Good morning,
I am new to dvc, but I believe it might be the right tool for us. We are interested in having an easy way to store large files in our GitHub repos, which dvc does extremely well, and the added benefit of versioning is nice.
I would like the opinion of someone who is more experienced with dvc about a workaround I wanted to use to “solve” the Google Drive auth issues.
Background:
I wanted to setup dvc using a Google Drive folder, but that now requires a Google Cloud Project and we would have to ask several permissions to our University (and likely pay extra). To circumvent this I wanted to use our QNAP NAS, but that doesn’t work as another user also pointed out in a dvc-ssh issue (asyncssh.misc.ProtocolError: Too many authentication failures · Issue #47 · iterative/dvc-ssh · GitHub).
Workaround:
To solve this, I directly added as dvc remote a folder on my mac which is stored in the Google Drive directory, as such I hacked my way around the drive auth issues. For our team this would be fine as I setup dvc to point to a symlink which would be valid on all unix systems.
Question:
Would this workaround work just fine long term or will it come to haunt me in the future? I am asking because I was one of the naive guys who cloned git repos in an iCloud folder and later discovered that iCloud was messing up my repo, and I want to avoid this with dvc.
While I am here, I would like to ask some clarification regarding dvc gc. From the docs I understood that this command should allow me to clear old data that is no longer needed. Since we don’t have huge datasets, but also we don’t have much cloud storage, we are interested in keeping only a short history of the changes (e.g., a year or so). Would dvc gc allow me to do this? I assume that once history is deleted, past commits will point to data which is no longer there, so I would need to be careful.
Thank you all in advance for the help!