*protects and fixes your data*

*protects and fixes your data*

Attached: 1511349318837.png (276x209, 50K)

Other urls found in this thread:

github.com/zfsonlinux/zfs/pull/6900
breden.org.uk/2009/05/10/home-fileserver-zfs-file-systems/
serverfocus.org/zfs-best-practices-guide
github.com/zfsonlinux/zfs/issues/7401
docs.oracle.com/cd/E23824_01/html/821-1448/gkkih.html
github.com/maharmstone/btrfs
twitter.com/SFWRedditGifs

*takes up way more space than just the file*
*can't easily resize the volume*

>*takes up way more space than just the file*

what the fuck are you even talking

replication right

And a huge cache

Supposedly we're finally going to get the ability to remove vdevs. Only took em a decade.
github.com/zfsonlinux/zfs/pull/6900

why would ZFS replicate your files?

Attached: 1522568339616.jpg (322x321, 12K)

Just brought up my first zfs server with a 14tb raidz2 pool with a hot space and cache. Enabled dedup and compression.
I get ~100Mb writes. Haven't tested reads yet but loading videos is instant.
Im running OmniOSce and followed the zfs best practices guide. All of the issues are memes or related to the shitty zfs implementation on linux.

Because its a le botnet made by le (((equality))) (((hires))) and is also a le botnet for le ebil nigger dindu dindu jew (((parentheses)))

get rid of raidz2, and get mirror (raid10)

you'll get better performance and a bit more space since it's only 4 drives

you can't really disable cache with ZFS, because unused ram is wasted ram and i'm not sure what you mean by hot space.

100 mb/s is normal on 1 Gbps

I'm using 7 4tb seagate enterprise drives. One is a spare and I lose 2 to redundancy.
I meant i have an ssd cache drive. Its only used for reads though. I am probably gping to get 2 more ssd's in a mirror as a log device.
I meant hotspare but apparently I can't into typing today.

you only need a log device if you're doing synchronous writes. Remember that using an L2ARC also increases the memory you need to devote to your regular in-memory ARC, since it increases the amount of metadata ZFS wants to keep in the cache. Then again if you're using dedup on parity RAID you either don't care or you're willing to just throw a few dozen more gigabytes of RAM at it if need be.

redundancy

I got an old supermicro mobo+cpu+ram combo off ebay for ~$80
The chipset can use up to 32gb of ddr3 but right now there's only 8gb.
I might upgrade it at some point 32gb of ddr3 ecc is only ~$130 last i checked which isnt that bad compared to ddr4 prices.

How the fuck is that gonna work? Doesn’t ZFS stripe it across them?

Last week I bought an 8TB HDD and I'd like to put zfs on it so I can safely store anime and movies. Does anyone know of any easy-to-follow zsf tutorials on how to format, partition and manage a zfs drive? Thanks!

Attached: 1511742420641.jpg (1280x720, 128K)

There's an indirection table. basically the vdev will still exist as a phantom in pointer chains inside ZFS's data structures, they just add new pointers that refer to new copies of the data on present vdevs. This is kinda suboptimal compared to btrfs's rebalancing, but its something.

you can use it on one drive but then the main features of it won't work, since there's no redundancy. The whole point of ZFS is using two or more drives with it so it can protect you from drive failures and bit rot.

breden.org.uk/2009/05/10/home-fileserver-zfs-file-systems/
It also goes over how to do ACLs
I also have a link to the zfs best practices guide: serverfocus.org/zfs-best-practices-guide

*deletes your data*
github.com/zfsonlinux/zfs/issues/7401

you're referring to parity as "replication" and you're complaining about redundancy?

the absolute fucking state of this shitty board

cache with ZFS is a big no-no

install freenas or nas4free, plug the drive in, and click on "add volume", the rest should be self explanatory

HAMMER2 > that zfs shit
Stable™ on Dragonfly 5.2, released this week.

Attached: 1493173547534.png (150x167, 11K)

Yeah but then I'd have to run a BSD

explain why it's better

I thought ZFS was awesome until I actually used it and it was a constant pain in the nuts. It's great in theory, but not very useful for private individuals. Leave that shit to data storage professionals if you ask me.

The best plan is just to keep your storage needs low enough to use single large drives. If you need more than 4-8TB, you probably have a data hoarding problem. Re-assess whether or not you really need all that shit. Backup your essential data to some cheapo 5200RPM external drive in case you get fucked over; there's literally no point using ZFS for redundancy. That shit is all about uptime, which only matters if you're running a server. For private data storage, just use btrfs, ext4, or even fucking ntfs (it just werks). ZFS is just asking for trouble for youself.

>needs 64gb ram

>you can use it on one drive but then the main features of it won't work, since there's no redundancy. The whole point of ZFS is using two or more drives with it so it can protect you from drive failures and bit rot.
Interesting. But I was told that it can do checksumming so it can notify me when something bad happens.
Thanks user, exactly what I needed.

