"Disk Failed", then "Disk Unplugged" errors

Questions about SNMP, Power, System, Logs, disk, & RAID.
Shmal
New here
Posts: 9
Joined: Wed Aug 13, 2014 1:03 pm

Re: "Disk Failed", then "Disk Unplugged" errors

Post by Shmal » Mon May 11, 2020 3:03 am

Icarus81 wrote:
Sat May 09, 2020 4:40 am
Hello everyone,

Since I also have the problem with my QNAP TVS-473 with the backplane. I want to know from someone. Runs the board without problems since you did the own repair?
I have had no problems with my 473 since doing the repair. Others have had positive results just by changing the ventilation so you could try that first if you have not already.

Icarus81
New here
Posts: 3
Joined: Fri May 08, 2020 4:09 am

Re: "Disk Failed", then "Disk Unplugged" errors

Post by Icarus81 » Mon May 11, 2020 7:49 pm

With the bypass of the MOFSET. Now bay 1, 3 and 4 works for me, but the 2 bay does not! Is there another MOFSET for this one?

Because on the board I only find 2 MOFSET components or is the second one responsible for bay 2.
backplane_473.JPG
Thanks a lot in advance.
You do not have the required permissions to view the files attached to this post.

choyj
New here
Posts: 2
Joined: Sun Mar 26, 2017 10:49 am

Re: "Disk Failed", then "Disk Unplugged" errors

Post by choyj » Tue May 12, 2020 6:56 am

Hansaplastique wrote:
Wed May 06, 2020 5:01 pm
My TVS-873 (bought in 2017) finally gave up completely on bays 3, 4, and 5.
After some back and forth haggling with QNAP support, my NAS now is at QNAP for repair ($590).
Even though QNAP claims that this is not a "known issue", they do have a test procedure.
Support stated that there isa 90% chance they'd send me a new one.
I'm in the exact same situation! TVS-873 bought in May 2017. Disks 3,4,5 went out last month and I rebuilt it, only to have 3,4,5 fail again a few days later. Currently using the fan method (i.e. pop the top off and blow a fan down on it) and it hasn't had a problem since, knock on wood.

I thought about replacing the 4957AGM MOSFETs myself, but thinking about it further... Why did all 3 disks go out at the same time? I highly doubt all 3 MOSFETs decided to misbehave at the same time. Whatever is driving the MOSFETs' gates is likely to be the real culprit. I wish QNAP would give us a root-cause on these failures.

Going to switch to the independent power solution when I get a chance.

WRT hotwiring the source and drain together on the MOSFETs... Peak power draw happens during start-up to spin the disk platters up to speed. That's why the disks are powered on via stages rather than all at once to avoid hitting the power supply hard. Anyone that's hot-wiring worry about the one-time stress it might place on the power-supply when you initially power it on? In my case, if I hot-wired 3,4,5 to always be on, it would potentially mean disks 1-5 would start up at the same time. Assuming a peak draw of 25 watts for each drive, that's 125 watts on the 250 watt power supply. While that's only half, other things are running, too, i.e. system board, fans, etc. I guess I could put my Kill-A-Watt on there and do finger-in-the-wind guesses to see if driving 5-drives together at power-on isn't pushing it too hard.

EasyGo
First post
Posts: 1
Joined: Thu May 07, 2020 9:22 am

Re: "Disk Failed", then "Disk Unplugged" errors

Post by EasyGo » Tue May 12, 2020 11:40 am

This happened to my 2017 vintage TS-831X some weeks ago. Original 8 x 4TB WD Red drives in RAID 10.

Drive 4 was showing an error for several months before I got around to trying to replace the drive. During an attempt to resurrect Drive 4 by scanning for bad spots, I got about 2 minutes into the scan and WHAM! Drive 4 went offline and so did Drive 8. After popping the drives out individually and putting them back in individually, Drive 4 would behave as if it were trying to rebuild, then fail and show unplugged. Drive 8 showed unplugged and wouldn't even try again when popped out and reinserted.

Both "failed" drives were tested in a PC and scanned for bad spots, and they passed all tests with flying colors. Still unconvinced that something was wrong with the QNAP box, I replaced Drive 4 and Drive 8 with new drives. Drive 4 attempted to rebuild, and like the old drive, failed in about 2 minutes. The new Drive 8 wouldn't do anything, showing unplugged.

