Summary
When I query a table with timestamps in Datalab, I get values that are one hour behind the ones I get directly from BigQuery.
BigQuery
The results in the BigQuery console show rows starting at 2014-01-01 00:00:00 UTC.
These are the absolute values from the table and they are correct.
Datalab
The results I get in Datalab for the same query are rows starting at 2013-12-31 23:00:00.000
Data
The data I'm using is in these public GCS buckets:
gs://solutions-public-assets/time-series-master/GBPUSD_2014_01.csv
gs://solutions-public-assets/time-series-master/GBPUSD_2014_02.csv
Summary When I query a table with timestamps in Datalab, I get values that are one hour behind the ones I get directly from BigQuery.
BigQuery The results in the BigQuery console show rows starting at 2014-01-01 00:00:00 UTC. These are the absolute values from the table and they are correct.
Datalab The results I get in Datalab for the same query are rows starting at 2013-12-31 23:00:00.000
This looks like it might be a regression of https://github.com/googledatalab/datalab/issues/1195 I have tested in Datalab 1.2.20180511 and 1.2.20180818.
Data The data I'm using is in these public GCS buckets: gs://solutions-public-assets/time-series-master/GBPUSD_2014_01.csv gs://solutions-public-assets/time-series-master/GBPUSD_2014_02.csv