julianmorley
commented
6 years ago I think this folds into the regular Archive workflow:
- Audit queries PCC for records of archive copies
- Audit doesn't find what it's looking for, e.g. no record of an archive copy in AWS
- Audit tells Archive, "hey, check out this Moab! Is it everywhere it needs to be?"
- Archive checks Preservation Policy, sees that the Moab should exist in AWS, OCSA and Local Archive.
- Archive queries all endpoints. AWS says "file not found", but OCSA and Local both have it.
- Archive creates a new archive copy from Online moab, sends it to AWS (or copies it from Local Archive, sends it to AWS).
- Archive updates PCC with a record saying, yes, it's in AWS.
- The next time Audit runs, it sees all the records it expects to see in the PCC. Good job, team!
If an Archive object is bad (corrupted) or missing, a recovery process should generate a new archive copy from Moab to add or replace the storage copy.
It may be sufficient to simply initiate the full archive process as other Archive endpoints will ignore a new copy if it matches what they already have.