This message was deleted.

https://druid.apache.org/docs/latest/operations/api-reference.html#segment-loading-1 -> Indicates only the load status of the local cache

Thanks Amatya. Could we use this API /druid/coordinator/v1/loadstatus for historical rolling updates? Say that I have a cluster with 3 historicals in a single tier (replicant = 2), and I want to perform a rolling update to the 3 historicals. Step 1: replace historical#1 - I want to confirm that all the segments have been downloaded from deep storage, so that there is no downtime if I’m proceeding to replace next historical instance. Could I use the coordinator API for such purpose?

GET /druid/coordinator/v1/loadstatus?full&computeUsingClusterView

would be better. Since you have two replicants for each segment, restarting a historical would mean that a few segments will have fallen to a replication of 1. Ensuring that the historicals load 2 replicas for each segment by using the above API, before killing the next historical would ensure that data is always available throughout the upgrade