[0.17.74] Hang on SIGTERM during startup.

[Hornwitser]

I have a unit test that send SIGTERM to the Linux headless server upon receiving the following message on stdout

Code: Select all

RemoteCommandProcessor.cpp:131: updateTick(0) changing state from(CreatingGame) to(InGame)

When running this test on Travis-CI it ocasionaly hangs instead of initiating the shutdown causing the test to fail.

The process is spawned with following parameters (where the ports and password are randomly generated)

Code: Select all

factorio/bin/x64/factorio --config test/temp/integration/config.ini --start-server-load-scenario test_scenario --port 64808 --rcon-port 62353 --rcon-password <...>

test scenario_contains only a control.lua file with following code

Code: Select all

script.on_init(function() print('test_scenario init') end)

and config.ini contains read and write data directories:

Code: Select all

[path]
read-data=factorio/data
write-data=test/temp/integration

There's nothing printed out on stdout when the server hangs. I added code to send a second SIGTERM if the factorio server hangs like this and what happened then was that it printed "Received SIGTERM, shutting down" in response to it but still stayed hanged. You can see an example of this behavior in this test result. At the very bottom is the output from the test that failed (apologies for it being difficult to read, it's parsed.)

[movax20h]

Do this happen with 0.17.73 or 0.17.72 too?

[Hornwitser]

I have seen this happen on 0.17.73.

[Klonan]

Can you provide a log file?

Does it happen on 0.17.74?

[Hornwitser]

The linked test log shows the log outputted on stdout, though in a somewhat unreadable form. If there's something else outputed to factorio-current.log then that has been lost (do tell me if there's anything of interest there and I'll dump it too for the next time it happens.) It also shows it's running on Factorio 0.17.74 (build 47523, linux64, headless).

I saw a simmilar issue occur on a modded server running in a virtual machine. But there it didn't even print out the "Received SIGINT, shutting down" message, I've attached the log for it, but it just ends, systemd killed it after a timeout.

[Rseding91]

Thanks for the report however without a way to reproduce this there's nothing we can do.

My best guess is it's some issue specific to your system otherwise we would have more people here complaining that there's an issue with the game crashing.

Sorry.

[Hornwitser]

Still happens on a semi regular basis on 0.17.79. It would be easy to add code that for example dumps a back trace of all threads when this is detected to happen.