[SES-909] Timeouts and federated queries

Thu, 12 Jul 2012 07:44:44 +0200 (CEST)

Issue reported on mailing list by Jan Stette

Basically, it looks to me like query timeouts (hence presumably query cancellations) are not taken into account in federated queries? If I write a (admittedly unreasonable) test query like this:

SELECT * WHERE {
  SERVICE <...some URL...> {
     ?s ?p ?o .
     FILTER(!isBlank(?s))
     FILTER(!isBlank(?o))
   }
  SERVICE <...some URL...>{
     ?s ?p ?o .
     FILTER(!isBlank(?s))
     FILTER(!isBlank(?o))
  }
}

...then this will cause the client process to hang, even when a query timeout is set.

This is a nonsensical query of course, but I think it's all too easy to accidentally produce a query that "accidentally" produces a lot of triples and a lot of calls to nested query patterns... And that fact that rogue queries can run and consume resources forever is of course a big problem in a live production system.

openRDF.org Issue Tracker

[SES-909] Timeouts and federated queries