>install freenas or nas4free, plug the drive in, and click on "add volume", the rest should be self explanatory
OK! Will give that a try too.

With a single device it can tell you that a file is corrupted, but it can't do anything about it. You need more than one drive for it to fix things for you. With a mirror or RAID Z array if a drive coughs up bad data, ZFS takes care of asking the other drives for enough copies of it (or parity data) for it to reconstruct things for you. Whatever program was reading from the drives never even knows there was an error.

OK user. Thanks for letting me know!

I can't afford another 8TB drive right now but can afford a smaller one. Is it of any use if I chain the two in some kind of a JBOD array?

Alternatively, what FS do you suggest I use on the 8TB drive? I'd like to be able to open it not only under Linux (my primary OS) but also under WIn/Mac.

Thanks for any advice!

Attached: 1494478724451.jpg (1500x1500, 197K)

*stops being developed after its sole creator gets jailed*

But this isn't ReiserFS

What is my best option for building a storage array if I have a bunch of HDDs of differing capacities

Attached: 1516676327691.jpg (500x373, 45K)

Hans Zfs murdered his wife.

Doesn't matter. People that like implementing file systems are a special breed. He could be arrested any minute now.

see

throw them away and get HDDs with the same size

Attached: 1505150231457.jpg (400x400, 137K)

I'm not going to throw away 15TB of storage

Just use btrfs buttboy

>btrfs
Several issues with btrfs:

1) No encryption - I'd like my data to be completely encrypted. Not worried about gov but thieves that might steal my shit. Some ruidimentary form of encryption (AES256) is a must.

2) I should be able to open the volumes under Windows and macOS. I occasionally use Windows. In case of an emergency, I'd like to be able to have files accessible under other OSes.

Unfortunately btrfs is simply not ready.

>no encryption
LUKS/dm-crypt, brainlet.

>open volumes under Windows and macOS
lmao, brainlet.

cache with zfs is perfectly sound. It's even mentioned in the best practice guide.
You don't want to use files or make a single drive log vdev though.

If you use them in an array they will operate at the speed of the slowest drive and the size of the smallest. So it might not be worth it if you have one drive which is significantly smaller or slower than the rest.

this

>t. FS dev

see

>I should be able to open the volumes under Windows and macOS.
There's no good way to do this. All the advanced filesystems are on Linux/BSD only. You can share them over the network with Samba or the like though, but with windows your only options for a locally-attached disk are FAT32 and NTFS, neither of which you want.

why should I care which filesystem I use?
taking into account that I'm not a paranoid autist

>no encryption
block level encryption > filesystem level encryption
if you really need fs-level encryption, use encfs
>open under windows and macos
it's either cross-platform compatibility or a good filesystem, not both

I understand that. But zfs has both of the features that I wanted. There are zfs drivers for both the Win and MacOS and you can use those to access your disks in case of emergency.

>it's either cross-platform compatibility or a good filesystem, not both
That's true. But my use case is really to just be able to open the disk in case my main Linux machine goes down.

For example, if my linux partition is fucked and I have to boot into Win, I can't access shit if I have everything under btrfs/LUKS/LVM etc.

Also, encrypting zfs seems straight forward... it works on "pools" level.

docs.oracle.com/cd/E23824_01/html/821-1448/gkkih.html

Anyway, these are the primary reasons why I've been interested in zfs lately.

I have no idea what ZFS is or does sorry.

If you somehow manage to trash your OS to the point that it won't boot, then boot from a live image and fix it instead of watching cartoons you god damn manchild.

*data disappears*

>insults
not an argument.

I'm not arguing. I'm informing you that you are a stupid manchild.

Attached: smug_stocking003.png (244x189, 52K)

>posts anime images
>calls others manchildren
oh sweetie... quit projecting so hard.

I don't need a windows partition because I might break my OS.

Attached: smug_panty006.png (310x307, 124K)

You don't need a Win partition because you don't have a job.

*uses gigabytes of memory*
*destroys your data if your memory isn't ECC*
*has mediocre support due to licensing issues*
Just use btrfs unless you need raid5/6

Whatever (You) need to tell yourself, manbaby.

Attached: smug_panty015.png (494x493, 295K)

>I should be able to open the volumes under Windows
github.com/maharmstone/btrfs

>*uses gigabytes of memory*
don't use dedup and you don't need piles of memory
>*destroys your data if your memory isn't ECC*
factually inaccurate, the "scrub of death" is a retarded meme
>*has mediocre support due to licensing issues*
Support in the sense of working on many platforms? It's on the Linux or BSD of your choice. Support in the sense of commercial paid support? That shit is and always has been worthless.

>explain why it's better
For one thing, it isn't known to corrupt your data.
And how's that a bad thing? Linux's block layer is a clusterfuck.

thanks user! will try it.

>no encryption
LUKS/dm-crypt. Supported by Dragonfly.
Contrary to btrfs, HAMMER2 isn't shit.
>open volumes under Windows and macOS
KYS