I5 2500k Server (NAS?)

They recalled all of those, any Sandy Mobo on the market won’t have that problem
The 2500k actually has quick sync

1 Like

Ah right you are, thought quicksync was an ivy bridge introduction.

In any case wasn’t great quality the first couple of generations.

1 Like

Sort of, but remember that garbage in, garbage out

1 Like

ZFS always returns correct data. If it can’t (because pool status doesn’t allow for correction of errors), it will tell you and (once errors and corruption won’t stop)will enter a “lockdown” to prevent further corruption of data.

ZFS can only validate and protect the data you gave it. If your memory is corrupt or you copy files over network and data gets corrupted in transit, that’s bad.

I don’t think Wendell is very fond of UnRaid. It’s an option, but UnRaid is the more casual approach, where TrueNAS is generally more appealing to professionals and enterprise.

That Unraid vs. TrueNAS comparison is very bad. It’s old (Scale was in Beta back then, why not use Core?) . Very unprofessional review and some things are just wrong.

1 Like

Just run FreeNAS Core and be done with it, you don’t need a HBA but I would recommend getting another NIC than the built in.

What would you recommend? Building something new, e.g. AM4 and so on, or would you go the ARM route (if possible)

I don’t really have a need for a spare room PC, or at least I don’t know what I’d be doing with one besides maintenance

Why not TrueNAS Scale?

A more general question: As @Ruklaw has pointed out, my system is indeed rather old. How would I proceed if the mainboard for instance dies? Is it easily possible to restore the RAID array on new hardware?

Also regarding backups (I know a RAID is not a backup), but I find it odd that one puts all the energy into ZFS and its bit-rot protection and then stores the backup on an external drive. This somehow defeats the purpose in my opinion. Any ideas regarding that or would you technically need two computers?

The general consensus appears to be that FreeBSD is preferred over Linux when it comes to ZFS but you’re of course free to choose whatever variant you want.

If the mainboard/CPU/* dies you just move it to another system and import the array(s).

You usually use off-site backups, nothing limits you from using ZFS on that remote system or whatever filesystem you want to use.

1 Like

Hiya, it’s so easy it will amaze you.

Just some small observations with Core:

I had strange issues with USB sticks, never with SSD.
I’ve had issues with non-Intel NICS, never with Intel.

As I got more serious with using it, I made sure I gave the machine no excuses to fail, and more importantly waste my time trying to fix. I now have 3 Core servers with a fourth on the way (self built), only with this fourth I’m going to wing it with non-ECC, mainly cos it’ll only run Scale for it’s container thingys.

If the machine is just laying around, install core with any old HDDs and SSDs lying around…play. :slight_smile: TrueNAS Core is one of the most stable systems that I’ve ever seen, even on really old gear (15+ years old).

1 Like

Thanks yet again!

I will probably purchase a couple of drives and, set it up and see how everything works. There are two remaining questions though:

  1. If I ever decide to move to a larger array I would either need to systems, use a backup or purchase one with enough SATA slots in order to transfer the data, right?
  2. Are used drives an absolute no-go? They are considerably cheaper which would enable me to purchase large enough drives. The downside here is that I assume anyone can claim that the drives are refurbished/recertified and I don’t really know which state they are in. Would it be ok to perform some kind of check (which one?) as well as looking at the S.M.A.R.T data?
1 Like

a

ZFS pool can be plugged into other systems without problems. Auto-detecting the pool and importing should be immediate if ZFS can import during boot. And use the typical YFS expansion options to add new drives to the pool. Expansion may be limiting depending on raidz or mirror. I just plug in two new drives, click two buttons and I have 16TB extra space on the pool.

For me, yes. I don’t trust SMART data on ebay or whereever. And drive hours are drive hours…clock is ticking and life is more or less finite. If you get enterprise drives, you get good TB/$.

P.S.: If ZFS prevents you from expanding the pool (well, RAIDZ-limitations depending on your config/expansion), best practise is to recreate the pool from backup. Just another reason to keep these around :slight_smile:

2 Likes

With ZFS I´d say it´s fine when you get a good deal. It does not trust the drives anyways. Do regular scrubs. Make backups. Value proposition is harder to answer because if it dies much sooner than a new one would have, did you really save a buck? Depends on what you paid for it in the end.

1 Like

What @Exard3k says :+1:

The only difference is I’d say, if you were just ‘playing’ with TrueNAS, get some super cheap second hand drives. This sounds wrong, but you’ll likely get some duds - and this is when you’ll experience a failure, what it looks like and how to fix it!

Buying used hard drives is the same as a used car - it could have all the service stamps in it, it could show true mileage, but you never really know how it’s been treated. But like I say above, get some some small capacity drives and have a practical play. As an example, you could even get 4-6 250-500GB used drives. Test them all, then make a RAIDZ2 (can have up to 2 failed drives), then make sure you have email notification on and just wait. Eventually you’ll get an email that one is failing, practice safely removing it (via the GUI) and safely replacing it (again, via the GUI).

Hope that helps!

2 Likes

Just an additional, some enterprise SATA drives are surprisingly cheap, just need to run the numbers.

1 Like

I was asking about the used drives because I found some Seagate 10TB drives that were classified as factory rejects for 60% of their initial value with 2y warranty by the shop. However, I inquired if I would have the 5y warranty provided by the manufacturer and they were not able to answer…

Seagate Exos 16TB is the same price as their 10TB version in my region. I’d rather get +60% instead of used -40% price on 10TB

1 Like

Might be, but those 320 bucks per drive for the 16TB are a bit expensive for my taste, especially considering I will need at least three.

(10TB is at roughly 260 new ans the uses ones are at 150)

It’s a matter of choice. Even new drives may fail early and I’ve seen both new and old working really well in my past. You may get a car with 100.000km run, doing well. But every statistic says HDDs failure rate skyrockets after 4 years. Some run 10 years straight and never complain.

Buying from a somewhat reputable commercial seller who also does shipping&packaging properly, is way better than getting some seemingly cheap stuff from your random ebay seller.

It’s a hard decision for all of us here, because we usually get multiple drives for our NAS/server and even rather cheap storage per TB, gets to the budget limit really quick.

Failed drive won’t get your pool destroyed, because you have redundancy (of one drive or more) in ZFS. That’s the good news :slight_smile:

1 Like

Agree 100% with this post.

A NAS is for storing your important stuff.

A 12 year old platform with known sata controller issues is not an ideal base if you’re planning a build today.

You can get something much newer secondhand for peanuts.

Motherboard capacitors etc do eventually die.

2 Likes

There are no issues with the sata controller and it’ll do fine for now

1 Like

Thanks for all the comments again. I have not entirely made up my mind on which drives to purchase, but I will see what comes up.

I don’t want to put much money into this build because if I do intend to make it nice and fancy it will probably never happen and I assume it is better to have a NAS system with a few downsides than not having one at all…