FreeNAS Without the RAID

[pro lamer]

I want a Linux free program that manage separate NTFS formatted HDDs.

First a disclaimer: my SnapRAID literacy is none. I've just googled for a moment and found their page https://www.snapraid.it/faq#fs

Having said that they advertise they support any filesystem. SnapRAID is CLI but they mention

the Elucidate external GUI, and a plugin for OpenMediaVault.

I've never heard of the former. The latter: a quick Google-foo said OMV supports NTFS somehow. I don't know the details.

In other words: you may consider learning more about SnapRAID and whether it suits you.

Sent from my phone

 

[carlmart]

OK, it seems I was wrong on two things:

1) Apparently I can use FreeNAS with separate individual HDDs without any problems whatsoever. With desktop HDDs. Do you agree?

2) The zfs format is very good, and apparently even better than NTFS.

Now the question is to find a recommended 10TB HDD desktop, which it doesn't seem to be as common as NAS versions. Is the WD BLACK very good?

 

[carlmart]

Why shouldn't I use a NAS HDD if I'm planning to assemble a server with individual separate drivers?

Desktop HDDs are very scarce in the 10TB size.

 

[garm]

without any problems whatsoever.

Dspends on your definition of “problem”, ZFS will not be able to heal data.

The zfs format is very good, and apparently even better than NTFS

There is really no comparison..

Why shouldn't I use a NAS HDD if I'm planning to assemble a server with individual separate drivers?

Because of the arguments @Arwen stated..

 

[carlmart]

What do you mean by ZFS will not able to heal data? How does NTFS heal data?

 

[danb35]

2) The zfs format is very good, and apparently even better than NTFS.

This makes it sound like you consider NTFS the epitome of filesystems. It isn't. Not by a long shot.

What do you mean by ZFS will not able to heal data?

Exactly what he said. If data on disk is damaged, and you have no redundancy, ZFS won't be able to heal that data.

How does NTFS heal data?

It doesn't.

 

[carlmart]

Exactly what he said. If data on disk is damaged, and you have no redundancy, ZFS won't be able to heal that data.

So if ZFS doesn't heal the data, and neither NTFS does, how different would that be from what I already have on my desktop? Which is two 4TB HDs holding my data?

My intention with the server is to have the HDDs, non-raided, on a separate box, at least three if they are 10TB, with another 10TB HDD turned off that could be used as backup if any HDD shows any sign of problem.

Only one of the HDDs will probably be used daily to play the videos or music, usually at night, so perhaps the HDDs should be in sleep mode, at least two of them. I could arrange to place files that will be played daily on the HD that will never turn off. Maybe that way I could prevent future problems.

 

[danb35]

with another 10TB HDD turned off that could be used as backup if any HDD shows any sign of problem.

So your idea is that you won't bother backing things up unless/until you suspect there's a problem? Hope your data isn't very important--as you've already been told, drives do fail without warning. A much more sensible (though still not recommended) solution would be to put those four drives into a single RAIDZ pool. That way, all your storage is pooled, you have redundancy, and a single drive failure will not result in data loss.

But with that said, yes, FreeNAS will let you do that. You'll be fighting the way it's designed to work, and it still won't let you share ext4/NTFS disks (which you'd previously mentioned as an absolute requirement), but nothing's stopping you from having more than one pool.

 

[carlmart]

I was wrong on NTFS being an absolute requirement.

The world that has been described to me with RAID, without RAID, with redundancy, without it, seems to leave only one option to "absolutely" safe: to have 100% redundancy. Which means spend twice the money I was planning on HDDs. That is what is not an option.

So it seems the system I use now, which is burning my video files in DVD-DL or BD disks, will continue to be. The error rate I get with such a system along the years, which is about 10 years since I started, has been much, very much below the rates I have been told for HDDs problems, even with total redundancy.

So I will have MY redundancy in optical disks, and the practical side of being able to still play things from the server. Relatively safe as it seems it will be.

A single RAID pool is still not recommended, so why would I use it?

The problem I have to solve now is finding reliable 10TB desktop HDDs, which doesn't seem to be a popular size for desktops.

