TVS-951X RAM issues

[nishantsri25]

Hi Guys:
Couple of months ago, I bought the TVS-951X. It came with 2G of factory installed RAM which I upgraded to 2x16G (32GB) using Kingston RAM model KVR24S17D8/16. The model is the recommendation on Kingston's website for TVS-951X. Original installed RAM was ADATA AD4S240022G17-BSSE (2GB).

All, went well and QTS detected 32G of memory and all of the functionalities worked as expected until I realized that the files copied from my local machine to QNAP are getting corrupted. I verified by comparing SHA256 HASH. While trying to figure out the issue, I investigated:
- Network interfaces
- SMB/NFS protocols settings
- RAID - tried re-initializing from scratch
- Disabling Q-tier etc.

Finally, I found out that the real problem is to do with the RAM. if I switch back to 2G factory installed RAM, the problem goes away. After fiddling with RAM installation, I observed:
16G RAM in Slot 1 only - WORKS
16G RAM in Slot 2 only - DOESN'T WORK
16G RAM in both Slot 1 and 2 - DOESN'T WORK
16G Kingston in slot 1 and 2G ADATA in Slot 2 - WORKS
2G ADATA in slot 1 and 16G Kingston in Slot 2 - DOESN'T WORK

where:
WORKS = No file corruption
DOESN'T WORK = File is corrupted

Note that it does not matter which 16G module I use for my testing. They both behave the same as per the above observation.

Has anyone experienced this problem? Seems like the Slot 2 is the problem and it does not do well with 16G RAM module.

I have reported this to support and it's being investigated.

Thanks,
-N

[dolbyman]

I have 10GB in mine, 2GB (stock) and 8GB works great, make sure you retain the original RAM or QNAP will refuse any support (even via ticket)

[nishantsri25]

Thanks for the quick response.
I do have the original ADATA RAM. And QNAP support has been really helpful so far.
I will let you know how it goes.

[nishantsri25]

So, I figure out what the problem was. It seems like this model is incompatible with RAM with CAS latency of 17-17-17.
I switched to a 16-16-16 RAM and it all worked well in my preliminary test.

The RAM model I used is:
https://www.gskill.com/specification/2/ ... cification

I will update if something changes.

[nishantsri25]

Final update...after almost a week of testing, everything is working as expected with these new Gskill RAM.
I learned/realized how much problem an incompatible RAM can lead to. With the Kingston RAM, I was having file corruption as well as RAID issues. Most of the time, after a restart, the RAID will start re-building for no reason. I was also having trouble with Container station.

However, now with the new RAM module installed all of those issues have vanished. I have 2 CentOS based VMs and one Confluence instance running within Container Station and they are working as expected.

Support has been really great. They were willing to RMA the unit based on all the test results. But I'm glad we were able to address this without the hassle.

My configuration:
Storage: 5x2TB WD Red HDD in RAID6 + 4x500GB WD Blue SSD in RAID5 with Q-Tier
RAM: 32 GB DDR4 2400

[gabeyd]

Interesting thread, thanks.

I'm having problems with my TVS-951X corrupting my RAID array, see my thread here:
viewtopic.php?f=25&t=155420

I'm going to try swapping back to the factory 2G Adata RAM and see if my RAID array stays stable after a rebuild with the factory RAM installed.

I'm using Crucial 32GB 2x16GB CT2K16G4SFD824A. It's also CAS 17, but isn't the 2G Adata factory stick CAS 17 also? Not sure CAS latency is the issue...

Have you been running fine still with the Gskill RAM?

Did you ever run MemTest86 on your Kingston 32G kit? My Crucial kit ran fine for 24 hours with 0 errors.

[gabeyd]

And that was it. So, at least my TVS-951X is not compatible with Crucial 32GB CT2K16G4SFD824A. Or, that RAM is marginally failing.

The ONLY symptom of bad RAM was raid scrubbing finding blocks to repair. I never noticed any corruption of files, but there could be some.

I ordered the same Gskill 32GB kit that you got, hopefully it works for me too.

[gabeyd]

Yes, the gskill ripjaws F4-2400C16D-32GRS is working for me. No more raid errors!

[treylo]

Hopping into this thread. I have a TS-451 which I upgraded to 8GB of Crucial RAM. Over the last few years, I've regularly noticed file corruption when copying to/from USB drives and over the network. I have another TS-451 with a 4GB RAM upgrade but the problems always seemed isolated to the NAS with the 8GB upgrade. Based on this thread, I'm wondering if the RAM is at fault.

