[0.18.31] Ongoing unspecific crashes, generally related to Fluids or Electric Circuits

[oekelly]

Hi All,

In both my current and previous (abandoned) save I have been experiencing ongoing unspecific crashes, generally related to Fluids or Electric Circuits. The crashes generally do not occur at the start of a game, but become more and more frequent as the game progresses. I've checked the logs after some of these crashes and there doesn't appear to be any obvious fault, although the stack trace almost always has a Fluid or Electric Circuits component. Additionally, the crashes seem to only occur when in "busy" areas of the factory where these types of entities are found, when out at quieter locations the crashes feel a lot less frequent.

Logs and save file attached; please let me know if there is anything else I can add to assist.

[movax20h]

You have a number of mods installed. Would be good to reduce number of mods to easy troubleshooting or include a zip file with all mods in this save.

[Rseding91]

Based off the CPU you're running and the "random" crashing in areas that are being run in multiple threads I suspect you have the Ryzen CPU bug and need to update your BIOS/Chipset drivers and or remove any overclocking you might have.

[oekelly]

Thanks @Rseding91,

I am using a Ryzen: AMD Ryzen 5 3600X 6-Core Processor with an MSI X570-A PRO motherboard, running Ubuntu 20.04 LTS. My BIOS version is on H.70 (released 01/09/2020), and there are no other non-beta versions available at this stage. My OS packages are all up to date (linux-firmware 1.187, amd64-microcode 3.20191218.1ubuntu1).

I don't beleive I have enabled any overclocking, but I will drop to BIOS and double check all my settings now.

Just in case it is of any relevance, i'm also using an AMD Radeon GPU: Radeon RX 5700 XT Saphire Pulse, using the Mesa drivers (mesa-vulkan-drivers 20.0.4-2ubuntu1).

[movax20h]

AFAIK AMD Ryzen 3600X should not be affected by the infamous (rare) Ryzen SMT issues under Linux, they only affected Ryzen 1800X manufactured before 25th week of 2017.

To me the fact that it does crash in the same spot usually, is indicator of a bug, but I would suggest excluding a hardware fault first, by running some other stress tests on CPU and memory, checking cooling, thermals and not using overclocking for a moment.

The microcode package you are using is afaik most recent one. But the Ryzen bug the rsedig was refering to was a hardware issue during manufacturing that needed a CPU replacement (AMD is providing replacement for free, if you happen to be one affected, and has the Ryzen 1000X series manufactured before 25th week of 2017, and running Linux).

[oekelly]

Thanks all, I can confirm that I have double checked all the BIOS settings and I have no overclocking (CPU or RAM) enabled.

[movax20h]

Thanks all, I can confirm that I have double checked all the BIOS settings and I have no overclocking (CPU or RAM) enabled.

Then I guess, you need to find a way to make it crash reliably.

Does it crash when you interact with the game (i.e. placing / removing an entity by you or bots)? If so, try to narrow it down, and do a save just before (<1 second) the crash to make it reproducible by devs.

If it happen without your interaction, it should be similarly possibly to narrow it down and save just a second before the crash.