Eu tenho um pipeline de feixe que lê do PubSub e grava no BigQuery depois de aplicar alguma transformação. O pipeline falha constantemente com um NPE. Estou usando o feixe SDK versão 0.6.0. Alguma idéia do que eu poderia estar fazendo de errado? Estou tentando executar o pipeline com um DirectRunner.

java.lang.NullPointerException
at org.apache.beam.sdk.io.PubsubUnboundedSource$PubsubReader.ackBatch(PubsubUnboundedSource.java:640)
at org.apache.beam.sdk.io.PubsubUnboundedSource$PubsubCheckpoint.finalizeCheckpoint(PubsubUnboundedSource.java:313)
at org.apache.beam.runners.direct.UnboundedReadEvaluatorFactory$UnboundedReadEvaluator.getReader(UnboundedReadEvaluatorFactory.java:174)
at org.apache.beam.runners.direct.UnboundedReadEvaluatorFactory$UnboundedReadEvaluator.processElement(UnboundedReadEvaluatorFactory.java:127)
at org.apache.beam.runners.direct.TransformExecutor.processElements(TransformExecutor.java:139)
at org.apache.beam.runners.direct.TransformExecutor.run(TransformExecutor.java:107)
at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
at java.util.concurrent.FutureTask.run(FutureTask.java:266)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
at java.lang.Thread.run(Thread.java:745)