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.
I've read a few posts but have not found an answer to the following.
Is there any way to force an RN104 into Raid 1 mode when populated with four HDDs?
And if there is what is the pattern of the array. ie disk 2 mirrors disc 1 and disc 4 mirrors disc 3 or some other arrangement?
I am about to populate an RN104 with larger NAS drives in stages and prefer to keep them as Raid 1 when I insert the second pair.
I understand the first pair will be configured for Raid 1, but don't want the addition of the second pair to trigger a change to Raid 5. I'm happy to accept the lower array size that this would cause.
Budget and whether you want to have to manually spread shares across the volumes are two main drivers if drive sizes and rotation speeds are consistent.
Expanding RAID-10 might be more complicated than expanding two RAID-1 volumes.
One thing to think about is whether you actually need two RAID-1 arrays, or if one will do. Drive prices (in $/TB) are pretty flat right now between 8 and 12 TB. Getting one pair of larger drives now, and leaving two bays empty for the future might also be a good strategy.
@StephenB 's solution will give you two separate volumes. But you'll have to manually split shares and deal with two size limits between the two volumes. It sounds to me more like you want RAID10 -- two concatinated drives mirrored by two other concatinated drives. While the ReadyNAS can do that in FlexRAID mode, I don't believe it can go from RAID1 to RAID10 since the underlying MDADM can't go directly between them. Via SSH, you would have to convert the RAID1 to a single drive RAID0 (which is what a ReadyNAS JBOD drive is "under the hood"), and you could then convert to RAID 10 either via SSH or the GUI. If you are comfortable doing it, here my article on downsizing a RAID: Reducing-RAID-size-removing-drives-WITHOUT-DATA-LOSS-is-possible. Here is another useful article that's not ReadyNAS specific: Converting-raid1-to-raid10-online.
Frankly, this takes a long time and backing up and starting from scratch as RAID10 is probably the better solution if the two volume solution isn't what you are looking for.
Many thanks to both StephenB and Sandshark for their prompt and informative replies.
On balance I believe the pair of concatenated drives would be the optimum solution, but it raises some questions that I will ask for clarification:
1. Is the best way forward to use drives from the same manufacturer and of the same capacity?
2. I suspect that proceeding in stages, ie a pair of drive now in Raid 1 followed later by a another pair of drives - which I think would entail backing up the raid 1 pair then configuring the four drives as a pair of concatenated drives in raid 10 and then re-establishing the contents from the backup would be a quite long process?
3. A better solution, since I have the contents of the RN104 already backed up, would be to acquire the four drives and then proceed thus avoiding the intermediate step of an initial two drives in raid 1? Of course this raises the question of a rather large expenditure of cash that I was hoping I could spread across a few months.
1. Is the best way forward to use drives from the same manufacturer and of the same capacity?
With RAID-10 should be the same speed and it would be simplest if they were the same capacity.
If budget is a constraint, I'd proceed step-wise. Personally I wouldn't rule out have two RAID-1 volumes. You can simply add the the second volume, and shift a couple of shares over to the new volume. Balancing the space between them isn't that difficult.
Another thing to look at is the results of drive failures. For both configurations, you can lose one drive and certain combinations of two and lose nothing. With RAID10, there are conditions under which you could lose two drives and lose everything. With two separate volumes, the worst with two drives failed is losing one of the two volumes. You didn't ask about it, but RAID6 can be grown from your current volume (and in steps, RAID5 then RAID6) and can lose any two drives and still retain all data. But it can be slower.
Budget and whether you want to have to manually spread shares across the volumes are two main drivers if drive sizes and rotation speeds are consistent. If you have a lot of data that's archive and can be on on volume, and other data that is more active for the other volume, then it makes it easier to deal with multiple volumes, since one needs little "care and feeding".
Budget and whether you want to have to manually spread shares across the volumes are two main drivers if drive sizes and rotation speeds are consistent.
Expanding RAID-10 might be more complicated than expanding two RAID-1 volumes.
One thing to think about is whether you actually need two RAID-1 arrays, or if one will do. Drive prices (in $/TB) are pretty flat right now between 8 and 12 TB. Getting one pair of larger drives now, and leaving two bays empty for the future might also be a good strategy.
Well, a bit of a delay but I sourced a pair of iron wolf 10TB drives and loaded them into the NAS. It recognised them and began a syncing activity, which I wondered about since the drives were brand new with no data. After about 26 hours the box went quiet and I was able to start loading some data that I wanted access to across the network. No drama, things went well, and the data is available on all the devices I have on the network. I think that the pair of drives are operating in Raid 1 cofiguration as I got a message to the effect that the "syncing" was being done in X-Raid?
I'm now at the point where I want to add the next two drives, and have decided to let the box do its thing as I add them.
I believe this is done by attaching the drives (phyically) with the box running? If so do they get added one at a time or can I shorten things by powering down attaching the new two drives then rebooting and letting the box do its thing?
I understand I won't get exactly what I was intending to do as described in earlier posts , but will get a larger capacity since the full array of drives will be changed to a different Raid configuration (Raid-5?).
I'm electing to go this way because I have a full backup of the data on three 4TB drives (currently) and will maintain a backup of the NAS using the 4TB drives that were originally populating the NAS, as data gets added to the NAS in the future.
