unRAID 6.2 - Screenshot Dual Parity support

As near as I know, LT hasn't released the specifics as to whether or not you can use different sizes for the two parity drives.    I suspect that both parity drives will have to be the same size; but depending on the implementation that may not be required => it would be possible to only require that each parity drive be >= the largest data drive.

 

 

 

Thanks, garycase - I appreciate the response - we'll just have to wait and see what LT comes up with.

 

I am interested in how it will be implemented. If it continues the Unraid tradition of "plugging a couple of drives in there - whatever you got" and them being spanned into a larger pseudo Parity drive (pooled) - that's cool, or a double redundant Parity (not for me but still useful for others).

 

If the drives aren't pooled into a larger pseudo Parity drive, I personally don't see the utility of a redundant Parity drive - I'm not that paranoid, but I am not keeping photos and other irreplaceable data on my server anyway and I am sure other people are. If they could be pooled and that allowed you to use two smaller drives (maybe as a stopgap or emergency scenario) to replace or extend a larger Parity drive - that would seem to just make Unraid more flexible and accessible to folks with limited funds (and especially if doing this allowed you to use larger data drives!). I realize that this is probably more risky than some would be willing to accept. I don't hope to experience a double or triple drive failure - then I will really understand what redundant parity is really all about!

 

Something I have always thought would be awesome would be "targeted protection". Where you identified what shares you wanted to protect versus those more "expendable". But that would be more like a ZFS Snapshot of a backup of whatever files/shares you identified. That way, critical files could be protected outside the parity level protection. So, if you HAD to lose a drive, and parity was insufficient, you would have a second chance at restoring critical data once you corrected the failure(s). Just not as nice as replacing the dead drive with a new drive and letting Unraid restore the volume.

Unqualified Spectator, It will not be 2 drives pooled into a larger Single Parity drive.

There you go again...stompin' on a dream...FINE.    No worries...

Unqualified Spectator, It will not be 2 drives pooled into a larger Single Parity drive.

Unqualified Spectator is the only one I have ever heard that even wanted this, and no discussions about Dual Parity here for years or elsewhere with other kinds of fault-tolerant systems, ever mention anything like this.

Unqualified Spectator, It will not be 2 drives pooled into a larger Single Parity drive.

Unqualified Spectator is the only one I have ever heard that even wanted this, and no discussions about Dual Parity here for years or elsewhere with other kinds of fault-tolerant systems, ever mention anything like this.

If you really want this then it is possible to do it with a SATA adapter that allows two drives to be set up as RAID0 and the result to be presented to unRAID as a single drive. 

 

I was experimenting to see if I could repurpose two older drives to avoid buying a larger parity drive.  It worked Ok, but I found that using 2x4TB drives gave slightly less space than a 8TB single drive (I assume due to some overheads in setting up the RAID0 array).  Using a 4TB and a 5TB drive together worked OK.

Another parity drive that doesn't provide any additional fault-tolerance is just a wasted port/bay. Buy a bigger drive!

Another parity drive that doesn't provide any additional fault-tolerance is just a wasted port/bay. Buy a bigger drive!

 

I tend to agree -- HOWEVER, the experimentation that several folks did with this was primarily done last year to provide an 8TB "drive" to use as parity when using shingled 8TB data drives ... this was before there had been enough experimentation with these drives to confirm that the mitigations Seagate as done on them to overcome the limitations of the shingled technology actually work very nicely even when used as parity drives.

 

There are now two things that eliminate the need to combine drives in this way:

 

(1)  It's been shown that the shingled drives work fine, even as parity, in the vast majority of UnRAID use cases;

and

(2)  If you don't want to use a shingled drive, there are now standard PMR 8TB drives available

 

I was experimenting to see if I could repurpose two older drives to avoid buying a larger parity drive.  It worked Ok, but I found that using 2x4TB drives gave slightly less space than a 8TB single drive (I assume due to some overheads in setting up the RAID0 array).  Using a 4TB and a 5TB drive together worked OK.

 

Depending on the RAID card, there may be a setting to not truncate the drive space to the nearest 1G or 10G. If the space is NOT truncated, 2x4T would be a bit larger than an 8T drive. But for cards lacking the option, they typically DO truncate the size resulting in something slightly smaller than needed for parity.

