The azure loadbalancer health probe shows a warning, how to get rid of the warning.

Question

Looking at the metric for one of the NVA components its loadbalancer.

We are seeing 2 weired things.

• data path availability is not always 100, it drops regularly, configured to TCP and port 22 for every 5 seconds interval.
• the health probe shows a warning:

Unhealthy threshold, otherwise known as the property numberOfProbes, is not respected. Load Balancer health probes will probe up/ down immediately after 1 probe regardless of the property's configured value.Learn more.

how do i ensure the data path availability is always 100% and how to get rid of the warning.

Answer 1

Hi Ramesh Kallikuppa GOPALAKRISHNA,

Welcome to the Microsoft Q&A Platform! Thank you for asking your question here.

If you are seeing a significant drop in data path availability, it could be due to various reasons such as network congestion, misconfiguration, or issues with the backend pool VMs. To troubleshoot this issue, you can refer to the documentation on troubleshooting Azure Load Balancer backend traffic for more information.

Troubleshoot Azure Load Balancer resource health, frontend, and backend availability problems | Microsoft Learn

 

The warning message indicates that the unhealthy threshold, which is the number of probes that must fail before a backend pool member is marked as unhealthy, is not being respected. By default, the unhealthy threshold is set to 1, which means that if one consecutive probes fail, the backend pool member is marked as unhealthy. However, if you are seeing the warning message, it means that the backend pool member is being marked as unhealthy after only one probe failure.

• As mentioned in Known issues, the "numberOfProbes" property isn't respected.
• So the value defined in "numberOfProbes" does not take effect.
• If you would like to have a property to control the number of successful or failed consecutive probes necessary to mark backend instances as healthy or unhealthy, please leverage the property "probeThreshold" instead.
  • You should use API version 2022-05-01 or higher if you would like to use this property.

The value of "numberOfProbes" was not configurable using Portal and would stick to the default value of "1"

• I can see the support for this in Azure CLI and Azure Powershell
• az network lb probe
  • Add-AzLoadBalancerProbeConfig

Cheers,

Rohith

If you have any further queries, do let us know.

If the answer is helpful, please click "Accept Answer" and "Upvote it."