Getting the Source Code¶
This project is open source. You can get the code on GitHub.
How failures are dedicated¶
Failures are identified based on the two things"
- SMART health status
- Known bad attributes
Monitoring SMART health status is easy. SMART will flag a disk as being a failure and indicate a failed test.
However, there are some other attributes that do not necessarily flagged by SMART as a failure but are known to be associated with failing disks. BackBlaze (see https://www.backblaze.com/blog/hard-drive-smart-stats/) has determined that the following can be used to flag drives as likely failing soon:
- SMART 5 – Reallocated_Sector_Count. (1-4 warning, > 4 replace)
- SMART 187 – Reported_Uncorrectable_Errors. (> 1 replace)
- SMART 188 – Command_Timeout. (1-13 warning, > 13 replace)
- SMART 197 – Current_Pending_Sector_Count. (> 1 replace)
- SMART 198 – Offline_Uncorrectable. (> 1 replace)