-
Type: Task
-
Status: Resolved
-
Priority: Major
-
Resolution: Fixed
-
Affects Version/s: None
-
Fix Version/s: 8.3
-
Component/s: Nuxeo Drive
-
Tags:
-
Sprint:nxfit 8.3.3
-
Story Points:5
The VCS implementation of the batched remote scan, see NXP-19482 and DocumentBackedFolderItem#getScrollBatch relies on a "homemade" implementation of a scroll search and doesn't allow to sort the documents by path because it relies on a queryAndFecth core query.
Yet on the Drive side we would rather have the documents sorted by path to easily build the remote tree information in the local storage (need the FileSystemItem path).
That's why we now rely on the Elasticsearch scroll API exposed by the ElasticSearchService, see NXP-19631, in the case of a synchronization root and if Elasticsearch is enabled of course.
See ESSyncRootFolderItem#getScrollBatch.
This way we use the native ES scroll which allows to return the documents sorted by path.
Note: we consider that when scrolling through the descendants of a synchronization root its sub-tree must be indexed. This is most likely the case when performing the initial remote scan or in the case of a newly registered folder.
- depends on
-
NXP-19631 Expose Elasticsearch Scroll API in ElasticSearchService
- Resolved
-
NXP-19388 Setup a bench for the Nuxeo Drive optimized remote scan
- Resolved
-
NXP-19482 Drive: Optimize remote scan execution by using a scroll API
- Resolved
- is required by
-
NXP-19659 Limit number of threads calling DocumentBackedFolderItem#scrollDescendants
- Resolved
-
NXP-19443 Drive: optimize ScrollDescendants operation by avoiding DocumentModel loading
- Open
-
NXDRIVE-307 Remote scan: use new API to fetch the descendants of a folder
- Resolved
-
NXDRIVE-441 Test improved Remote Scan execution with volume
- Resolved