There was a conflict on DefaultEventHandler, but I reviewed the patch.
Should the default for log.cleaner.delete.retention.ms be 24 hours instead of 1 hour ?
2.1 Should the check for dedup buffer be
config.dedupeBufferSize / config.numThreads > Int.MaxValue
3. DefaultEventHandler (There was a conflict, maybe you already handled this)
Need to check for null payload in the following trace-
trace("Successfully sent message: %s".format(Utils.readString(message.message.payload)))))
Should this be reading message.key instead ?
print(" key: " + Utils.readString(messageAndOffset.message.payload, "UTF-8"))
Should probably check for null here in getData well -
ByteBuffer buf = message.payload();
6. OffsetMap6.1 If I understand correctly from getPosition(), it seems that the probe length will change arbitrarily each time. What is the advantage of doing this VS picking a fixed probe length that is relatively prime to the total number of entries that the hash table can fit in ? The purpose of this property is so that every slot in the hash table can be eventually traversed.
6.2 Why does attempts increment by 1 and not by 4 ?
The purpose of dumpLogs config is not clear from the command line option description.