merca wrote:Great - that sounds like a nice setup.
As iLovefs noted, raid0 is not "nice", it is dangerous.
My setup of the main pool is:
- Code: Select all
[ihecc:~] root# zpool status zeus
pool: zeus
state: ONLINE
scan: scrub repaired 0 in 10h34m with 0 errors on Mon Nov 16 23:53:52 2015
config:
NAME STATE READ WRITE CKSUM
zeus ONLINE 0 0 0
raidz1-0 ONLINE 0 0 0
disk1 ONLINE 0 0 0
disk3 ONLINE 0 0 0
disk4 ONLINE 0 0 0
disk2 ONLINE 0 0 0
raidz1-1 ONLINE 0 0 0
disk15 ONLINE 0 0 0
disk14 ONLINE 0 0 0
disk12 ONLINE 0 0 0
disk7 ONLINE 0 0 0
logs
disk5s4 ONLINE 0 0 0
cache
disk5s5 ONLINE 0 0 0
errors: No known data errors
[ihecc:~] root# zpool list zeus
NAME SIZE ALLOC FREE EXPANDSZ FRAG CAP DEDUP HEALTH ALTROOT
zeus 21,8T 15,2T 6,60T - 27% 69% 1.00x ONLINE -
disks are all 3TB, log is 8 GB and cache is ca 50 GB. One vddv is in the Thunderbay via Thunerbolt 10, the other vdev is in a eSATA chassi. Loads of small jpegs, quite a few database dumps and a lot of time machine blocks (64MB).
The backup-pool is 6*4TB in a raidz2 setup with compression and dedup.
merca wrote:What sort of speeds are you seeing?
When I scrub, the average rate over the whole scrubbing 470MBps. If I remember correctly it tops just below 600MBps according to "zpool iostat 60".
merca wrote:Turning off atime. Will it still function like a normal OSX hfs+ drive?
No problems what I have seen.
merca wrote:ashift=12 - how much diskspace do you lose with this?
The important issue is not the diskspace, the important issue to align to modern disks which are 4096b/sector. Otherwise the speed will suffer greatly. Also, with the cost of 3TB disks I basically do not care that much about wasting diskspace.
merca wrote:does everything work nicely with OSX - or are there gotchas one needs to be aware of?
It sits on my server so my clients do not see the zfs. So no issue there.
It would be nice to have MacOSX to recognize zfs a bit more, for example as a potential Time Machine "disk".
I used USB3 when I stared playing around with ifs but it wa painful with all the disk renumbering. I have since changed to eSATA and Thunerbolt and now it just works.