This topic has been marked solved and closed to new posts due to inactivity. We hope you'll join the conversation by posting to an open topic or starting a new one.
@jak0lantash wrote: Based on my experience, most complaints are about the lack of capacity, not the lack of redundancy.
Yes, and my suggestions above would preempt at least some of those complaints.
I'd be very annoyed if I added disk 6 to get capacity and then discovered the system decided I should get dual redundancy instead - forcing me to do a factory reset to change it. If I was expecting RAID-6, I'd be equally annoyed.
I think it's a bad idea for the system to automatically do either RAID mode, since once started the action can't be undone.
I think it's a bad idea for the system to automatically do either RAID mode, since once started the action can't be undone.
Bingo.
Ideally all of this could be resolved (for me) if OS6 waited to create the data volume until the setup wizard, then setup wizard could ASK and PROVIDE INFO as to which raid mode is most appropriate for the user.
Additional info could be linked to an appropriate KB article that explains the available raid modes and why you would choose (or not choose) a particular mode.
Said article could refer to the forum here for further questions or discussion.
If the setup wizard provided this and the option to choose space versus redundancy and explained in simple terms to the user, I think all sides of the discussion could be fairly satisfied.
After all, everyone was a noob once, and in a situation where an early decision about raid mode will have long lasting repurcussions and cannot easily be changed later on in the future.
Everyone complains about space vs redundancy, until they lose their data due to multiple disk failures.
The only RAID array I have running is made of USB sticks and is exclusively for experiments. Yet, there is very little chance I ever lose any data (important data anyway). Even with a RAID6, you shouldn't rely only RAID redundancy. Redundancy is not a backup, but it does help increase availability. If your RAID array breaks and you lose data, it means your backup plan is flawed, not that you should increase redundancy. OVH just had a massive outage, multiple redundancy didn't save them, their backup did.
Anyway, we can keep talking, it's not gonna change X-RAID (or Flex-RAID). I agree with some choices and disagree with others.
Yes, backup is a must. But when you're shoving around 20, 30, and higher TB in arrays the recovery time is painful. So while backup is important, not letting it die because you popped a second drive while another you didn't noticed was bad for weeks rebuilds is a consideration. That little bit of extra parity could save weeks or more if recovery from offsite backups or other sources. But of course you can only do so much to automate good behavior and choices. 😄
There's a reason why all these RAID modes are deployed - each strikes its own balance between availability, capacity, and performance. So there will never be agreement on exactly what the rules should be, because different people will value availability, capacity, and performance differently.
I'm fine with simply aligning with the documentation, but I remain convinced that simply applying the rules automatically is a bad approach.
I found the link to the manual I assume that you were referencing and the note that pertains to this issue on page 24;
For systems with at least two disks and up to five disks, X-RAID reserves the capacity of one disk for data protection. (If there are six or more disks, the default format is RAID 6, which reserves two disks.) The actual space reserved for data protection is distributed across all disks.
While this statement is incorrect, instead of correcting the documentation, I vote (not that it matters) to correct the behavior to match the document.
Note, the statement says exactly what I mentioned previously, the 'default format', it does not mention or specify expansion, which is a whole different situation, which has also been discussed.
on page 37, there is a small section regarding adding 'protection', I suggest this be modifed to say either 'redundancy' or 'parity' to be more in alignment with standard raid terminalogy.
This section indicates how to switch from raid 5 to raid 6.
An X-RAID volume that includes two or more disks is automatically formatted to protect against the failure of one disk. Except for very large ReadyNAS systems (for example 12-bay systems), if you want to protect your data against the failure of two disks, you must switch to Flex-RAID and select RAID 6. (Very large ReadyNAS systems use RAID 6 by default.) To use RAID 6, you must install four or more disks. For more information about how to switch to Flex-RAID, see Change from X-RAID to Flex-RAID on page 26.
In reference to backups, that is whole nother discussion of which there are already very many on this and other forums.
