I've tried to figure out this problem, but I'm no mod developer and my troubleshooting skills are limited. We are using a custom modpack based on Create: Astral. It came with Chunky, which I'm using to pregenerate worlds to reduce lag. However, when generating, the TPS drops rapidly, leading to a crash. This has gotten faster as more chunks have been generated. It appears to be much worse in the nether. Any players exploring ungenerated chunks also cause a crash. TPS is a pretty steady 20 normally and we have an 8GB server which is rarely maxed out so it doesn't seem to be a need for more RAM.
I don't know how to interpret the Spark report in figuring out the cause of the crash. It seems to have to do with the scoreboard, but I don't know why generating new chunks is requiring so much interface with the scoreboard. Here is a Spark profiler run during ordinary gameplay with players online while an individual explored the nether (leading to a crash) and the accompanying crash report. Here is a spark profiler just run for a few seconds this morning while no players were online, as chunky generated chunks in the nether and crashed (crash report). It takes just a minute or two for the server to crash after starting chunk generation at this point.
Any advice would be useful, I'm usually decent at troubleshooting my way through things but this is a step beyond me. Our strategy if we can't figure this out is to pre-generate chunks in single player and then copy/paste them onto the server, but that is a REALLY janky solution.
Great news - you nailed the issue. C2me was the source of the problem. We still see the branch on the Spark profiler dealing with scoreboard things, but it's not ballooning astronomically when Chunky or players generate new chunks. If you have time and a theory about it, I'd be interested to know why you think that might have been happening, since I'm aspiring to not have to reach out for troubleshooting help going forward. Otherwise, just thanks so much for what you do - you saved us a lot of effort on our part.
I've tried to figure out this problem, but I'm no mod developer and my troubleshooting skills are limited. We are using a custom modpack based on Create: Astral. It came with Chunky, which I'm using to pregenerate worlds to reduce lag. However, when generating, the TPS drops rapidly, leading to a crash. This has gotten faster as more chunks have been generated. It appears to be much worse in the nether. Any players exploring ungenerated chunks also cause a crash. TPS is a pretty steady 20 normally and we have an 8GB server which is rarely maxed out so it doesn't seem to be a need for more RAM.
I don't know how to interpret the Spark report in figuring out the cause of the crash. It seems to have to do with the scoreboard, but I don't know why generating new chunks is requiring so much interface with the scoreboard. Here is a Spark profiler run during ordinary gameplay with players online while an individual explored the nether (leading to a crash) and the accompanying crash report. Here is a spark profiler just run for a few seconds this morning while no players were online, as chunky generated chunks in the nether and crashed (crash report). It takes just a minute or two for the server to crash after starting chunk generation at this point.
Any advice would be useful, I'm usually decent at troubleshooting my way through things but this is a step beyond me. Our strategy if we can't figure this out is to pre-generate chunks in single player and then copy/paste them onto the server, but that is a REALLY janky solution.
Go to the server.properties and set max-tick-time to -1
If there are still issues, make a test without:
c2me: Concurrent Chunk Management
and/or
fabricord
>> Link to Curseforge <<
Thank you so much for this advice, I will test this out tomorrow and get back to you.
Great news - you nailed the issue. C2me was the source of the problem. We still see the branch on the Spark profiler dealing with scoreboard things, but it's not ballooning astronomically when Chunky or players generate new chunks. If you have time and a theory about it, I'd be interested to know why you think that might have been happening, since I'm aspiring to not have to reach out for troubleshooting help going forward. Otherwise, just thanks so much for what you do - you saved us a lot of effort on our part.