[KAFKA-333] mirroring - enable non-random partitioning in embedded producer

• Type: New Feature
• Status: Resolved
• Priority: Major
• Resolution: Invalid
• Component/s: core
• Assignee: Unassigned
• Reporter: xiaoyu wang
• Created: 18/Apr/12 20:18
• Updated: 07/Feb/15 23:50
• Resolved: 07/Feb/15 23:50

Description

Currently the producer for mirroring uses the random partitioner. It will be very useful if we can specify partitioner there. One use case is when we aggregate ad server logs by mirroring the local kafka cluster on each ad server, we want to partition ad logs based on where they come from - the ad server id.