I'm seemingly completely unable to pregenerate my world...
MC 1.21.x
PluginHi, I’m trying to pre-generate LODs for my world on a 1.21.8 Paper server using the latest DHS. I’ve been attempting this for the past 3–4 days and I keep hitting one of two outcomes: either the pregen appears to stall completely (not a short pause — I’ve left it 8+ hours with the exact same “Processed LODs” value and no change to the Last Modified time on the .sqlite), or the ETA is massively high (55–65+ days, and not a short-term estimation - stayed around this value for ~18hrs) and doesn’t seem to line up with what I’m seeing others report.
Hardware-wise the server is an i9-12900K. I normally allocate 12GB, but I also tried pushing it as high as 100GB just to rule out memory. I’ve tested
I’ve tried two approaches:
Wiping the world and using DHS to do full generation (chunks + LODs). This ran the longest without “hard failing” and the processed number did keep increasing, but the ETA sat around 55–65 days for a very long time (18+ hours), which seems way off compared to other reports I’ve seen.
Removing DHS, pregenerating chunks with Chunky (radius 10,000 to match /worldborder set 20000), then reinstalling DHS and running LOD-only generation. This is the approach I’d prefer, but it consistently stalls around ~7%. When it stalls, there’s no error message — “Processed LODs” stops moving, time remaining goes to "Infinite" and the data.sqlite file stops changing (size + modified time). There’s also no useful log output right when it stops, so I don’t have an error log to attach.
I’m using these JVM args as well: https://exa.y2k.diy/garden/jvm-args/
Any help appreciated as right now the only thing DHS means to me is "Don't H*ng Self"
Hardware-wise the server is an i9-12900K. I normally allocate 12GB, but I also tried pushing it as high as 100GB just to rule out memory. I’ve tested
scheduler_threadsI’ve tried two approaches:
Wiping the world and using DHS to do full generation (chunks + LODs). This ran the longest without “hard failing” and the processed number did keep increasing, but the ETA sat around 55–65 days for a very long time (18+ hours), which seems way off compared to other reports I’ve seen.
Removing DHS, pregenerating chunks with Chunky (radius 10,000 to match /worldborder set 20000), then reinstalling DHS and running LOD-only generation. This is the approach I’d prefer, but it consistently stalls around ~7%. When it stalls, there’s no error message — “Processed LODs” stops moving, time remaining goes to "Infinite" and the data.sqlite file stops changing (size + modified time). There’s also no useful log output right when it stops, so I don’t have an error log to attach.
I’m using these JVM args as well: https://exa.y2k.diy/garden/jvm-args/
Any help appreciated as right now the only thing DHS means to me is "Don't H*ng Self"
Was this page helpful?
