[OAK-3818] Provide a way to check if all blob references are valid

• Type: New Feature
• Status: Resolved
• Priority: Major
• Resolution: Duplicate
• Component/s: blob
• Epic Link: datastore resilience
• Assignee: Chetan Mehrotra
• Reporter: Chetan Mehrotra
• Created: 21/Dec/15 15:53
• Updated: 21/Dec/15 16:52
• Resolved: 21/Dec/15 16:52

Description

For sanity and health check it would be good to have a way to check if all blob references are valid i.e. there are no orphan references in the repository.

We have most of logic in current BlobStore GC logic. So just need a way

1. Iterate over whole repository and collect blob references
2. Iterate over whole BlobStore and collect blob entries
3. Compare the two and entries which are present in #1 and not in #2 would be considered as orphan

Apart from check this job can also generate a report

1. Total size of valid blobs
2. Total size of garabage

This check can then we exposed as a JMX operation

Issue Links

• duplicates: OAK-3184 Consistency checker for data/blob store