I acquired SATA extension cables and connected the original Drive 4 and Drive 8 with those and an external power supply. The drives came up, the RAID set rebuilt (after scanning Disk 4 for bad spots and finding none) and all the lights were green! I ran them like this for a couple of weeks, and once I was convinced that it was somewhat reliable, I decided to try and fix the backplane instead of continuing with the external drives and power supply.

This board also shares the same MOSFET configuration as many of the other boards mentioned, having one for each drive. One half of the dual MOSFET switches the 12V line and the other side switches the 5V line. As chripopper noted earlier in the thread with his device, there are solder pads on each side of each MOSFET, on mine marked FBxx instead of Bxx (I remember one was FB16, I didn't write down which ones I jumpered). FB usually means "ferrite bead" which is basically a jumper with a noise-suppression bead on it. It appears that the board was designed to potentially have a bead/jumper installed there instead of the MOSFET sequencing circuit. This made me feel a bit better about jumpering across these pads to bypass the MOSFET on the affected channels, so that's what I did.

The power supply is only 250 watts in my 831X, so I'd be a bit dubious about just jumpering all 8 drives to immediately spin up at turn-on, but after a couple of weeks the jumpers on two channels seem to have worked. Incidentally, I considered that there may have been some drop in voltage or something that caused the initial problem so I cleaned the interconnects and power supply connectors with DeOxit D5. I guess time will tell if it holds up. If any more channels fail, I will likely first verify that the supply voltage is stable and sufficient, and if so, replace the MOSFETs.

I am rather flabbergasted to be having essentially an end-of-life failure of a 3.5 year old piece of very expensive equipment.

Icarus81
New here
Posts: 3
Joined: Fri May 08, 2020 4:09 am

Re: "Disk Failed", then "Disk Unplugged" errors

Post by Icarus81 » Tue May 12, 2020 2:37 pm

Icarus81 wrote:
Mon May 11, 2020 7:49 pm
With the bypass of the MOFSET. Now bay 1, 3 and 4 works for me, but the 2 bay does not! Is there another MOFSET for this one?

Because on the board I only find 2 MOFSET components or is the second one responsible for bay 2.


Thanks a lot in advance.
I have now purchased SATA data extension cables and SATA power splitter cables. So I could find out that the power supply on Bay 2 works but the SATA data port (7 pin) does not work.

Does anyone have a hint what I can do now?

Exarkun77
New here
Posts: 3
Joined: Thu Feb 06, 2020 9:06 pm

Re: "Disk Failed", then "Disk Unplugged" errors

Post by Exarkun77 » Tue May 12, 2020 4:06 pm

MagicStuff wrote:
Fri Nov 08, 2019 3:52 am
Hi,

I'm french and will make an effort to be understandable.

I have read the Komissar's post and now I have my TS-431 fully functional.
Here the fix for this model.

My problem was on the 4th bay with the "Disk unplugged" error several times by minute.

To confirm my problem, I have powered the 4th disk with external PSU like this :
20191105_213604.jpg

20191105_213614.jpg
With this mounting, all is OK. So it's a power default like explained by Komissar.

My back plane with the MOSFET of the third and fourth bay in red circle:
20191105_224102.jpg

The MOSFET is a NTMD6P02 : https://www.onsemi.com/pub/Collateral/NTMD6P02R2-D.PDF

With a multimeter, the document above and a documentation about SATA power, my conclusion is :
- D1 S1 G1 is for the 12V
- D2 S2 G2 is for the 5v

Si, I maked short circuits between D1 and S1, S2 and S2 to inhibt the MOSFET.
Trick : there is two drain (D1 and D1) and (D2 and D2). So, you can choose any drain or overflow with tin between D1 and D1 or D2 and D2 it's not a problem.
My soldering iron is not thight so my welds are not very ** but functional :)
20191106_211312.jpg

The backplane in place :
20191106_211449.jpg

Thanks for the trick. My NAS was shutdown for 2 years because of this default.

I'm open to questions about this model if necessary.

Thanks !

