-
Type: Improvement
-
Status: Open
-
Priority: Minor
-
Resolution: Unresolved
-
Affects Version/s: 2.5.9
-
Fix Version/s: Not yet scheduled version
-
Component/s: Remote watcher
-
Sprint:nxMS 9.10.1, nxDrive 11.1.10
-
Story Points:3
While syncing a large tree that takes time, we still check for remote changes on a regular basis. It may introduces slowdowns:
- The request on the server, which is already working hard, may be slow and memory/time consuming. For a synchronization taking 3 hours, we have 360 calls to RemoteWatcher._handle_changes(), that means 360 calls to the server for a given client that is already syncing and where new changes will not be taken into account until the original sync is done.
- When the sync seems blocked or run for hours, it is even worse.
- We saw timeouts with that request in case of server overload, and so Drive will end on an not useful error, so for the sake of all, just skip it.
We could add a syncing flag or check queues size to prevent or enable that call.