I was experimenting to see if I could repurpose two older drives to avoid buying a larger parity drive.  It worked Ok, but I found that using 2x4TB drives gave slightly less space than a 8TB single drive (I assume due to some overheads in setting up the RAID0 array).  Using a 4TB and a 5TB drive together worked OK.

 

Depending on the RAID card, there may be a setting to not truncate the drive space to the nearest 1G or 10G. If the space is NOT truncated, 2x4T would be a bit larger than an 8T drive. But for cards lacking the option, they typically DO truncate the size resulting in something slightly smaller than needed for parity.

I do not remember seeing such a setting on the card I tried!

 

It could still be useful if repurposing smaller drives as part of upgrading drive sizes.  For example if one already has a 6TB parity, then coupling that with a 3TB drive (giving a 9TB notional size) would still allow you to start using 8TB data drives.  Of course this does assume you have a free physical slot for the drive, but you might be doing something like adding a 8TB drive to replace a couple of 3TB drives.  In such a case you could use one of the 3TB drives to increase the size of the parity to 9TB (6+3) without increasing the overall physical count of drives used.

I do not remember seeing such a setting on the card I tried!

 

It could still be useful if repurposing smaller drives as part of upgrading drive sizes.  For example if one already has a 6TB parity, then coupling that with a 3TB drive (giving a 9TB notional size) would still allow you to start using 8TB data drives.  Of course this does assume you have a free physical slot for the drive, but you might be doing something like adding a 8TB drive to replace a couple of 3TB drives.  In such a case you could use one of the 3TB drives to increase the size of the parity to 9TB (6+3) without increasing the overall physical count of drives used.

 

The Areca cards have the needed configuration. Might look at the ARC-1200, 1231, 1261, and 1280 which can often be found on eBay at a good price.

 

Not sure about your 6T+3T concept. RAID cards are going to expect drives of the same size. Likely the 6T would be truncated to 3T and you'd gain nothing.

 

I have been using 2x3T drives as my parity for years. And am prepared to go to 2x4T when/if I jump to 8T drives. I confirmed that it is larger than an 8T drive on Areca controllers.

 

Another option is to use your 4T+4T setup with existing card (slightly smaller than 8T drive) as parity and create tiny HPAs on your 8T drives. This is quite easy to do (so long as the drives don't have any data on them). The space lost is trivial.

I do not remember seeing such a setting on the card I tried!

 

It could still be useful if repurposing smaller drives as part of upgrading drive sizes.  For example if one already has a 6TB parity, then coupling that with a 3TB drive (giving a 9TB notional size) would still allow you to start using 8TB data drives.  Of course this does assume you have a free physical slot for the drive, but you might be doing something like adding a 8TB drive to replace a couple of 3TB drives.  In such a case you could use one of the 3TB drives to increase the size of the parity to 9TB (6+3) without increasing the overall physical count of drives used.

The Areca cards have the needed configuration. Might look at the ARC-1200, 1231, 1261, and 1280 which can often be found on eBay at a good price.

 

Not sure about your 6T+3T concept. RAID cards are going to expect drives of the same size. Likely the 6T would be truncated to 3T and you'd gain nothing.

 

I have been using 2x3T drives as my parity for years. And am prepared to go to 2x4T when/if I jump to 8T drives. I confirmed that it is larger than an 8T drive on Areca controllers.

 

Another option is to use your 4T+4T setup with existing card (slightly smaller than 8T drive) as parity and create tiny HPAs on your 8T drives. This is quite easy to do (so long as the drives don't have any data on them). The space lost is trivial.

Thinking about you are correct - I testing was with 2x4TB as an alternative to a 8TB drive.

 

I had not thought of the HPA approach to slightly reduce the size of the 8TB drive (one is normally more worried about getting rid of HPA rather than adding it ).  Might need to think that is worthwhile to whether I should just bite the bullet and put a native 8TB drive in for parity.  However as I already have the hardware for the 2x4TB approach it is definitely worth considering.

 

 

Unqualified Spectator, It will not be 2 drives pooled into a larger Single Parity drive.

