Details
-
Improvement
-
Resolution: Duplicate
-
Major
-
Cheshire-Cat
-
None
Description
As number of nodes in a cluster scales up, XDCR source nodes/target nodes pinging ns_server using the legacy bucket API (pools/default/buckets/<name>) could start causing some more workload on ns_server.
It may be worthwhile for XDCR to contact both source and target nodes using the streaming bucket API (pools/default/bucketStreaming/<bucketName>)
As well as the normal pools/default --> (/poolsStreaming/default)
Attachments
Issue Links
- relates to
-
MB-45983 XDCR - coordinate topologyChangeDetector effort to lessen load on ns_server
- Closed