 

[danb35]

A single RAID pool is still not recommended, so why would I use it?

Because it still beats no redundancy.

We assume that people use FreeNAS because they care about their data, and therefore are willing to invest adequate resources to protect it. You clearly don't and aren't. That's up to you--it is, after all, your data--but don't expect a lot of help in doing what most of us consider a very foolish thing.

 

[Chris Moore]

A single RAID pool is still not recommended, so why would I use it?

You have a language problem. You are not understanding what you are being told. We say one thing and you conclude something totally opposite.

 

[Heracles]

Hey Calmart,

Lets try to restart this from the beginning...

What are you ready to accept as the consequence of a single defective hard drive ?
--You accept to loose eveyrthing ?
--You accept to loose only what was on that drive ?
--You do not want to loose anything despite that failure but accept to be high risk until the drive is replaced ?
--You do not want to loose anything despite that failure and also want to remain safe until the drive is replaced ?

If you accept to loose everything, you will end up with something that is not recommended because it is too risky and people always say they accept the risk until bad things happen. FYI, the basis of the solution would be a serie of 1 drive vDev grouped in a single pool should you wish to do it despite our recommendation.

If you accept to loose what was on that drive, you will end up in a setup that is not recommended because you will not benefit from any of FreeNAS's major points. FYI, the basis of the solution would be a serie of pools, each with a single drive vDev. Again, that is only if you want to do it against our recommendation.

If you wish to be protected from a single failure and ready to expose yourself to higher risk, you will end up with something that is not recommended because that protection is too low and will betray you down the road. Because your drives are of different sizes, you will also waste space. FYI, the basis of the solution would be a RaidZ1 pool and again, not recommended but do as you wish.

Should you wish to be safe even after loosing a single drive, you need to look at Raid10 or RaidZ2. From here you can have something that will follow recommendations but to design that solution, you must be ready to sacrifice the equivalent of at least 2 drives. You also need to know that FreeNAS will always re-format your drives before using them with ZFS. As such, you will have to empty these drives first.

Hope this help you understand a little better what the situation is,

 

[carlmart]

You have a language problem. You are not understanding what you are being told. We say one thing and you conclude something totally opposite.

No language problem at all. I understand perfectly what everyone is explaining. Just do not agree with your conclusions.

You do not seem to care for my experience using HDDs, particularly since I started watching temperatures and checking HDDs once a week their health with CrystalDiskInfo.

I also mentioned a friend of mine who has been dealing with HDDs professionally, in large production studios for 30 years, being in continuous connection with other technical directors and manufacturer, NAB exhibits and all. He also doesn't agree with your conclusions. He asked me if you think I'm assembling a server for a production company, which certainly is not my case: is a home server.

 

[garm]

He asked me if you think I'm assembling a server for a production company, which certainly is not my case: is a home server.

Please explain what the difference would be in your opinion?

 

[carlmart]

Please explain what the difference would be in your opinion?

So my friend is right then. Then the problem is not mine if you can't see the difference.

 

[garm]

So my friend is right then. Then the problem is not mine if you can't see the difference.

The problem is yours, I don’t have a problem with my server. You should examine why you want FreeNAS, as previously stated by others. If you want to lower the probability of data loss, zfs and data redundancy is your best bet. And FreeNAS is probably the best implementation. But you are free to do what ever you want with your stuff.. if you want to accept the risk of storing it without redundancy or on an inferior file system, by all means do.
You have had the benefits of ZFS and redundancy explained to you but you keep drawing strange conclusions from what we say, so please tell us what you want to achieve apart from what you already have stated?

 

[danb35]

*plonk*

 

[carlmart]

Of course the problem is mine.. All I want is a server, and it doesn't have to be necessarily FreeNAS.

So I get to "strange conclusions" to you, because I do not agree and question what you recommend? You even call what I have in mind "inferior", because I question the reality of your "excessive precaution system", to call it somehow. What about inverting the concepts and claiming (because we can only put claims here) that you go over the top, by equalizing your home needs to those of a video production company?

It's fine and perfectly OK to think your server system is what you want, but you do not seem to accept, even antagonize, putting in doubt or questioning some of your ideas. OK, you may have a point with your proposal, but I also have a point with mine. It's not a question of language, it's a question of having an open mind.

