Why did you choose freenas over nas4free?

[jgreco]

Well, you can think whatever you want, of course, but around here we're pretty much interested in doing things correctly to insure data integrity. A lot of the NAS4Free crowd are repurposing old random gear, and if and when they lose their data, ....?thenwhat? Of course they're very happy when they discover NAS4Free will run on their ten year old gear. Your first car was probably a used car, and you were very happy with it ... at least for a little while.

Now, if you can't afford a new car, you learn to live with it, and probably even convince yourself that you really like it and that you prefer it over having a new car. That's an "informed choice" of some sort, I guess, but I don't think I agree that this is really the sort of informed choice you imply that it is.

Around here, we pretty much beat on people to have systems that actually meet the specs of ZFS, including using ECC, setting up SMART, paying attention to redundancy and backup/replication issues, etc. The system is designed as an appliance from the ground up. If you're used to just loading "a Linux distro" onto a box and having a go at it, NAS4Free will feel more familiar because it is more along those lines. If you've ever used a storage appliance, though, you'll see the design intent behind FreeNAS and you'll probably recognize it as something that is more viable as an enterprise offering.

No one here will say you have to use FreeNAS. However, a little clue for anyone looking to jump ship to NAS4Free. Read up on ZFS and hardware requirements /here/. No one over at NAS4Free can honestly say that it is safe to use non-ECC memory in your ZFS filer - though a bunch of people have said exactly that. There is risk. We debate the risk here. There's a massive thread containing all the ins and outs here. After reading it, you'll have enough information to make your own informed decision, rather than trusting the comments from some random NAS4Free user who says "sure you can use non-ECC" because his experience with his one box is that he's never identified a problem caused by his selection of non-ECC. Will non-ECC work? Of COURSE it does. It's like driving in a car without a seatbelt. Of course you can do it and you always get to your destination safely ... unless one day, you have an accident. At which point the seatbelt might have saved you.

 

[Storms]

thanks jgreco for your informative reply

My drives have arrived. I've decided I'm going to trial Nas4Free with some sample data. Then I'm going to do the same tests with FreeNas. These are the tests I'm going to do (using RAIDZ1). If you guys have any suggestions about additional tests then let me know

1. Simulate a USB drive failure; see how it handles replacing the USB key with another one
2. Simulate a hard disk failure and replace with a new one
3. Measure data speeds on my home network; e.g watching a movie and transferring files

 

[HoneyBadger]

RAIDZ1

Run whatever OS you like, but please don't run RAIDZ1 on large consumer-grade drives. It's not a good recipe for safe data storage.

 

[anodos]

Run whatever OS you like, but please don't run RAIDZ1 on large consumer-grade drives. It's not a good recipe for safe data storage.

One more suggestion is to use appropriate server hardware that is well-supported by FreeBSD.
You may want to address the following:

1. How is ZFS configured under both of them?
2. What are the differences between default samba, nfs, afp, etc configurations?
3. What are the available tools for managing ZFS?
4. How are you notified of critical errors (i.e. disk failure, failed replication tasks, SMART errors, etc.)
5. How are you notified of ZFS scrub results?

(1) - (2) will probably account for any performance differences. (3)-(5) are important for long-term data security.

 

[pschatz100]

@Storms
Don't berate the folks here on the forum for being committed to FreeNAS. Many posters over on the NAS4Free forum are just as committed to their choice.

I originally started with FreeNAS 0.7. In fact, I stayed with 0.7 for a very long time after FreeNAS 8 was released because 0.7 provided the services I wanted at the time. I didn't make the update to FreeNAS 8 until FreeBSD jails and minidlna were supported. As it turns out, I REALLY LIKE jails. When I looked at NAS4Free back then, it didn't offer support for jails so I didn't go down that road.

My perpsective on FreeNAS versus NAS4Free:

The choice depends upon what you want to do. FreeNAS targets a user putting together a moderately sophisticated NAS, and the software design assumes that you have a greater amount of system resource (ECC memory, and more of it for the most part.) The FreeNAS Webgui provides easy access to a wider array of advanced tools than does NAS4Free. And jails are supported "natively".

