We seem to have narrowed down the server issue causing the php-fpm process crashing to load-spikes caused by our Movim instance.
It isn’t entirely clear though if this is a problem with the development version of it that we run, or maybe the blog part is being targeted by LLM scraper bots.
As a way forward the plan is to finally run Movim in a Podman container, which makes it easier to control the resources used, but since this isn’t an officially supported way of running Movim in production, it needs a bit of extra work to get running.
We might also try to put the blog part behind the Anubis AI scraper protection, like we already do for our Lemmy.
We hope to have it back online at the end of the weekend, but it depends on what unforeseen issues come up.
You must log in or # to comment.
Containerization completed, seems to run for now.
The load spike issue still seems to happen from time to time, but it is now isolated and can’t take down the main server like before.
We will monitor it a bit and see what happens and maybe make some additional improvements later on.