RAID 5 slows down after a few weeks

I don't know how to troubleshoot this without buying expensive parts, like a RAID controller. Even if I did that, it might take so long (three weeks) for the problem to reoccur that I probably won't be able to use the store return windows. My transfer speeds are fine for weeks and then slow down to a fluctuating 20 to 30 mbps. Then I turn the NAS off for five seconds and turn it back on and I can transfer at a steady 113 mbps again. Is there anything in the diagnostics that should give me some idea of what the problem is?

That gap on the right was where I turned off the power. It happened even before I added this SATA expansion card (https://www.newegg.com/p/1FR-00BS-000D3?Item=9SIAX3GETV7720) and expanded from a three drive array to four drives. Meaning it was the same using both the motherboard SATA ports and the card.

Specs:

Ryzen 3 1200

GIGABYTE GA-A320M-S2H

EVGA 400 N1, 400W

Corsair Vengeance LPX 2x8GB DDR4 2400MHz

Samsung SSD 840 EVO 120 GB (system drive)

4x WD Red 12 TB (RAID 5)

+ that SATA expansion card linked above

Zitat von mi-hol

From reading "transfer speed" I assume network (what kind?) and a client are involved.

what is the reasoning for assuming the root cause is on the server?

I'm accessing the files through Windows 10 on my home network. The previous time, I reset the router, which didn't help. Another time, I was able to fix it by removing the network switch that goes to my PC, which solved it that time.

NAS > router > network switch > PC

I just kind of assumed it was the NAS this time because it was also happening on a computer in another room, which didn't have that problem the previous time, when I still used the switch in my room. I'll check the network when it happens again in probably another couple of weeks.

Someone else told me it might be the power supply, but I feel like 400 Watts should be plenty for four hard drives, a CPU, SSD, three fans and two DDR4 sticks. Yesterday, I turned off Monitoring in the Power Management settings of OMV, believing that maybe the thing automatically selecting the "appropriate CPU levels" might be causing it. Is that definitely not related?

Zitat von mi-hol

I assume you are using a wired LAN, no wireless networking involved at all?

1.) I'm asking because what happened to me was that a laptop connected to LAN cable occasionally used slow WiFi due to a BIOS issue involving hibernation.
2.) The second PC was switched off at the time the issue occurred? Otherwise the (usually) tiny CPU in router could be the root cause.
Since I switched to a recent router with dual core CPU, my users get much better performance.

Yeah, a wired LAN. Neither of the computers has WiFi.

No, the other PC was being used by someone.

Perhaps it is the router. Makes me regret restarting the NAS before testing/resetting the network. I just didn't want to bother the other person using his computer. Will just have to see again in a couple of weeks. Seems to happen sooner when I copy or remux many GB of data in a short period.

Zitat von geaves

Possibly, this is something I never do, I have a switch all my ethernet connections are connected to the switch with a single connection from the router to the switch to give internet access.

Whilst most modern router ethernet connection should function as a switch my experience is that most of the time they appear to act as a hub, which means each ethernet connection gets 'polled' when data is transferred.

Interesting. Okay, then I'm putting my other switch to use.

I don't have Ethernet cables long enough to make it less complicated. Yes, I had the NAS plugged right into the router before. Need to just wait and see.

THis morning the NAS was completely inaccessible, and it's not the first time. Resetting the network didn't help; only turning off the NAS resolved it. Is it possible that this device my brother uses to charge all 120-something batteries on his bicycle sent bad magnetic signals through the case of the NAS? He keeps it right beside the NAS.

I don't think it's the network. I shut it down for twenty minutes and even tried plugging the NAS into the other network switch.

router > network switch x NAS > my computer

Instead of

NAS > router > network switch > computer

And still the only thing that fixed it was restarting the NAS. It's frustrating that there seems to be no way to tell in the OMV menu if a hard drive is bad. Before I restarted the NAS, it kept making a noise every fifteen to forty seconds that I haven't been able to find on YouTube so far. It's far less frequent after the restart, which makes it harder to troubleshoot. I thought of plugging only one hard drive in at a time to find out which of them is making that noise, but how can I do that if it takes sometimes days or weeks for the NAS to slow down? I wouldn't have used RAID 5 if I had known it was going to be this much trouble.

So, what can I do? Just restart whenever it slows down and wait for the bad drive to die?

Actually, after the restart it IS still frequent enough that I can test it? Here's the sound:

So I can try plugging the drives in one at a time to see which of them is making that noise. But is there any way to confirm the slowdowns are a mechanical issue?

Took the NAS to my room where I could monitor it better. Whenever that noise came, the transfer speed slowed down to practically nothing. So I do have to RMA. Never had to do that with a hard drive in my life.

Figured the noise and the read speeds slowing to nothing temporarily was probably one of the hard drives rebooting due to insufficient power, so I plugged the NAS into my PC's PSU to test it my theory. The noise stopped. Then I went and bought another Corsair CX650M at Best Buy and replaced the 400 watt EVGA with it. Just watched a movie over the network for the first time in weeks, without any stutter and freezing. Hopefully this fixes the other issue that I've had far longer (a year) of the NAS severely slowing down after days or weeks too, but time will tell. My speeds since replacing the PSU have been a steady 112 to 113 mbps and sometimes fluctuating between 93 and 107.

Happening again. My transfer speeds fall from around 113 Mbps to 3 Mbps, then stay at about 20 briefly, then jump back up.

To reiterate, I have four 12 TB hard drives in RAID 5, controlled by an SSD. I'm doing SMART tests on all of them, which will take 22 hours, but those tests haven't told me anything useful in the past. The SATA expansion card that they are all connected to is only PCIe 2, so 500 Mbps divided between the five storage devices. I wanna upgrade to a PCIe 3 SATA card. Unfortunately, the motherboard (GIGABYTE GA-A320M-S2H) only has one PCIe 3 slot, and because Ryzen CPUs have no integrated graphics I can't access the BIOS without a graphics card (PCIe 3). RAM is 2 x 8 GB DDR4.

Started happening again probably not an hour after uploading 4 GB of videos to the internet. That took about two hours, so I'm wondering if those crappy upload speeds being extended so long put some stress on the array. Had no problems for a month, after replacing the power supply.

What would you do, without spending hundreds of dollars? How would you troubleshoot it?

Hoping it's just a bad SATA cable, but doubt it.

Someone told me to read my dmesg and /var/log/. And to "dd some big file to /dev/null," whatever that means, to see if it's the network or hard drive.

I barely remember how we set this up. When I plug the monitor in, it asks for my username and password, but won't let me type in the password. Says, "To access the system visit the OMV web control panel," followed by six digits of numbers and letters, a colon and the IP address. Talking about this, I assume:

I don't see anywhere to enter those Linux commands. I know he must have meant directly, meaning not over my browser, but I can't do anything directly because of the passwords not being entered.