If you want this to be applied to all jobs that run on that agent then yes. Otherwise you can set it up on the task level as well

I would suggest opening dev tools (F12) and then filtering in the UI and see what is sent there

Hi @<1585078752969232384:profile|FantasticDuck7> , I think you can pass this in bash setup script when the docker spins up

Can you elaborate on how you did that?

Please follow the instructions.

Hi @<1652845271123496960:profile|AdorableClams1> , you set up fixed users in your docker compose, I would check there

@<1664079296102141952:profile|DangerousStarfish38> , can you provide logs please?

Hi @<1523701491863392256:profile|VastShells9> , I would suggest the following form to contact ClearML - None

Then you'd need to change the image in the docker compose or to spin up the webserver individually and make the necessary changes in the docker compose. Either way, you need a backend to work with the web ui

I don't believe this is part of the open documentation. In the enterprise there is an admin panel, SSO integration and RBAC on top of all the user management system. All of this is managed via an API like everything else in the system.

May I ask why you need docs on this?

Hi @<1668427989253099520:profile|DisgustedSquid10> , Unfortunately the open source has not programmatic user API, you can however remotely access your server and edit the user file live.
If user management is key, then the enterprise has full SSO integration including RBAC and of course API access

It's supported 🙂

I assigned both the pipeline controller and the component to this worker. Do I rather need to create two agents, one in services mode for the controller and then another one (not in services mode) for the component (which does training and predictions)? But, this seems to defeat the point of being able to run multiple tasks in services mode...

Yes. Again, the services mode is for special 'system' services if you will. The controller can run on the services agent (although not necessary...

Also, in the link above there is the warning

Do not enqueue training or inference tasks into the services queue. They will put an unnecessary load on the server.

I am not using the dedicated

services

queue on the server but I am doing training and inference in the pipeline component.

Steps of a pipeline should have dedicated queues with relevant resources to them

Really depends on how you want to set up your pipeline. I suggest going over the documentation and watching the youtube videos for a better understanding.

Hi @<1523701066867150848:profile|JitteryCoyote63> , you mean a global "env" variable that can be passed along the pipeline?

Hi @<1718437476634202112:profile|SteepAnt48> , you certainly can! But you already have HPO capability built in:
None
None

In the config file it should be something like this: agent.cuda_version="11.2" I think

And the experiments ran on agents or locally (i.e pycharm/terminal/vscode/jupyter/...)

Hi @<1734020162731905024:profile|RattyBluewhale45> , are they running anything? Can you see machine statistics on the experiments themselves?

Also, if you open Developer Tools, do you see any errors in the console?

Can you share a screenshot of the workers page?

MelancholyElk85 , it looks like add_files has the following parameter: dataset_path
Try with it 🙂

What version is the server? Do you see any errors in the API server or webserver containers?

I'm guessing this is a self deployed server, correct?

If you go into the settings, at the bottom right you will see the version of the server

@<1734020162731905024:profile|RattyBluewhale45> , can you try upgrading to the latest version of the server? 1.16.2 should have a fix for this issue

Hi @<1526734383564722176:profile|BoredBat47> , do you mean new debug samples or old ones? Please note that older debug samples were registered to the previous URL

You'd have to change the URLs in elastic itself

Also in network section of developer tools. What is returned to one of the 400 messages?