Sylvain (from France)
How is your TS-431 now after the mosfet bypass you did? It's already months since you did the bypass. Is it still working good without Disk 4 getting disconnected?

microsolder
New here
Posts: 5
Joined: Fri May 15, 2020 12:49 am

Re: "Disk Failed", then "Disk Unplugged" errors

Post by microsolder » Fri May 15, 2020 1:50 am

These described issues of a NAS rejecting a disk are not limited to QNAP devices, and not even only to NAS devices. In fact, they are common to all electronics using a high-side PMOS switches, like some LCD TVs.

This particular case of QNAP backplanes with failing PMOS switches is made more manifest by what looks like a design issue in the gate driver of the switch. The case at hand I fixed about a year ago is a QNAP TS-431
that began rejecting disks about two years ago. Changing the failing disk helped for a few weeks or months, but then the new one was also rejected. It started with bay 3 and then was also affecting bay 4.

A closer look at the backplane revealed a NTMD6P02 dual PMOS driven by a 2N7002. The design issue is that the driver circuit has been designed for a rather high resistance level, about 100 kohm, which would be sufficient
except for the surface leaks in a very tight board layout. For that reason the 2N7002 driver is not in all devices able to pull the switch gate down sufficiently. For that reason the channel resistance of the switch is
relatively high, leading to heating of the switch and premature failure by high temperature accelerated channel conductivity degradation (see Wikipedia for NBTI).

If the switch fails, it must be replaced, which is relatively easy for one with experience in surface mount component rework and having the proper tools. Also, the component (NTMD6P02) is readily available and cheap.
The issue is that this does not help in long term. What is required is also to replace the two 100k resistors (marked R67 and R70 on the board of one switch and corresponding resistors in the other) with ones
of lower resistance, like 10 kohm. That keeps the gate pulled lower and the channel resistance and heating lower which improves the lifetime of the switch.

The TS-431 I fixed and modified a year ago has now been in service for a year without any further issues. Unlike disabling the power-off in fixes mentioned earlier in this thread this modification retains the device's
ability to sleep the disks saving both power and the disks.

dolbyman
Guru
Posts: 18134
Joined: Sat Feb 12, 2011 2:11 am
Location: Vancouver BC , Canada

Re: "Disk Failed", then "Disk Unplugged" errors

Post by dolbyman » Fri May 15, 2020 1:54 am

thanks for the great tech analysis

could you maybe upload a picture of the components in question? (resistor labels might be different on other layouts, but at least a picture would give an indication for what to look for)

I wonder if heatsinks could prolong the life of existing components (might be an easier beginners fix than SM soldering)

microsolder
New here
Posts: 5
Joined: Fri May 15, 2020 12:49 am

Re: "Disk Failed", then "Disk Unplugged" errors

Post by microsolder » Fri May 15, 2020 2:37 am

The image of TS-431 back plane, identical to one I worked on, is in posting by MagicStuff » Thu Nov 07, 2019 9:09 pm (page 27 of this thread).
Photos 3 and 4 show the components. The drivers (2N7002) are the SOT-23 transistors Q2 (for bay 4) and Q3 (for bay 3). For Q2 the resistors
are R67 and R74 next to C79, and for Q3 the resistors are R64 and R70 next to C77 (marking for R70 is in group B shown at the edge of the board).
backplane.png
Heat sink won't help if the component is already failing. If the channel resistance is low (gate drive voltage difference is large enough)
no heat sink is needed. For this component the channel resistance is typically below 0.05 ohm.

(The wires in the image were used to monitor the 12V and 5V lines for few hours to analyze the issue.)
cool.png
tripping.png
These images show what happens in the 5V supply. The gate voltage is about 3.2V because of the leak to the gate line (likely from the PCB
surface current). When the device is cool, the Vgs of about -1.8V is sufficient for the switch loss of about 0.04 V (upper trace).
When the device begins warming up, the gate resistance begins rising and in this case at about 8 minutes the loss has risen to about 0.1V
at which time the heating is sufficient to set on the drift to failure. At about 12 minutes the 5V line drops below the 4.75 V which is the
point at which the disk begins to reset (the ticking sound) and soon the controller rejects the disk and shuts it off (12:30).

Disk beginning to reset after the failing 5V line may cause bad sectors or even damage the disk. The reason for the damage is not a bad
disk but the mentioned issue in the power delivery electronics in the back plane.

Another NAS in which a similar behavior develops over time in Buffalo LS-WTGL/R1 a.k.a LinkStation Pro Duo. In that case, the
dual PMOS switch is not in production anymore and there is no replacement with matching pinout. A working switch can be
constructed out of discrete PMOS devices.
You do not have the required permissions to view the files attached to this post.

