[sf-lug] disk stiction & "old": Re: How not to do main storage (was: Belated writeup of last SF-LUG meeting)

[Michael Paoli]

disk stiction & "old"

It's rather like a weighted average, goes approximately like:
power on hours and nature of usage and environmental conditions
power cycles
power off hours and environmental conditions

Various things go on with spinning (and stationary) rust:

spinning - things wear, contaminants wear/spread, lubricants get sticky
more head movement and/or more vibration, more wear, more opportunity to
spread contaminants
warmer/hotter things wear/spread more quickly

power cycles - power up/down is somewhat stressful on the hardware
(thermal changes and other stresses)

power off - things still age/degrade, but mostly more slowly
colder generally degrades more slowly, warmer/hotter more quickly
longer and colder, more likely to stick, if power-off is warmer and
shorter less likely to stick - but contaminants spread/leak more quickly
when warmer/hotter

And yes, I've taken "dead" drives that were no longer spinning or
otherwise able to read nothing, and "brought them back to life".
I've generally had roughly 50% (+-20% or so) success rate with that.
Generally have drive well warm, connected, and powered up and well
settled in its normal operating environment, the as quickly as
feasible, remove, quick wrist flick action (trying to get platters to
rotate about spindle - so do a quick fast flick rotation of whole
enclosure as if the platters were spinning and dragging whole enclosure
with it - then rapidly stop, then reverse - back and forth a few to
several times), then as quickly as feasible get it reconnected.
Repeat as needed up to 10 times.  If the disk spins up and can be read,
immediately proceed to backup any data one wishes to access/recover.
I've found that such "brought back to life" drives are highly short
lived.  I'd say >60% of the time once powered down again they not only
don't spin up again ... but they never spin up again and can no longer
manage to coerce them into spinning up again by again using those same
techniques.

Some will try further techniques opening drive to attempt recovery -
I've never bothered.  I figure if the "wrist flick" method doesn't do
it, after that, probability of not fscking things up and rather to
quite irreversibly and getting it working by opening drive - is
pretty low - so I've never bothered/attempted.

Anyway, that's what I've generally found in dealing with lots of hard
drives for over 40 years.

On Fri, Sep 8, 2023 at 6:06 PM Akkana Peck <akkana at shallowsky.com> wrote:
>
> Rick Moen writes:
> That reminds me of one other thing I wanted to ask. In another thread you said
>
> > "old hard drives that get powered down may not come up again upon
> > restart (the "head stiction" problem)".
>
> Is that oldness measured in years, in power cycles, or in total spinning time? How old is old?
>
> I do my backups (home system, not talking enterprise here) to a small number of external USB hard drives, rotating them with one always offsite. The oldest is maybe 10 years old. They're powered off except when actually backing up. Whichever one is offsite goes months without being powered up.
>
> Should I be worried about stiction, and retire these disks when they reach a certain age? Should I replace them all with SSDs now that SSDs have gotten sufficiently cheap? Is there even any data on long-term SSD longevity and how it compares to spinning disks?
>
> A web search on head stiction found mostly hand-waving and anecdata.