Details
-
Improvement
-
Status: Resolved
-
Low
-
Resolution: Fixed
-
None
-
None
Description
Deploying 1.2 against a cluster whose clocks are not synchronized will cause every request to timeout. Suggest adding a cross_node_timeout option defaulting to false that users must explicitly enable after installing ntpd. Otherwise we fall back to the pessimistic case of assuming the request was forwarded to the replica instantly by the coordinator.