NETGEAR is aware of a growing number of phone and online scams. To learn how to stay safe click here.
Forum Discussion
RN314 re-syncing even though sync completed
RN314 from 2014... came with 4x1TB Toshiba MG03ACA100 drives. Noisy as hell, and out of space. So, in process of upgrading to Seagate ST4000DM004 4TB drives. This is for personal use, so not "mission...
Though Marc_V is correct in saying that sometimes spontaneous resyncs happen (and they need to be investigated when they do) - that isn't what is happening to you. This is a normal part of the vertical expansion process. As it happens, I am vertically expanding my own NAS at the moment, so this is top-of-mind.
jimk1963 wrote:
Is it safe to swap out the 3rd drive now? Or do I have to wait for this extra resyncing to finish? Why would it show all the drives as Green if it is still in the process of resyncing???
You need to wait for the second ("extra") resync to complete.
You have two different RAID groups on your system as a result of vertical expansion. When you inserted the second disk, the original 4x1TB RAID group first needed to be rebuilt. After that a new RAID-1 group of 2x3TB is created, and concatenated to the original group in your volume. The NAS for some reason reports both resync operations separately, and that can give you an incorrect impression that the process is complete.
You will see this again when you install the next two drives (which will take longer to complete than the first one, since there are more data blocks to process). The first RAID group gets rebuilt, the new one will be re-shaped (horizontally expanded).
If you have ssh enabled on the NAS, you can watch the progress with
watch cat /proc/mdstat
You'll see both RAID groups for the data volume (md126 and md127), and the status of the one being resynced. There are two much smaller RAID groups (md0 and md1) - md0 is the OS partition; md1 is the swap partition.
Stephen! Thanks again, incredibly helpful. I did let the second resync complete, which took over 10 hours. At midnight, dropped in the third disk, completed in about 5 hours again and Is now on a 20+ hour second sync, exactly as you described. Altogether this process is going to take around 3 full days, which really surprises me. Seems horribly inefficient. I'm wondering now if it would have been better to just copy everything to my larger NAS, replace all 4 drives at once, then just copy everything back. Only talking about 3TB. Seems that would have been way faster than this RAID rebuilding process.
As you continue through the process, you are much better off doing the hot swap, as that lets you and the NAS know what is going on as it happens (removal, insertion, and re-sync are logged). As an EE, you should know that the power pins on the drives are not all the same length, expressly to safely support hot swapping. Hot swapping with a cable where it may not go straight on can still be an issue. But in a NAS, they go in straight.
BTW, "DM" drives are a poor choice for a NAS (or any RAID). They wll likley suffer from premature failure and may cause other issues along the way. There is a reason NAS purposed drives were created.
jimk1963 wrote:
I'm wondering now if it would have been better to just copy everything to my larger NAS, replace all 4 drives at once, then just copy everything back. Only talking about 3TB. Seems that would have been way faster than this RAID rebuilding process.
Often that is more efficient, as that method only builds the array once.
In your case you are resyncing the 1 TB RAID group 4 times, and the upper 3 TB RAID group 3 times. Each resync involved reading or writing each sector in the group at least once. So you are talking about 4*4TB for the original RAID group, and 2*3TB+3*3TB+4*3TB for new one. That's 43 TB of disk I/O.
The other method would have required 16 TB of disk I/O to create the raid array, plus 12 TB of total disk I/O to make and restore the backup.
Note that the RAID groups are "underneath" the file system - providing raw block storage. So the resync process takes the same time no matter how much data is in the volume.
Sandshark wrote:
BTW, "DM" drives are a poor choice for a NAS (or any RAID).
In particular, the ST4000DM004 uses SMR technology: https://blocksandfiles.com/2020/04/15/seagate-2-4-and-8tb-barracuda-and-desktop-hdd-smr/
That's a pretty hot topic now (and Western Digital has also done that with their 2-6 TB NAS purposed drives). Arguably the situation with WDC is worse. Though I believe WDC did tune their SMR in the WD Reds to accomodate RAID (note this requires some speculation, but there is some RAID performance testing out there that supports that idea). Seagate doesn't recommend using DM drives in RAID arrays.
Personally I'm keeping on open mind on whether SMR can be adapted for RAID, but have chosen not to use SMR drives in my own NAS.
FWIW, I agree with Sandshark that hot-swap is preferred - the OS then detects the disk removal/insertions and doesn't need to figure out what changed when the NAS was powered off.
