HCX - NE appliance VM service pipeline down/up event

Products

• VMware HCX

Issue/Introduction

To document a known issue

Symptoms:

An HCX Network Extension (NE) appliance VM may experience intermittent service pipeline flap events where the service pipeline is reported as down and then reported as up again during the next poling cycle 1 minute later. Another key signature is that during the service pipe down event both the transport (fou) tunnel and IPsec tunnel that the service pipe line is carried inside of are both reported as up between the onPrem and cloud NE appliance pair

The NE service pipeline involves IP connectivity between the onPrem NE appliance VM and cloud NE appliance VM pair

The NE service pipeline down event is typical only seen on one of the NE appliances in a given NE appliance pair at either the onPrem or cloud site 

The NE service pipeline events can be observed either on the HCX Manager UI or vCenter UI at the following locations:
 

HCX Manager UI >> Administration >> Alerts

vCenter UI >> [select NE VM] >> Monitor >> Tasks and Events >> Events

Cause

The poling for NE service pipeline availability between an onPrem and cloud NE appliance pair is currently too aggressive. If only a single probe is lost or delayed it results in a service pipeline down event

For comparison, the IPsec tunnel requires 3 consecutive failed probes to report the IPsec tunnel as down in order to avoid false positives due to temporary CPU or network utilization spikes or data loss in the network path between sites

NOTE - this issue does not apply to the HCX Interconnect (IX) appliance VM that is used for migrations. The service pipeline for an IX appliance has a different use case which is to connect to the local WAN Optimization (WAN-Opt) appliance VM at the same site 

Resolution

Fixed in HCX 4.10.0

Workaround:
None

Additional Information

Impact/Risks:
There is no impact to NE operations or network traffic