Re: [smartmontools-support] Which attributes can estimate number of bad blocks?

[Gandalf C. <gan...@gm...>]

2016-08-02 14:32 GMT+02:00 Robert S <ro...@sp...>:
> My thresholds are based mostly on the fact I don't get to monitor the most
> of the hard drives. I get to see them once a month, if that. I also know the
> data is not backed up often, if at all, so I tend to be very conservative.
> It is cheaper to replace than to do data recovery.
>
> And for those I can monitor regularly, they are in servers where scheduling
> downtime is painful, and involves unpaid overtime for me (salaried). Having
> a server fail in the middle of production due to a dying hard drive is not
> good career-wise. Again, a whiff of problems and I replace. Since these are
> hot plugable in a raid, I don't have to do unpaid overtime. Things just get
> a little slow while the raid rebuilds.

These disks are on production servers.
Yes, I have backups, but as everybody, backups are never updated in
real time and recovering
a failed 3TB RAID-6 would become a real pain in the ass, as I have 22
virtual machines over it.
But recovering a 600GB disks in this raid-6 would take 28 hours and
these are very expensive disks.
Changing a working disks with another would mean trashing away about
300-400$ and I want
to be sure that this disk has really some issues.