java.lang.IllegalStateException: Could not find type for id: 32269

[David Kimdon]

Hi,

I've started trying to do OLAP on our production graph cluster and am running into a problem with really simple queries (like g.V(1)).  The problem always looks like this (the id is always the same):

                  (0 + 80) / 3957]org.apache.spark.SparkException: Job aborted due to stage failure: Task 50 in stage 0.0 failed 4 times, most recent failure: Lost task 50.3 in stage 0.0 (TID 229, 4.19.20.147): java.lang.IllegalStateException: Could not find type for id: 32269

at com.google.common.base.Preconditions.checkState(Preconditions.java:197)

at com.thinkaurelius.titan.graphdb.types.vertices.TitanSchemaVertex.name(TitanSchemaVertex.java:45)

at com.thinkaurelius.titan.hadoop.formats.util.input.current.TitanHadoopSetupImpl.getTypeInspector(TitanHadoopSetupImpl.java:50)

at com.thinkaurelius.titan.hadoop.formats.util.TitanVertexDeserializer.<init>(TitanVertexDeserializer.java:39)

at com.thinkaurelius.titan.hadoop.formats.util.GiraphInputFormat.lambda$setConf$0(GiraphInputFormat.java:48)

at com.thinkaurelius.titan.hadoop.formats.util.GiraphInputFormat$RefCountedCloseable.acquire(GiraphInputFormat.java:70)

at com.thinkaurelius.titan.hadoop.formats.util.GiraphRecordReader.<init>(GiraphRecordReader.java:33)

at com.thinkaurelius.titan.hadoop.formats.util.GiraphInputFormat.createRecordReader(GiraphInputFormat.java:38)

at org.apache.spark.rdd.NewHadoopRDD$$anon$1.<init>(NewHadoopRDD.scala:133)

. . .. 

I opened up and OLTP shell and looked around:

gremlin> g.V(32269)

Could not find type for id: 32269

Display stack trace? [yN] y

java.lang.IllegalStateException: Could not find type for id: 32269

at com.google.common.base.Preconditions.checkState(Preconditions.java:197)

at com.thinkaurelius.titan.graphdb.types.vertices.TitanSchemaVertex.name(TitanSchemaVertex.java:45)

at com.thinkaurelius.titan.graphdb.types.vertices.TitanSchemaVertex.toString(TitanSchemaVertex.java:152)

at 

. . . 

 gremlin> g.V(32269).valueMap()

==>[:]

gremlin> g.V(32269).label()

==>vertex

Well, this vertex didn't look like something that is helping me and after googling related errors it looks like data corruption (and, in fact data corruption problems is why I want to do OLAP).  I decided to try removing the vertex.  Now I have removed it:

gremlin> g.V(32269).drop()

gremlin> g.tx().commit()

==>null

gremlin> g.V(32269)

gremlin> 

But alas the graph computer still encounters the same error.  After reading more titan source code it looks like this is a schema element so deleting it was more dangerous than I had thought it would be at the time though I haven't seen any collateral damage.  Perhaps I removed something else, schema and data are perhaps not in the same space.

Now I am wondering how to proceed.  One idea is to swallow the exception so that I can succeed at reading in the pieces of the schema that work:

diff --git a/titan-hadoop-parent/titan-hadoop-core/src/main/java/com/thinkaurelius/titan/hadoop/formats/util/input/current/TitanHadoopSetupImpl.jav

index 05dda3e..2087f65 100644

--- a/titan-hadoop-parent/titan-hadoop-core/src/main/java/com/thinkaurelius/titan/hadoop/formats/util/input/current/TitanHadoopSetupImpl.java

+++ b/titan-hadoop-parent/titan-hadoop-core/src/main/java/com/thinkaurelius/titan/hadoop/formats/util/input/current/TitanHadoopSetupImpl.java

@@ -47,8 +47,12 @@ public class TitanHadoopSetupImpl extends TitanHadoopSetupCommon {

                 assert k instanceof TitanSchemaVertex;

                 TitanSchemaVertex s = (TitanSchemaVertex)k;

                 if (sc.hasName()) {

-                    String name = s.name();

-                    Preconditions.checkNotNull(name);

+                    try {

+                        String name = s.name();

+                        Preconditions.checkNotNull(name);

+                    } catch (Exception ex) {

+                        // ??

+                    }

                 }

                 TypeDefinitionMap dm = s.getDefinition();

                 Preconditions.checkNotNull(dm);

but . . I'm not sure how safe this would be.

Thoughts?

Thanks for any ideas,

David

[Sarah Masud]

I am facing a similar issue, any workaround for this?