One idea: is it possible to store usable credentials in advance and place them in a volume that the ClearML containers can access and then use?

I SOLVED IT, NO NEED TO READ FURTHER 😄

I'm a chump and didn't read the docs: None

Oh, I think I got overexcited and didn't look at this closely. So this ACCESS/SECRET key pair is on the agent-services container.

I can see that agent-services is simply a container running clearml-agent daemon --queue services --docker
and hey! that's what I was going to have to add to achieve getting a services worker!

I do have a question: Is there a way I can get the CLEARML_API_ACCESS_KEY and CLEARML_API_SECRET_KEY programmatically to give to this service? Or do I need to create these in the UI? If I create them in the UI and then destroy/re-create the EC2 instance the docker-compose.yml is running on, would that invalidate those keys?

I did a quick local experiment and observed that credentials created from the UI indeed become invalid if you delete the ClearML volumes.

• starting docker-compose locally
• creating a set of credentials from the UI
• hardcodign those credentials into the docker-compose file
• restarting
• the agent-services container started up and successfully became a registered worker
• I killed the docker-compose and deleted the volume folders
• restarted the docker-compose (with the same hard-coded credentials)
  The resulting logs from the agent-services container were this:

Using environment access key CLEARML_API_ACCESS_KEY=NL0V1RKGB0TN57L3OZRK
Using environment secret key CLEARML_API_SECRET_KEY=********

clearml_agent: ERROR: Failed getting token (error 401 from 

): Unauthorized (invalid credentials) (failed to locate provided credentials)

Oh my word, is this it? None

Can you set these to any strings and have them function as API keys?

^^^ For my own notes: this is the web request made by the frontend to create a set of credentials

This is already part of the docker-compose file,
https://github.com/allegroai/clearml-server/blob/master/docker/docker-compose.yml

@<1541954607595393024:profile|BattyCrocodile47> you mean like environment variables?

yes 🙂

The question I'm exploring remains: is it possible to acquire that initial set of ClearML API keys programmatically so that the manual steps of 1-4 above can be avoided for an initial deployment?

That could work! Is that an option? Something that lets me spin up the ClearML and get a services worker to connect to it without manual steps.