RN314 re-syncing even though sync completed

Also, thanks for the SSH command line to watch the reshaping. Still on 3rd disk, has been 24 hours so far with roughly 3 hours to go. I suppose the 4th disk tomorrow is going to take 36 hours if I extrapolate. Wow. This odyssey will end up taking 4 full days to complete, with only a few hours of "idle" time due to this third swap finishing in the middle of the night.

---

jimk1963 wrote:

 

I realize there's this NUT concept where ethernet-connected boxes can be relayed a message by the box that's USB-connected to the UPS, but that literally means that all networking gear in the chain, up to and including the router, have to be battery backed too or that NUT command is 100% useless. Correct me if I'm wrong. 

If the NAS are connected to the same switch, then all you need to do is protect that switch.  While unmanaged switches aren't layer-3 routers, they do learn the mac address destinations, and will send the ethernet frames directly to the destination (never reaching your router or upstream switches).  That works with all traffic.  If I am watching a video stored on my NAS on my media player, I can turn off my router and the video continues to play just fine.

That said, I do have multiple UPS (I've picked up too many NAS over the years...).  Another aspect is that you don't get much run-time if you try to power too much stuff (even with the 1500VA model).

One longer-term option on the switches is to replace them with ones that can be powered with PoE.  Then you can use one UPS to protect your router and the network (assuming one PoE switch is co-located with the router).  If the NAS are using remote monitoring with a different UPS, then make sure that the network UPS is sized to give longer run time.

FYI, the less expensive CP1500AVRLCD model will also work with ReadyNAS.

Thanks StephenB, you answered a question I meant to ask - namely, will the local switch connected to all 3 NAS boxes communicate the shut-down message directly, without requiring a trip to the router. WIth that revelation, and since all 3 NAS plus the XS716E sit within a foot of each other, I will try to make a single UPS work. Using the AC adapter power specs on all 3 NAS plus the switch, I get 250+92+60+63 = 465W. This is way overestimated, as per the data sheet the actual power consumptions are of course much lower than the AC bricks are capable of, e.g., the bricks incorporate derating to ensure no issues. The 900W/1500VA should therefore easily run long enough for the system to shuttle a shut-down message over ETH, and allow for all 3 boxes to gracefully shut down before finally killing the switch (if power stays off that long). I have no critical need for auto-restart, I'm simply looking to protect the boxes from instantaneous shut down when they may be in the middle of something.

Will stay with the PFC (pure sine wave) version vs. the AVR version, thanks for pointing out the savings. Reason is, I may eventually relocate some gear to an area occupied by Home Theater devices. If/when I do that, a PFC ensures I don't need to worry about any weird issues with audio noise/compatibility/etc. that fake sine wave boxes sometimes introduce. 

Re: the final (4th) disk, I did a hot swap this morning per yours and SandShark's advice. Still made me nervous, notwithstanding the theory of multiple-length pins, inrush current prevention, etc. But it worked just fine. System has been rebuilding for 12 hours now, and claims to have 20 hours to go. I've found on the 3 previous disks that the 2nd reshaping ends up taking a lot less time than what the NAS estimates. Not sure why, but I would guess that by morning this system will be fully updated. 

Fun learning experience! Next up - swap out the fan (tomorrow) and RAM (Friday).

Job complete:

1) 4x1TB Toshiba HDDs (super noisy!) replaced with 4x4TB Seagate Compute (SMR-based, not recommended for NAS)

2) 2GB DDR3 replaced with 4GB Hynix HMT351S6BFR8C-H9 DDR3 (Amazon), DDR3 PC3-10600S

3) Delta Fan (noisy!) replaced with Noctua NF-A9 FLX 3-pin fan

HDD upgrade elapsed time: 90 hours (roughly 4 days). Powered down NAS for first 3 HDD swap-outs, hot-swapped the 4th drive after getting over my fear.

Biggest lesson learned: RAID5 rebuild times are enormous when swapping in one new drive at a time. And each swap takes progressively longer as StephenB explained earlier. Fourth disk swap took 33 hours to fully reshape drives. Yes, 33 hours. Where is that kind of information in the Netgear literature? As StephenB also concurred, would have been a lot faster to simply copy everything to a second NAS, swap all 4 drives at once, then copy everything back. And would have resulted in something like 28TB in writes vs. my one-at-a-time method that needed roughly 43TB in writes. This was a lot more involved than I had imagined.

