Enabling NSX on cluster cause high CPU load

[Dmitry_G]

I'm testing NSX in my nested vSphere lab, when enable NSX on cluster each nested esxi starts to use 100% of 1 pCore assigned to nested ESXi.

esxtop shows that netcpa process use 100% of resource of 1 pcore.

Also I try to reproduce that situation on physical hardware, and the result was the same, netcpa uses 100% of one pcore.

Is it normal situation? Why NSX uses so much resource.

[James__M]

You'll have to give more details:

-is hardware on vSphere and NSX HCL?

-physical hardware specifications

-NSX version

-vSphere components installed, versions and baseline functionality (is environment falling apart or functional and super fast)

-How far have you gotten with the NSX deployment...did you fully deploy solution or stop at the host preparation step when you noticed high CPU load

-are there any other problems in environment or with host/s

-is the solution licensed properly

-have you checked the netcpa log file on the ESXi host?

-have you checked the hostd log file on the ESXi host?

[a_krenev]

HI,

that was a known issue at least for NSX 6.1.3 + vSphere 6, which is now resolved in 6.1.4 update (VMware NSX for vSphere 6.1.4 Release Notes)

[Dmitry_G]

Hi!

Thnx for your replay. As I can see in that article only said about fixed issue with CPU usage on NSX manager and not on ESXi hosts. anyway I'll test this version in my lab.

[a_krenev]

I suppose such kind of fix could be excluded from the release notes because nested environment is not officially supported.

Please let me know if it worked for your lab.

[aalbar]

I had the same issue using NSX 6.1.3 on nested environment. I re-deployed NSX using 6.1.4 and it works like charm!

My compute cluster is using vSAN, so one thing I did this time was to set the vSAN network on different port-group than the NSX VXLAN zone. I though that vSAN traffic on top of NSX would cause this high CPU, especially in nested environments were IP packets usually get DUP-licated