[2.0.15] Corrupt map: unknown trivial-smoke prototype ID 100.

[Geebs]

Hi, I can't load this map or any of the 3 autosaves, all give the above error.

I've attached the log file and the main save as well as a crash dump file from a bit earlier than the saves.

Any help would be greatly appreciated, thanks

[dannyboy]

I've just experienced the same bug on my server. I launch my game with a script which automatically overwrites the map file with the latest autosave if it is newer and restarts the server if it had a nonzero exit code. While playing, the server crashed for some reason and tried several times to reload without success, wiping the server backlog off the screen, so I have no idea what caused the initial crash.

I am running Factorio 2.0.20 on Linux.

[Geebs]

Just to update on my own troubleshooting, I created a mod to populate the trivial-smoke prototypes up to ID 100 with dummy prototypes mostly out of curiosity not expecting to actually fix the save but more to see what other errors there might be. I then got an error about a trivial-smoke with ID 255 so made the mod populate up to 255 (max instances). After loading with this I then got an error about unknown decoratives (can't remember the exact wording) so I did the same with these, making dummy optimized-decorative prototypes up to ~60,000 (max 65,535). Loading with this gave me an "Invalid bool loaded from input file." error so I assume the save is truly borked. My friend luckily had a backup from a couple of days prior to the corruption so I just accepted the loss and moved on.

[Rseding91]

Thanks for the report however unless you know of a way to take a fresh save file and get it into this state I can only assume some hardware fault broke the save data as it was being written to the zip and to disk. This error means "you have hardware issues or something/someone corrupted the zip file on disk" - something outside of our control - something we have no ability to do anything about.

There's a reason it doesn't say "please report this on the forums" like the crash message does - because it's not our error to fix (although I wish we had the ability to fix players hardware )

[dannyboy]

Thanks for the report however unless you know of a way to take a fresh save file and get it into this state I can only assume some hardware fault broke the save data as it was being written to the zip and to disk. This error means "you have hardware issues or something/someone corrupted the zip file on disk" - something outside of our control - something we have no ability to do anything about.

Thanks for taking a look. The reason I replied to this bug report is that all of my autosaves have the same error, "Corrupt map: unknown trivial-smoke prototype ID 160," which seems highly unlikely to be due to a hardware error. In my case, the server (Factorio; my server OS/hardware remained operational) crashed for some reason and subsequently restarting it with any of the autosaves results in this error. It seemed to me that Factorio saved something naughty into its savefiles.

I've attached my earliest autosave in the hopes that the Factorio team might be able to analyze it and delta it against the latest save (thisland.zip in my prior reply).

[dannyboy]

It's happened again, this time on 2.0.21-5. All five autosaves fail with the same error:

Code: Select all

   1.600 Info Scenario.cpp:178: Map version 2.0.21-5
   2.719 Warning Map.cpp:358: Map loading failed: Corrupt map: unknown entity prototype ID 1896.

Interestingly, a friend was on the server when his computer crashed. The server log scrollback has since fallen off of my screen, but it shows him disconnecting, but the server still saw a peer (2) and continued to autosave for several hours. Thereafter, the server did not shut down gracefully on ctrl+c. I copied the latest autosave file over the map and restarted to be greeted with the error above.

Is it reasonable to assume that each autosave would contain data from the same areas in RAM? I can't otherwise conceive of how all of my autosaves are being corrupted with the same issue.

[Rseding91]

Is it reasonable to assume that each autosave would contain data from the same areas in RAM? I can't otherwise conceive of how all of my autosaves are being corrupted with the same issue.

Yes