Idempotent producer doesn't recover after partition temporary unwritable

[eghilt]

During a couple of rebalances, some partitions became temporarily unwritable due to the number of available replicas falling below the minimum required. Applications utilizing idempotent producers, which sent messages to the affected topics, were unable to recover from this issue and continued to discard messages, even after the replica problem was resolved on the Kafka server side. The kafka client applications only recovered after a restart.
Applications without the idempotent producer activated were able to recover immediately once the server-side issue was fixed, without a restart.

Error logs:

• reason='Local: Inconsistent state' code='Local_Inconsistent'
• reason='Local: Purged in queue' code='Local_PurgeQueue'
• reason='Broker: Broker received an out of order sequence number' code='OutOfOrderSequenceNumber'
• "Unable to reconstruct MessageSet"

"Local_PurgeQueue" and "Unable to reconstruct MessageSet" errors stopped after a few minutes, but "Local_Inconsistent" and "OutOfOrderSequenceNumber" didn't stop until the instance was restarted

Confluent.Kafka Version=2.5