"Disk Failed", then "Disk Unplugged" errors

[Hansaplastique]

I just found out some interesting that may or may not be helpful to others;
I did get these unexpected "Disk Failed" and "Disk Unplugged" errors as well, as described in my previous post.

As I was tinkering with my QNAP (TVS-873), I opened the case and noticed a more stable environment pretty quickly after that (the array was actually booting just fine).

Adding an additional (very big) fan on top of the open case, actually made all components drop significantly in temperature (see table below).
As of that moment, my QNAP is running very stable, and I'm not longer seeing these errors.
Removing the fan however, made the array less stable again within a few minutes, and closing the case brought me back to the disappearing disks.

Now the "normal - closed case" temperatures and "extra cooled - open case" are in a very normal range, I would think anyway.

Code: Select all

        Normal cooling       Extra cooling
CPU     ~ 43 C               ~ 30 C
Disks   ~ 38 - 44 C          ~ 32 - 38 C
M2SSD   ~ 36 C               ~ 50 C

* all temps in Celsius

Even QNAP support confirms that these temperatures are quite normal, either way.
The PSU however becomes pretty hot.

Also note that I have a old TS869Pro running, with the exact same brand/model disks, and that one shows a higher CPU temp (different CPU, so no surprise there) at 52 C, and the disks are in the 35 - 38 C range (closed enclosure, no extra cooling) and runs just fine.

Since QNAP will charge an absurd amount of money for it's RMA services, I'm wondering if an extra fan could an alternative the solution.
Any advice/insight is most welcome.

[Shmal]

To summarize, the issue is HW related. At this stage there is no SW/FW fix. If your "Disk Unplugged", your NAS is on warranty and you see symptoms described above - contact QNAP. If it is not possible just follow the provided clues.

Thanks for posting this! I had a 473 die and RMA'ed, now a year and a bit later it is showing the same symptoms (last time they just swapped the backplane). I believe, following your clues, that I understand what to do but would appreciate a quick message if you are still watching this post to confirm I am on the right track / have identified the correct part. Or if anyone else has the details, please email me.

shmalspam at gmail.

[Toxic17]

I curious as to why people are wanting tech advice from an email rather than just asking to post the fix in the forums for everyone to see. Asking for a guide for a fix via email is going to solve what for everyone else?

This is a community forum. So we share information on the forum

Sent from my GM1913 using Tapatalk

[Celser]

Say Kommisar, can you send me the instructions to fix this problem. I to have a nas with a dead backplane that is out of warranty. Also, have you tried using a IGBT instead of a MOSFET?, anyway, thank you.

celser@hughes.net

[dolbyman]

if at any point..anyone actually gets a reply to their email..could that person please just post the guide?

email requests for the method seem silly

[Shmal]

Based on Kommisar's post we gave it a shot today and it looks to be working. Disk 3 is on, and raid is rebuilding. If you are going to attempt this on your own boards, read and understand Kommisar's post so you can trace your own board. Verify your problem is identical before trying this solution. Here is what was needed for my specific scenario (disk 3 failure, eventually leading to no power).

For reference, my board is labelled TS-473 BP V1.4 (QY04):

image5.jpeg

The mosfet that seems to be the issue is on the opposite side, and is labelled 4957AGM. Google for pinouts.

image3.jpeg

That IC contains 2 mosfets. Tracing the pins looks like one feeds the 12V sata pins and the other feeds the 5V sata pins (at least for bay 3, other bays may need a different fix). Shorting the source and drain of each will bypass the mosfet and supply power permanently to the sata ports. One or both could be faulty, in my case it looks like only the 5V fix was needed but we did both. You could also just replace the IC with another 4957AGM, although it may still fail again (something caused it to fail originally).

Shorted:

image0.jpeg

Make these changes at your own risk. Modifying a circuit should be your last resort! RMA it if you are not familiar with what is shown here (or don't have a friend who is!).

Might not hurt to ask QNAP support if they will offer up a replacement board for sale (it's an easy enough swap), assuming they have a solution to the fault.

Thanks Kommisar for the diagnostic work!

[Hansaplastique]

Thanks Shmal for posting, and thanks Kommisar for the diagnostic work and fix!

[Celser]

When I was talking to Qnap, the tech said I could not replace the backplane myself because it had to be re-initialized............ that's some bull if that's not true. I bought a different NAS but would lie to get my original one back up for back-up purposes.

[dolbyman]

if it is just a(or two) failing power rails for the drive, nothing has to be reinitialized on fix

[Hansaplastique]

@Celser; QNAP support said the same thing to me and wanted to charge me $530 + shipping.

[hbunjes]

With my TVS-473 it seems to be temperature related too. I will try to place a heat sink on the 4957AGM. Anybody tried that already?

[Hansaplastique]

I have the (seemingly) temperature related version as well. I have not tried the Komissar-fix yet.

Instead I grabbed a small USB fan (one of those really cheap ones, you'd normally plugin to the USB slot and place on your desk), and pointed it at the QNAP's CPU and the 2 SSD's I have installed (right side). It runs at a very low speed, so I can hardly hear it. For now, I have my case open (for now).

This seems to work very well - temperature drops for CPU and SSD's of more than 10 degrees C and zero disks issues.

[hbunjes]

The interesting question is: Which component must be cooled? If a backplane swap fixes it, it should be a circuit on the backplane.
Does your USB fan also cool the backplane?

[dolbyman]

temp could just be a warping issue of the component, you could try to reflow the pins of the mosfet

[hbunjes]

Yip thats right.

I will try both as soon as I get my TVS-473 back. I sent it to my dealer who sent it to QNAP in the netherlands. But nothing happens so I will probably have to reclaim it.
Thank you!

I will report my progress (or frustration) as soon as I get the device back.