I don't think anyone in this discussion has stated or believes that any raid level is concidered a backup, so while I completely agree that backups are a critical necessity, they have zero relevance to the discussion of whether the default raid with 6 or more disks should be raid 5 or raid 6.
And while we all have our individual opinions on which is most appropriate for our own situations, I would HOPE that most of you would agree that raid 6 is a safer (but not necessarily the most efficient) default choice *for those who do not yet know enough to determine their own decision*.
Note, the statement says exactly what I mentioned previously, the 'default format', it does not mention or specify expansion, which is a whole different situation, which has also been discussed.
If I use XRAID, then did some expansion, and then need to do a factory default, it'd be a huge issue if my volume size ended up smaller than what I had before the default.
So if expanding from 5 to 6 disks automatically preserves single redundancy, then defaulting that 6 disk array better deliver RAID-5 too. Otherwise users end up w/o enough space to restore their backups.
I counter that, it is as easy and inconvenient to delete a default raid6 data volume and switch to raid 5, as it is to delete a default raid5 data volume and switch to raid6.
Either way, someone will be inconvenienced equally.
The question is not who is inconvenienced, but what is the best default for those who don't know how to do choose for themselves.
For everyone else, they can choose for themselves and if that choice is different from the default, they understand and know how to make that change.
I only spoke about backups because there was mention of losing data. RAID redundancy helps preventing the loss of volume.
Afaik, X-RAID has been behaving this way for years, so I think the documentation should be corrected.
On the point about the difference between expansion and volume creation, it's already different. (As per 6.5.0) If factory default a 12 bays with 5 disks then add a 6th, you'll end up with a RAID5. From there, if you FD, you'll have a RAID6.
On the point about the difference between expansion and volume creation, it's already different. (As per 6.5.0) If factory default a 12 bays with 5 disks then add a 6th, you'll end up with a RAID5. From there, if you FD, you'll have a RAID6.
And I think that's a problem that needs fixing - especially if the new 8-bay desktops have the same behavior.. My "ask first" proposal would be enough I think (as long as it applies to both setup and expansion).
As is hopefully clear, I'm more concerned about the user not getting what they expect (or need) than the detail of when dual-redundancy should kick in.
On the point about the difference between expansion and volume creation, it's already different. (As per 6.5.0) If factory default a 12 bays with 5 disks then add a 6th, you'll end up with a RAID5. From there, if you FD, you'll have a RAID6.
Expansion of XRAID, only expand the capacity of volume. If you want to add parity, you need to change it to Flex-RAID(disable XRAID mode), then select the free drive and click "Add Parity" button.
Expansion of XRAID, only expand the capacity of volume. If you want to add parity, you need to change it to Flex-RAID(disable XRAID mode), then select the free drive and click "Add Parity" button.
You might be missing the point.
Imagine a customer has an XRAID volume of 20TB ( with 6x4TB) and he needs to do a factory default to take care a problem. Something that happens every day to somebody.
With these rules he will end up with only 16 TB (dual redundancy). That is confusing, and probably not enough storage to reload his data. He'll contact support or post here, and then learn that he needs to destroy the volume, create a new one, resync again, and then restore his data. He won't be happy.
A few months ago this scenario would only happen with rackmount owners using XRAID - only a small fraction of the customer base. However, with the new 8-bay desktop units, it will begin happening to more users.
Sorry @Leia, but your statement is incorrect. Adding a 7th disk to a 6 disks RAID5 under X-RAID reshapes to RAID6 (as per 6.5.0). You can see the results of my experiment in the link posted previously.
