On 09/18/2018 01:41 AM, tudor@xxxxxxxxxxxxxxx wrote:
--------snip------
We do have a constraint in lvm2 to require the raid6 minimum for N to
be 3.
Configuring a raid6 LV (an array by MD terms) with 2 data stripes is
suboptimal for performance,
because data striping is minimal in this case. In addition, the
metadata overhead is maximal
for parity, P- and Q-syndromes being half of the brutto size of the
raid6 LV.
My apologies. I googled "brutto size" but came up with nothing. I do
have a basic understanding of P and Q syndromes so I think I have a
vague understanding of your meaning.
I have to apologize, I meant gross size
I understand it's suboptimal, but not non-existent. In particular why
is this a restriction and not just a warning?
The lvm2 code assumes data stripes to be more than parity/syndrome stripes.
Performance aside, I could also argue that there's a use case for a
minimal (and not degraded) set before expansion.
Sure you could, when expansion is planned one can create the raid6 set
with the intended total stripes though.
That aside, this is a constraint for such use cases and we may
eventually get rid of it.
Regards,
Heinz
Cheers,
Tudor.
_______________________________________________
linux-lvm mailing list
linux-lvm@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/
_______________________________________________
linux-lvm mailing list
linux-lvm@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/
