1. Home
  2. Docs
  3. Getting started on Snowplow Open Source
  4. Setup Snowplow Open Source on GCP
  5. Setup Google Cloud Storage (GCS) Destination
  6. Configuration options

Configuration options

Cloud Storage Loader specific options

  • --inputSubscription=String The Cloud Pub/Sub subscription to read from, formatted as projects/[PROJECT]/subscriptions/[SUB]
  • --outputDirectory=String The Cloud Storage directory to output files to, ends with /
  • --outputFilenamePrefix=String Default: output The Cloud Storage prefix to output files to
  • --shardTemplate=String Default: -W-P-SSSSS-of-NNNNN The shard template which will be part of the filenames
  • --outputFilenameSuffix=String Default: .txt The suffix of the filenames written out
  • --windowDuration=Int Default: 5 The window duration in minutes
  • --compression=String Default: none The compression used (gzip, bz2 or none), bz2 can’t be loaded into BigQuery
  • --numShards=int Default: 1 The maximum number of output shards produced when writing

Dataflow options

To run on Dataflow, Beam Enrich will rely on a set of additional configuration options:

  • --runner=DataFlowRunner which specifies that we want to run on Dataflow
  • --project=[PROJECT], the name of the GCP project
  • --streaming=true to notify Dataflow that we’re running a streaming application
  • --zone=europe-west2-a, the zone where the Dataflow nodes (effectively GCP Compute Engine nodes) will be launched
  • --region=europe-west2, the region where the Dataflow job will be launched
  • --gcpTempLocation=gs://[BUCKET]/, the GCS bucket where temporary files necessary to run the job (e.g. JARs) will be stored

The list of all the options can be found at https://cloud.google.com/dataflow/pipelines/specifying-exec-params#setting-other-cloud-pipeline-options.


The tests for this codebase can be run with sbt test.


You can run the job locally and experiment with its different parts using the SCIO REPL by running sbt repl/run.


The output of the Snowplow Google Cloud Storage Loader can be loaded into BigQuery. This becomes particularly useful if you choose to load the Snowplow bad stream into Cloud Storage. Indeed, loading that data from Cloud Storage into BigQuery is straightforward and makes it easy to investigate bad rows.