10a) Factory Default with 5x1TB in a twelve bays unit. md127 is a RAID5 of 5 partitions of 1TB (4TB capacity, 1TB parity). BTRFS volume capacity is 4TB. Add a 1TB HDD, the chassis now contains 6x1TB. md127 expands to a RAID5 of 6 partitions of 1TB (5TB capacity, 1TB parity). BTRFS volume capacity is 5TB. Add a 1TB HDD, the chassis now contains 7x1TB. md127 reshapes to a RAID6 of 7 partitions of 1TB (5TB capacity, 2TB parity). BTRFS volume capacity is 5TB. 11111 -FD-> RAID5 (11111) 11111 + 1 = 111111 -HE-> RAID5 (111111) 111111 + 1 = 1111111 -HE-> RAID6 (1111111)
Which causes some interesting situations where the volume can be composed of RAID6 and RAID5 and/or RAID1 arrays. Which would obviously not be recreated the same way after a Factory Default.
Just to clarify - there are other use cases where the array isn't recreated the same way, but is created larger. I'm not concerned about those - only the ones where the recreated volume is smaller than the original XRAID.
Just to clarify - there are other use cases where the array isn't recreated the same way, but is created larger. I'm not concerned about those - only the ones where the recreated volume is smaller than the original XRAID.
Just to clarify too - the cases I mentioned earlier would result in a smaller volume after Factory Default.
@jak0lantash wrote: Sorry @Leia, but your statement is incorrect. Adding a 7th disk to a 6 disks RAID5 under X-RAID reshapes to RAID6 (as per 6.5.0). You can see the results of my experiment in the link posted previously.
I can confirm that. I moved a 6 x 3TB XRAID(RAID5) volume from a Pro6 to a 4800V2, both running OS6.7.1, and added another 3TB drive with XRAID enabled. It converted the volume to RAID6, adding no additional space. That was what I expected, but perhaps not what all would want.
BTW, that re-sync took 10 days to complete (extended a bit because some backups occurred during the re-sync). That's double the original estimate, and the estimate didn't start to update to something close to the actual until it hit the 50% completion point.
in fact i think there is some kind of magic involved in order to keep redundancy whilst redistributing data and (double) parity amongst both a new disk and existing disks.
OK, so here is a situation I've wondered about. I just created a new volume with 4 x 6TB + 2 x 4TB drives. I just put them in and let it factory default to XRAID/RAID5. It created a 6 x 4TB layer and then a 4 x 2TB one, using all the available space, as I expected and wanted.
But let's say a wanted dual redundancy. If I turned XRAID off, destroyed the existing volume, created a RAID6 volume (which would only use 6 x 4TB), then turned XRAID back on, would it then add the 4 x 2TB layer using the remainder of the 6TB drives? If I let XRAID create a volume from 3 x 6TB and 2 x 4TB, then turned XRAID off, could I then add the last 6TB and select "Add Parity" even though standard Flex-RAID can't create that two-layer RAID5 on it's own?
Of course Iwouldn't want it to be RAID6. I think it would be a terrible use of the drives, with so much of the space dedicated to dual redundancy. That's why I believe the current default of RAID5 for 6 drives is better than a default of RAID6 and that any change should only be to have a choice.
The new FlexRAID supports multiple RAID groups (and vertical expansion) - though I haven't tried it. But I think it would allow you to create dual redundancy using all the space.
If the option for multiple RAID groups in the same volume was there when I was running some experiments with the new FlexRAID options before comitting my new NAS to production, it was not obvious how to do so. An article on how to use these new functions is clearly called for.
If the option for multiple RAID groups in the same volume was there when I was running some experiments with the new FlexRAID options before comitting my new NAS to production, it was not obvious how to do so. An article on how to use these new functions is clearly called for.
The new features of Flex-RAID came in 6.7.0 iirc.
For the doc, I don't see that coming anytime soon, if ever. The old features aren't even documented.
For the doc, I don't see that coming anytime soon, if ever. The old features aren't even documented.
Yep, it was 6.7.1 I was using for the experimentation. I don't doubt that the multi-volume capability was there, it's just not obvious that it is or how to use it, and I had a limited amount of time to fool around. Undocumented features (real ones like this) do nothing to attract new customers and frustrate old ones.
Kudos to the engineers that are putting advanced functions like this into the firmware. Raspberries to Netgear management for failure to acknowlege, much less feature them.
