Currently, all topology and alert probes run in their own system processes. As such, each probe opens a separate connection to the graph backend.
In order to prevent conflicting updates in the graph resulting from unordered event processing, there should be a global mechanism (lock) to synchronize access to the graph.
Currently, all topology and alert probes run in their own system processes. As such, each probe opens a separate connection to the graph backend.
In order to prevent conflicting updates in the graph resulting from unordered event processing, there should be a global mechanism (lock) to synchronize access to the graph.