HCX - MON (/32) host route removal post vCenter DRS based vMotion

Products

• VMware HCX

Issue/Introduction

• HCX NE MON /32 host route gets removed during vCenter DRS based vMotion event of Cloud workload VM.
• HCX Mobility Optimized Networking (MON) configures (/32) host route on target NSX DLR which may get removed during vCenter DRS based vMotion event for a Cloud workload VM sitting over MON enabled extended segment.
  
  Below logs will be generated under HCX Cloud manager app.log (/common/logs/admin):
  

[NetworkStretchService_SvcThread-402336, j: 8ff2d229, s: e90ed51f, , TxId: ########-####-####-####-############] INFO  c.v.v.h.n.ConfigureHostRoutesOnEdgeJob- Removing {"network":"1<MON enabled VM IP>\/32","next_hops":[{"admin_distance":1,"scope":["\/infra\/segments\/hcx-ne-########-####-####-####-############"]}],"enabled_on_secondary":false,"resource_type":"StaticRoutes","id":"########-####-####-####-############","display_name":"HCX Policy based MON for Subnet <MON enabled VM subnet>","path":"\/infra\/tier-1s\/<T1-router-name>\/static-routes\/########-####-####-####-############","relative_path":"########-####-####-####-############","parent_path":"\/infra\/tier-1s\/<T1-router-name>","unique_id":"########-####-####-####-############","realization_id":"########-####-####-####-############","marked_for_delete":false,"overridden":false,"_create_time":############,"_create_user":"<username>","_last_modified_time":############,"_last_modified_user":"<username>","_system_owned":false,"_protection":"NOT_PROTECTED","_revision":0}

 

 

Environment

VMware HCX

Cause

When vMotion gets performed for a workload VM sitting over extended segment, the HCX Manager wait for certain amount of time before declaring the Cloud VM IP address as stale.
Note: Once its declared to be stale then the existing host route needs to be removed from NSX DLR which is an expected workflow to avoid inactive/stale routes from routing base.
As part of current implementations, when Cloud/dst vCenter is unable to send vMotion event completion response within 1min for the workload VM, HCX Manager removes the configured static/host route against that VM.

IMPORTANT: vCenter DRS vMotion event depends upon a given infrastructure and may take additional time to complete all vMotion related workflow.

Resolution

This issue is fixed in HCX 4.9.1 or later versions.

Workaround:
MON functionality needs to be toggled to re-instate the Network Extension job followed by re-addition of static routes against that workload VM.

Additional Information

Refer HCX - Host Route (/32) missing after enabling MON

Impact/Risks:

• The HCX Migration and DR services remain unaffected.
• The MON enabled workload VM for which the (/32) host route gets removed will face connectivity issues.