[2.0.28] Crash of server loading some saves (Chunk::load)

[b14ckyy]

I have a save with a friend that has 84 play hours on it so far. We started it via local hosting from our PCs but I decided to set up a Headless Server on my Custom build OpenMediavault Media-Server.
After some trouble with the ballistic missiles mod that caused the game to crash if you shoot a missile out of map bounds https://mods.factorio.com/mod/ballistic ... 26369d9da7 everything was working fine for a few hours but now I can't load ANY recent save of this map anymore. The original saves result in a server crash loop after startup. After taking the same safe file on PC, loading it there and save as a new one, the crash loop is gone but the server still crashes when a player connects and the map is prepared for download.

To reproduce:
- Run Headless Server in docker
- Use the provided save game
- start server (INFWG_repair.zip should start up fine, the other one should crash loop the server)
- try to log in with a client to that Headless Server
- on INFWG_repair.zip the server will crash when map is prepared for download to client (at different stages each try)

TS steps done:
- server clean installed from ground up
- removed all mods
- tried different save games - Older save with ~54hrs of play time works fine. New saves and other maps also work just fine with no crash
- tried the same save on different PCs to launch locally - all fine there

We could bite in the sour apple and use an old save but a save that works locally and on hosted multiplayer in Windows should not cause a server crash on a headless server I guess.

[Rseding91]

This sounds suspiciously like faling/failed hardware. The headless linux build of the game is identical code wise to the non-headless and windows build of the game in every way except the headless linux build doesn't let you enable graphics mode.

on INFWG_repair.zip the server will crash when map is prepared for download to client (at different stages each try)

That almost always signals bad hardware. "random issues" should never happen in a deterministic game like Factorio.

[b14ckyy]

I am very sure the hardware is 100% fine. This server is running for months and months overall with no issues. I will run a Memtest soon when I can though.

BUT. I solved the issue with the server crashing by just fully copying the Game data folder from my Desktop to the server. There have been some additional files that should not be needed by the server as far as I know (but what do I know?) and now the server does NOT crash anymore with the same save games as before. This speaks against a Hardware failure and the server is running for 2 days now.

BUT I get a new issue. When we connect through the Public network (my friend has to), we always get a desync. Even I get a desync when I connect through the public server list or via IP address despite being in the same local network as the server. All worked for a short time and then suddenly these desync issues happened. This is all super frustrating and makes no sense to me

I will set it up from scratch with a more recent save game and then take one file after another from my PC to see if anything changes, to find the crash reason.

That almost always signals bad hardware. "random issues" should never happen in a deterministic game like Factorio.

Did you try to run one of the save games on a Headless for testing or reproducing the issue? That would either confirm or fully disprove a hardware fault, before I take my server out of its place to check it and swap memories. I can also try to run it in a VM on my PC for testing if that helps.

btw the crash happens when the server is pre-compressing the map state as a ZIP to allow the client to download it. That ZIP process is not determistic as the data change any second and the compression file order is always different.