I created https://issues.apache.org/jira/browse/TINKERPOP-2806 to enhance the GraphManager interface to allow providers to opt into receiving notifications when scripts/queries are processed.
Ideally what we would like as a provider is to be able to tell when a query starts, completes, errors or encounters a timeout so we can provide metrics showing success / error rates by graph. Currently as a provider the gremlin-server is a black box where you have no insight to what is being processed. TINKERPOP-2806 proposes adding some default methods to the GraphManager interface which will allow providers to opt into handling the lifecycle of script/query execution. I created a prototype of what this could look like at https://github.com/apache/tinkerpop/compare/3.6-dev...tkolanko:tinkerpop:3.6-dev?expand=1 Inside AbstractEvalOpProcessor and TraversalOpProcessor I am getting the GraphManager from context and calling the appropriate method depending where in the lifecycle the script/query processing is
