By default, the latest versions of Redis do not timeout client connections, however, if a timeout t is set on the server, then a PE that reads/writes to Redis will get disconnected if there are no tuples for t seconds.
This is a problem in environments where whomever is managing a Streams application doesn't have control over the Redis setup.
Am I right in that a workaround would be to:
Run the redis set.
Check the error code to see that the connection was lost.
Try to reconnect based on the error code.
I believe we should automatically try to reconnect in the event of a connection failure like this. Operators in the messaging toolkit could potentially serve as an example for reconnection policies.
By default, the latest versions of Redis do not timeout client connections, however, if a timeout t is set on the server, then a PE that reads/writes to Redis will get disconnected if there are no tuples for t seconds.
This is a problem in environments where whomever is managing a Streams application doesn't have control over the Redis setup.
Am I right in that a workaround would be to:
I believe we should automatically try to reconnect in the event of a connection failure like this. Operators in the messaging toolkit could potentially serve as an example for reconnection policies.