I'm about to start experimenting with FreeNas in order to prepare for when my real NAS arrives. Does anyone with experience have any helpful advice before I dig in?
Specs:
Front Backplane : SAS-846EL1
Rear backplane : SAS-826EL1
Boot:Supermicro SSD 16GB SATA DOM
Mobo : X8DTl-3F
Cpu: 2x Intel® Xeon® Processor E5645(12M Cache, 2.40 GHz, 6-Core)
Ram : 48GB ECC
Psu : 1400W
Storage: 15TB to start. Will add 3TB every 6months until I reach 60TB.
Last bump and I'm off to fucking around
I had horrible permission errors when using FreeNAS SMB shares between Linux and Windows
>Storage: 15TB to start. Will add 3TB every 6months until I reach 60TB.
How are you setting up your drives? You can't really add to a zpool if you're trying to keep it redundant, at least not one drive at a time.
hello again hording man-child. read the documentation, really. find out exactly what you want to do, how you want to share data in your network. learn to tell the difference between zvols, datasets, etc. plan your storage ahead. just adding a drive every month ain't gonna work just like that.
really, read the zfs documentation, it's well enough written for tech illiterates with good examples.
that's because i assume you are not reading the documentation and just click around hoping your trial and error gets you somewhere.
I liked OpenMediaVault a lot better.
Dedup is a trap for most datasets.
Ignore it unless you need to have multiple different forms of a single object or structure of file.
Compression is always good when you have redundancy.
When will brainlets learn? I hope this wont be storing any critical information.
>FreeNAS
What's the point of it besides on disk support of zfs?
comfort in functionality and administration, when you get older you might understand
I'm not setting up a RAID array. Each HDD will have its own zpool. I already know that if the drive dies inside the NAS then there's no recovery, and that's fine. It's highly unlikely both my original and backup will die.
that's weird.
I'm using FreeNAS via VM and I'm having no trouble so far.
>find out exactly what you want to do
My primary goal is to get this running as backup storage for my desktop. Since it's not Raid I'll need to find a way so that my data is sync'd every 12 hours or so.
I wonder though. Let's see my desktop's HDD's starts getting errors and begins corrupting files. Can those corrupt files be detected by ZFS?
>really, read the zfs documentation
Reading up on FreeNAS first, but yeah I definitely have to know how zfs works since I've only ever dealt with NTFS.
What do you mean? I mean, it's not critical in the sense that my original HDD's are in it. If a HDD dies, then I can just replace it and redo my backup
Freenas is for fags to stupid to configure things
Protip: learn FreeBSD directly
I was a user of FreeNAS 10 and while it's really comfy to use for everyday stuff, everything "out of the box" you'll want to add will be a pain in the ass and most likely will require fixing when you upgrade FreeNAS, even across minor updates.
For example in FreeNAS 10 the default locale is set to C, I wanted to set it to something with UTF-8 capabilities and despite doing the correct process for FreeBSD the setting was not persistent across reboots.
I finally found where to modify it to make it persistent across reboot (a FreeNAS only file) yet the next update broke it.
Filed a bug with them that where I asked them to make an easy and persistent way of setting the locale system wide permanently and was closed willy-nilly as it wasn't a required feature.
FreeBSD allows you to do everything that FreeNAS does and better, also you get to learn an operating system.
dedup is also a security issue if used in prod. there was a talk at the ccc "Memory Deduplication: The Curse that Keeps on Giving "
dedup can lead not only on datasets but any type of store to massive performance drops
man i remember that they had an issue with the certificate ID and you couldn't access your frontend anymore if you used a certificate from the CA from your freenas, that was some garbage to fix ...
that aside learning your way around the cli would be very useful.
>My primary goal is to get this running as backup storage for my desktop. Since it's not Raid I'll need to find a way so that my data is sync'd every 12 hours or so.
you can accomplish that in many ways. read first into zvols, zpools and datasets and chose based on that
>Can those corrupt files be detected by ZFS?
no. if you put corrupted data on any filesystem it will not know that it is corrupt, it will handle it like any data, give it a checksum and store it across multiple sectors.
>i configure everything by hand and i like to automate everything from scratch because i have no life and i like to give my life a sense of meaning
can't you go shitpost in a desktop thread?
>no. if you put corrupted data on any filesystem it will not know that it is corrupt
I see. Guess there's no way around that unless I used my NAS as my primary storage, which I can't do because that's too expensive right now. I'd need to buy 2x the amount of HDD's.
set1 of 30TB hdd's to copy onto an empty set.
set2 of 30TB hdd's to act as set1's backup
then I'll have a remaining 30TB of NTFS drives which I guess can be wiped and put into the nas too.
>can't you go shitpost in a desktop thread?
this, I've got too many other things going on to be tweaking config files all day.
>I see. Guess there's no way around that unless I used my NAS as my primary storage, which I can't do
sure you can, you migrate the first batch of data to the new drives, reuse the old ones for a new pool and copy the rest to the reused drives in your storage. chances are turning a machine off that has been running 24/7 with refurbished consumer drives, that they never come back again, but then again the first batch of data has been migrated, you should be alright.
having backup and general storage on the same devices is pretty dumb. buy 10tb drives and do offline backups.
when it works, it works and doesn't fuck it. when it doesn't work it will drive you out of your fucking mind trying to figure out what's wrong
that's my experience.
>I'm not setting up a RAID array. Each HDD will have its own zpool. I already know that if the drive dies inside the NAS then there's no recovery, and that's fine. It's highly unlikely both my original and backup will die.
Bad idea. Pool them. Do not use ZFS the same way you use windows storage, this caught me good back when I was a winfag.
ZFS is a paradigm shift in terms of storage and you should be using pooled storage.
It's not just about identifying damage, it's also about repair. If you have no redundancy you will have to manually restore files upon checksum errors. Individual pools are hard to manage over a single zpool
ZFS also also crazy smart in the event of dataloss, which is why you want to pool.
>Rebuild fails with no redundancy at 95%
Lose 5% of your files that it hasn't rebuilt yet and restore them from backup becasue it tells you what's missing
>Entire vdev dies in a pool that was gradually expanding
Only the files and chunks of files written to the damaged vdev are destroyed, restore from backup, easy peasy
>Gradual rot of data with redundancy from intensive writes means ZFS automatically rebuilds bad data for you, with no user intervention
Individual zpools require user input for every little error and your backups are often older copies of the same data which may not be what you want to restore every single time a hard disk fucks up.
just rsync to an external once a week that is cycled with a second drive stored offsite.
>30tb rsyncing every week once
sounds like a totally smart idea
>implying your kawaii anime uguu waifu shis is enterprise grade data
yeah, neck yourself