NAS4Free is an older architecture that maintains support for low end hardware configurations, but at the expense of higher end features. Yes, you can install extensions for certain popular services. Yes, you can now use jails - but you have to install a separate jail manager. Yes, you can use ZFS but it is more difficult to set up and manage, and you have to run special tools to tweak and optimize memory for best performance. Not for the faint of heart.

For building a modern NAS with good management features and using ZFS, FreeNAS is the way to go. For re-purposing old hardware to build a small system using UFS, NAS4Free might be a choice. However, if you plan to use ZFS with NAS4Free, you will find yourself using more command line tools and the system requirements approach those of FreeNAS if you want decent performance. At the end of the day, it can be done but you won't have the nice integrated 'system' experience that you get with FreeNAS.

 

[cyberjock]

When I was "up for grabs" for a NAS OS, FreeNAS was on 8.0.4 and NAS4Free was on 9.2. At that time NAS4Free was running on a current kernel, FreeNAS wasn't. NAS4Free supported v5000 pools, FreeNAS was on v15. I still went with FreeNAS because it had more potential for future expansion since it has a commercial side.

I won't argue that FreeNAS is better than NAS4Free or vice versa. Just like with AMD and Intel, WD and Seagate, AMD and Nvidia, each has advantages and disadvantages. As long as they advantages align with you and the disadvantages aren't a critically big deal, that's what you should probably go with.

We've recommended NAS4Free quite a bit here for people that just won't buy 8GB+ of RAM. And if you go to their forums they really hate FreeNAS. There's definitely animosity between NAS4Free and FreeNAS from their side. I've heard the stories, but it's not my place to discuss it, so I won't.

 

[jkh]

However, I think the big selling point of NAS4Free is that it uses a more up to date version of freeBsd. This means better security, performance, stability, more command line options, etc

Please justify that statement. Which specific version, merged from which branch on which date? I don't see any evidence to support such an assertion in the actual bits being shipped. FreeNAS 9.2.1.x is based on FreeBSD 9.2. FreeNAS 9.3 is based on FreeBSD 9.3. Both have been also merging code from the FreeBSD-stable branch, as a review of the code on github will demonstrate to anyone who cares to look. By what evidence do you make that assertion?

It sounds to me like you've already decided on NAS4Free in any case and that's a perfectly fine decision for you to make - no one here will try to convert you to a religion that doesn't meet your criteria, but please don't make claims like that without the evidence to back them up. They are both open source projects, the source code is freely available, no hand-waving is necessary - URLs and branch tags can be cited instead.

 

[jgreco]

Please justify that statement.

Guessing he means that "FreeBSD 10" > "FreeBSD 9". We see this here sometimes. Actually at this point, with FreeBSD 10.0 being the current RELEASE, the truth is "FreeBSD 9" > "FreeBSD 10" for the meanings of "greater than" that we care about.

It is worth noting that FreeBSD typically supports at least two and usually three major releases at the same time.

For a data storage appliance, please explain any advantage that there might be to being on the bleeding edge, and compare/contrast that to the risks of "new" code. FreeBSD 10 is inherently LESS stable than FreeBSD 9; it is where they throw all the new features and where the most innovation is going on. "New features" and "innovation" translates to risk. It is also currently at a .0 release, though .1 is coming RSN.

From my point of view, I expect my data storage appliance to be able to store and retrieve files over the network. I've got some Sun 3/60's (~~= 1988? 25 years old!) around here somewhere that can do that. This is basic UNIX capability. For an appliance, we might want new drivers - which can often be backported, new features like an improved Samba - which are usually version agnostic, and ZFS - which FreeNAS has historically maintained semi-separately by cherry-picking updates anyways.

What I don't really want is a filer that is inadvertently unstable because some dev decided that it was So Fricking Important To Run The Very Latest Possible Release. For an appliance without a particularly good reason to be on the very newest major rev of the underlying OS, it seems like it'd be irresponsible to be running on anything other than a stable, safe, well-understood release of the underlying OS.

For a home user whose data might not be so valuable, the benefit to such conservatism is relatively lower, of course.

In any case, we don't really have anything against NAS4Free here. But as I previously said, come here first and at least read up on what we've found for solid hardware, and ECC, and some of the other resources we have. Being well-informed about this stuff is a good idea.

