StackOverflowException: new Error()

[cj-the-affable]

Description

In at least one situation, Error(IntPtr error) ctor causes a StackOverflowException. For my specific case, it was because the TransactionTimeout was set to a value that was too large.

Client Version: 1.4
OS: Windows 10

Stack Trace:

Confluent.Kafka.dll!Confluent.Kafka.Impl.Librdkafka.error_destroy(System.IntPtr error) Line 1301
Confluent.Kafka.dll!Confluent.Kafka.Impl.Librdkafka.error_destroy(System.IntPtr error) Line 1301
Confluent.Kafka.dll!Confluent.Kafka.Impl.Librdkafka.error_destroy(System.IntPtr error) Line 1301
Confluent.Kafka.dll!Confluent.Kafka.Impl.Librdkafka.error_destroy(System.IntPtr error) Line 1301
Confluent.Kafka.dll!Confluent.Kafka.Impl.Librdkafka.error_destroy(System.IntPtr error) Line 1301
Confluent.Kafka.dll!Confluent.Kafka.Error.Error(System.IntPtr error) Line 67
Confluent.Kafka.dll!Confluent.Kafka.Impl.SafeKafkaHandle.InitTransactions(int millisecondsTimeout) Line 495
Confluent.Kafka.dll!.Confluent.kafka.Producer.InitTransactions()

How to reproduce

using Confluent.Kafka;
using System;

namespace SimpleProject
{
    class Program
    {
        static void Main(string[] args)
        {
            const int TOO_LARGE_VALUE = 1000 * 60 * 60 * 5;
            var clientConfig = new ClientConfig()
            {
                BootstrapServers = "localhost:9092",
            };

            var producerConfig = new ProducerConfig(clientConfig)
            {
                TransactionalId = "test-trx-id",
                TransactionTimeoutMs = TOO_LARGE_VALUE
            };
            using (var producer = new ProducerBuilder<string, string>(producerConfig).Build())
            {
                producer.InitTransactions(TimeSpan.FromHours(5));
            }
        }
    }
}

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]

i took a quick look at the code and didn't see how this was possible. thanks for the code snippet - will investigate.

[midnightriot]

@mhowlett I'm encountering this same issue when making calls to IProducer<>.SendOffsetsToTransaction() after my transaction has timed out and I get an old Epoch error right before the stack overflow.
I took a look at the code and think I have located the issue here:

confluent-kafka-dotnet/src/Confluent.Kafka/Impl/LibRdKafka.cs

Lines 1299 to 1301 in 98f019a

	private static Action<IntPtr> _error_destroy;
	internal static void error_destroy(IntPtr error)
	=> error_destroy(error);

error_destroy is calling itself instead of the private _error_destroy resulting in the stack overflow.

I'd be happy to put in a PR for this just let me know.

[mhowlett]

cool, thanks! we're doing a v1.4.4, i'll include this fix. feel free to open a pr if you want (or not, doesn't matter).

[midnightriot]

I'll leave it to you then thanks.

[mhowlett]

resolved in 1.4.4-RC1