uktrade/jupyters3

Create bucket if it does not exist

Closed this issue · 1 comments

remche commented

It could be handy to create the bucket if it does not exist (opt-in behavior ?). In our case, we would like to create a bucket per JupyterHub user, on first time they run a singleuser instance.
I'm not familliar with Tornado but I could try to do a PR if needed.

Thanks for you clean work on this !

michalc commented

Hi @remche,

Thanks for the suggestion. However, at the moment, I think I'm going to say no to this. Essentially because:

  • I don't foresee needing this feature myself

  • I would also recommend against having a bucket per user, and instead a single bucket, but having a key prefix per user. For one thing, there is one global namespace for S3 buckets across all AWS accounts, so someone else could have taken whatever bucket name you might come up with. For another, there is never a need to "create" a prefix like you need to create a bucket. You just PUT objects with keys that starts with that prefix. So there is less to setup.

Thanks,

Michal