As an aside I ran a full gamut of tests from Seatools on the (barracuda) drive the NAS was showing as faulty and it passed all of them with no indication of bad sectors of other errors. This is the second Barracuda drive that a Netgear NAS box has indicated problems with that subsequently passed all of Seatools tests. Can anyone throw some light on why Netgear hardware seems to be so fussy?
It recognised them and began a syncing activity, which I wondered about since the drives were brand new with no data. I think that the pair of drives are operating in Raid 1 cofiguration as I got a message to the effect that the "syncing" was being done in X-Raid?
RAID creates virtual disk(s) from your physical disks, and the btrfs file system runs on top of that virtual disk. Since RAID is creating raw storage, the sync process is always the same (it doesn't matter if the drives are empty or full). With RAID-1, it is copying every sector on the first disk to the second. It gets more complicated with RAID-5, but it is still either reading or writing every sector on all the drives in the volume.
I'm now at the point where I want to add the next two drives, and have decided to let the box do its thing as I add them.
I believe this is done by attaching the drives (phyically) with the box running? If so do they get added one at a time or can I shorten things by powering down attaching the new two drives then rebooting and letting the box do its thing?
You add one drive at a time (hot-inserting them), and wait for the sync to complete. The system won't process two at a time, even if you add them together.
The fastest way is actually to start over - putting all drives in place, doing a factory default, reconfigure the NAS, and then restore the data from your backup. Your approach is doing multiple syncs - 20+30+40 TB of disk I/O (90 TB altogether, with 70 TB left to go). Starting over would just take 40 TB of disk I/O, since the factory default will create the RAID array in one go. So it will be nearly twice as fast.
But if you aren't in a rush, you can proceed one-at-a-time. The sync will take longer for each new disk (since the total disk I/O goes up with each disk you add).
I understand I won't get exactly what I was intending to do as described in earlier posts , but will get a larger capacity since the full array of drives will be changed to a different Raid configuration (Raid-5?).
Since your disks are of equal size, the volume will end up RAID-5. Things get more complicated if you increase the disk size in the future - for example, if you upgraded two drives to 16 TB you'd end up with a single volume with two RAID groups - 4x10TB RAID-5 and 2x6TB RAID-1. That still gives you single redundancy. XRAID manages all that for you.
The volume capacity rule is "sum the disks and subtract the largest". You'll end up with 30 TB (~27 TiB).
Note that if you want dual-redundancy (protection from loss of two disks), you can do that also. The volume size would shrink down to 20 TB. After adding the third disk, you'd switch to FlexRAID, and add the fourth for redundancy. If you start over (which again is faster), you'd switch to FlexRAID right away (before the system is done resyncing), destroy the initial volume, and create one that is RAID-6.
RAID-6 would give you about the same read performance, but write performance would be slower. Also, you'd need to upgrade all four disks if you ever want to increase your storage. Personally I use XRAID single redundancy. I have good backups, and value the extra capacity and expansion flexibility more than the extra redundancy.
As an aside I ran a full gamut of tests from Seatools on the (barracuda) drive the NAS was showing as faulty and it passed all of them with no indication of bad sectors of other errors. This is the second Barracuda drive that a Netgear NAS box has indicated problems with that subsequently passed all of Seatools tests. Can anyone throw some light on why Netgear hardware seems to be so fussy?
Dis you run the write zeros (full erase) test?
I haven't found the NAS to be particularly fussy. However, several Barracuda desktop drives use SMR (shingled magnetic recording) technology - and that includes the ST4000DM004. SMR is not well suited for RAID - when you write a lot of data, the write speed slows down to a crawl. Re-syncing for instance can take a very long time, especially if you are loading data onto the NAS during the process. https://blocksandfiles.com/2020/04/15/seagate-2-4-and-8tb-barracuda-and-desktop-hdd-smr/
Personally I don't recommend using SMR drives in any ReadyNAS.
Many thanks for the very detailed reply - not only did it answer my questions but also gave me a better understanding of the Raid process.
I'll add the new drives one at a time and allow the syncing to occur. At the moment doing things at a slower pace isn't a problem.
The only problem I do forsee, perhaps, is mains power outages - which are rare where I live. We occassionaly get momentary outages if the weather gets stormy, but I have UPS powering the NAS that is good for several minutes, certainly enough time to organise a controlled power down.
If this is done does the syncing start over or recommence from where it left off after power down from the NAS power switch. Either way it would not be a problem, but I don't want to lose what is currently on the NAS if it is avoidable.
It should handle a controlled shutdown ok, but it is an unavoidable risk.
But you do have a full backup (since you just loaded the files onto the NAS), so just make sure you keep that.
BTW, I generally do full tests of disks before I add them - including a full erase/write zeros test. Another risk is that a disk will fail during the resync, and the testing does reduce the odds of that.
It should handle a controlled shutdown ok, but it is an unavoidable risk.
The key word here is controlled. A NAS that is "protected by an UPS" includes that UPS being monitored by the NAS and a shut-down being initiated if the UPS battery gets low. An UPS not monitored by the NAS is not fully protect in the NAS from a power failure to the full extent it can be.
