Control VM failed to sync all routes and routing table is flushed on hosts

Products

• VMware NSX Data Center for vSphere

Issue/Introduction

• Routed traffic get interrupted with static and dynamic routes synched from Control VM
• Control VM version is less than VMware NSX Data Center for vSphere 6.4.5
  
  

Cause

The issue applies to Control VM deployed in the environment and if control VM loses connectivity to local netcpa agent and once it reconnects back, we need to perform full sync of routes to controller once again.
The issue is encountered when a Control VM is configured with both static and dynamic routing (OSPF or BGP)

On the Control VM support logs:

**** AUDIT 0x3d02 - 76 (0000) **** I:00193be3 F:00000002
vmw_sync.c 2267 :at 22:02:12, 12 August 2020 (47641680 ms) Sync layer info: Type: VMCI, send EOM
**** AUDIT 0x3d02 - 76 (0000) **** I:00193be6 F:00000002
vmw_sync.c 2267 :at 22:02:12, 12 August 2020 (47641680 ms) Sync layer info: Type: LOCAL, send EOM
**** AUDIT 0x3d02 - 77 (0000) **** I:4134dd50 F:00000040
vmw_sync.c 1984 :at 19:51:24, 12 August 2020 (1979119922 ms) Sync layer debug: Sync buffer misconfiguration: can’t add EOM

Resolution

This is a known issue with prior to VMware NSX Data Center for vSphere  6.4.5. The issue is resolved in VMware NSX Data Center for vSphere 6.4.5 and above.

Workaround:
Upgrade of DLR control VM to 6.4.5 and above
Add or delete a one or two static routes in Control VM (DLR), Control VM will send static and dynamic (OSPF or BGP) routes to the Controller again

Additional Information

Impact/Risks:
Connectivity issues for the VMs hosted on NSX infrastructure