VMware NSX
VMware NSX-T Data Center
When a user leverages Simplified UI to perform Traceflow, Policy API framework may schedule MP API call to different manager nodes. The NSX Manager node that is scheduled with Traceflow creation API writes the timestamp at which the Traceflow request is expected to be finished into Corfu DB, and other NSX Manager nodes will read this timestamp to determine whether the received Traceflow observation message corresponds to a stale Traceflow request or not. If the received observation message corresponds to a stale or non-existent Traceflow request, it will be ignored. Thus, when clocks of NSX Manager nodes are out of synchronization, NSX Manager nodes other than the one where Traceflow creation API is scheduled may ignore the received observation messages. When Policy API framework schedules the result querying API to the NSX Manager nodes other than the one where Traceflow creation API is scheduled, empty result is returned.
This is a condition that may occur in a VMware NSX environment.
To resolve this issue, ensure that the NSX Manager nodes' clocks are synchronized. Clock synchronization can be achieved by leveraging NTP (see NSX documentation) or just the system clock manually on each node.