nathankw
commented
4 years ago I found that the default Cromwell conf file /cromwell/jes_template.conf in the corresponding Docker image (broadinstitute/wdl-runner) sets the location to us-central1, which is causing this issue since I'm specifying a network in a different region. So the main questions are thus:
1) Can location be parameterized in the default Cromwell conf file? 2) How can we supply our own Cromwell conf file ?
Thanks
Hello, I am curious to see what the default cromwell conf file is (the one referenced in wdl_runner.py as CROMWELL_CONF if the user doesn't overwrite the env variable). The reason is that I want to possibly use a customized one to hopefully fix an error related to not finding the specified network.
You'll notice in the command below that I use a specific network: gcloud beta lifesciences pipelines run \ --location us-west2 \ --regions us-west1 \ --pipeline-file GenomicsInTheCloudBook/wdl_runner/wdl_runner/wdl_pipeline.yaml \ --inputs-from-file WDL=${WDL},WORKFLOW_INPUTS=${INPUTS},WORKFLOW_OPTIONS=${OPTIONS} \ --env-vars WORKSPACE=$BUCKET/wdl_runner/test/work,OUTPUTS=$BUCKET/wdl_runner/test/output \ --logging $BUCKET/wdl_runner/test/logging2 \ --network "https://www.googleapis.com/compute/v1/projects/$MY_PROJECT/global/networks/$MY_NETWORK" \ --subnetwork regions/us-west1/subnetworks/$MY_SUBNET
But the pipeline always fails with the log file indicating this error:
""" PAPI error code 3. Execution failed: creating instance: inserting instance: Invalid value for field 'resource.networkInterfaces[0].network': ''. The referenced network resource cannot be found. """ So while the cromwell server VM is up and running just fine, it's unable to submit jobs - seems like only the default network may be supported - and or GCP project doesn't not enable a default network.