 

[jkh]

Guessing he means that "FreeBSD 10" > "FreeBSD 9". We see this here sometimes.

I am obviously quite familiar with FreeBSD release roadmap, but how does that relate to NAS4Free vs FreeNAS? There is no official NAS4Free release based on FreeBSD 10, just as there is no official FreeNAS release based on FreeBSD 10, so how is that point anything but irrelevant to the current discussion? That's why I asked for justification for those earlier claims.

 

[jgreco]

(I'm in the enjoyable position of being able to call some of you iXers N00bs today.)

You're definitely too N00b to have been around for it, but FreeNAS 8 was based on FreeBSD 8 while NAS4Free somethingorother was based on a FreeBSD 9 system. I don't remember exactly when this was, but I sure as hell remember it because I kept having to defend the conservative approach to selecting an OS for an appliance. And I keep defending the conservative choice of underlying OS. Seems like it happened as recently as last Friday.

What I hear him saying is that he wants a FreeNAS release based on FreeBSD 10, because FreeBSD 10 is "newer" (actually meaning numbered-higher, though you, me, and some of the rest of us know that 9.3 is actually *newer* than 10.0).

Now you and I both understand why that's a frellin' crock, but to the general public, there will always be some question as to why there's a lag. But to people who are used to running the very latest and greatest, the conservative approach can be confusing.

 

[jkh]

(I'm in the enjoyable position of being able to call some of you iXers N00bs today.)

You're definitely too N00b to have been around for it, but FreeNAS 8 was based on FreeBSD 8 while NAS4Free somethingorother was based on a FreeBSD 9 system.

Argh! Man, we are seriously talking past one another today! I don't give a frig' about the past in this discussion - it's not relevant to comparing the two solutions as they exist today! I'm sure if I delve far enough into the past, I can find some networking appliance based on FreeBSD 2.0.5 while another was based on FreeBSD 2.1.0, and we can debate the merits of the two, but we're not here to engage in nostalgia or rehash ancient, outdated arguments. All I asked was for the most recent claims to be substantiated using current statistics and current data since anything else is completely pointless.

 

[jgreco]

Argh! Man, we are seriously talking past one another today! I don't give a frig' about the past in this discussion - it's not relevant to comparing the two solutions as they exist today!

Meaningless. People get an impression and then they stick with it, Jordan. When people are comparing FreeNAS and NAS4Free, history is SO important for SO many reasons - not the least of which is that a whole lot of people who have recommended "FreeNAS" in the past are actually referring to the product now known as NAS4Free. And there's a problem there, because at one point some of the NAS4Free people made a big kerfuffle over the fact that they had a "newer" FreeBSD OS at the core. That meaningless tripe is undoubtedly still being regurgitated in various spots (see date on that link!), so even if it isn't true anymore, it is an issue in the minds of users.

Most of the users here couldn't identify the lineage of a FreeBSD derived appliance or where branches happened and/or merges happened. No one's likely to give you a researched answer to your request for justification. It isn't you and I who are talking past one another - you and I both know that the claim is irrelevant and also not true, at least today. The problem is that you're failing to recognize that users are getting this information from somewhere, and that someone needs to set them straight, not challenge them to a code diff. You're failing to actually communicate something meaningful, useful, and/or to the people who need straightening out.

So, you're welcome for the numerous times I've explained this to users in the style I used above. I'm kind of tired of it and one day I'll write a sticky about it, I suppose.

 

[jkh]

OK, point taken - reason is useless here and I shouldn't even bother. Perhaps we should simply delete these sorts of discussion threads as pointless exercises in comparative idiocy then? I don't see any more value here than in the average "vi vs emacs" discussion, and those certainly get nuked whenever they appear in development mailing lists.

 

[anodos]

I like the idea of a sticky. It will at least increase the chances that the next Gizmodo-style 'NAS shoot-off' will be able to plaigarize some accurate info.

Emacs? Is that a type of cheeseburger?

 

[jgreco]

OK, point taken - reason is useless here and I shouldn't even bother. Perhaps we should simply delete these sorts of discussion threads as pointless exercises in comparative idiocy then? I don't see any more value here than in the average "vi vs emacs" discussion, and those certainly get nuked whenever they appear in development mailing lists.

