![]()
How I Used Solaris OS and ZFS to Solve My Mac OS X Storage ProblemHere is a procedure for setting up the Solaris 10 OS with a four-disk, ZFS RAID-Z drive configuration to create a multi-terabyte file server for Macintosh computers. Community-Submitted Tech Tip How to Leave Comments or Tag Pages1. Register. |
|
Thanks Kevin, this is a really nice summary. Would you mind sharing the exact model of your SATA controller?
Cheers,
Comments (8)
Oct 25, 2007
DCJackson says:
Ditto, great article. And I really would like to know the make/model of your SA...Ditto, great article. And I really would like to know the make/model of your SATA controller.
Nov 27, 2007
KevinMcAleer says:
Hi, My SATA controller is a supermicro 8 port: Supermicro 8 Port SATA 2 Card (PC...Hi,
My SATA controller is a supermicro 8 port: Supermicro 8 Port SATA 2 Card (PCI/PCI-X) AOC-SAT2-MV8 available from scan.co.uk: http://www.scan.co.uk/Products/ProductInfo.asp?WebProductID=258223
Oct 25, 2007
Bards says:
Nice, I dont mean to nitpick but there are a few things I would like to comment...Nice,
I dont mean to nitpick but there are a few things I would like to comment on;
Do you realize that your 120gb SATA drive is now a single point of failure ? If it dies I'm not sure you would be able to recover the ZFS stuff on your 4 x 500gb pool. I don't know enough (yet !) about ZFS to be able to determine that you can indeed recover, or whether there is some metadata that you should be backing up on the system drive. To be sure, you should add another 120gb SATA drive and mirror it using SVM or better a hardware raid controller.
By my reckoning 4 x 500gb drives in a raidz configuration should give around 1.5TB of storage. Here is a quote from ZFS best practices;
"A RAID-Z configuration with N disks of size X with P parity disks can hold approximately (N-P)*X bytes and can withstand P device(s) failing before data integrity is compromised"
In your case (4 - 1) * 500gb = 1500gb.
Given that 500gb disks are not 500gb, more like 475gb raw and potentially even less when you put a filesystem or something useful on it. I'm interested to know how that could translate to 1.8TB. At a guess I'd say that 4 x 475g is closer to your 1.8TB than 3 x. Could it be that your data is only striped and not RaidZ ? If that were the case then you could well be one disk failure away from loosing everything. And with 5 disks in the system the chances are significantly increased.
Let us know.
Nov 27, 2007
KevinMcAleer says:
Hi Bards, Yes your correct about the single point of failure on the root disk. I...Hi Bards,
Yes your correct about the single point of failure on the root disk. I plan to mirror this, or at least backup the drive so I can recover it.
Your also correct about 1.5Tb of storage, although ZFS reports the total amount of available space as you use it, so to begin with I had 1.8TB although in reality it was at least only 3/4 of this. This is because ZFS has a variable strip size, depending on the amount of data being written. Also if compression is turned on this can account for the 'extra' space. I have compression enabled.
My setup is a raidZ:
pool: storagepool
state: ONLINE
scrub: none requested
config:
NAME STATE READ WRITE CKSUM
storagepool ONLINE 0 0 0
raidz1 ONLINE 0 0 0
c1t0d0 ONLINE 0 0 0
c1t1d0 ONLINE 0 0 0
c1t2d0 ONLINE 0 0 0
c1t3d0 ONLINE 0 0 0
errors: No known data errors
NAME USED AVAIL REFER MOUNTPOINT
storagepool 612G 755G 209G /export/storagepool
Hope this clarifies my setup.
I plan to add a few more disks to the storagepool after christmas as I plan to upgrade to solaris express developer edition, as it contains xVM and I want to have a few virtual machine images on there.
Thanks for the interest and comments!
Mar 06, 2008
rogermialkowski_2977 says:
He'll experience down time (since he'll have to replace the drive and reinstall ...He'll experience down time (since he'll have to replace the drive and reinstall Solaris) but he won't lose the ZFS pool. Upon reinstallation, Solaris will recognize any ZFS filesystems.
Sidenote, don't forget to scrub your filesystems regularly. I recently had a drive developing read errors but only realized it after doing a scrub. Now I do a scrub once a week since I have old SCSI drives.
Mar 20, 2008
KevinMcAleer says:
Hi Roger, A few weeks ago my main boot disk failed. Panic! I replaced the disk ...Hi Roger,
A few weeks ago my main boot disk failed. Panic!
I replaced the disk with a new one and installed solaris again. Once the installation was complete solaris detected the ZFS storagepool and I was able to import it. I had to use the -f option to force the import as solaris thought that this pool was being used by another system (the old, dead installation).
I scrubbed the disk and no errors were detected; the pooled storage disks are much newer than boot disk which was donated to me.
I've been using ZFS now for several months and I've got to take my hat off to Sun for creating a great filing system. My data is pretty safe.
I'm building another solaris box this weekend so I can use the ZFS send/recieve commands to replicate my most important data accross to a backup server.
Thanks for all your comments!
-Kev
Nov 27, 2007
KevinMcAleer says:
Hi DCJackson, My SATA controller is a supermicro 8 port: Supermicro 8 Port SATA ...Hi DCJackson,
My SATA controller is a supermicro 8 port: Supermicro 8 Port SATA 2 Card (PCI/PCI-X) AOC-SAT2-MV8 available from scan.co.uk: http://www.scan.co.uk/Products/ProductInfo.asp?WebProductID=258223
May 12, 2008
wildman72 says:
I was wondering what kind of throughput your getting on the AOC-SAT2-MV8. I hav...I was wondering what kind of throughput your getting on the AOC-SAT2-MV8.
I have the same card as with Solaris 10 1/08 I'm getting around a sustained 40MB/s with 2 drives in RAIDZ--but on one install I did I actually got burst speeds of 110MB/s but like I said they were just burst speeds not sustained.
The same drive mounted and with the RAIDZ produces a consistent 80MB/s. Weird. I've reinstalled from bare-metal more times than I would care to admit.