conda-incubator / conda-store

Data science environments, for collaboration. ✨
https://conda.store
BSD 3-Clause "New" or "Revised" License
137 stars 44 forks source link

[ENH] - Limit conda-store storage requirements #791

Open pavithraes opened 3 months ago

pavithraes commented 3 months ago

Feature description

While discussing the Nebari backup and restore functionality one thing that was discussed was the issue with ever growing conda-store storage requirements in production deployments. This is one of the core factors that breaks production Nebari installations, conda-store eventually runs out of disk space and Nebari becomes mostly non-usable until an admin goes in manually and deletes a whole bunch of builds one by one and also bounces to conda-store pod.

We have talked about have a mechanism where maybe only the last 1 or 2 builds of an environment are actually kept on disk and older ones are deleted. If someone wants to go back to an older build and activate it then it could be either pulled and unpacked from the s3 bucket tar.gz or it could be recreated from the lock file. This overlaps with https://github.com/conda-incubator/conda-store/issues/676 and with designing a conda-store backup and restore for Nebari.

Value and/or benefit

More stable and acceptable storage requirements in the default installation.

Anything else?

Originally shared by @dharhas on Slack

pavithraes commented 3 months ago

xref: #591