[wellylug] Using UUIDs to identify filesystems (was Re: File recovery after a crash.)

[Daniel Pittman]

Jonathan Harker <jon at jon.geek.nz> writes:
> MDM Productions wrote:
>> Xav: All the other drives are there, and all have UUID's except for hda6.
>> Will that be restored automagically once we get the show on the road?
>
> UUIDs are optional. Their only advantage is if you introduce another hard
> disk, and your Linux kernel decides to shift what used to be your /dev/sda
> to /dev/sdb or something, and your /dev/sda6 partition becomes /dev/sdb6
> which fouls up your /etc/fstab. Using UUIDs instead would prevent that from
> happening.

I think it is worth noting that using the UUID, or a similar hardware
independent identifier, is well worth the trouble these days.

You are more or less correct in saying that, right now, you have to introduce
another hard disk for the device order to change.

"Hard disk" is a pretty flexible term now, though: a USB key is the absolutely
most common extra hard disk to add — and usually turns up before the rest of
the SATA devices, in my experience.


It is also notable that asynchronous scanning for devices, such as SATA, is on
the way: you can expect that within the next couple of years all your SCSI
hosts will be scanned asynchronously.

That means that if you have two HBAs — and I do, in my stock-standard x86
laptop, as do many of the other laptops and desktops I deal with — you
probably have the risk that device order is unpredictable on that hardware.

Regards,
        Daniel

-- 
✣ Daniel Pittman            ✉ daniel at rimspace.net            ☎ +61 401 155 707
               ♽ made with 100 percent post-consumer electrons
   Looking for work?  Love Perl?  In Melbourne, Australia?  We are hiring.