NETGEAR is aware of a growing number of phone and online scams. To learn how to stay safe click here.
Forum Discussion
RN214, query on a 2nd vertical expansion
Hello Forum
Simply query, just checking. Can I carry out a “vertical expansion” more than once?
I have a RN214 ( 6.10.3, X-RAID) which already went from 4x 4TB to 4x8TB, and I am just about to do it again, going (in stages) to 4x 12TB.
If I remember well, OS6 will create another RAID for the “expanded portion” of the disks. This is “unseen” in the normal use, but it can be seen in the volume log, which now has RAID 0 and RAID 1 which contain md0 and md1 respectively.
But, if I remember right, that will cause a double resinch operation (maybe just after the 2nd disk is swapped?)
So, if it is possible to expand again, and that would create a RAID 2 & md3, woudn’t that require a 3rd resinch? ( that may take weeks, lol ).
Many thanks in advance, Berillio
15 Replies
Start with XRAID. You can still choose RAID0 as the initial volume, then it will automatically do exactly what you are planning for the expansion. The OS sometimes disallows your ability to switch from FlexRAID to XRAID. Nothing you plan should trigger that, AFAIK, but there is no need to take the chance.
There is no need to reset anything. In fact, you shouldn't. All configuration is stored on the drives, so a reset would do nothing if done without the drives and wipe the drives (making putting them back in useless if you find you forgot something) if done with them.
GRRRRR
I have been refreshing the page every 30 mins since I posted it, but I never saw any reply until few minutes ago (the following day) when I refreshed “the entire tree”. Last night (say 20h ago), I started copying back the files and it has done ~4tb by now. I suppose I could “delete the volume” and reboot. The disks are now formatted and it shouldn’t go into “safe mode” which is what happened yesterday, and meant restarting EVERYTHING passwords, emails, fixed network node etc..
Then I could start again keeping the X-RAID, at moment it doesn’t allow me to switch “Cannot select X-RAID on RAID 0 or RAID 10” because I don’t have the redundant disk
Sandshark, the Soft Manual says that X-RAID doesn’t support RAID 0, but infact when I started, X-RAID was there (as default), even after I selected RAID 0 (albeit it wasn’t highlighted in green), and it was me who switched it o Flexi-RAID. So…. I should have left it on and maybe it would have gone green when I added the 3 rd disk.
StephenB, I will follow advice and go 6.10.9… now ( I stopped the data copying already)
If you still have all the drives from the original volume and you want things to be set up the same as they were, you could put those drives back in and save the configuration, then restore it on your new volume. The volume name must be the same to do that. If you had apps installed, you should re-install them before restoring, so if you had apps you don't plan to re-install, don't backup "everything", just do the users & groups and share access. If you have ReadyCloud users you are removing, then this is probably not a good idea.
I believe you are correct that it would have gone green when you added the third drive. At the Linux level, a RAID0 created by the OS is actually a RAID that's missing a member (so no redundancy), but it's flagged that the OS won't complain abo0ut that missing element. A single drive RAID1 is a RAID1 with a missing member and a 2-drive or more is a RAID-5 missing one, So it's not showing green because there is no redundancy (I missed that in my experiments).
remove all disks from the 214; (do I need a reset ?)
Well, the system will do a fresh install when you insert the two 2x12 disks, which is the same as a reset.
7) convert the Flexi-RAID to X-RAID.
Not sure you'll be able to do this, you might end up stuck in FlexRAID. Not the worst outcome, but not ideal.
But at moment, this RN214 is on 6.10.3 (and so is the other 214, the 424 is on 6.10.4)
Personally I'd upgrade both to at least 6.10.9 There is no need to do this sequentially.
6.10.10 does have some security fixes, but it also removes the ability to install apps from the web ui. If you don't care about that, then you should go to 6.10.10.
OK. If you have another home for the replaced 8TB drives, then the cost efficiency is very different.
As far as your alternate plan, I doubt that will be a lot faster (you still have 3 syncs to do, plus all the copying) and it will definitely be more operator intensive.
