Automatic Reconnect Crashes MQTT when `client.reconnect()` is called about paho.mqtt.cpp

[fpagliughi]

Awesome, thanks. I'll have a look. Probably over the weekend.

But in general, think what might work better for you - automatic or manual updates. It's not too tough to do it manually; many of the example apps demonstrate it. But the auto/backoff algorithm is usable in most situations.

[fpagliughi]

I haven't figured out if the C lib fires the "connection lost" callback if it gets a clean disconnect from the server (i.e. receives a DISCONNECT packet). At first I assumed it did, then I thought it didn't. Now I'm not sure. I'm currently trying to set up the test broker to figure it out and/or check the C code to confirm.

That means you may need to set an on_disconnect handler via set_disconnected_handler() to make sure you're always being informed when the connection is lost.

I'll confirm ASAP, since I need to resolve #458.

[fpagliughi]

What version of the Paho C client are you using? Something that sounds just like this was recently fixed in the C v1.3.13 client.

[fpagliughi]

Oh, actually, this may be different. What platform? What compiler/version?

I can’t say that the expected behavior would be exactly what you want. Perhaps the library might ignore an explicit request to reconnect if the automatic option is turned on? I’d have to look at the implementation.

But either way, it definitely should never hard crash.

[lizziemac]

As of this morning, I've reproduced it on two separate OS's, both on ARM platforms. If you want to reproduce, I just modified the multithr_pub_sub.cpp example (attached with my changes) - multithr_pubsub.txt

MacOS (just tested this morning)

• Using MacPorts with MQTTVersion 1.3.13
• Compiler is clang

$ clang --version
Apple clang version 14.0.0 (clang-1400.0.29.202)
Target: arm64-apple-darwin23.1.0
Thread model: posix
InstalledDir: /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin

• Backtrace:

* thread #4, stop reason = signal SIGABRT
  * frame #0: 0x000000018ba0111c libsystem_kernel.dylib`__pthread_kill + 8
    frame #1: 0x000000018ba38cc0 libsystem_pthread.dylib`pthread_kill + 288
    frame #2: 0x000000018b948a40 libsystem_c.dylib`abort + 180
    frame #3: 0x000000018b9f06d8 libc++abi.dylib`abort_message + 132
    frame #4: 0x000000018b9e07ac libc++abi.dylib`demangling_terminate_handler() + 320
    frame #5: 0x000000018b68b8a4 libobjc.A.dylib`_objc_terminate() + 160
    frame #6: 0x000000018b9efa9c libc++abi.dylib`std::__terminate(void (*)()) + 16
    frame #7: 0x000000018b9f2a48 libc++abi.dylib`__cxxabiv1::failed_throw(__cxxabiv1::__cxa_exception*) + 36
    frame #8: 0x000000018b9f29f4 libc++abi.dylib`__cxa_throw + 140
    frame #9: 0x00000001004c7e94 libpaho-mqttpp3.1.dylib`mqtt::async_client::reconnect() + 324
    frame #10: 0x0000000100004588 harbor.out`publisher_func(std::__1::shared_ptr<mqtt::async_client>, std::__1::shared_ptr<multithr_counter>) + 292
    frame #11: 0x000000010000e540 harbor.out`decltype(static_cast<void (*>(fp)(static_cast<std::__1::shared_ptr<mqtt::async_client>>(fp0), static_cast<std::__1::shared_ptr<multithr_counter>>(fp0))) std::__1::__invoke<void (*)(std::__1::shared_ptr<mqtt::async_client>, std::__1::shared_ptr<multithr_counter>), std::__1::shared_ptr<mqtt::async_client>, std::__1::shared_ptr<multithr_counter> >(void (*&&)(std::__1::shared_ptr<mqtt::async_client>, std::__1::shared_ptr<multithr_counter>), std::__1::shared_ptr<mqtt::async_client>&&, std::__1::shared_ptr<multithr_counter>&&) + 84
    frame #12: 0x000000010000e4a0 harbor.out`void std::__1::__thread_execute<std::__1::unique_ptr<std::__1::__thread_struct, std::__1::default_delete<std::__1::__thread_struct> >, void (*)(std::__1::shared_ptr<mqtt::async_client>, std::__1::shared_ptr<multithr_counter>), std::__1::shared_ptr<mqtt::async_client>, std::__1::shared_ptr<multithr_counter>, 2ul, 3ul>(std::__1::tuple<std::__1::unique_ptr<std::__1::__thread_struct, std::__1::default_delete<std::__1::__thread_struct> >, void (*)(std::__1::shared_ptr<mqtt::async_client>, std::__1::shared_ptr<multithr_counter>), std::__1::shared_ptr<mqtt::async_client>, std::__1::shared_ptr<multithr_counter> >&, std::__1::__tuple_indices<2ul, 3ul>) + 76
    frame #13: 0x000000010000d91c harbor.out`void* std::__1::__thread_proxy<std::__1::tuple<std::__1::unique_ptr<std::__1::__thread_struct, std::__1::default_delete<std::__1::__thread_struct> >, void (*)(std::__1::shared_ptr<mqtt::async_client>, std::__1::shared_ptr<multithr_counter>), std::__1::shared_ptr<mqtt::async_client>, std::__1::shared_ptr<multithr_counter> > >(void*) + 84
    frame #14: 0x000000018ba39034 libsystem_pthread.dylib`_pthread_start + 136

Linux (initial PR) - Debian Bullseye

• running off of master branch (I probably shouldn't, and will likely address after this to run on 1.3.13 or greater)
• Building with CMake & clang under the hood, version is what is provided in Bullseye:

[bullseye (oldstable)](https://packages.debian.org/bullseye/clang) (devel): C, C++ and Objective-C compiler (LLVM based), clang binary
1:11.0-51+nmu5: amd64 arm64 armel armhf i386 mips64el mipsel ppc64el s390x

[lizziemac]

Sounds good, thanks! I was having some issues with the connection_lost callback being called for manual (yet autoreconnect working) but I'll take another stab at it.

[lizziemac]

Thanks! I'll look into implementing that