-
Notifications
You must be signed in to change notification settings - Fork 41
Endpoint recovery
When an endpoint leads to a non recoverable error for a technical reason (only), the endpoint is banned. Various technical reasons:
- connection drop
- invalid response
- ...
cassandra-sharp is able to recover endpoint in case of failure automatically.
The service is setup as soon a a cluster a created with the recoverable attribute set to true (see recoverable in the transport configuration).
The service tries to recover an enpoint every 60 seconds (something actually not configurable). If no endpoint needs recovery, the service won't run.
If the transport configuration specifies recoverable=true then connection recovery is performed behind. When the endpoint is back, the endpoint is allowed again.
If recoverability is not requested, then the endpoint won't be reused anymore until the cluster shutdown.
The recovery service is stopped when ClusterManager.Shutdown() is called.