DHS Crash on join
Plugin version: 1.21.4:0.8.0-SNAPSHOT
Mod version: 2.3.0-b-deb
After joining the server the client receives a message that the terrain is downloading and shortly after the server crashes.
21 Replies
can you send the full logs of both server and client?
This is the full log from the server. Server boots, I join, and around 1 minute later, shortly after it starts downloading, it crashes.
Here are the client logs:
Not a lot of information to go on here, but I'm thinking you are hosting your server through some sort of web panel and the server is being killed by the panel for using too much RAM. How much RAM do you have allocated?
4GB with 75% being regularly in use. I will try to allocate more and see if that helps. Thank you
You can also try reducing the number of generator threads in DHS' config file to lower the demand. With only 4 gigs I'd maybe stick to a single worker
I am hosting it on my own hardware so I can increase it no problem!
👌
I'll let you know if the situation improves.
Was the guess about using a panel correct? Or is it living inside of some other management software?
I am running it in a docker container using this container: https://github.com/itzg/docker-minecraft-server
It's not a web panel, but close enough 😄 Pretty much the same thing just without a GUI
Okay cool, that effectively what Pterodactyl does, which is the panel I am most familiar with
I guess the Docker daemon would log it if it was a forcefull kill
Kernel and syslogs would also be worth checking
It detects the crash and immediately reboots. The log ends at the server log I sent earlier
The container yeah, but the Docker daemon is hopefully not crashing
I suppose not hahaha, I wouldn't know where else to see logs that might related to this in this OS I am using. I'm slowing working on migrating to a different OS with better management.
On the Linux distros I am used to, "dmesg -T" will give you the kernel log. Any processes the OS kills will show up there
For Docker I suppose it would be somewhere under /var/log, probably a file or directory starting with "docker"
Ah nice
One moment
Doesn't get more straight forward than that
Nice
Awesome stuff, thanks for the help!
No worries 👌
How much RAM would you recommend per worker?
Each worker will process an area of 4x4 chunks for each LOD they generate, and it's completely up to the server itself to decide when to unload these chunks again afterward, which means it's a question with many different answers. I would recommend testing it with a few different values in your own setup, and making a decision based on that.
When DHS has requested a chunk load specifically for LOD generation, it also requests an unload with saving disabled after, but again this is completely up to the server to actually execute.
So in practice if we compare the result of running DHS on 4 of the most popular server solutions, Spigot, Paper, Purpur, and Folia, you're likely to get 4 very different results.