Windows Server Summit 2024
Mar 26 2024 08:00 AM - Mar 28 2024 04:30 PM (PDT)
Microsoft Tech Community
LIVE

S2D performance very very poor

Copper Contributor

We're facing a problem of very slow performance with an S2D three node cluster

 

Lenovo SR650 7X05

Windows Server 2019 17738

384Gb RAM

4 HDD 6TB ST6000NM0115

2 NVMe 900GB PX04PMB096

2 Mellanox ConnectX-4

 

The only clue is this "Threshold Exceeded" warning on two of the NVMe physical disks:

Get-PhysicalDisk

DeviceId FriendlyName       SerialNumber         MediaType   CanPool OperationalStatus        HealthStatus Usage
-------- ------------       ------------         ---------   ------- -----------------        ------------ -----
0        ThinkSystem M.2 VD b344a54d77360010     Unspecified False   OK                       Healthy      Auto-Select
2002     ATA ST6000NM0115   ZAD29F2X             HDD         False   OK                       Healthy      Auto-Select
2005     PX04PMB096         8CE3_8E07_0503_B800. SSD         False   {OK, Threshold Exceeded} Healthy      Journal
3006     PX04PMB096         8CE3_8E07_0503_D600. SSD         False   {OK, Threshold Exceeded} Healthy      Journal
3004     ATA ST6000NM0115   ZAD29YNC             HDD         False   OK                       Healthy      Auto-Select
3002     ATA ST6000NM0115   ZAD29V4X             HDD         False   OK                       Healthy      Auto-Select
2006     PX04PMB096         8CE3_8E07_0503_C000. SSD         False   OK                       Healthy      Journal
1006     PX04PMB096         8CE3_8E07_0503_D200. SSD         False   OK                       Healthy      Journal
3003     ATA ST6000NM0115   ZAD29S6W             HDD         False   OK                       Healthy      Auto-Select
1004     ATA ST6000NM0115   ZAD29SE4             HDD         False   OK                       Healthy      Auto-Select
2004     ATA ST6000NM0115   ZAD29VAR             HDD         False   OK                       Healthy      Auto-Select
2003     ATA ST6000NM0115   ZAD29SMK             HDD         False   OK                       Healthy      Auto-Select
3001     ATA ST6000NM0115   ZAD29YL6             HDD         False   OK                       Healthy      Auto-Select
2001     ATA ST6000NM0115   ZAD29YK2             HDD         False   OK                       Healthy      Auto-Select
1003     ATA ST6000NM0115   ZAD29Z0A             HDD         False   OK                       Healthy      Auto-Select
1001     ATA ST6000NM0115   ZAD06HEJ             HDD         False   OK                       Healthy      Auto-Select
3005     PX04PMB096         8CE3_8E07_0503_D100. SSD         False   OK                       Healthy      Journal
1005     PX04PMB096         8CE3_8E07_0503_D300. SSD         False   OK                       Healthy      Journal
1002     ATA ST6000NM0115   ZAD29YQ5             HDD         False   OK                       Healthy      Auto-Select

 

Testing the disks with Lenovo SSD tool I see that the Life Remaining of the disks is still at 99%:

 

(c)Copyright Lenovo 2016.

Portions (c)Copyright IBM Corporation.

 

SSDCLI -- Display SMART Info                    v:7.3.2[Tue Jan  9 15:24:37 2018]

-------------------------------------------------------------------------

1   PN:00YK145-01GT679 SN:S3YAM3EN FW:300BT11L

    Number bytes written to SSD: 98028.9GB

    Number bytes supported by warranty: 17520000GB

    Life Remaining Gauge: 99%

    SSD temperature:22(c)      Spec Max: 70(c)

    PFA trip: No

    Warranty Exceed:No

 

2   PN:00YK145-01GT679 SN:S3YAM3FN FW:300BT11L

    Number bytes written to SSD: 96597.2GB

    Number bytes supported by warranty: 17520000GB

    Life Remaining Gauge: 99%

    SSD temperature:22(c)      Spec Max: 70(c)

    PFA trip: No

    Warranty Exceed:No

 

    2 Device(s) Found(SATA:0  SAS:0  NVME:2)

 

Any help is really appreciated

Alex

 

4 Replies

Thanks for the report.  What build are you using?

 

Steven Ekren

Senior Program Manager

Windows Server

Microsoft

17738

Could you let us know what workload you are running? Also, can you email us the output of Get-SDDCDiagnosticInfo (trying to see what your configuration is to root cause)? (I will dm you my email)

 

Adi Agashe

Program Manager

Windows Server

Microsoft

I've sent you what requested.

We have 52 VMs, with mixed operating system that goes from windows Server 2008 R2 to 2016 and some Ubuntu Linux.

The whole size is 6TB of data.

Thanks

Alessio