mojira.dev
MC-243037

Server crash when teleported near ungenerated chunks (with datapack)

The crash happened when I teleported next to the ungenerated chunks (like 2 chunks away from the ungenerated parts of the world). Now it's happening every time server is started and I log in again in the same place. From the deobfuscated stacktrace it looks like some infinite recursion involving liquids and playing sounds (weird).

The main thing the datapack does is increasing the lower bound of the world to y=-128. It worked for me for hours before teleporting.

Haven't reproduced it yet, but I'll try. Maybe the stacktrace will be helpful to find the cause. I'm attaching crash logs from RC1 (obfuscated and deobfuscated) and RC3 (obfuscated) and the datapack. Server log has no errors till the crash:

 

[08:53:01] [Server thread/INFO]: okarkalic joined the game
[08:53:03] [Server thread/WARN]: Can't keep up! Is the server overloaded? Running 2177ms or 43 ticks behind
[08:54:17] [Server Watchdog/FATAL]: A single server tick took 60.00 seconds (should be max 0.05)
[08:54:17] [Server Watchdog/FATAL]: Considering it to be crashed, server will forcibly shutdown.
[08:54:17] [Server Watchdog/ERROR]: This crash report has been saved to: /home/mc/18/./crash-reports/crash-2021-11-27

 

Edit: looking at the stacktraces I believe this may have been caused by lava and water flowing out of a "wall" between the old chunks and newly generated chunks (when you set the min y of the world to -128 in the datapack, the old chunks still update only to -64, this could be treated as a separate bug). I'm attaching screenshots of these fluids made before I'm getting kicked. The screenshost have been made at ca. y -100

Attachments

Comments 8

Thank you for your report!
We're tracking this issue in MC-63590, so this ticket is being resolved and linked as a duplicate.

That ticket has already been resolved as working as intended, which means this is not considered a bug and won't be fixed. Please do not leave a comment on the linked ticket.

If you haven't already, you might like to make use of the search feature to see if the issue has already been mentioned.

Quick Links:
📓 Bug Tracker Guidelines – 💬 Community Support – 📧 Mojang Support
📓 Project Summary – ✍️ Feedback and Suggestions – 📖 Game Wiki

The watchdog is triggered by a bug (infinite recursion) and not a recoverable lag, so this ticket should stay opened.

I believe that the first time this occurred I even got "stack overflow" error when I was kicked. I changed the max-tick-time to 6 mins (300000) and the same happened.

Hello,

As you said, this does not seem to be recoverable lag, so it will be kept open.

However, note that the StackOverflowError you saw is likely caused by MC-142590.

Quick Links:
📓 Bug Tracker Guidelines – 💬 Community Support – 📧 Mojang Support
📓 Project Summary – ✍️ Feedback and Suggestions – 📖 Game Wiki

Do you have an un-upgraded world where this occurs? We would need this to act. Thanks!

Please note that upgrading worlds using experimental datapacks is not supported.

I hope that a link to Google Drive will be ok, I packed the whole server folder and it's a little more than 100 MB. The world included is vanilla 1.17, made without datapacks, trimmed by mca-selector. In the datapack folder I've just added a 1.18 datapack named "wielipack" which mainly changes the lower bound of the world to -128. Steps to reproduce:

  1. in that folder: java -jar server.jar (port is set to 25566)

  2. wait till the server is loaded

  3. log in on some account (set render distance to 22)

  4. tp player 1000 100 2000 (these are the exact coords where this happens)

  5. don't move, wait half a minute, though you can rotate yourself to see how the world loads

  6. after half a minute server becomes unresponsive and eventually crashes

https://drive.google.com/file/d/1bSQClEHpYnRSmFi_ERrNiqWqVkrdVJVc/view?usp=sharing

Changing the world height of a pre-existing world is unsupported.

Oskar

(Unassigned)

Plausible

Crash, Performance

1.18 Release Candidate 3

Retrieved