You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This issue impacts both scalability and makes releases more difficult.
Releases
It's not recommended to run multiple instances of southpaw as they share the same backup location. Southpaw uses RocksDB to keep track of the state of the world and periodically commits changes to an S3 path to ensure quick recovery from a reboot. A 2nd running instance could corrupt the S3 backups or produce unexpected results such as duplications in the output topic. Note, additional care needs to be taken while running in a container orchestration system such as Kubernetes, such as, but not limited to, a dedicated host to prevent background container duplications. We'd like to be able to run multiple instances for deployments to keep deployments standard as well as have zero downtime during release events.
Scalability
Due to the constraint of a single node processing messages, it's not yet possible to scale this component horizontally to reduce latency during periods of bursty load.
The text was updated successfully, but these errors were encountered:
This issue impacts both scalability and makes releases more difficult.
Releases
It's not recommended to run multiple instances of southpaw as they share the same backup location. Southpaw uses RocksDB to keep track of the state of the world and periodically commits changes to an S3 path to ensure quick recovery from a reboot. A 2nd running instance could corrupt the S3 backups or produce unexpected results such as duplications in the output topic. Note, additional care needs to be taken while running in a container orchestration system such as Kubernetes, such as, but not limited to, a dedicated host to prevent background container duplications. We'd like to be able to run multiple instances for deployments to keep deployments standard as well as have zero downtime during release events.
Scalability
Due to the constraint of a single node processing messages, it's not yet possible to scale this component horizontally to reduce latency during periods of bursty load.
The text was updated successfully, but these errors were encountered: