Description of problem:

Following the bug: https://issues.redhat.com/browse/OADP-2681

The datauploads resource was fixed and it spread evenly as much as possible. The backup has been significantly improved from 50min to 25min (Tested with 100 pvs with different data usage)

The datadownloads need to improve as well. The datadownloads queue per worker will be shorter and the restore time should be fasterOADP_1.3.0-Bug2681.txt

Version-Release number of selected component (if applicable):

OCP 4.12.9

ODF 4.12.9-rhodf
OADP 1.3.0-138

How reproducible:

Steps to Reproduce:
1. Create a namespace with 100 PVs using different data usage (mix namespace)
2. Run datamover backup
3. Delete the 100 pvs namespace
4. Run datamover restore

5. Monitor datadownloads resource (oc get datadownloads)

Actual results:

Datadownloads are not spread evenly across the workers.

Expected results:

spread datadownloads evenly as much as possible

Additional info:

Example for datadownloads spread:
worker000-r640 : 29
worker001-r640 : 54
worker002-r640 : 13
worker003-r640 : 3
worker004-r640 : 0
worker005-r640 : 1

worker000-r640 : 30
worker001-r640 : 52
worker002-r640 : 14
worker003-r640 : 3
worker004-r640 : 0
worker005-r640 : 1

Example for datauploads before the fix:
worker000-r640 : 22
worker001-r640 : 0
worker002-r640 : 39
worker003-r640 : 0
worker004-r640 : 39
worker005-r640 : 0

datauploads after the fix:

worker000-r640 : 19
worker001-r640 : 20
worker002-r640 : 21
worker003-r640 : 15
worker004-r640 : 12
worker005-r640 : 13