Well, I was already moving the data out, so I was committed to that. Data safety was the greatest concern, surely it was more operator intensive. Done now, anyhow.
Now all 4 hdd have been Read-Write-Verify’d, and all data has been moved to two of the 12TBs. Now the procedure to follow could be:
- remove all disks from the 214; (do I need a reset ?)
- add 2x 12tb and start;
- choose Flexi-RAID and set it to RAID 0 with full capacity (no protection, unique data);
- move the data back there (which again may take few days);
On the SoftwareManual, Flexi-RAID with “2 or more disks” and RAID 0 “Can I add a disk for data protection? – Yes. (Additional disk provides single redundancy and converts the volume to RAID 5.)” .
- 5) hot-add the3rd disk which would invoke the 1st resinch;
- 6) hot add the 4th diskwhich shouldinvoke the 2nd resinch;
- 7) convert the Flexi-RAID to X-RAID.
But I could also convert from FlexiRAID to X-RAID after I copied the data, before (5) and (6) (adding the 3rd and 4th disks): the Manual quotes “If your system contains only one volume, you can easily switch from Flex-RAID to X-RAID. Data on the Flex-RAID volume is preserved when you switch to X-RAID.”
Which of the two procedure should I choose?
On a different topic, I generally try to follow the “If ain’t broke, don’t fix it” rule. But at moment, this RN214 is on 6.10.3 (and so is the other 214, the 424 is on 6.10.4). OS6 is now on 6.10. After (1) (NAS with no data), it could be a good time to upgrade the OS, if so, it may be better to do it sequentially 6.10.3 to 6.10.4, then to 6.10.5 etc.
But,is it really NECESSARY to upgrade the OS and, if so, also NECESSARY to do it sequentially?
Thanks in advance, Berillio
If I remember well, OS6 will create another RAID for the “expanded portion” of the disks. This is “unseen” in the normal use, but it can be seen in the volume log, which now has RAID 0 and RAID 1 which contain md0 and md1 respectively.
Almost.
md0 is the OS volume (4 GB total), and that is mirrored on every disk with RAID-1. md1 is a small swap partition.
The data RAID groups start with md127 and count downwards. They are concatenated into one volume by BTRFS.
Since you've expanded once already, you should see both md126 and md127 in btrfs.log and mdstat.log. Volume.log should only show md127 (down in the df -i and df -h sections).
I did not say but the RAID is near full, 21.00TB shown on the pie in Volumes (with 0 snapshots). For Info, the SCRUB I just run took almost 32 hours. Is that roughly the time which will be needed for any resinch operation? I just thought of another way of doing this which maybe quicker,
- a) buy the other 2x 12 TB ( 2days, say, to receive them)(actually I just ordered them);
- b) fit my existing 12TB on a PC which I have, which has 6 SATA slots;
- c) copy/move all the 20TB on these 2x 12TB. Than may need 21,000,000 / 80 (MB/sec) = 262500 seconds = 73hours, ~4days (I can start doing this while waiting for the new hdds to be delivered);
- d) ReadWriteVerify test the hdd(s) once thy arrive (~15h each);
- e) remove the 8TB drives frome the RN214, fit 2 of them to the PC, and copy over the contents of one of the 12TB which is now free;
- f) fit the NAS with 3x 12TB hdd and copy the data (another 4 days). After that, the data is availlable in a RAID in a protected state;
- g) wipe the 12TB left on the PC and fit it to the NAS (that would be an horizontal expansion). That will do ONE SINGLE resinch and should have a simple RAID 5 on the 4 hdd ( not a RAID+ RAID 1+ RAID 2 raid structure
Wouldn’t that be faster and simpler than doing the 3rd vertical expansion as I described earlier?
Hopefully, you've not unboxed the 12TB drives, because your plan is very inefficient cost wise. Updating all 4 8TB to 12TB only adds 12TB of storage at the price of 4 x 12TB drives. For less money (and fewer re-syncs), you can swap out two for 24TB drives and get a 16TB increase with the option of later swapping another, and then later another, for yet another 24TB for an additional 16TB from each. I believe 24TB to be the best size-to-cost ratio available today. On Amazon, a 24TB Ironwolf Pro is the same price as a 20TB.
