Backup During Rebalance - GSI count mismatch

Possible regression/test blocker while working on automation for backup during rebalance.

Test Flow:

1. deploy AWS cluster
2. import travel-sample bucket
3. Start back of travel-sample bucket
4. terminate/restart random cluster node
5. Wait for backup to complete
6. validate doc count, GSI etc. contained in backup

If the affected node includes the index service; the GSI observed in the backup is lower than expected.

gsi count in backup not matching. Expected 46 != Observed 30 

Terminating/restarting a non-index node results in no test failure with the expected 46 GSI in the backup.

Is this expected behaviour for index node failure?