Should retry when the error occurs.kafka.(*Client).Produce: fetch request error: topic partition has no leader

[stonever]

Describe the bug

Kafka is encountering this error for some reason, but the client should support retries. I have configured the number of retries, but it is not working.

My code:

w := &kafka.Writer{
		Addr:                   kafka.TCP(option.Broker...),
		Topic:                  topic,
		Balancer:               &kafka.Hash{},
		Async:                  true,
		RequiredAcks:           kafka.RequireAll,
		Transport:              transport,
		MaxAttempts:            100,
		AllowAutoTopicCreation: true,
		Logger: kafka.LoggerFunc(func(s string, i ...interface{}) {
			log.With(ctx).Info().Msgf(s, i...)
		}),
		ErrorLogger: kafka.LoggerFunc(func(s string, i ...interface{}) {
			log.With(ctx).Error().Msgf(s, i...)
		}),
	}

[Modestying]

I also meet this question . I'm not familiar with kafka .... Why this question happen ? Kafka config error or others ?

[nimesh-zepto]

Kafka throw this error, and it is not retryable. Attaching the error which is returned in this case

If you see the error is of type protocol.TopicPartitionError which is produce.Error and the library did not retry for this type of error.

The library only try for https://github.com/segmentio/kafka-go/blob/0cbe406d4f176d0d55eb8bd0997c2bba20fe4b9e/error.go#L138C16-L138C25 and Network error

Library should handle this error too.

[wyzzzl]

Is there any updates on this? I meet the same problem.