Redis streams are filling up the cache with legacy transactions. In a separate thread in the Files service, truncate messages every hour (or day?) based on the following process/rules:

• Query Redis for all active channels
• For each active channel, query the API service for the latest checkpoint (or one prior?)
• Truncate the stream (channel) up to the set sequence number of the checkpoint

Things to consider:

• Optimize the lookups to only trim logs that need trimming (could require more persistence in API)
• Consider auto-truncating after creating a new checkpoint in the Files service to avoid a separate thread potentially creating race conditions