Part of the purpose of the forums is to help answer questions new users (or potential new users) have. They are often FILLED with misinformation, usually not maliciously, but rather just because they've done some research and were guided wrong by some blowhard blogger, or because they've taken their own experiences and tried to project that onto a FreeNAS build, or any number of other newuser class errors that come from not being you or me or someone else who already knows this stuff.

Now I'm not exactly the poster child for kindness and patience with n00bs, but I can and do try to put myself in their position and generally I am willing to give some sort of a shove in the right direction. Generally speaking the level I'm at is high enough that I'm never going to find the forums personally useful for getting help with a problem; I stick around because I like the crowd and occasionally see someone with a problem I find interesting. I also appreciate FreeNAS and it's a way to help say thanks to the devs.

So please take a bit of advice: Step away from the keyboard, Jordan. The question was being handled and there's no point at taking offense at a question that was valid but was incorrect in its assumptions. Deleting threads or calling them pointless exercises in comparative idiocy isn't helping anything. We try to impart clue here. The process may be painful but anyone willing to learn usually gets their question answered, or at least a big shove in the right direction.

 

[rogerh]

Well I don't use vi every day and have never got round to starting emacs; but then this is not a "development mailing list".

 

[jgreco]

Control-x control-c!!!!

 

[depasseg]

Control-x control-c!!!!

what about right-click?

 

[anodos]

Control-x control-c!!!!

I prefer to keep my religion separated from my software. :)

That being said, I've started teaching one of my boys to type using VI. He got interested one day when I was setting up a test server in our living room. I opened up VI, typed out a paragraph and had him copy it. We went back and forth for about an hour. One nice thing about VI is that I don't have to worry about him accidentally finding out how to close the session. :)

 

[Storms]

Here is my review and test results; bear in mind i'm only concerned with the NAS aspects of each system. I dont care about jails or any other features they may have. Before the tests I was on the fence about which way to go. I set up a RAIDZ pool for the tests.

Documentation and Community
Nas4Free - Small community and the documentation in parts is outdated and has mistakes
FreeNas - Large community and the documentation excellent

Speeds
The speed test was done by transferring a file from linux mint machine to the nas box. The linux mint machine was connected over WiFi to the router. The NAS box had a wired connection
Nas4Free - Got about 45 Mbps
FreeNas - Got about 45 Mbps

Installation and first boot
Nas4Free - easy installation, fast. No problems.
FreeNas - easy to set up usb stick but on first boot it kept hanging at 'waiting up to 5 seconds for ixdiagnose to finish'. waited over 20 minutes and it was still hanging. Eventually after several reboots got into freenas (i unplugged everything except the usb flash drive).

Web GUI
Nas4Free - very simple UI with just the basics. Very intuitive.
FreeNas - Much more options and features in the UI which makes it tricky to navigate. Also it's not obvious what a button does until you hover over it (e.g. volume status button is just a a button with a rectangle picture on it). The 'Alert ' is very useful to quickly determine the health of the system.

Simulating a damaged Disk
For this test, I took a disk offline. Shutdown the system. Replaced the disk with a another one. Then 'replaced' it in the OS.
Nas4Free - Took the disk offline via the UI, but was unable to replace it via the Ui. Had to replace it on command line using zpool replace -f. Then had a healthy volume again.
FreeNas - Took the disk offline via the UI, but was unable to replace it via the Ui. Had to replace it on command line using zpool replace -f. Then had a healthy pool again.

Simulating a damaged USB
For this test i backed up the confguartion and installed the OS onto a different USB stick and then restored the config
Nas4Free - Worked fine system booted up quickly, I updated the config and was back to status quo
FreeNas - Once again kept hanging at 'waiting up to 5 seconds for ixdiagnose to finish'. When I updated the config it said pool was degraded. For some reason the disk I had replaced in the last test was showing up as unavailable again. Gave up trying to get it Healthy again.


Conclusion
Due to the problems I had with FreeNas; I'm going to go with NAS4Free. When it comes to replacing a faulty drive it has to be easy and fault free. Afterall that is the whole point in having redundancy.