Latest update causes SIGIOT errors

[Brackston]

Not sure what happened but the last update leaves bots in map after a server crash, removing the artifacts (leftover bots from previous crash) while bots are running crashed our server. Not sure of the problem. our Server had plenty of ram (32Gig) so a SIGIOT is strange,

[nixnoxus]

.. leaves bots in map after a server crash

This this problem has existed for some time.

See https://github.com/joe7575/signs_bot/pull/21

so it must not necessarily be related to the last update

[nixnoxus]

.. removing the artifacts (leftover bots from previous crash) while bots are running crashed our server.

this problem I did not know until now and could actually be related to the last update

[debagos]

Yes, we must distinguish here between two things. 1) Fatal crash. 2) The leftover bots after a fatal error.

Number 2 happens when the server dies without shutting down gracefully. @nixnoxus explained the problem very detailed in #21 But number 1, the fatal crashes, I can't really explain them. All information I have is that a sub process of Minetest dies and then the main process panics and kills itself instantly with a SIGIOT, which is basically a abort signal (MT returns that when a sub process dies). So no shutdown is executed, the whole server is instantly death.

I am not too sure if this really is a problem with this mod, because a mod should never be able to crash the Lua process and furthermore there are signs that using nested interrupts in Mesecons Lua Controller crash the server the same way.

[joe7575]

It's very unlikely that the bugfix is the cause of your SIGIOT. The bug fix only handles the case, when a bot places a bot box. Check the diff

[debagos]

Here are two crash logs. But most times I don't get any log message at all...

2022-03-26 21:43:11: ERROR[Server]: An unhandled exception occurred: std::bad_alloc
2022-03-26 21:43:11: ERROR[Server]: In thread 7fa18ef38700:
2022-03-26 21:43:11: ERROR[Server]: /home/juri/github/minetest-5.5.0/src/server.cpp:86: virtual void* ServerThread::run(): A fatal error occurred: std::bad_alloc

2022-03-25 21:07:57: ERROR[Server]: An unhandled exception occurred: basic_string::_M_construct null not valid
2022-03-25 21:07:57: ERROR[Server]: In thread 7fa2198c5700:
2022-03-25 21:07:57: ERROR[Server]: /home/juri/github/minetest-5.5.0/src/server.cpp:86: virtual void* ServerThread::run(): A fatal error occurred: basic_string::_M_construct null not valid

[joe7575]

A std::bad_alloc is something you should report in the minetest repo or in the forum.

[debagos]

Hence this seems like an engine problem, we should close this issue. I created an issue in the Minetest repo: https://github.com/minetest/minetest/issues/12163

[joe7575]

Thanks for the info