Druid hadoop indexer failing if hadoop cluster is in different time zone than hadoop client.

I am trying to hadoop indexer from hadoop client machine. hadoop client machine is in different time zone and hadoop server is in different time zone.

I am getting following exception. I am using druid version 0.7.3

Caused by: java.lang.NullPointerException

at io.druid.indexer.HadoopDruidIndexerConfig.getBucket(HadoopDruidIndexerConfig.java:356)

at io.druid.indexer.IndexGeneratorJob$IndexGeneratorMapper.innerMap(IndexGeneratorJob.java:208)

at io.druid.indexer.HadoopDruidIndexerMapper.map(HadoopDruidIndexerMapper.java:95)

… 9 more

you can try specifying the timezone you wish to use.

when submitting you can use -Duser.timezone to control the timezone used during submission.

you can set “mapreduce.map.java.opts” and “mapreduce.reduce.java.opts” to have the same -Duser.timezone configuration to control the timezone used during map/reduce phases.

-Michael-

Hi Jai,

It is recommended to run all your druid nodes and hadoop cluster with UTC timezone.

as Michael pointed out, you need to set timezone to UTC on all the druid nodes as well as in mapreduce properties.