noooo waaaaay... that was it!? :))) Big thanks! :)
The result is also correct now.


On 18.03.2015 22:49, Vasiliki Kalavri wrote:
haha, yes, actually I just confirmed!
If I flip my args, I get the error you mention in the first e-mail. you're trying to generate a graph giving the edge list as a vertex list and this is a way too big dataset for your memory settings (cmp. ~15m edges vs. the actual 400k).

I hope that clear everything out :-)


On 18 March 2015 at 23:44, Vasiliki Kalavri <> wrote:
Well, one thing I notice is that your vertices and edges args are flipped. Might be the source of error :-)

On 18 March 2015 at 23:04, Mihail Vieru <> wrote:
I'm also using 0 as sourceID. The exact program arguments:

0 /home/vieru/dev/flink-experiments/data/social_network.edgelist /home/vieru/dev/flink-experiments/data/social_network.verticeslist /home/vieru/dev/flink-experiments/sssp-output-higgstwitter 10

And yes, I call both methods on the initialized Graph mappedInput. I don't understand why the distances are computed correctly for the small graph (also read from files) but not for the larger one.
The messages appear to be wrong in the latter case.

On 18.03.2015 21:55, Vasiliki Kalavri wrote:
hmm, I'm starting to run out of ideas...
What's your source ID parameter? I ran mine with 0. 
About the result, you call both createVertexCentricIteration() and runVertexCentricIteration() on the initialized graph, right?

On 18 March 2015 at 22:33, Mihail Vieru <> wrote:
Hi Vasia,

yes, I am using the latest master. I just did a pull again and the problem persists. Perhaps Robert could confirm as well.

I've set the solution set to unmanaged in SSSPUnweighted as Stephan proposed and the job finishes. So I am able to proceed using this workaround.

An odd thing occurs now though. The distances aren't computed correctly for the SNAP graph and remain the one set in InitVerticesMapper(). For the small graph in SSSPDataUnweighted they are OK. I'm currently investigating this behavior.


On 18.03.2015 20:55, Vasiliki Kalavri wrote:
Hi Mihail,

I used your code to generate the vertex file, then gave this and the edge list as input to your SSSP implementation and still couldn't reproduce the exception. I'm using the same local setup as I describe above.
I'm not aware of any recent changes that might be relevant, but, just in case, are you using the latest master?


On 18 March 2015 at 19:21, Mihail Vieru <> wrote:
Hi Vasia,

I have used a simple job (attached) to generate a file which looks like this:

0 0
1 1
2 2
456629 456629
456630 456630

I need the vertices to be generated from a file for my future work.


On 18.03.2015 17:04, Vasiliki Kalavri wrote:
Hi Mihail, Robert,

I've tried reproducing this, but I couldn't.
I'm using the same twitter input graph from SNAP that you link to and also Scala IDE.
The job finishes without a problem (both the SSSP example from Gelly and the unweighted version).

The only thing I changed to run your version was creating the graph from the edge set only, i.e. like this:

Graph<Long, Long, NullValue> graph = Graph.fromDataSet(edges,
new MapFunction<Long, Long>() {
public Long map(Long value) {
return Long.MAX_VALUE;
}, env);
Since the twitter input is an edge list, how do you generate the vertex dataset in your case?


On 18 March 2015 at 16:54, Mihail Vieru <> wrote:

great! Thanks!

I really need this bug fixed because I'm laying the groundwork for my Diplom thesis and I need to be sure that the Gelly API is reliable and can handle large datasets as intended.


On 18.03.2015 15:40, Robert Waury wrote:

I managed to reproduce the behavior and as far as I can tell it seems to be a problem with the memory allocation.

I have filed a bug report in JIRA to get the attention of somebody who knows the runtime better than I do.


On Tue, Mar 17, 2015 at 3:52 PM, Mihail Vieru <> wrote:
Hi Robert,

thank you for your reply.

I'm starting the job from the Scala IDE. So only one JobManager and one TaskManager in the same JVM.
I've doubled the memory in the eclipse.ini settings but I still get the Exception.



On 17.03.2015 10:11, Robert Waury wrote:

can you tell me how much memory your job has and how many workers you are running?

From the trace it seems the internal hash table allocated only 7 MB for the graph data and therefore runs out of memory pretty quickly.

Skewed data could also be an issue but with a minimum of 5 pages and a maximum of 8 it seems to be distributed fairly even to the different partitions.


On Tue, Mar 17, 2015 at 1:25 AM, Mihail Vieru <> wrote:
And the correct SSSPUnweighted attached.

On 17.03.2015 01:23, Mihail Vieru wrote:

I'm getting the following RuntimeException for an adaptation of the SingleSourceShortestPaths example using the Gelly API (see attachment). It's been adapted for unweighted graphs having vertices with Long values.

As an input graph I'm using the social network graph (~200MB unpacked) from here:

For the small SSSPDataUnweighted graph (also attached) it terminates and computes the distances correctly.

03/16/2015 17:18:23    IterationHead(WorksetIteration (Vertex-centric iteration (org.apache.flink.graph.library.SingleSourceShortestPathsUnweighted$VertexDistanceUpdater@dca6fe4 | org.apache.flink.graph.library.SingleSourceShortestPathsUnweighted$MinDistanceMessenger@6577e8ce)))(2/4) switched to FAILED
java.lang.RuntimeException: Memory ran out. Compaction failed. numPartitions: 32 minPartition: 5 maxPartition: 8 number of overflow segments: 176 bucketSize: 217 Overall memory: 20316160 Partition memory: 7208960 Message: Index: 8, Size: 7
    at org.apache.flink.runtime.operators.hash.CompactingHashTable.insert(
    at org.apache.flink.runtime.operators.hash.CompactingHashTable.buildTable(
    at org.apache.flink.runtime.iterative.task.IterationHeadPactTask.readInitialSolutionSet(
    at org.apache.flink.runtime.operators.RegularPactTask.invoke(