Result:  writes are showing more up-and-down variability now (Windows Explorer copy/paste), guessing due to this drive type. That's my "real world" test. Reads are max'ing out the ETH link at about 115-120 MB/s. Attached are further results from ATTO and NAS Performance Tester v. 1.7. StephenB explained why I'm seeing 230MB/s reads on NAS PT and ATTO, which is roughly 2X what I get with Windows Explorer, but I've forgotten why already. Not very happy with writes, they are definitely inferior to previous Toshiba drives. Of course, I replaced 7200 RPM with 5400 RPM (well technically, 5900 RPM says Seagate but ReadyNAS reports 5400 RPM). So I expected the chance of some drop, but not this dramatic up and down roller coaster effect. Old drives had 64MB cache, new Seagates have 256MB "variable" cache, not sure if that's contributing to the up-and-down write speeds.

Noise Results:  QUIET!!! If these drives don't up and die on me, as the experts here believe they may, this box is practically dead silent, even at full speed. I literally have to get within a foot to hear anything. No more Woody Woodpecker from those damn Toshiba drives. The Noctua fan helped immensely as well, fan noise is gone. Disks are running right around 35C with "Cool" profile enabled. The CPU seems to be running a bit hot at up to +60C when fully busy, don't recall it getting that high previously so will watch that. Overall, could not be more pleased.

Finally, purchased a CyberPower CP1500PFCLCD UPS, arrives in a few days. Plan to follow StephenB's advice to connect the 3 NAS, 1 PC and 1 10GbE switch (all devices connect directly to this switch) to the UPS, and program everything for graceful shutdown.

Will monitor drives for health, and will schedule a 3 month full disk check per StephenB's advice. In fact, guess I should run that now as well...? Anyway thanks very much StepenB and SandShark, you guys are superb.

Here's a Windows Explorer file copy example. Copying from a Core i7-6700 PC with EVO 970 Plus SSD (PCIe 3.0 x4 adapter card with 3GB speeds) to the RN314. This example is copying two 1 GB zip files. Transfer starts out at 110-115 MB/s then drops all the way to 0 for several seconds (even though graph shows >0), then back up to full speed, then back to near-zero the second time, then back to full speed again. Must be something about these drives that either Windows or the NAS doesn't like, I guess.

With SMR drives, you will see fast transfer until the cache fills, then a slow-down while it moves data to the actual platters.  Non-SMR will do that some, too, but it's worse with SMR.  But if you don't normally copy large or multiple files to the NAS or just don't care (like with an unattended backup in the wee hours), it can be tolerable.  I am a bit surprised that your memory upgrade didn't help that any, since the NAS will use some for drive buffering.

My NAS are in a storage room (that's air conditioned here in Florida, USA, but still stays warmer than the rest of the house most of the year) and are rack mount.  You think yours was loud?  You should hear a 12-bay rack mount with enterprise drives during a scrub when the fans are at a fairly high speed.

---

Sandshark wrote:

With SMR drives, you will see fast transfer until the cache fills, then a slow-down while it moves data to the actual platters.  Non-SMR will do that some, too, but it's worse with SMR.  But if you don't normally copy large or multiple files to the NAS or just don't care (like with an unattended backup in the wee hours), it can be tolerable. 

---

Exactly so.  But I can't tell from the screenshot whether you are seeing the effects of SMR or not.   There are other possible reasons for a short drop-off in speed (the PC might simply be doing something else for a couple seconds).

I'd be more concerned if the slow down was a lot longer.  

---

StephenB wrote:

---

Sandshark wrote:

With SMR drives, you will see fast transfer until the cache fills, then a slow-down while it moves data to the actual platters.  Non-SMR will do that some, too, but it's worse with SMR.  But if you don't normally copy large or multiple files to the NAS or just don't care (like with an unattended backup in the wee hours), it can be tolerable. 

---

Exactly so.  But I can't tell from the screenshot whether you are seeing the effects of SMR or not.   There are other possible reasons for a short drop-off in speed (the PC might simply be doing something else for a couple seconds).

 

I'd be more concerned if the slow down was a lot longer.  

---

Thanks StephenB , the PC is not doing much at all during these transfers, based on activity monitoring with Task Manager. CPU was at idle as was network access and RAM usage, threads, etc. Looks quite different than the old Toshiba 1TB drives in this respect. Will keep investigating.