Hendrik Boom <hendrik@???> wrote:

>> Indeed, but guess what you get when you buy any storage appliance !
>> The last drives I bought were used 6TB ones - apart from the cost, I figured they’d be “run in”.
>
> I don't know what you mean by this.
> I haven't been able to guess and what is “run in”?

Buy any new storage appliance, and you’ll get new drives, all the same model, and almost certainly from the same manufacturing batch - so plenty of scope for common mode failures that could cause multiple drive failures in a short space of time. As mentioned, ideally you want drives from different batches to avoid this issue. With my work hat on, we deal with safety critical systems where we have massive redundancy - but a big concern is common cause failures that could affect multiple nodes and thus render the redundancy moot.

“Run in” comes from the automotive world. Going back, not too far, it was normal practice to “run in” a new (or newly overhauled) engine - avoiding prolonged runs at high speed and/or high power until the piston rings and bores had had change to wear to fit each other. Not running in a new engine could significantly reduce it’s life till it next needed an overhaul by not allowing the machining in the bores to wear down correctly, which can significantly affect the oil retaining properties, hence lubrication, and hence future wear of piston rings and bores.

The last drives I bought were used (presumably from some large bit barn having a periodic hardware refresh, so I figure could be from different batches (I bought them some time apart, and they had significantly different power on hours), and would be well past any early mortality phase (like many products, they tend to follow the classic bathtub curve https://en.wikipedia.org/wiki/Bathtub_curve).



Steve Litt <slitt@???> wrote:

> Take my one man business, for instance. Troubleshooters.Com has a guy
> skilled in ext4, bash, Python, POSIX etc, but nobody skilled in LVM or
> RAID. I'm obviously not going to hire an employee to do LVM and RAID,
> so upon disk problems I'd need to retain the services of a contractor
> proficient with LVM and RAID, and probably that consultant will take a
> day or two to get here. So the day or two I lose simply using ext4 is a
> wash.

Well I would suggest that your’s is not a typical business. I might also suggest that someone already proficient in "ext4, bash, Python, POSIX etc” could pick up madam and lvm in a few minutes, an hour or two tops if they are any good - assuming they want to of course. I would go as far as suggesting that learning madam and lvm (and setting up a mirrored pair) could be done is less time that it takes to change the disk every few years.

And you don’t have to use lvm. I use it because it makes life a lot easier whenever I want a new volume (e.g. when creating a new VM), or to resize a volume. The OPs question and resulting answers demonstrate what a p.i.t.a. it is to make such changes with raw partitions (with or without raid).

> But wait, there's more!
>
> I have a shellscript called bupsky with which I can back up the current
> directory just by typing bupsky. So on a hot project that's rapidly
> changing, I type bupsky every few minutes.

Like I said, some businesses will have different ideas - but again I would suggest that your’s is not typical. In any business I’ve known, relying on someone (even myself) performing a manual step like that multiple times in a day, and when busy, is asking for trouble sooner or later.


As mentioned already, we have the advantage over many in having this rich set of tools available to us - and having the freedom to choose which tools we use, and how.


Simon