

What we found was the process FileSyncSvc.exe
Was Reading every File’s $REPARSE_POINT
And writing to the heatstore.edb file
We found this is a normal case when the FileSyncSvc.exe has an unclean shutdown so needs to rebuild the DB
Side effect of mitigation:
Since heatstore DB is removed, it takes sometime to rebuild the DB during which tiering may be slow, which can result in a higher disk utilization and low tiering efficiency for a few hours. Once the rebuilding process completes, the tiering moves at its full speed. Recall of already tiered files should not be affected by this.