2

It seems that in Cosmos DB Gremlin API, query timeout is set to 30 seconds. Is there a known way to increase it?

  • Timeout is ignored in gremlin-driver 3.4.3
  • Same behavior with :remote config timeout in Gremlin Console 3.4.3
  • Increasing throughput from did not help either
// graph has >200000 vertices
String superSlowQuery = "g.V().repeat(identity()).times(32)";

time(() -> client.submit(superSlowQuery).all().get());
// OperationCanceledException after 31560ms

time(() -> client.submit(superSlowQuery).all().get());
// OperationCanceledException after 35081ms

RequestOptions requestOptions = RequestOptions.build().timeout(120 * 1000).create();
time(() -> client.submit(superSlowQuery, requestOptions).all().get());
// OperationCanceledException after 31656ms
dwitry
  • 21
  • 1
  • 1
    did you find an answer to this, cannot see in 3.4.3 where it states this is ignored as looking for an answer to this as well – Mark West Nov 17 '19 at 18:00

1 Answers1

0

As described in documentation 30 second timeout is predefined, and there is no indication that it is configurable:

Traversal timeout — 30 seconds — Traversal will be canceled when it exceeds this time. Cosmos DB Graph is an OLTP database with vast majority of traversals completing within milliseconds. To run OLAP queries on Cosmos DB Graph, use Apache Spark with Graph Data Frames and Cosmos DB Spark Connector.

dwitry
  • 21
  • 1