HD failure.... What's a few bad sectors...

[JSolo]

Relative tells me their pc is acting up. Do a quick look. Drive is a wd60ezrz, ~4 yrs old (maybe 5).



Quick look turns into a 6 hr project while salvageable data is backed up (~600GB). After restoring to new drive, do a binary comparison to determine which files were damaged.

There's more, but here's a snippet.. I guess motorweek has seen better days.



Looks like just some of the motorweek files took the hit. Other media content intact. I suppose it could be much worse. User profile folder is mapped to this drive. Fortunately those files are intact too. Should probably schedule something to run a weekly smart check.

 

[stoopid]

PSA: Each spot data can be stored on a drive is called a sector, and over time as they get written and erased their ability to accomplish this wanes. Eventually they can no longer do this reliably and the drive will flag them as 'bad'. The "uncorrectable sector count" I believe is the SMART entry that tracks this. The drive will know not to use the sector moving forward, but any data that was expected to be there that isn't is lost. The file that contained that data won't usually be readable because there's a section of data suddenly missing ('corrupted'). There's not really a way to recreate the lost bits, although some advanced techniques beyond my paygrade claim to be able to find hints of what might have been there last. Recovery services are expensive though.

Once you start seeing bad sectors it's best to get the data off the drive asap/replace, it's kinda like one tire going bald means you're likely needing to replace the others soon too. Most of the time (generalization) sectors fail from use, so like balding tires it's only a matter of time before more worn sectors are found.

 

[JSolo]

^^Good analogy with the tires. Although unlike tires, drives don't need to be replaced in pairs.

I generally try to rotate drives every 3-5 years. I got a few spares here.. An assortment of 6-8tb drives.

 

[stoopid]

^^Good analogy with the tires. Although unlike tires, drives don't need to be replaced in pairs.

Well... you can argue that drives in RAID that were deployed at the same time will fail at roughly the same rate. I've seen servers with a drive failure followed by multiple more in the weeks/months that follow. Typically when the first one fails I suggest purchasing two or more, knowing this is likely the outcome. It's important to get the replacement installed and rebuilt in the array before the next one fail [so having them on hand versus waiting for one to arrive by mail is important], or we're recovering from backup if there's no additional hot swap. Larger companies with larger arrays will sometimes have raid+1/2/3 as a safety net.

I also usually quote for a new server at that point too, if the customer seems nervous about any other hardware possibly failing [99% of the time these failures happen at the 3+ year mark from daily use, so a new server is certainly warranted especially after 5 years].