-
Type: Improvement
-
Status: Resolved
-
Priority: Blocker
-
Resolution: Fixed
-
Affects Version/s: None
-
Component/s: BlobManager
-
Release Notes Summary:A new "nuxeo.bulk.action.garbageCollectOrphanBlobs.sample.modulo" property to trace samples of blobs to remove
-
Release Notes Description:
-
Tags:
-
Team:PLATFORM
-
Sprint:nxplatform #107
-
Story Points:2
Today, the result of a Full blob GC in dryRun mode is just an aggregation of blobs and sizes.
It's difficult to blindly decide to run the GC for real if the number of deletions is very high.
The dryRun mode could output few blob identifiers that will be deleted, so we can manually check and make sure it's correct.
A possible implementation is to trace at warn level instead of debug every 1000 items a blob id in GarbageCollectOrphanBlobsComputation#compute.
Also, in this ticket we should provide some mongo db queries to assert that everything is correct:
- make sure blob keys are up to date:
-- should not return anything db.default.findOne({"ecm:blobKeys": {$exists: false}, "content.data": {$exists: true}})
- make sure a blob traced for deletion is not referenced:
...
- is related to
-
NXP-32378 Improve Full GC traces
- Resolved