Health Monitors

Overview

Health monitors continuously probe pool members and automatically remove unhealthy instances from rotation. When a member fails the configured number of consecutive probes, it is marked OFFLINE and receives no traffic until it recovers. This guide covers all supported monitor types, their configuration parameters, and best practices for different application types.

Prerequisites

An active Polystack account with appropriate permissions
Access to the Polystack Dashboard or CLI configured with credentials
API credentials sourced (source openrc.sh)

Supported Monitor Types

Type	Protocol	Use Case
`TCP`	Layer 4 connection check	Any TCP service — databases, custom protocols
`HTTP`	GET request + response code check	Web applications and REST APIs
`HTTPS`	Encrypted GET request check	HTTPS backends requiring secure probes
`PING`	ICMP echo	Basic host reachability test
`UDP-CONNECT`	UDP connection check	UDP-based services
`TLS-HELLO`	TLS handshake check	Verify TLS is functional without full HTTPS probe

Configuration Parameters

Parameter	Description	Recommended
Delay	Seconds between consecutive probes	`5–10`
Timeout	Seconds to wait for probe response before counting as failed	`3–5`
Max Retries	Consecutive failures before marking member DOWN	`3`
Max Retries Down	Consecutive successes before restoring a DOWN member to ONLINE	`3`
URL Path	Path for HTTP/HTTPS monitors (must return expected code)	`/health`
Expected Codes	Comma-separated HTTP status codes accepted as healthy	`200`

Set Delay to at least twice the Timeout value to prevent overlapping probes during slow response periods. For example: Delay=10, Timeout=5.

Create a Health Monitor

Dashboard
CLI

Navigate to pool health monitor

Open your load balancer, select the Pools tab, and click the pool name. Select the Health Monitor sub-tab. Click Create Health Monitor.

Select monitor type

Select the type that matches your backend service. HTTP is appropriate for web applications; TCP is appropriate for databases and custom protocols.

Set timing parameters

Field	Recommended Value
Delay	`10`
Timeout	`5`
Max Retries	`3`
For HTTP monitors, set URL Path to a dedicated health endpoint (e.g., `/health`)
that returns HTTP 200 only when the application is fully ready.

Confirm creation

Health monitor is ACTIVE. View member operating status to verify probes are running.

Create HTTP health monitor

openstack loadbalancer healthmonitor create \
  --name hm-http \
  --delay 10 \
  --timeout 5 \
  --max-retries 3 \
  --type HTTP \
  --url-path /health \
  --expected-codes 200 \
  pool-http

Create TCP health monitor

openstack loadbalancer healthmonitor create \
  --name hm-tcp \
  --delay 10 \
  --timeout 5 \
  --max-retries 3 \
  --type TCP \
  pool-db

Create HTTPS health monitor

openstack loadbalancer healthmonitor create \
  --name hm-https \
  --delay 10 \
  --timeout 5 \
  --max-retries 3 \
  --type HTTPS \
  --url-path /health \
  --expected-codes 200 \
  pool-secure

View Member Health Status

After creating a health monitor, verify members are being probed and showing the correct operating status:

List member operating status

openstack loadbalancer member list pool-http \
  -c name -c address -c operating_status

Expected healthy output:

+--------+-------------+------------------+
| name   | address     | operating_status |
+--------+-------------+------------------+
| web-01 | 192.168.1.5 | ONLINE           |
| web-02 | 192.168.1.6 | ONLINE           |
+--------+-------------+------------------+

All members show ONLINE operating status — they are healthy and receiving traffic.

Manage Health Monitors

openstack loadbalancer healthmonitor show hm-http

openstack loadbalancer healthmonitor set hm-http \
  --delay 5 \
  --timeout 3

openstack loadbalancer healthmonitor delete hm-http

Health Check Endpoint Design

For HTTP health monitors, implement a dedicated /health endpoint in your application:

What a health endpoint should check

A good health endpoint verifies that all critical dependencies are available:

Database connection is alive
Cache layer (Redis, Memcached) is reachable
Required configuration is loaded
Application can accept requests (not in restart/initialization state)

Return HTTP 200 only when all checks pass. Return HTTP 503 when any dependency is unavailable.

What a health endpoint should NOT do

Do NOT make the health endpoint expensive (no full DB queries)
Do NOT require authentication (health probes run without credentials)
Do NOT check external dependencies not under your control (external APIs, third-party services)
Do NOT return 200 during graceful shutdown — return 503 to drain traffic before stopping

Next Steps

Pools

Manage the pools that health monitors are attached to.

Troubleshooting

Resolve OFFLINE member status and health probe failures.

Create Load Balancer

Create a complete load balancer setup with health monitoring from scratch.

Listeners

Configure the protocol and ports that route traffic to your pools.

Core Services

Other Services

Overview

Supported Monitor Types

Configuration Parameters

Create a Health Monitor

Navigate to pool health monitor

Select monitor type

Set timing parameters

Confirm creation

View Member Health Status

Manage Health Monitors

Health Check Endpoint Design

Next Steps

Pools

Troubleshooting

Create Load Balancer

Listeners

​Overview

​Supported Monitor Types

​Configuration Parameters

​Create a Health Monitor

Navigate to pool health monitor

Select monitor type

Set timing parameters

Confirm creation

​View Member Health Status

​Manage Health Monitors

​Health Check Endpoint Design

​Next Steps

Pools

Troubleshooting

Create Load Balancer

Listeners

Overview

Supported Monitor Types

Configuration Parameters

Create a Health Monitor

View Member Health Status

Manage Health Monitors

Health Check Endpoint Design

Next Steps