OpenSSL 3 error 0A000126 can cause an offset reset

[emasab]

Description

An unclean SSL close, possibly caused by a TCP reset, causes a RD_KAFKA_RESP_ERR__SSL error. If it happens during offset validation it can cause an offset reset because it's treated as permanent.

How to reproduce

cause an unexpected EOF during a OffsetForLeaderEpoch call.

Checklist

Please provide the following information:

• librdkafka version 2.1.0, 2.1.1
• Apache Kafka version: 3.4.0
• librdkafka client configuration: unimportant
• Operating system: unimportant
• Provide logs (with debug=.. as necessary) from librdkafka
• Provide broker log excerpts
• Critical issue

 %3|1683846206.615|FAIL|... REDACTED...: SSL handshake failed: error:0A000126:SSL routines::unexpected eof while reading (after 5002ms in state SSL_HANDSHAKE)
%4|1683846206.616|OFFSET|... REDACTED...: offset reset (at offset INVALID (leader epoch 41), broker 25) to offset BEGINNING (leader epoch -1): Unable to validate offset and epoch: Local: SSL error: Local: Partition log truncation detected

[RefaelP23]

@emasab Is there an estimate on when this fix will be released?
We noticed this issue with the .Net client as well.