Thanks much StephenB and SandShark. A few hours on this forum with you guys is about 10x more helpful than poring over internet pages. Regarding the HDD choice made here:
- RN314 is for personal archival (photos, videos, work product) and is not a "real-time" NAS in my home network
- Archived info is duplicated across RN528X, RN212, RN314 and cloud storage (not all files, but many)
- RN528X and RN212 use Toshiba Enterprise-grade HDD's designed specifically for NAS (drives provided directly by Netgear)
- Am aware of NAS-specific drives like IronWolf, etc...
- Reasons I chose this drive:
- It's actually on Netgear's approved NAS drive list - do you find that surprising? I did
- With multiple back-ups, my fear of losing data isn't all that high; and with RAID5 I would hope that no more than 1 drive at a time would fail, such that I could recover from it in any event. Is this wrong-headed logic?
- Drives are on a very good sale, around ninety dollars each. For this old NAS, with only 1GbE ports and an Atom 2.1 GHz processor, I decided this would be a good "experimental" box to put to the test these complaints/concerns with SMR. If it fails, worst case I'm out a few hundred bucks
- Wanted lower noise. The 1TB Toshiba's are horrendous, got really tired of Woody Woodpecker all day long. Could have went with WD Reds, but in 30 years of PC ownership I've only had two HDD failures - both were WD. And the Reds also have this SMR issue, and they're more expensive for questionable gain
- Other than reliability worry, my only other worry is this weird issue with SMR writes. Unclear if these drives will exhibit serious write delays, that would kinda suck. I have 30 days to try out these drives, so we'll see
- One possibly silly thing I'm also doing is: (1) swapping out the 2GB DDR3 for a 4GB, and (2) swapping out the loud, crappy Delta fan for a Noctua A9 Flx (3-pin 92mm). More RAM probably does nothing but some have claimed a little "snappier" performance; the fan will definitely help, I've put Noctua's in the XS716E and other Netgear boxes that come with these sh#t Delta fans and they've worked spectacularly
If I'm trying to build a reliable, future-proof box, wouldn't even consider these drives. In fact, what I really want to do, and may do on one of these NAS boxes eventually, is move to enterprise-grade SSD. It's so damn expensive right now that I'm not willing to pull the trigger. But eventually that's the almost-holy grail - silent, ultra fast... it's just a question of reliability.
jimk1963 wrote:
- With multiple back-ups, my fear of losing data isn't all that high; and with RAID5 I would hope that no more than 1 drive at a time would fail, such that I could recover from it in any event. Is this wrong-headed logic?
Multiple backups of course should eliminate any need for data recovery. But depending on RAID redundancy alone (which you are not doing) is IMO a mistake. Some considerations are:
- Power surges (for instance nearby lightning strikes) can damage multiple disks (and UPS or surge protectors might not be enough to block the power surge). And of course there are other physical risks (fire, flood, theft).
- System crashes (and unclean shutdowns) can result in lost writes that can result in an out-of-sync array.
- Often disk issues are latent and don't show up until you later on try to read or write a sector. So the time window for "more than one drive at a time" can be quite long in practice. Scheduling disk tests (and scrubs) on a regular basis can reduce that window, so I recommend people do that.
- Disks are generally identical, are ordered and installed together, and are subject to identical loads under near-identical conditions. So it should be that surprising if they often tend to fail in rapid succession.
- The volume is unprotected during a disk replacement, and the resync requires reading every sector of the remaining disk. If those reads fail, then the RAID array fails and data is lost.
jimk1963 wrote:
- It's actually on Netgear's approved NAS drive list - do you find that surprising? I did
I know it is there, and it surprises me too.
jimk1963 wrote:
but in 30 years of PC ownership I've only had two HDD failures - both were WD.
Actually I had the opposite experience, and had a run of Seagate failures some years back. So I tend to stick with WD. Though one reason I've stayed with the CMR Reds is that I like the power consumption - which results in cooler temps than the comparable Seagates.
I don't use the SMR variants, and there is one WD Red CMR model that hasn't worked out that well for me (which is the WD60EFRX).
jimk1963 wrote:
- Other than reliability worry, my only other worry is this weird issue with SMR writes. Unclear if these drives will exhibit serious write delays, that would kinda suck. I have 30 days to try out these drives, so we'll see
Please follow up and let us know either way.
NETGEAR Academy
Boost your skills with the Netgear Academy - Get trained, certified and stay ahead with the latest Netgear technology!
Join Us!
