M365Con 2027 – free online conference · 11–22 Jan 2027 · 120 sessions m365con.net

Health probe

Concept

A periodic TCP, HTTP or HTTPS check that a load balancer sends to each backend instance to decide whether it is healthy and should receive new traffic; Azure probes originate from the address 168.63.129.16.

✦ CONCEPTHealth probeM365.FM
Title graphic: Health probe (Concept)

Description

A health probe is a periodic check that a load balancer sends to each backend instance to decide whether that instance is healthy and should receive new traffic. The check can use TCP, HTTP or HTTPS. In Azure, the probe traffic originates from the address 168.63.129.16. Without a health probe, a load-balancing rule has no way of knowing the state of each endpoint, so the probe is a required part of how a load balancer decides where to send requests.

In Azure Load Balancer, a health probe checks whether application instances in the backend pool are reachable and responding. The Standard SKU supports TCP, HTTP and HTTPS probes. A TCP probe counts as successful when the three-way handshake to the configured port completes. HTTP and HTTPS probes send a GET request to a given path and count only a status 200 as success. Any other response is treated as a failed check, which is why the probed path should be a lightweight endpoint that reliably returns 200 when the application is able to serve traffic.

A probe is defined by a small set of settings: protocol, port, path (for HTTP and HTTPS), interval, and a threshold of consecutive successes or failures. The threshold determines when an instance is marked healthy or unhealthy, so a single missed response does not necessarily take an instance out of rotation. When a probe fails, the load balancer stops sending new connections to that instance. This affects inbound traffic only. Probes can also be used deliberately, for example to steer load away from an instance or to drain connections before planned maintenance.

Several practical conditions must be met for a probe to succeed. The probe source address 168.63.129.16 must be allowed in network security groups and in local firewalls on the virtual machine. The probed port must be open on the VM, and the instance itself must be running. A common cause of an instance appearing unhealthy is a firewall rule or network security group that blocks the probe address, even though the application works for real clients. The documented limits are also worth knowing: HTTPS probes do not support client-certificate authentication, and HTTP probes do not support hostnames.

Health probes sit at the start of the traffic flow described in the M365.FM episode "Azure Load Balancer — Simply Explained". According to the episode, a client sends a request to the load balancer's frontend IP, the load balancer checks for a matching load-balancing rule, and a health probe identifies the available backend virtual machines. The request is then forwarded to a healthy VM using a hashing algorithm, and the response returns to the client through the load balancer. In this sequence, the probe result determines which backend instances are eligible to receive the request at all.

When designing a probe, the main decisions are which protocol to use and what the probe should represent. A TCP probe confirms only that a port accepts connections, while an HTTP or HTTPS probe can reflect whether the application itself is responding with a 200 status on a chosen path. The interval and threshold settings trade off how quickly a failing instance is removed against how tolerant the setup is of brief interruptions. Keeping the probe configuration aligned with the actual application behavior, and with firewall and network security group rules, is what allows the load balancer to route new traffic only to instances that can handle it.

Overview compiled with AI from M365.FM episodes and the facts on this page. You can improve it with the edit button below.

TypeConcept

Frequently asked questions

Questions marked with AI were written with AI from the facts on this page.

What does a health probe do in Azure Load Balancer? (AI)

It periodically checks each backend instance using TCP, HTTP or HTTPS to determine whether the instance is healthy. When a probe fails, the load balancer stops sending new connections to that instance.

What counts as a successful probe? (AI)

A TCP probe succeeds when the three-way handshake to the configured port completes. HTTP and HTTPS probes send a GET request to a given path and count only a status 200 as success.

Why might a healthy VM show as unhealthy? (AI)

The probe source address 168.63.129.16 must be allowed in network security groups and local firewalls, the probed port must be open on the VM, and the instance must be running. A blocked probe address is a common reason for a failed check.

Glossary

Community wiki

Short info above, community knowledge below: add practical tips, examples or corrections. New contributions are reviewed first; trusted members publish directly.

Experts and companies for this topic

Nobody has listed this topic yet. Add it to your profile.

Maintained by Mirko Peters (Microsoft MVP) · Last updated: