We are building a data catalog using CKAN, and we intend to index ClearML dataset via the ClearML SDK. Currently, we hardcoded the clearml.conf file into CKAN, but this means that all "get" and "create" datasets would be uploaded using a single ClearML credential that is independent of the user who is logged on. I would like to use the ClearML credentials based on the user who is currently logged in. Is there a way to query ClearML without using a pre-defined config file, but instead use the credentials of whoever is currently logged in?
Hi @<1583266807878258688:profile|PanickyCoyote41> , I'm not sure I follow. Can you elaborate a bit more into what exactly you're planning to do? Which credentials are you talking about that aren't ClearML's?
When running all tasks users submitted into a queue with a single agent?
I do not want to have a single clearML credential to facilitate all communications to clearML. Instead I want each user to use their own credentials. Is there a way to do that?
@<1583266807878258688:profile|PanickyCoyote41> instead of a clearml.conf file, you can use environment variables
Hi @<1523701087100473344:profile|SuccessfulKoala55> ,
Is there any way I can pass my ClearML credentials together with my query to ClearML without referring to the credentials stored in the clearml.conf file?
I can't use environment variables to store my credentials since my catalogue will be hosted on a server, so there won't be any environment variables that is accessible by all users.
The ClearML SDK can read credentials from the cleamrl.conf file or from environment variables. Why would you need to change them constantly?