Start a Local Agent without Default Storage Labels

Current behavior

Starting a LocalAgent will add storage labels by default, which can be confusing for new users and cluttered for users with many agents. Typically, starting a LocalAgent will produce the following by default:

[2020-09-01 15:34:50,612] INFO - agent | Starting LocalAgent with labels ['my-machine.local', 'azure-flow-storage', 'gcs-flow-storage', 's3-flow-storage', 'github-flow-storage', 'webhook-flow-storage']

Proposed behavior

Starting a Local Agent should automatically exclude the following labels: ‘azure-flow-storage’, ‘gcs-flow-storage’, ‘s3-flow-storage’, ‘github-flow-storage’, ‘webhook-flow-storage’

These labels should have some option to turn off by default or be removed by default.

Implemented here: https://github.com/PrefectHQ/prefect/blob/5edb9b39bb53e2cb11d37832f3f3f56bb1a3bc6a/src/prefect/agent/local/agent.py#L97

Example

prefect agent start -t “TOKEN” Produces:

[2020-09-01 17:56:00,216] INFO - agent | Starting LocalAgent with labels ['my-machine.local', 'azure-flow-storage', 'gcs-flow-storage', 's3-flow-storage', 'github-flow-storage', 'webhook-flow-storage']
[2020-09-01 17:56:00,216] INFO - agent | Agent documentation can be found at https://docs.prefect.io/orchestration/
[2020-09-01 17:56:00,216] INFO - agent | Agent connecting to the Prefect API at https://api.prefect.io
[2020-09-01 17:56:00,532] INFO - agent | Waiting for flow runs...

But can clutter the UI:

Potential Results: Agent management will be decluttered for users with many Agents. Quality of life change - specifying execution environments via flow affinity is a bit easier to navigate.

Please let me know if there is a current solution or reasoning I’m not aware of!

[clanzett]

Or maybe add some information about best practice regarding multiple agents on different hosts working on different runs of the same flow or something like that…

[flavienbwk]

I perfectly agree with @kmoonwright. Prefect shouldn’t force me to use labels and these should be disabled by default (such as how GitLab Runners system is working for example).