[6.10.3] Bad spare value from cache disks with Unraid

[Wirman]

Hi,

i have two Samsung Evo 970 Plus as cache drives. After a reboot the second disk could not be found and assigned to a pool.

I have taken the disk out of the server and put it in an nvme adapter and have run crystal info on it. There it says Critical warning 9 and available spare 0. (Threshold 10)

I have run the smart test with the diskSpeed docker in unraid on the first cache disk and there it also says only 30 available spare.

Why is this value there also so bad?

Here are the other values:

SMART Data as of   Mar 9, 2023 1:13 AM

Temperature 42 Celsius

Warning Comp. Temperature Time 0

Critical Comp. Temperature Time 0

Temperature Sensor 1 42 Celsius

Temperature Sensor 2 40 Celsius

Critical Warning 0x00

Available Spare 30%

Available Spare Threshold 10%

Percentage Used 2%

Data Units Read 7,002,678 [3.58 TB]

Data Units Written 55,581,189 [28.4 TB]

Host Read Commands 68,939,664

Host Write Commands 1,155,659,216

Controller Busy Time 20,175

Power Cycles 144

Power On Hours 6,125

Unsafe Shutdowns 70

 

What can i do to prevent this? Or is it a problem with the disks? (but it contains both disks, is that probable?)

Thanks in advance for your help

 

Wirman

 

 

[JorgeB]

This is the info given by the device, nothing you can do to change that.

[Wirman]

Yeah i know. But the spare value starts with value 100. My NVMe in my Laptop which is 7 years old has still a spare value of 100.

I am just wondering why the spare value of the EVO 970 Plus is so bad in combination with unraid.

The NVMes are only about a year old. One of them is apparently dead, and with the other one it is probably only a question of time.

 

So i am wondering if i can do something to prevent the disk of getting bad available spares.

[JorgeB]

Some NVMe go through the spare faster than others, even same model devices, like some fail sooner than others.

[Wirman]

Okay, but it is not a specific problem of the Evo 970 Plus? Maybe only a bad batch? 

I will send the nvme through RMA and hope the next one will be better.

[JorgeB]

There have been some issues with some Samsung NVMe devices, don't remember exactly the affect models, but that one might be one of them.

[Wirman]

Thanks. Do you recommend any other NVMe or brand?

[JorgeB]

Samsung is usually good, there have been some issues but from what I remember they are solved with a firmware update.

[Wirman]

okay, then i will try the new nvme from rma process with the firmware update.

[mgutt]

21 hours ago, JorgeB said:

they are solved with a firmware update.

Evo Plus never had a firmware bug. 

 

On 3/9/2023 at 2:14 AM, Wirman said:

I have run the smart test with the diskSpeed docker in unraid on the first cache disk and there it also says only 30 available spare.

Why are you using a container. Simply click on the pool name. Smart tests and values are integrated in Unraid.

 

On 3/10/2023 at 11:53 AM, Wirman said:

Maybe only a bad batch? 

Must be. Let Samsung replace it and done.

[mgutt]

On 3/13/2023 at 10:12 AM, mgutt said:

Evo Plus never had a firmware bug. 

It seems I have to correct myself. There exists a different Evo Plus version, which uses the same controller as the 980 Pro and has because of that the same bug:

https://www.reddit.com/r/DataHoarder/comments/x8arle/psa_samsung_980_pro_users_in_china_are_observing/

Quote

According to posts in the SSD sub in Tieba (Chinese equivalence of Reddit), the following are all affected:

980 Pro

PM9A1

970 Evo Plus with the new controller/nand