Using kill causes the program to crash

[NeilZhy]

Required information

Operating system:
E.g. Ubuntu 18.04 LTS

Compiler version:
E.g. GCC 7.4.0

Eclipse iceoryx version:
v2.0.3

Observed result or behaviour:
I use the script to kill my process for stress test

for i in {0..10000}
do
    pid=$(printf "%d" $(ps | grep test-demo | cut -c 1-6))
    for j in {0..5}
    do
        kill -15 $pid
    done
    sleep 10
done

my process will get the error and crash

_ZSt9terminatev(/lib/libstdc++.so 0x14186000)
_ZN3iox12ErrorHandler17reactOnErrorLevelENS_10ErrorLevelEPKc(/usr/lib/libiceoryx_hoofs.so 0x140a0000)
_ZN3iox12ErrorHandler14defaultHandlerENS_5ErrorERKNSt3__18functionIFvvEEENS_10ErrorLevelE(/usr/lib/libiceoryx_hoofs.so 0x140a0000)
_ZN3iox12errorHandlerENS_5ErrorERKNSt3__18functionIFvvEEENS_10ErrorLevelE(/usr/lib/libiceoryx_hoofs.so 0x140a0000)
_ZN3iox3cxx8internal7RequireEbPKciS3_S3_(/usr/lib/libiceoryx_hoofs.so 0x140a0000)
_ZN3iox10concurrent12PeriodicTaskINS_3cxx14MethodCallbackIvJEEEE3runEv(/usr/lib/libiceoryx_posh.so 0x13fb1000)

Expected result or behaviour:
no crash

[mossmaurice]

@NeilZhy Thanks for raising this issue.

If you demangle the symbols it becomes more readable:

std::terminate()(/lib/libstdc  .so 0x14186000)
iox::ErrorHandler::reactOnErrorLevel(iox::ErrorLevel, char const*)(/usr/lib/libiceoryx_hoofs.so 0x140a0000)
iox::ErrorHandler::defaultHandler(iox::Error, std::__1::function<void ()> const&, iox::ErrorLevel)(/usr/lib/libiceoryx_hoofs.so 0x140a0000)
iox::errorHandler(iox::Error, std::__1::function<void ()> const&, iox::ErrorLevel)(/usr/lib/libiceoryx_hoofs.so 0x140a0000)
iox::cxx::internal::Require(bool, char const*, int, char const*, char const*)(/usr/lib/libiceoryx_hoofs.so 0x140a0000)
iox::concurrent::PeriodicTask<iox::cxx::MethodCallback<void> >::run()(/usr/lib/libiceoryx_posh.so 0x13fb1000)

Could you try to find out which condition no longer holds in PeriodicTask? In general, I think it's not a good idea to kill applications forcefully from the outside but instead exit gracefully to prevent such issues. I assume some OS resource is already gone.

[NeilZhy]

@mossmaurice thanks for your reply.
sorry I can not find out which condition no longer holds in PeriodicTask, I check my log and do not find the problem.
the kill script is from our testet for stress test.

[elBoberido]

@NeilZhy do you gracefully shut down your applications, i.e. register a signal handler?

Can you check whether the problem persists with latest main or at least with the v2.95.3 tag?

[NeilZhy]

@elBoberido thanks，i will check our program for graceful shut down， and i will test the least version。

[elBoberido]

@NeilZhy did you have the chance to check for the graceful shut down and/or the v2.95.3 tag?