The only reasons to be doing what you are is if you suspect all the 8TB drives are in need of replacement or if you are using FlexRAID (which requires all drives be of equal size). But if you are using FlexRAID, the incremental syncs StephenB listed are not what you will see.
Hopefully, you've not unboxed the 12TB drives, because your plan is very inefficient cost wise.
I agree that would be more cost-effective to get fewer larger drives.
Though I believe berillio is in the UK, not the US. Looking at amazon.co.uk, the most cost effective combination I see is 3x16TB+8TB. That adds 16 TB at ~£49/TB gained (£784 total cost).
My suggestion above would give him a much smaller storage gain, but would also give him enough capacity in the PC to back up his volume going forward. That could be modified to go with 2x16TB+2x8TB, with a 16TB and one or two 8 TB drives in the PC for backup. Cost would be the same (assuming no exchange fee on the disks), and it would give him more space in the volume than my original suggestion.
On Amazon, a 24TB Ironwolf Pro is the same price as a 20TB.
FWIW, I am seeing 20 TB at about $400, and 24 TB at about $480. Cost per TB is the same (perhaps that is what you meant)?
Wouldn’t that be faster and simpler than doing the 3rd vertical expansion as I described earlier?
The number of RAID groups isn't really relevant to the speed, as every RAID group needs to be resynced when you replace a disk. I usually size the process by totalling up the disk reads/writes.
Just hot-swapping the disks in the NAS requires reading or writing every sector in the volume at every step (including the parity blocks). So
- 32 TB when you swap the first disk
- 40 TB when you swap the second
- 44 TB when you swap the third
- 48 TB when you swap the fourth
164 TB of total disk activity, which will take quite a while. Note each step takes more time than the one before it.
As far as the RAID is concerned, your process takes
- 36 TB to sync the initial 3x12TB volume
- 48 TB to sync and expand to 4x12TB
So 84 TB for that bit.
But copying the data to the 2x12 disks at the beginning requires 42 TB of disk I/O (21 TB of reads and 21 TB of writes), the second transfer to the 8 TB drives requires another ~20 TB of copying (assuming about 10 TB of data on the source 12 TB drive). And 42 more TB to transfer the data back to the NAS. A total of 114 TB of disk I/O for the copying.
So your process requires more disk I/O than hot-swapping, not less. Though this is distributed across two devices, so in practice the wall clock time is likely similar.
That said, copying the data to the PC first is safer, because you have no RAID protection during the hot-swap process. If one of the 8 TB drive fails during the hot-swap process you lose all your data. Personally I think that maintaining data safety is more important than the amount of time the process takes.
What you could maybe consider is leaving two of the 12 TB drives in the PC. Then you'd stop the NAS expansion at 2x12TB + 2x8 TB. You'd gain only 4 TB of storage, but that would still give you 25% free space. This would look like
- insert 2x12 TB in the PC and copy the data from the NAS (42 TB of disk I/O)
- hot swap one 12 TB drive and and resync (32 TB)
- hot-swap the second 12 TB drive and resync (40 TB)
- add at least one of the 8 TB disks to the PC to give more free space there, and back up the NAS to the PC on schedule
Total disk I/O is 114 TB, so less than either of the two methods above. Later on, you can continue the expansion by hot-swapping two more 12 TB drives to the NAS (and maintaining enough disk space in the PC to keep the files backed up).
FWIW, I am ignoring the disk test time, since you'd be doing that no matter what path you pick. Personally I run the extended read test, followed by a full (single-pass) write test on each disk before I put it into my NAS. That would take more than 15 hours with a 12 TB drive. I have had some out-of-the-box drives that pass one of these tests but fail the other.
NETGEAR Academy
Boost your skills with the Netgear Academy - Get trained, certified and stay ahead with the latest Netgear technology!
Join Us!
