On Fri, Jun 21, 2013 at 10:03 PM, Rick Stevens ricks@alldigital.com wrote:
To me, it is a bit surprising how the failure of the hard disk can be predicted (and with a time projection)!
One of the lines you listed was the "reallocated sector count". Basically, when you buy a 500G drive, it really has more space than that. Some of the sectors are reserved and not reported to the system. When the drive detects a bad sector, it relocates it to one of the reserved sectors. Once it has relocated a certain number, it reports the drive as failing (possibly because it is out of reserved sectors to use).
I have already done a full backup. Should I wait for some days to check out whether the alarm is true or not?
Given there were a bunch of reallocated sectors, I'd replace it as soon as possible and then destroy it.
Thanks, Chris, for your very useful clarification -- I now understand better the things at issue.
The warnings from smartctl aren't guarantees the drive is dying, but that there's a high probability that it will at some point in the very near future. If you get an alert, the standard recommendation is to back up your data and get a new drive ASAP.
Think of the warning as the "check engine" light on a car. If it goes on, it could mean something as innocuous as you're out of windscreen washer fluid to "your gearbox fell out on the tarmac two miles back." Either way, you need to look at the output from the ODB2 scanner to see what's really wrong. The Linux equivalent of the OBD2 scanner is the "smartctl -a -H /dev/sdX" command.
Based on that info, you can decide if you must replace the drive immediately or if you can wait a bit. Given the relatively low cost of new drives, I'd replace it sooner rather than later. In fact, when I see a "deal" on drives, I'll buy a couple so I can have spares for just this sort of situation. I believe in a belt and suspenders as far as my data is concerned.
Thanks, Rick. I am going to follow your advice.
Paul