[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [linux-lvm] lvm over raid5 - sizes



Scott Serr wrote:
> I have alot of ideas going through my head.  Here is my situation:
>
> I've had 5 IDEs with 4 partitions each.  Each partition is in a RAID5
> md:stripped.
> md0 = hda5, hdb5, hdc5, hdd5, hde5
> md1 = hda6, hdb6, hdc6, hdd6, hde6
> md2 = hda7, hdb7, hdc7, hdd7
> md3 = hda8, hdb8, hdc8, hdd8
>
> md0-3 are in an LVM2 volume group.  This makes it really nice when
> expanding or adding another disk because I just pull say md3 out of the
> volume group... rebuild md3 bigger and then add it back.
> I just upgraded, replacing hde with a bigger disk and while the pvmoves
> are taking FOREVER...  I thought what is the right size for the RAID5
> chunk-size?
>
> I have chunk-size = 128k right now.
>
> The pv extents size is (always?) 4MB.
>
> Am I just shooting myself in the foot by not making my RAID chunk size
> atleast 1MB?  Can someone give me a reason to increase it?


You're shooting yourself in the foot with RAID 5 and a crazy setup like this in the first place. I sure hope this isn't for a production box :)

Just get more disks and use RAID 1.  It's faster, simpler, and more
robust.  You can still do all of the tricks that you're looking for.

Hey, maybe one of the disks has failed.  Standard RAID 5 behaviour as
designed in this case is to slow to a crawl.  Alternatively maybe you're
just experiencing what is known in the 'biz as "The Butterfly Effect".

With most striping, keep the chunk size as large as possible.  It's a
real waste to spend 10ms waiting for the disk platter to rotate, and then
only read 128kB of data.  Set it as high as possible.
--
Sam Vilain, sam /\T vilain |><>T net, PGP key ID: 0x05B52F13
(include my PGP key ID in personal replies to avoid spam filtering)


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]