Subject: Repeated Command timed out / Connection is closed after periodic health check (instance IDs: <ids>, timestamps: <times>)

Hi Railway support,

We observe repeated Command timed out immediately followed by Failed to create streams client: Connection is closed in our realtime worker logs. This happens around the same window as our periodic Redis health check. When this occurs, pooled connections tear down and the streams consumer disables, causing HTTP fallback.

Attached: full server logs (with timestamps), CLIENT LIST snapshot and Redis diagnostic outputs.

Please check for any TCP/egress limits, proxy-level timeouts for blocking XREADGROUP commands, or connection kill policies that might close pooled connections during repeated connect/disconnect sequences.

Thanks.