Kafka Library Crashes when using ProducerConfig with SetLogHandler / SetErrorHandler

[albertherd]

Description

When creating a ProducerConfig and your application is running containerized (using the default Dockerfile from VS 2022), when a log is generated and is passed to the API, the application just terminates. It even ignores the try-catch (maybe there is no exception being raised), the application just terminates.

• I reproduced this in nuget package version 1.8.2
• OS is windows, but Docker container is Linux
• You don't need a Kafka instance running
• No logs are emitted
• Tested on different machines, both Windows 10 and Windows 11

How to reproduce

• Clone my repo - https://github.com/albertherd/KafkaProducerConfigIssue
• Run it
• When the swagger UI comes up, run the main endpoint
• The app will try to communicate with a non-existant kafka instance, will try to log and the app will immediately exit.

You can see it life in action here:

The offending code is the following

new ProducerBuilder<string, string>(producerConfig) .SetValueSerializer(new JsonSerializer<string>(schemaRegistry)) .SetLogHandler((_, message) => Console.WriteLine("I am a log")) .SetErrorHandler((_, error) => Console.WriteLine("I am an error")) .Build();

If you do NOT run it containerized, it will fail with expected behaviour.

Checklist

Please provide the following information:

• A complete (i.e. we can run it), minimal program demonstrating the problem. No need to supply a project file.
• Confluent.Kafka nuget version.
• Apache Kafka version.
• Client configuration.
• Operating system.
• Provide logs (with "debug" : "..." as necessary in configuration).
• Provide broker log excerpts.
• Critical issue.

[mhowlett]

this has been reported before.
i'll need to look into it.
seems like librdkafka is segfaulting

[vpa1977]

I have got following stacktrace on Ubuntu 22.10, 1.9.3 when using debugger.
No crash when running without it.
Could it be something runtime-related?

 thread #1, name = 'KafkaCrashRepor', stop reason = signal SIGTRAP
  * frame #0: 0x00007f0ac735fba1
    frame #1: 0x00007f0b2e6210fe librdkafka.so`rd_kafka_log0 + 670
    frame #2: 0x00007f0b2e630feb librdkafka.so`rd_kafka_broker_set_error + 667
    frame #3: 0x00007f0b2e632c2c librdkafka.so`rd_kafka_broker_fail + 316
    frame #4: 0x00007f0b2e632fa7 librdkafka.so`rd_kafka_broker_conn_closed + 71
    frame #5: 0x00007f0b2e63748e librdkafka.so`rd_kafka_recv + 638
    frame #6: 0x00007f0b2e65ac50 librdkafka.so`rd_kafka_transport_io_event.constprop.6 + 464
    frame #7: 0x00007f0b2e65b793 librdkafka.so`rd_kafka_transport_io_serve + 147
    frame #8: 0x00007f0b2e641fc4 librdkafka.so`rd_kafka_broker_ops_io_serve + 276
    frame #9: 0x00007f0b2e643e30 librdkafka.so`rd_kafka_broker_serve + 1008
    frame #10: 0x00007f0b2e64464d librdkafka.so`rd_kafka_broker_thread_main + 717
    frame #11: 0x00007f0b2e6bc2f7 librdkafka.so`_thrd_wrapper_function + 23
    frame #12: 0x00007f0b408bd402 libc.so.6`start_thread(arg=<unavailable>) at pthread_create.c:442:8
    frame #13: 0x00007f0b4094c590 libc.so.6`__clone3 at clone3.S:81
(lldb) ip2md 0x00007f0ac735fba1
MethodDesc:   00007f0ac740c128
Method Name:          Confluent.Kafka.Producer`2[[System.__Canon, System.Private.CoreLib],[System.__Canon, System.Private.CoreLib]].LogCallback(IntPtr, Confluent.Kafka.SyslogLevel, System.String, System.String)
Class:                00007f0ac733df58
MethodTable:          00007f0ac740c220
mdToken:              00000000060002C6
Module:               00007f0ac7060f78
IsJitted:             yes
Current CodeAddr:     00007f0ac735fba0
Version History:
  ILCodeVersion:      0000000000000000
  ReJIT ID:           0
  IL Addr:            00007f0b3c0419a4
     CodeAddr:           00007f0ac735fba0  (MinOptJitted)
     NativeCodeVersion:  0000000000000000

[deNoor]

Since this issue is going to be the main one for consumer .SetLogHandler stealth crashes, let's mention here:

• happens in net6.0 and net7.0, but works in net5.0;
• happens with dotnet debugger attached and works without it.

I've tested with:
Visual Studio 2022 17.6.5,
Docker Desktop 4.21.1 (Windows 11, WSL2 mode),
container images mcr.microsoft.com/dotnet/runtime (5.0, 6.0, 7.0),
librdkafka 1.5.2, 1.9.0, 2.2.0 from https://packages.confluent.io/clients/deb buster/main amd64,
Confluent.Kafka 1.5.2, 1.9.3.

[henioStraszny]

I've tested on Debian and Confluent.Kafka 2.3.0 with the same result.