We have a console app which executes 24/7, and after a few days/weeks, the disk was full, and after investigating it, it looks as the disk metadata was corrupted. We needed to run chkdsk to fix the disk.
The app does more than ~1M I/O operations a day: Files/Folders creation, deletion, permissions add/remove. Files and folders created are empty, and we even checked and they didn't consume a lot of disk space (not even 1%).
This lead me to investigate the issue, there I found that the MFT could be the one who consumes all space.
Can someone help and elaborate more if this is the correct issue?
If so, anyway to avoid it in the future ?
when you have many Operations in a day it can happen that the MFT consumes all the space. You are not filling up the Drive with data but with the information where the data is stored on the platter. When you run chkdsk it tries to restore the MTF with the data that is actually on the disk. The MTF is only edited when the corresponding file gets overwritten (after deletion) by the user/OS. For every new file that is not overwritten, there is a new part of disk space allocated on the drive. The best solution is to switch to a ZSF raid System their you can store millions of files with ease. Because ZSF works very differently from NTFS.