NAS4Free Basic Questions

[renjithc]

Hi,

I've a truck load of questions here. They are simple ones but appreciate your inputs. A quick recap on what got me here - I've till now had my roughly 6.xTB of sata on (3x2TB NTFS, 2x1TB HW RAID1 NTFS) on my Windows 2003 ML110. Common sense suggested I move my 5.5TB of data to raid1 considering the disks are aging & I want a better chance at having my data intact incase of a disk failure. And I found NAS. Thats FreeNAS7, FreeNAS 8 & NAS4Free. Atleast I've figured out I should stick to NAS4Free. (HW Specs for my NAS Box - HP ML110 G5, 4GB, 3x3TB Seagate[NEW], 4GB USB for NAS4Free, 4port SATA2 Card, 1kVA APC UPS)

Now comes my questions - its not being lazy but my limited newfound know-how on FreeBSD\Nas4Free\UFS\ZFS will take me ages to figure out things & may cause valuable data loss. Here goes:
1) ZFS RAIDZ1 or SoftRAID RAID1 - Which one is faster, practical & reliable for 3x3TB thats 5.5TB of Single Partition? In my basic tests I got the feeling SoftRAID was slightly faster. Is this me or something I should be aware of? Again I'll sacrifice 10% performance for better accuracy at finding errors \ disk failures and the lot - if ZFS is as good as I hear.

2) Say I'm going with ZFS RAIDZ1 - I'm quite paranoid when it comes to recovering data from Software based RAID (I had the HW RAID mentality) in the event of the foll. failures. Say
(a) One Drive Fails - Is it as simple as buying a new HDD of same capacity, add it as spare Virtual Disk, use 'add_spare' to existing ZFS Pool, 'replace' failed disk and everything sync's magically? In case of 3TB will it really happen in a reasonable time frame & should I expect anything to go wrong at FreeBSD OS or NAS4Free Level?
(b) System Failure\Raid Card Failure - Re-Installing NAS4Free may be necessary. In this case will NAS4Free be able to import all my RAID Pools automatically\sync? Or I should add them one by one under VD, Pools and create them again? Without any loss of data?
(c) NAS4Free OS Crash. Thats as simple as re-installing new OS & resotoring the last good configuration to get all my pools?
If my above understanding is wrong - pls. suggest the correct method. All I want is safe storage, 5.7TB Single Partition from 3x3TB Hard-Disks, Ability to recover them safely in the event of any of these failures. Mainly all these storage would be used as Windows Shares.

3) Deleting ZFS Pools - Whats the reliable method to delete ZFS RAIZ\Strip Pools. Whenever I delete the pool, remove the virtual disk (VD)- still I can see these pools in the information\current config\syncronistion tabs? Even after removing \ restarting .. etc. At times they magically disappear after some restart.

4) Disk Identification Name - I'm getting lost with the whole ada0 to ad10 naming culture for disks detected. In my test's - at times of replacing disks or changing sata ports - these naming's seems to change and mess up with existing ZFS & SoftRAID Pools? How does this disk identification & naming work? And how will this be affected - say I rearrange all physical connection to sata ports & then do a restore of Nas4Free - will everything mess up or remain clean because its able to locate HDD serials or some other identifiers?

5) What does the ZFS -> Configuration -> Synchronisation TAB really do? Auto-import RAIDZ sets when adding existing ZFS HDDs?

Appreciate your thoughts and I get to build by NAS Box Safely.

Thx

[armandh]

I may be able to answer a very small part of your question

ZFS has a protection against bit rot [unknown data deterioration] as it corrects damaged files on reads
I have never found the bit of extra overhead a problem
both soft UFS raid and ZFS raid are easy to recover from non drive hardware failure IF YOU SAVED THE CONFIGURATION FILE
of course saving it on several client computers is ideal [its a simple web gui down load]

lastly, get a good UPS, as a power glitch on write can screw up the file journal
I prefer an [always on] double conversion type

PS
it wont hurt to label all the drives but once installed they are usually not moved about

[shakky4711]

Addition PLUS for ZFS--> NAS4Free, FreeBSD 9 and OpenIndiana (Which is additionally able to run as a Live-CD system at the Desktop version) are able to handle ZFS Pools v28, so you could plug in the disks to one of such systems, import the pool and it is up again. So you are not coupled to one specific system.

ZFS is designed for storage, the reliability and data integrity were the absolute goals at development. Standard Software Raid5 is declared as obselete from the FreeNAS developers a long time ago, so do not waste your time with it. BTW, nearly 90% from all crashed arrays at the old FreeNAS forum were standard Raid5 arrays...

Best regards
Shakky

[renjithc]

Thanks guys - Alright I'm convinced on ZFS Single Parity RAIDZ & not planning on changing the drives unless the raid card fails. I'm still concerned about the recovery scenarios.

Ok - when you say protection against data rot - are you referring to data scrub? Is this automated or manual job?

[aaronb]

1. ZFS tends to want a bit more processing power power and ram than a soft-RAID5. In a sense, the more processor and ram you give ZFS, the better it will run. That being said, ZFS gives a lot of features that RAID5 does not have while including everything RAID5 does have. One of the biggest speed advantages you will find with ZFS is initial setup and also re-silvering (if you fail one disk). ZFS is aware of what is on the file system and will only run the re-silver operations on those disk blocks. RAID5 will have to calculate the entire disk. On you 3x3TB array - RAID5 probably takes some time to format. ZFS takes a few seconds as it recognizes there is no data to duplicate.

2a. Both ZFS and RAID5 have procedures to replacing a failed disk. There is documentation for NAS4free on how to do this for both. ZFS will re-silver faster than RAID5, however if you have a good quantity of data it will still take a while. Since your OS installation and data disks are seperate, there is no reason to believe that a failed ZFS drive will affect the FreeBSD install.

2b & c. Simple to replate the OS and retain your data on your ZFS pool. I was playing with my NAS4Free install and corrupted my OS on the USB (monkeying with files on the stick). I reinstalled NAS4Free, went to ZFS area, hit the import button twice and sync'd config. ZFS pool and all data were again available immediately. Just had to finish configuring my networks shares and user accounts.

3. Best practice... Delete your volumes/datasets. Then delete the pool. Then delete the vdevs.

4. The ada0, ada1, etc is how Unix systems label the disks. They exist in the filesystem under /dev/. It takes a bit of reading, but the short part is that everything in Unix is treated as a file - disks, sound cards, processors, etc. So, when the system detects a disk, it adds it to the /dev/ file system in order of detection. I believe that when you create the VDEVs for ZFS it labels the disks in a way that re-ordering them in the hardware shouldn't matter. However, I haven't tried it.

5. It is the process for the answer to 2b & c about. It allows you to import a ZPOOL configuration that exists but the system lost. Quick experiment - put in clean install of NAS4Free on another USB and then go in and reconnect your existing pool on that install.