Kafka spout stops processing data

[Mike Thomsen <mi...@gmail.com>]

When we deploy a topology that uses the kafka spout (we're on 0.9.2 and
using the bundled spout), it seems to only pull the messages that were
there when it started. So if we have 5,000 messages already in, but are
pushing more into the queue as it deploys, it seems to only grab those
first 5,000 messages. When we redeploy, it finds the new messages.

Any ideas on what is happening here?

Thanks,

Mike