| 1 |
Andrew Savchenko <bircoph <at> gmail.com> writes: |
| 2 |
|
| 3 |
|
| 4 |
> Ceph is optimized for btrfs by design, it has no configure options |
| 5 |
> to enable or disable btrfs-related stuff: |
| 6 |
> https://github.com/ceph/ceph/blob/master/configure.ac |
| 7 |
> No configure option => no use flag. |
| 8 |
|
| 9 |
Good to know; nice script. |
| 10 |
|
| 11 |
> Just use the latest (0.80.7 ATM). You may just nerame and rehash |
| 12 |
> 0.80.5 ebuild (usually this works fine). Or you may stay with |
| 13 |
> 0.80.5, but with fewer bug fixes. |
| 14 |
|
| 15 |
So just download from ceph.com, put it in distfiles and copy-edit |
| 16 |
ceph-0.80.7 in my /usr/local/portage, or is there an overlay somewhere |
| 17 |
I missed? |
| 18 |
|
| 19 |
> If raid is supposed to be read more frequently than written to, |
| 20 |
> then my favourite solution is raid-10-f2 (2 far copies, perfectly |
| 21 |
> fine for 2 disks). This will give you read performance of raid-0 and |
| 22 |
> robustness of raid-1. Though write i/o will be somewhat slower due |
| 23 |
> to more seeks. Also it depends on workload: if you'll have a lot of |
| 24 |
> independent read requests, raid-1 will be fine too. But for large read |
| 25 |
> i/o from a single or few clients raid-10-f2 is the best imo. |
| 26 |
|
| 27 |
Interesting. For now I'm going to stay with simple mirroring. After |
| 28 |
some time I might migrate to a more agressive FS arrangement, once |
| 29 |
I have a better idea of the i/o needs. With spark(RDD) on top of mesos, |
| 30 |
I shooting for mostly "in-memory" usage so i/o is not very heavily |
| 31 |
used. We'll just have to see how things work out. |
| 32 |
|
| 33 |
Last point. I'm using openrc and not systemd, at this time; any |
| 34 |
ceph issues with openrc, as I do see systemd related items with ceph. |
| 35 |
|
| 36 |
|
| 37 |
> Andrew Savchenko |
| 38 |
|
| 39 |
|
| 40 |
Very good advice. |
| 41 |
Thanks, |
| 42 |
James |
Archives