Monitor disk usage (per user and per site; split by transient/persistent; optionally weighted by #users who want persistent).
(Total: 4.00 h)
- Showing usage in Workbench
- Preliminary garbage collection results
- Set up a data manager service that generates reports for qr1hi periodically.
- Deliver reports as JSON in Log objects. (At least while our metadata/event/log systems are being sorted out.)
- Show on Workbench
- Workbench "user list" / "group list" pages will show latest usage numbers for each user/group
- Workbench "show user" / "show group" pages will show last few usage reports for that user/group
- Garbage collection & replication-checking results -- some attainable subset of...
- Report statistics (blobs with less/more than required replication)
- Report hashes of blobs with less than required replication
- Next candidates for deletion (oldest not mentioned in any wanted collection)
- Report cache window vs. % free space: "we can get 30% free space by deleting everything >54 days old" (this graph might conceivably show interesting usage patterns)
- Candidates appear in which unwanted collections (if any)? Which of those collections are currently intact (i.e., will get their first "holes" in next GC run)
- (?) Report of affected collections for each user (but we'd need "replication=0" instead of "don't want" in order to know this, right?)