1 |
On Sat, Jun 22, 2013 at 7:04 PM, Mark Knecht <markknecht@×××××.com> wrote: |
2 |
> I've been rereading everyone's posts as well as trying to collect |
3 |
> my own thoughts. One question I have at this point, being that you and |
4 |
> I seem to be the two non-RAID1 users (but not necessarily devotees) at |
5 |
> this time, is what chunk size, stride & stripe width with you are |
6 |
> using? |
7 |
|
8 |
I'm using 512K chunks on the two RAID5s which are my LVM PVs: |
9 |
md7 : active raid5 sdc3[0] sdd3[6] sde3[7] sda4[2] sdb4[5] |
10 |
971765760 blocks super 1.2 level 5, 512k chunk, algorithm 2 [5/5] [UUUUU] |
11 |
bitmap: 1/2 pages [4KB], 65536KB chunk |
12 |
|
13 |
md6 : active raid5 sda3[0] sdd2[4] sdb3[3] sde2[5] |
14 |
2197687296 blocks super 1.2 level 5, 512k chunk, algorithm 2 [4/4] [UUUU] |
15 |
bitmap: 2/6 pages [8KB], 65536KB chunk |
16 |
|
17 |
On top of this I have a few LVs with ext4 filesystems: |
18 |
tune2fs -l /dev/vg1/root | grep RAID |
19 |
RAID stride: 128 |
20 |
RAID stripe width: 384 |
21 |
(this is root, bin, sbin, lib) |
22 |
|
23 |
tune2fs -l /dev/vg1/data | grep RAID |
24 |
RAID stride: 19204 |
25 |
(this is just about everything else) |
26 |
|
27 |
tune2fs -l /dev/vg1/video | grep RAID |
28 |
RAID stride: 11047 |
29 |
(this is mythtv video) |
30 |
|
31 |
Those were all the defaults picked, and with the exception of root I |
32 |
believe the array was quite different when the others were created. |
33 |
I'm pretty confident that none of these are optimizes, and I'd be |
34 |
shocked if any of them are aligned unless this is automated (including |
35 |
across pvmoves, reshaping, and such). |
36 |
|
37 |
That is part of why I'd like to move to btrfs - optimizing raid with |
38 |
mdadm+lvm+mkfs.ext4 involves a lot of micromanagement as far as I'm |
39 |
aware. Docs are very spotty at best, and it isn't at all clear that |
40 |
things get adjusted as needed when you actually take advantage of |
41 |
things like pvmove or reshaping arrays. I suspect that having btrfs |
42 |
on bare metal will be more likely to result in something that keeps |
43 |
itself in-tune. |
44 |
|
45 |
Rich |