Beddhist
Starting out
Posts: 13
Joined: Fri Dec 29, 2017 5:36 pm

Re: "Disk Failed", then "Disk Unplugged" errors

Post by Beddhist » Sat May 16, 2020 1:21 pm

Beddhist wrote:
Tue Apr 28, 2020 10:32 pm
Ok, I have the drives working again, by increasing the fan speed to medium. But to get there involved total data loss on disks 3 & 4, with 2 being unknown at this stage. Disk 1 data survived a factory reset and format (ha! nothing was formatted, no matter what QTS said.) Disk 1 is back to normal, but I'm going to start a new topic now to get help creating volumes...
The disk failed again and I increased the fan speed to max. This seemed to fix the problem for now, but the noise is unacceptable.

So, I loaded Sylvain's picture of the jumpered board into my phone and from that a TV repair technician soldered jumpers on. It has now been running for several days with the fan on auto (slow).

Many thanks to Kommisar and MagicStuff.

brunisher
New here
Posts: 3
Joined: Sat Oct 18, 2014 9:31 am

Re: "Disk Failed", then "Disk Unplugged" errors

Post by brunisher » Sat May 23, 2020 3:55 am

Hi - I've seen the process described by Kommisar and MagicStuff *thanks again for the details!), and may have to go through it for a unit.

I am curious though: what does QNAP do on returned/fixed units: do they replace the failing MOSFET? Do they change the backpanel for another one with a different design?

I know it's more work and requires a bit more tooling than just soldering, but am wondering if replacing the MOSFET isn't a more reliable approach than just bypassing it. Maybe a safer one as well? (Unsure what's the full impact of bypassing the MOSFET)

Beddhist
Starting out
Posts: 13
Joined: Fri Dec 29, 2017 5:36 pm

Re: "Disk Failed", then "Disk Unplugged" errors

Post by Beddhist » Sat May 23, 2020 9:09 am

IIRC somebody mentioned near the start of this topic that this particular component is no longer produced, so replacing it will be difficult.

I suspect that they are used to control when the drive starts up upon boot. Worst case, all drives start at the same time. What effect this has on your PSU obviously depends on its electrical specs and what and how many drives you have. The effects could be cumulative and long term, who knows? Mine are on 24/7, so hopefully they will keep gong for a long time.

ghibli39
First post
Posts: 1
Joined: Sun May 24, 2020 8:10 pm

Re: "Disk Failed", then "Disk Unplugged" errors

Post by ghibli39 » Sun May 24, 2020 8:33 pm

Hello Kommisar, could you help me by sending the information to ghibli39@gmail.com My qnap TS-451A (TS-451 II BP V1.0 QX70) is presenting this fault at disk 4, thanks. The QNAP Netherland ask me 250 dollars and the shipping cost. Thank you so much
Top

dolbyman
Guru
Posts: 18134
Joined: Sat Feb 12, 2011 2:11 am
Location: Vancouver BC , Canada

Re: "Disk Failed", then "Disk Unplugged" errors

Post by dolbyman » Sun May 24, 2020 9:23 pm

just read the posts..email hasn't been neccessary in a while

brunisher
New here
Posts: 3
Joined: Sat Oct 18, 2014 9:31 am

Re: "Disk Failed", then "Disk Unplugged" errors

Post by brunisher » Tue May 26, 2020 1:24 am

Beddhist wrote:
Sat May 23, 2020 9:09 am
IIRC somebody mentioned near the start of this topic that this particular component is no longer produced, so replacing it will be difficult.

I suspect that they are used to control when the drive starts up upon boot. Worst case, all drives start at the same time. What effect this has on your PSU obviously depends on its electrical specs and what and how many drives you have. The effects could be cumulative and long term, who knows? Mine are on 24/7, so hopefully they will keep gong for a long time.
I assume it's the backplane that is no longer produced, but I guess the MOSFET is still available?

That's why I was curious what QNAP was actually fixing when it was still sending back "fixed" units: I'm not sure if they were replacing with the same backplane model but new MOSFETs. Or if they were sending back the NAS with an updated backplane altogether.

Post Reply

Return to “System & Disk Volume Management”