Summary
Add a quiltx bucket reindex command that takes an S3 prefix (bucket + path) and triggers reindexing of all objects under that prefix.
Motivation
Currently, indexing only happens via S3 event notifications (SNS) configured during bucket add. There is no way to manually reindex a subset of a bucket — for example after:
- Bulk uploads that bypassed event notifications
- Recovery from a failed or interrupted indexing pipeline
- Retroactively indexing objects that predate bucket registration
Proposed behavior
quiltx bucket reindex s3://my-bucket/some/prefix/
Option A:
- Find an existing Quilt indexing API that works on prefixes
- Call that from quiltx
Option B:
- Enumerate all objects under the given S3 prefix
- Trigger reindexing for each object (via the catalog's indexing API or by replaying SNS notifications)
Results
- Report progress (object count, errors, completion)
- Should support
--dry-run to preview what would be reindexed
Open questions
- What is the right mechanism to trigger reindexing — direct API call, SNS notification replay, or something else?
Summary
Add a
quiltx bucket reindexcommand that takes an S3 prefix (bucket + path) and triggers reindexing of all objects under that prefix.Motivation
Currently, indexing only happens via S3 event notifications (SNS) configured during
bucket add. There is no way to manually reindex a subset of a bucket — for example after:Proposed behavior
Option A:
Option B:
Results
--dry-runto preview what would be reindexedOpen questions