You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Jan 24, 2024. It is now read-only.
…eamnative#101)
### Motivation
Currently, a `ConnectionPool` singleton is used to maintain connections
to broker. Each broker has its unique connection. Thus, for different
Kafka clients that produce messages to the same topic, only 1 connection
will be used. This connection might also be used to forward other
requests. It harms the concurrency. What's worse, each connection should
have different sessions (e.g. authentication role) for different
clients.
### Modifications
- Refactor `ConnectionPool` to `ConnectionFactory`, which no longer
maintains a connection cache.
- For each `KafkaProxyRequestHandler`, maintain the following
connections:
- `leaderBrokers`: the map that maps the unresolved address to the
connection of leader brokers that PRODUCE and FETCH requests might be
sent.
- `metadataBroker`: the broker to query the metadata
- `groupCoordinator`: the broker for group coordinator requests
Sign up for freeto subscribe to this conversation on GitHub.
Already have an account?
Sign in.
This is the sub task for #57
The text was updated successfully, but these errors were encountered: