VROC - RAID5 - Parity Errors

[LBois]

I have an ASUS C422 PRO/SE motherboard with Intel VROC Premium chip installed, and an ASUS Hyper M.2 X16 Card.

The motherboard has an Intel W2135 CPU and this processor has been verified.

The board has Kingston 64GB Kit(4x16GB) - DDR4 2400MHz - 64 GB - DDR4 SDRAM - 2400 MHz DDR4-2400/PC4-19200 - 1.20 V - ECC - 288-pin - DIMM

The disks are NVMe Samsung 960 Pro 512GB X4. This is configured in RAID5 bootable with Microsoft Server 2016 Essentials and I'm constantly getting parity errors in my RAID5 and the system is sluggish and sometimes crashes. The sequential read speed is about 3000MB/s and the write speed is an abysmal 49MB/s. These disks have been verified with the latest firmware applied and they are in good condition as per the Samsung Magician disk utility and scanned with HDDScan. BTW the Intel VROC Premium allows bootable RAID with 3rd party SSDs, however after these errors; I decided to get Samsung PM953 NVMe disks which are part of your recommended list of SSDs for VROC.

Unfortunately, I am still getting parity errors with these new disks!!

I configured the RAID5 volume as follows:

VMD – Not Spanned

Close RAID Write Hole - Distributed

Verification – Done after initialization and subsequently on several passes to reduce the number of parity errors upon repair to no avail. The number of errors does not reduce; one pass of verification could have higher or lower number of errors from one scan to another. It makes no difference.

I don't know how to resolve this. I configured the RAID in the BIOS as per the ASUS instructions. I am stumped. Please advise.

[idata]

Hello,

Thank you for contacting Intel Customer Support.We will check with our team in order to provide you with the correct solutions for your platform. We will get back to you once we get the information. In the meantime, can we get some logs of the errors/crashes you're getting? Regards,

[LBois]

Unfortunately, I decided to rebuild in RAID10 to see if I would be experiencing any parity errors and so far, I do not have any. So I'm gonna stick with RAID10; I have been at this for a very long time and I can't afford to keep rebuilding the server (at least 12 times now), with different variances in the configuration. Besides, I am getting no support from ASUS. I escalated this issue to their engineering group and I keep getting the exact copy/paste resolution to "verify all the cables and make sure all is well connected" from different "engineers". I mean, COME ON! They don't know their own hardware; it's a PCIe card with NVMe disks ... there are no cables! I am sticking with RAID10. It is unfortunate that I cannot use the full potential of VROC Premium; I could have settled for VROC Standard and perhaps save some time and money. Even with RAID10 I did experience a massive error and completely lost the installation when the UEFI (Windows Boot Manager) completely disappeared and I had to reinstall the OS once again; I still don't know what happened there but for now, it is still stable and I am letting it run as is for a while. I still believe that the VROC and/or ASUS motherboard and/or their Hyper card are not ready for prime time.

[George2005]

Hi again,

please let me ask - which drives does your current RAID10 consist of? 960 Pro or PM953?

Thank you for your information.

[LBois]

PM953

[LBois]

To be honest, proper detection and RAID implementation of the Pro 960 was never a problem for me. I'm sure it would work. I don't believe the parity errors I experienced was due to the disks themselves since I experienced the same issues with the PM953's as I did with the Pro 960's. My problem was with RAID5 and perhaps I still will with the RAID10 but so far, so good.

[George2005]

OK, thank you very much for your valuable experiences!

[idata]

Hello Luc,

Okay, I see. Then I'm going to escalate this with our engineers and see if we can duplicate this set up and test it, it could take a few days so I'd appreciate your patience. Once we have any update I'll let you know!

Regards,

Carlos V.

Intel Support.

[GraniteStateColin]

idata, did Intel engineers ever find the root cause or a solution? This appears to be the same issue that's still occurring today, in early 2019. See my post with additional details for a very similar question at: https://forums.intel.com/s/question/0D50P00004Dbhf2SAB/getting-massive-parity-errors-for-raid5-arrays-with-more-than-3-drives-drives-are-fine-should-we-ignore-or-is-there-a-fix-for-this

Specifically, I think Intel's RAID5 support reports parity errors if there are more than 3 drives in a RAID5 array following a Verification and Repair run. The more drives in the array, the higher the number of parity errors reported. Same drives report no errors in a 3-drive RAID5 array or RAID1/10 array or in various SMART tests.