to consider downgrading my kernel then. Good grief.