OH MY GOD!!! SOMEONE ELSE!!! Oh my god it’s not just me!
This sounds very similar to the issue that has been dragging me for MONTHS. I was on Ubuntu 18, MySQL 5.7, Nginx and Lucee 5.4.4.38 on a cloud VM with 2GB RAM and 2 vCPUs.
Ubuntu and MySQL were at EOL, so it was time to move, so I went to the same 2GB RAM/2 vCPU, but went with Ubuntu 22.04 (and 20.04, same problems), MySQL 8, Apache (and Nginx! no difference) and Lucee 5.4.4.38. No changes to my code, no changes to traffic, no changes whatsoever and my server would run for a few hours, randomly, and then, with no identifiable or common triggers whatsoever would completely crash, halt all network traffic and require me to do a force stop and start of the cloud server. My server logs told me I was having “Out of Memory” events, which were tied to Lucee.
I installed FusionReactor, which didn’t help with performance or uptime but did tell me that there was a ceaseless, never-ending increase in metaspace and non-heap memory allocation and usage. I looked into memory leaks (tried to, anyway) and couldn’t find anything definitive, other than “classes” being opened and then rarely closed.
I was told by other posters that the relentless, never-ending, pathological, ever-increasing memory usage was, apparently, no big deal! Nothing to worry about and totally normal (which seemed, you know, on-its-face crazy to me? But, I don’t know anything about java memory stuff, after all). But I still think this memory thing is an issue, even if everybody else keeps blowing it off and doesn’t care. Perhaps this is also happening to you? Perhaps the effect is most pronounced in our somewhat unique environments and combinations of software?
In the hours and hours and hours and hours of my short life on this beautiful planet I’ve spent working on this issue, I did run across this article mentioning that MySQL 8 absolutely consumes more memory than 5.7 and this is most noticeable on smaller servers. Additionally, this article mentions that you can turn off the “Performance Schema” monitoring tools on small instances to free up some memory. I tried doing that, but it didn’t really help, it just added a few more hours of life before the inevitable server meltdown.
So – I so very much wish I had advice to offer, but what I can mostly say is – it’s not just you! This really is a problem! It is! There is something going on with Lucee + MySQL 8.0. It seems very possible, to me, that the combination of the two puts some kind of unique strains on server memory management.