The only way to do that today is by doing a raid 0 parity. The Areca cards work fine.  My Areca 1631 and 1680 cards take several smaller drives and create one larger raid 0 drive that unRaid sees.

 

I am able to use up old 2tb drives this way. Once we have dual parity, a bunch more will be consumed.  These spanned virtual drives won't show temperature on the unRaid GUI.

 

Edit. @bjp999 beat me to it.

 

Thinking about you are correct - I testing was with 2x4TB as an alternative to a 8TB drive.

 

I had not thought of the HPA approach to slightly reduce the size of the 8TB drive (one is normally more worried about getting rid of HPA rather than adding it ).  Might need to think that is worthwhile to whether I should just bite the bullet and put a native 8TB drive in for parity.  However as I already have the hardware for the 2x4TB approach it is definitely worth considering.

 

In the days before 3T drive support, I needed space and wanted to buy some 3T drives because they were cheaper per T than the 2T drives, and I wanted the denser storage. I knew Tom was working on 3T support, and it would eventually come. So I bought some 3T drives and added a 0.9T HPA so that the drives were at the max size unRAID would support (~2.1T). And used them for probably 6 months like that, before the 3T drive support was added to a 5.0 beta (~5b4). Worked fine, and removing the HPA and migrating data was not difficult. (Unlike me, there would be no reason to migrate data to remove the tiny HPA you would create.)

 

I did learn that some controllers didn't play nice with creating and removing HPAs, but that once they were in place, all controllers respected them. So you might need to use a motherboard port to create them.

The gigabyte thing has most people running scared of HPA, but in reality, as long as you are in control and don't do it on a drive with data you want to keep, adding and removing HPA is no big deal. The scary situation is when you don't know if your drive will suddenly get an HPA at the most inconvenient time.

 

HPA can be a good tool, not something to be feared and immediately removed.

HPAs can be used to "short stroke" a drive. We all know that drives are faster at the beginning of a preclear, parity check, or parity build than at the end. Access to the inner tracks is much slower. Creating an HPA causes the innermost to be unused and results in a faster drive on average. This was a good technique prior to SSDs to get high performance from certain disks at the expense of capacity.

Another parity drive that doesn't provide any additional fault-tolerance is just a wasted port/bay. Buy a bigger drive!

 

Respectfully, you are more likely to have spare ports and bays than 8+ TB Parity drives. That was my only point for wanting to RAID0 two drives (that you may have) so you could bring your server back online with protection while you figure out your next move (or go get a replacement drive after your brand new drive craps out unexpectedly). It is precisely having these kinds of options that makes unraid so awesome. It works with what you have - not necessarily requiring you to go buy a large assortment of expensive gear to use it.

 

Thanks to all for your responses. It's cool that some already tried Raiding 2 drives together and using that as a single parity and it worked. I didn't think about drive temp reporting in a RAID 0 configuration, either...

Another parity drive that doesn't provide any additional fault-tolerance is just a wasted port/bay. Buy a bigger drive!

 

Respectfully, you are more likely to have spare ports and bays than 8+ TB Parity drives...

You may be more likely. For others it could mean a complete redesign and replacing case and mobo just to get one more port/bay.

The real question for me personally with dual parity coming soon, is if parity+btrfs disks can be used to repair bitrot.

dual parity coming soon

 

 

So I saw this and I think we should make this a thing. Anytime anyone asks about unRAID's dual parity, even once it's released, we could correct them...

 

[some-user]

I have a question about dual parity...

 

[Me]

Are you talking about unRAID's new dual parity coming soon?

The real question for me personally with dual parity coming soon, is if parity+btrfs disks can be used to repair bitrot.

 

+1, would also love to hear more details!

at 6:33m in Linus's video it shows a login screen to unRAID webui?  I wonder if that means another change?

at 6:33m in Linus's video it shows a login screen to unRAID webui?  I wonder if that means another change?

 

I didn't review the video again, but once I changed my root password, it prompts me to log into the webui:

 

 

This is on 6.1.6

Thats the webUI via your browser.. it looked in video like the WebUI was replaced with some sort of Linux desktop login... I could be wrong

Went back to the video... now I see what you're saying. It looks like it's a more polished webui login to me. Guess we'll see. Some good things appear to be just over the horizon for unRAID. Very interested in what comes next.