Who is this for and what problem do they have today?

AutoMQ team

Why is solving this problem impactful?

Currently, the following types of data are stored in the KVRecord:

• Partition -> MetaStreamId: kafka$clusterId/$topicName/$partitionIndex -> $metaStreamId
• Node -> NodeMetadata: __automq_node/$nodeId -> $nodeMetadata.

KVs of different functions are isolated by the prefix of the key.

Expectations:

• Add namespace to KVRecord to more elegantly isolate (future) KVs used by different functions.
• Add epoch to KVRecord to avoid concurrent operations of PutKVRequest accidentally overwriting KVs.

Additional notes