You mention trying different combinations and

Finally, I found out that the real problem is to do with the RAM. if I switch back to 2G factory installed RAM, the problem goes away. After fiddling with RAM installation, I observed:
16G RAM in Slot 1 only - WORKS
16G RAM in Slot 2 only - DOESN'T WORK
16G RAM in both Slot 1 and 2 - DOESN'T WORK
16G Kingston in slot 1 and 2G ADATA in Slot 2 - WORKS
2G ADATA in slot 1 and 16G Kingston in Slot 2 - DOESN'T WORK

I am curious what your testing procedure was to determine "works" vs "doesn't work". I don't ALWAYs get corrupted files but I had enough of them that I started double-checking every single file after it was transferred and some are corrupt while others are fine. Are you just transferring a single large file back and forth and then running some kind of check on it? I don't have the raid-scrubbing issues, I only initially noticed it because I couldn't get some video files to play and then I started doing binary-comparisons and found that a bunch of files were different once they were copied. I now feel like I should come up with a specific process for testing and wondered if you had any suggestions. Thanks

[Yippym]

Hopping into this thread. I have a TS-451 which I upgraded to 8GB of Crucial RAM. Over the last few years, I've regularly noticed file corruption when copying to/from USB drives and over the network. I have another TS-451 with a 4GB RAM upgrade but the problems always seemed isolated to the NAS with the 8GB upgrade. Based on this thread, I'm wondering if the RAM is at fault.

You mention trying different combinations and

Finally, I found out that the real problem is to do with the RAM. if I switch back to 2G factory installed RAM, the problem goes away. After fiddling with RAM installation, I observed:
16G RAM in Slot 1 only - WORKS
16G RAM in Slot 2 only - DOESN'T WORK
16G RAM in both Slot 1 and 2 - DOESN'T WORK
16G Kingston in slot 1 and 2G ADATA in Slot 2 - WORKS
2G ADATA in slot 1 and 16G Kingston in Slot 2 - DOESN'T WORK

I am curious what your testing procedure was to determine "works" vs "doesn't work". I don't ALWAYs get corrupted files but I had enough of them that I started double-checking every single file after it was transferred and some are corrupt while others are fine. Are you just transferring a single large file back and forth and then running some kind of check on it? I don't have the raid-scrubbing issues, I only initially noticed it because I couldn't get some video files to play and then I started doing binary-comparisons and found that a bunch of files were different once they were copied. I now feel like I should come up with a specific process for testing and wondered if you had any suggestions. Thanks

I believe 'Doesn't Work' meant as the system didn't boot, due to the RAM he got it will apply the CAS Timing of 17-17-17 from the SPD. This would cause compatibllity issue for the system to boot. Until he has figured out that the CAS Timing of 16-16-16 is compatible, he can then use the new RAM.

If you are concerned with the RAM that is installed on your system, you should boot MemTest86 bootable on USB
https://www.memtest86.com/download.htm

Though faulty RAM would normally show symptoms of random restart with no causes, but running the test will help provide confirmation that the RAMs are good. The test normally last for hours, it will do unlimited cycles. I would do 2-3 cycles to ensure the 1st cycle wasn't a fluke.

[gabeyd]

I believe 'Doesn't Work' meant as the system didn't boot, due to the RAM he got it will apply the CAS Timing of 17-17-17 from the SPD. This would cause compatibllity issue for the system to boot. Until he has figured out that the CAS Timing of 16-16-16 is compatible, he can then use the new RAM.

If you are concerned with the RAM that is installed on your system, you should boot MemTest86 bootable on USB
https://www.memtest86.com/download.htm

Though faulty RAM would normally show symptoms of random restart with no causes, but running the test will help provide confirmation that the RAMs are good. The test normally last for hours, it will do unlimited cycles. I would do 2-3 cycles to ensure the 1st cycle wasn't a fluke.

The NAS boots fine with the "bad" ram, and in fact, I ran MemTest86 for 92 hours, which was 12 passes, with 0 errors.

Yet, my raid scrub was repairing blocks 100% of the time that I ran it with that RAM.

The other poster, was also able to boot and run fine, but was seeing file corruption...

It's not clear if these symptoms are "compatibility" or marginally bad RAM, but in either case, MemTest86 was not sufficient for proving "working" or not, and neither was just booting.

[Yippym]

Ahh misread the whole topic, this is news to me for RAM not working correctly with software even if the MemTest passed fine. Though MemTest will tell what timing you're currently using and you can compile a list of bad timing for the TVS-951X. There could be a odd race condition with RAM cycles, maybe the slower the RAM cycle the data is flagged corrupted by QNAP by software?

But memtest just prove the RAM works fine, I cannot see that the RAM being defected so QNAP is doing funny with the data.

Maybe just try sticking a native speed of DDR4-2133 for the CPU Memory controller, as faster RAM sometimes gives strange CAS timing combo when downclocking.