Re: index could no longer be used


jme...@...
 

Sorry!  I found it's my own fault.
The property is a String type, and is a mixed index.

while this went wrong,
gremlin>  g.E().has('contents','hdfs').values()
Could not find a suitable index to answer graph query and graph scans are disabled: [( contents = hdfs)]: EDGE
this is ok
gremlin>  g.E().has('contents',textContains('hdfs')).count()
gremlin
>  11


So ,when a String type with mixed type index
Query should use textContains.


On Tuesday, May 2, 2017 at 11:46:13 PM UTC+8, Jason Plurad wrote:
FYI, the Titan mailing list is here.
What does your index definition look like?
Are you using JanusGraph or is this a problem you can recreate on JanusGraph?

On Tuesday, May 2, 2017 at 11:38:37 AM UTC-4, Jhon Mernio wrote:
   Hello:

   I'm using titan with hbase and elasticsearch, all was ok at first, but yesterday when I was searching as usual, it worn 
   Could not find a suitable index to answer graph query and graph scans are disabled: [( contents = hdfs)]: EDGE
   like below:

gremlin>  index=mgmt.getGraphIndex('message') 
==>message
gremlin
>  index.getIndexStatus(mgmt.getPropertyKey('contents')
==>ENABLED
gremlin
>  g.E().has('contents','hdfs').values()
Could not find a suitable index to answer graph query and graph scans are disabled: [( contents = hdfs)]: EDGE

 the index was a mixed type index, using elasticsearch backhand 
 this is very strange, because in fact the index does exist, and before today, all queries alike this was ok.
 and when data inserts into the database, I could see the increase of indexed data num by using elasticsearch count API

 and the flowing query is ok

 gremlin>  graph.indexQuery("message","e.contents: hdfs").edges()

 Anyone have any idea about this problem?

 Thanks!

Join janusgraph-users@lists.lfaidata.foundation to automatically receive all group messages.