But it seems quite unlikely we will get to an agreement here.

 

[Tigersharke]

The prior post Heracle's post-528500 seems to lay it all out fairly well. I've been in unrecoverable situations even with ZFS because I didn't set things up correctly or I made some other catastrophic error, and most certainly because I did not comprehend how exactly to recover from my specific situation. So I had to start over fresh completely, or had to rebuild my home directory from next to zero, only luckily did I keep some configs on google drive and had printouts of some other things to use for recovery. It was minimal recovery, some things were permanently lost. I used FreeNAS in the past but even succeeded to make some fatal errors with it as well. I maintain hope that that data could be recovered but I have not made a serious attempt yet. I choose to 'live on the edge' so to speak, partly for something to do and partly because I am foolishly lazy in some respects. Any sort of backup to a FreeNAS box would have saved me, or without that, a much greater understanding (that FreeNAS in effect would relieve me of needing to learn in such great detail) of ZFS and recovery and setup.

In setting up a machine for someone else, I would choose to use FreeBSD, ZFS, and at least mirrored drives to give some minimum of fault protection. This would not be FreeNAS but simply a FreeBSD box that may have some hope of recovery in the event of something going wrong.

I would STRONGLY recommend a pair of disks, a total of two, which are mirrored, but this as a minimum. Skip FreeNAS, skip multiple redundancies, you could even simply grab another drive to attach by USB and copy stuff to it from time to time. Why spend more than you must if something so simple would work just as well? Less convenient, no ZFS advantages, and surely more reasons I'm not going to list. FreeNAS because its easy to install and has all these cool plugins but not one care in the world about whether any of those files, data, will be lost or damaged or become unrecoverable. The risk is certainly yours. Would not power filtering, uninterruptible power supply, and a firewall be necessary precautions as well, since they are potential causes of data loss too.

It makes very VERY little sense to ask for an opinion or help and yet firmly remain against all recommendations, those who reply are purely wasting their time. Eventually, a discussion such as this will devolve and any number of participants could be viewed as trolling, because of no resolution and continual argument/disagreement over reality. What size bullets shall one use to shoot themselves in the foot? Those answering might want to know how much of a foot is desired to remain afterward but first they'd say not to pick up the gun in the first place, let alone load it.

 

[garm]

it seems quite unlikely we will get to an agreement here.

I’m not looking for an agreement, im a semi professional system admin trying to guide lost souls trapped by the dunning-Kruger effect.

I build systems this way; I identify my needs, I research the alternatives and I try (implies testing) implementing something.

I have data going back more then 20 years, I have lost data due to corruption that propagated to backups before starting to use ZFS. I have data that are reproducible and I have data that are not reproducible. Reproducible data I store in either one or two copies depending on what I do with it. I store non-reproducible data in at least three different copies, in atleast two different geographical locations.

Now, all this is fine to some point. For this discussion, what ZFS contributes is accountability. ZFS has the ability to measure the validity of data on disk, ensuring that data beaing read is the same data that once was written. In the time periods I intend to keep my data, corruption is unavoidable so I use mirror pools in my present main server due to the fact that I wanted some flexibility in the drives I used (for the cost of price/TB). What you have been recommended is RAIDZ2 which lowers the cost of redundancy by the number of disks in your vdev as the cost is fixed at two drives. My present setup will likely never grow larger then 4 drives due to limitations of my chassi so RAIDZ2 and mirrors bare the same costs, with mirrrors having slightly less reliability due to the fact that I cannot loose two drives isn’t he same vdev. Of course if I put only two drives in production I would also use mirrors, in fact that is what I do with a small ssd pool that takes most of the writings to my server off the main storage pool.

My backup targets are mirror pools. Benefits of a mirror pool is again that data corruption can be mitigated at the cost of price/tb. Having said that, I also plan single drive pool backup targets in the near future, to lower the cost and mitigate the increase the risk of data loss per pool by increasing geographical locations.

Everything is a balance of price and risk. No one can make that decision for you. You need to understand what you are doing, make a conscious decision and live with the consequences.