When I run a task with Dataset.get the agent requests the dataset from a weird url. adamastor.gaiavf.local in this case.2022-10-03 17:50:17,556 - clearml.storage - ERROR - Could not download , err: HTTPConnectionPool(host='adamastor.gaiavf.local', port=8081): Max retries exceeded with url: /plants/.datasets/plants/plants.6eca8fc6690f4ef8aa6c6831fcea2a63/artifacts/state/state.json (Caused by NewConnectionError('<urllib3.connection.HTTPConnection object at 0x7fe746afe690>: Failed to establish a new connection: [Errno -2] Name or service not known'))Even though the agent’s config has different urls:api { # Notice: 'host' is the api server (default port 8008), not the web server. api_server: web_server: files_server: }And the env variables for clearml server are also set to the same urls.
How can I make it get the dataset from the right url?
Hi AdventurousButterfly15 , can you share the output of this small script?from clearml import Task session = Task._get_default_session() print(session.get_files_server_host())
When trying it I realized that my local clearml.conf had the old hostnames still ( adamastor.gaiavf.local ). Now your script returns the proper value of http://adamastor-office.periploinnovation.com:8081 . I will see if it works now!
