gcloud dataflow jobs update-options

NAME
gcloud dataflow jobs update-options - update pipeline options on-the-fly for running Dataflow jobs
SYNOPSIS
gcloud dataflow jobs update-options JOB_ID [--max-num-workers=MAX_NUM_WORKERS] [--min-num-workers=MIN_NUM_WORKERS] [--region=REGION_ID] [--unset-worker-utilization-hint] [--worker-utilization-hint=WORKER_UTILIZATION_HINT] [GCLOUD_WIDE_FLAG]
DESCRIPTION
This command can modify properties of running Dataflow jobs. Currently, only updating autoscaling settings for Streaming Engine jobs is supported.

Adjust the autoscaling settings for Streaming Engine Dataflow jobs by providing at-least one of --min-num-workers or --max-num-workers or --worker-utilization-hint (or all 3), or --unset-worker-utilization-hint (which cannot be run at the same time as --worker-utilization-hint but works with the others). Allow a few minutes for the changes to take effect.

Note that autoscaling settings can only be modified on-the-fly for Streaming Engine jobs. Attempts to modify batch job or Streaming Appliance jobs will fail.

EXAMPLES
Modify autoscaling settings to scale between 5-10 workers:
gcloud dataflow jobs update-options --min-num-workers=5 --max-num-workers=10

Require a job to use at least 2 workers:

gcloud dataflow jobs update-options --min-num-workers=2

Require a job to use at most 20 workers:

gcloud dataflow jobs update-options --max-num-workers=20

Adjust the hint of target worker utilization to 70% for horizontal autoscaling:

gcloud dataflow jobs update-options --worker-utilization-hint=0.7

"Unset" worker utilization hint so that horizontal scaling will rely on its default CPU utilization target:

gcloud dataflow jobs update-options --unset-worker-utilization-hint
POSITIONAL ARGUMENTS
JOB_ID
Job ID to operate on.
FLAGS
--max-num-workers=MAX_NUM_WORKERS
Upper-bound for autoscaling, between 1-1000. Only supported for streaming-engine jobs.
--min-num-workers=MIN_NUM_WORKERS
Lower-bound for autoscaling, between 1-1000. Only supported for streaming-engine jobs.
--region=REGION_ID
Region ID of the job's regional endpoint. Defaults to 'us-central1'.
--unset-worker-utilization-hint
Unset --worker-utilization-hint. This causes the job autoscaling to fall back to internal tunings if they exist, or otherwise use the default hint value.
--worker-utilization-hint=WORKER_UTILIZATION_HINT
Target CPU utilization for autoscaling, ranging from 0.1 to 0.9. Only supported for streaming-engine jobs with autoscaling enabled.
GCLOUD WIDE FLAGS
These flags are available to all commands: --access-token-file, --account, --billing-project, --configuration, --flags-file, --flatten, --format, --help, --impersonate-service-account, --log-http, --project, --quiet, --trace-token, --user-output-enabled, --verbosity.

Run $ gcloud help for details.

NOTES
This variant is also available:
gcloud beta dataflow jobs update-options