Is index creation API atomic?

Is index creation API atomic? As I might have multiple clients trying to create the same index at the same time, can I assume index creation is atomic which mimic POSIX file system?

In a way it is atomic, since the new index must be added to the cluster state which is controlled by the master node. Even if multiple requests are trying to add a new index to the cluster, only the first one handled by the master node will succeed - all the other requests will fail and return a "resource_already_exists_exception" to the clients.

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.