You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When starting a cluster, ensure that the chosen bucket is in the same region as the cluster, and warn the user if not.
Ensure that the bucket is not multi-regional (?)
Upon cluster deletion, confirm whether user is ok with cleaning out files from chosen bucket. If so, delete all uploads and outputs, but not the bucket itself.
Documentation:
Add notes about the potential cost premiums for not having the bucket in the same zone as the cluster (or having bucket be "interzone"?) and not deleting files after run ends.
Include instructions on manually checking the regions of buckets and clusters, and deleting files manually from a bucket.
The text was updated successfully, but these errors were encountered:
Upon cluster deletion, confirm whether user is ok with cleaning out files from chosen bucket. If so, delete all uploads and outputs, but not the bucket itself.
This was resolved by the bucket-monitor update. Currently, the bucket monitor will crawl the bucket every INTERVAL (which is configurable), and remove all files from a set of prefixes ("/uploads" and "/output" by default) that are older than 3 days (also configurable).
This should prevent files from building up and costing extra 💰💰💰.
We can also build a warning in the menu to help notify users that they might end up having interzon egress charges with a multi-regional bucket.
We cannot "ensure" that they do not have a mutlizone bucket, there is a use for that. I think the best we can do is inform them that a multizone bucket will cost extra money. This warning is addressed in #329
Functionality:
Documentation:
The text was updated successfully, but these errors were encountered: