topic Re: DC S4500 Pending Sector Issue in Solid State Drives (NAND)

DC S4500 Pending Sector Issue

vhan1 — Thu, 15 Feb 2018 10:10:54 GMT

Dear,

We have installed dozens of DC S4500 SSD in our servers. Several of them triggers smart attributes alert after warm up stress test.

Here is the output from Intel Datacenter tool of linux:

$> sudo isdct show -smart --intelssd 2

**********Some lines are skipped*************

- C5 -

Action : Pass

Description : Pending Sector Count

ID : C5

Normalized : 100

Raw : 12

Status : 18

Threshold : 0

Worst : 100

**********Some lines are skipped*************

Does raw value of smart attribute C5 indicate healthy problem of SSD? Is so, is file system resides on those SSDs are corrupted when these sectors are hit?

Thanks a lot!

Below is firmware version:

$> sudo isdct show -intelssd 2

- Intel SSD DC S4500 Series PHYS738000YP480BGN -

Bootloader : Property not found

DevicePath : /dev/sg2

DeviceStatus : Healthy

Firmware : SCV10100

FirmwareUpdateAvailable : SCV10111

Index : 2

ModelNumber : INTEL SSDSC2KB480G7

ProductFamily : Intel SSD DC S4500 Series

SerialNumber : PHYS738000YP480BGN

Re: DC S4500 Pending Sector Issue

idata — Thu, 15 Feb 2018 19:00:36 GMT

Hello vanbashan,

Thank you for your interest in the Intel® SSD DC S4500 Series. According to the Intel® Solid State Drive Data Center for SATA SMART Attributes – Application Note (https://www.intel.com/content/dam/support/us/en/documents/solid-state-drives/Intel_SSD_Smart_Attrib_for_SATA.PDF https://www.intel.com/content/dam/support/us/en/documents/solid-state-drives/Intel_SSD_Smart_Attrib_for_SATA.PDF), attribute C5 (pending sector count): Shows the number of current unrecoverable read errors that will be reallocated on next write. Regarding the properties that you shared, please notice that unless the Raw property of the device tends to the Worst property (in this case 100) at an abnormal pace, we can talk about an expected degradation of the SSD. In order to fully define what "abnormal pace" means, details such as workload, and system configuration must be taken into account. Your DeviceStatus appears as Healthy, and it is your best reference regarding the current state of the drive. I'm not sure what you mean by "is file system resides on those SSDs are corrupted when these sectors are hit?" Could you please rephrase your question? If you have further questions, please update your firmware version to SCV10111, run the test again and share the output from the Intel® SSD Data Center Tool version 3.0.10 (https://downloadcenter.intel.com/download/27497?v=t https://downloadcenter.intel.com/download/27497?v=t).Also, could you please provide more details regarding the tasks involved in the warm up stress test? What is the percentage of SSDs that have triggered the SMART attribute alert? I'll be waiting for your response.Regards,Andres V.

Re: DC S4500 Pending Sector Issue

vhan1 — Fri, 16 Feb 2018 02:41:20 GMT

Hi, Andres,

Thanks for your quick reply.

The warm up test is just fill up whole space once and then read out data and verify its checksum by FIO.

We observed 25% of drives (5 drives out of 20) with pending sectors above 0. Besides it, raw values of reallocated sector (0x5) are above 0.

"I'm not sure what you mean by "is file system resides on those SSDs are corrupted when these sectors are hit?" Could you please rephrase your question?"

If some sectors are unrecoverable when read, there should be data lost. Is my understanding correct?

However, I did not meet any application error right now. And these drives passed smartctl extended self test without any error. Is it a firmware bug?

I'd like to try upgrading firmware to see whether the issue can be solved. Is there any place to check firmware release notes of SCV10111 before upgrade?

Re: DC S4500 Pending Sector Issue

idata — Fri, 16 Feb 2018 16:51:18 GMT

Hello vanbashan,

Thank you for the information provided. Loss of data due to unrecoverable sectors is always a possibility, due to this, frequent data back-up and RAID configurations are always recommended. Regarding your inquiry about the firmware release notes, the available information regarding firmware version SCV1011 can be found in the latest Intel® Solid State Drive Data Center Tool – Release Notes (https://downloadmirror.intel.com/27497/eng/Intel_SSD_Data_Center_Tool_3_0_10_Release_Notes_330715-024.pdf https://downloadmirror.intel.com/27497/eng/Intel_SSD_Data_Center_Tool_3_0_10_Release_Notes_330715-024.pdf😞 The following changes are included in this firmware update:

Improved Random Read performance
Faster Secure Erase of the drive
Bug fixes from previous firmware SCV10100

Regards,

Andres V.

Re: DC S4500 Pending Sector Issue

vhan1 — Sat, 17 Feb 2018 02:40:01 GMT

Some disks with pending sector count (C5) above zero are failed without any response. I have to return them to resellers... 😞

Re: DC S4500 Pending Sector Issue

idata — Mon, 19 Feb 2018 17:40:41 GMT

Hello vanbashan,

I'm sorry to hear that some of the SSDs failed and had to be returned. Could you please tell me if you performed the suggested firmware update before the SSDs entered this non-responsive state? Also, you mention that you returned the defective devices to the resellers, will they replace the corresponding SSDs? In case you have any update or question, don't hesitate to contact us. Regards,Andres V.

Re: DC S4500 Pending Sector Issue

idata — Mon, 26 Feb 2018 18:43:25 GMT

Hello vanbashan,

I was wondering if you have any updates. Could you check the previous message? In case you consider that you need further assistance, don't hesitate to contact us. Regards,Andres V.

Re: DC S4500 Pending Sector Issue

vhan1 — Wed, 28 Feb 2018 15:40:24 GMT

Andres，

I was out of office in last few days. Sorry for the late reply.

Several disks have failed. One failed after firm update, but two failed just before firm update. Another one turned to brick during firmware update...

I have returned them to the reseller.

SC 35xx or 36xx series are very solid, but not so good for S4500. I suspect there is quality problem for SSDs in the same batch .

Re: DC S4500 Pending Sector Issue

idata — Wed, 28 Feb 2018 17:10:38 GMT

Hello vanbashan,

Thank you for sharing the update. In case you consider that you need further assistance, don't hesitate to contact us. Regards,Andres V.