Hi @<1830053010663804928:profile|SwankyMosquito41> , sure, please see answers below 🙂
- What do I need to put for CLEARML_{API,FILES,WEB}_HOST env variables for the docker compose step? would that be the docker network internal addresses ( None , ...) or my external url e.g. None ?I think the external. Are you getting errors when you run docker compose?
- Is it possible to just completely remove the fileserver docker container and fully rely on Azure for artifacts?Yes, you also don't have to use it in general. Just point any files server configuration to your Azure blob
- I have the issue that preview links for datasets direct me to my file server's host name with the azure blob storage path, however the dataset download correctly works (pulls from azure blob)Can you elaborate on this? What have you configured/ran?
- Do you need to put an SAS or a real account key into AZURE_STORAGE_KEY?account name & key, as stated in the docs - None
- In the web app portal I get an error message when trying to set the azure storage credentials in the corresponding settings page ("Update storage credentials failed") - is it correct to set the credentials there or should you rather set them as env variables for docker compose, or in the services.conf?The web app is totally unrelated as it saves locally on the user's browser. For this you need the SAS
- Should you replace the keys in the default secure.conf?No
- can you configure GIT to use a ssh key instead of username/password?Yes you can - None
I usually use the search function in the docs whenever I need something 😄
Hi @<1523701070390366208:profile|CostlyOstrich36> , thank you so much. I was able to resolve the preview and url issues, this was a configuration issue with CORS and my reverse proxy, also I had an SSL issue.
What I am currently wondering is, if there is a way to put tags programmatically on the items directly visible on the datasets pages? I know how to put tags on the dataset versions, but I could only find that I am able to manually tag datasets themselves at top-level... Can I have the tags of the first dataset version be automatically taken for the top-level item too? (I don't mean the project, but the "folder" that holds the dataset versions)?