Details
-
Bug
-
Resolution: Won't Do
-
Major
-
None
-
6.0.1
-
None
-
Untriaged
-
1
-
Unknown
Description
Queryport server has tcp keep alive timeout set to 1 second. But similar timeout is not configured on queryport scan client. Without this, scan client will wait for indexer upto until tcp read timeout occurs before it can try with replicas.
Adminport client (watcher) has 100 millisecond tcp keep alive timeout.
Note that the connections can be kept in the conn_pool indefinitely. So, keep alive should be reset once the scan using the connection completes.
Also, make the default queryport client timeout equal to default scan timeout (i.e. 2 mins)
Attachments
Issue Links
- is a backport of
-
MB-47759 [CBSE] Introduce TCP keep alive timeout on gsi scan client side for scan connections
- Closed