what does metaflow use RDS for? is it related to t...
# ask-metaflow
w
what does metaflow use RDS for? is it related to the UI?
1
a
metadata about runs and tasks is persisted in a Postgres database, that is typically RDS on AWS
w
what would happen if it's deleted? is it mandatory?
a
its technically optional in a sense that you can get started with Metfalow without metadata service, in that case that info is stored locally on your filesystem. But it quickly becomes necessary to have central metadata service and the database if you work with other people, want to share results and metadata about past executions with them
if you delete it, you basically lose the history of runs and associated artifacts
w
by associated artifacts, you mean the links from the runs to the blobs in S3? the blobs will still be in S3, just orphaned since we don't know what they are anymore?
d
correct — they would technically not be orphaned because the association also exists in S3 (each task has an associated file with it that relates the artifact name to the blob name (so the hash)). This is what metaflow uses when running (ie: when running it never reads the db, it only writes to it).