Quarkus Nagios

Nagios (https://www.nagios.org/) is a monitoring and alerting tool.

Nagios allows to add custom health checks via shell scripts. A simple way to monitor the uptime of a quarkus application is to curl the (SmallRye) health endpoints.

However, if you want to use more features of Nagios, such as specific error messages, performance graphs, and alert levels, you need to implement a custom script for each aspect you want to check. Scripts need to output their data in a specific format for Nagios to pick it up. Such a setup creates friction between Devs and Operation every time checks need to be modified.

This extension adds endpoints that report all Microprofile health checks in the Nagios format, so that in the Nagios server a small re-usable script around curl is enough to configure all checks.

Furthermore, this extension provides a custom implementation of the Microprofile HealthCheckResponse API, that allows to use more Nagios features:

  • 4 alert levels (ok, warning, unknown, critical)

  • export numerical results as performance data (allows Nagios to graph historic data)

  • re-usable check definitions with Nagios alert ranges

Installation

If you want to use this extension, you need to add the io.quarkiverse.nagios:quarkus-nagios extension first to your build file.

For instance, with Maven, add the following dependency to your POM file:

<dependency>
    <groupId>io.quarkiverse.nagios</groupId>
    <artifactId>quarkus-nagios</artifactId>
    <version>0.1.1</version>
</dependency>

Usage

Nagios Setup

Use this shell script to set up an active service check in nagios:

#!/bin/bash

url=$1

if [ -z "$url" ]; then
    echo "usage: $0 URL"
    exit
fi

declare -A statusmap=(
  ["OK"]="0"
  ["WARN"]="1"
  ["WARNING"]="1"
  ["CRIT"]="2"
  ["CRITICAL"]="2"
  ["UNKNOWN"]="3"
)

result=$(curl --proto-default https --fail --max-time 15 --connect-timeout 7 $url 2>/dev/null)

if [ -z "$result" ]; then
    result="WARN: Got no result from $url"
fi

status=${result%%:*}

if [ -z "${statusmap[$status]}" ]; then
    status="WARNING"
fi

echo "$result"
exit ${statusmap[$status]}

Nagios Health Checks

The Quarkus Nagios extensions works with all health checks. However, this extension provides an implementation of the Health Check API that allows to use additional nagios-specific features, such as performance graphs and additional status values.

@Wellness
@Singleton
public class QueueSizeHealth implements HealthCheck {

    @Override
    public NagiosCheckResponse call() {
        int queueSize = ...
        return QUEUE_SIZE.result(queueSize).asResponse();
    }

    private static final NagiosCheck QUEUE_SIZE = NagiosCheck.named("queue size")
            .performance()           // export as performance data
            .warningIf().above(30)   // warning range
            .criticalIf().above(100) // critical range
            .build();
}

Asynchronous Health Check Helpers

The extensions provides additional helper classes to simplify building responsive asynchronous health checks.

DelayedFailHealthCheck

The DelayedFailHealthCheck will hide a failing health check for a configurable amount of time. This is useful for health checks that are expected to fail for a short period of time, e.g. during a deployment, and will often repair themselves.

Example: a critical result will be reported as OK for one more minute, and as WARNING for 5 minutes.

Uni<NagiosCheckResponse> actual = ...;

return actual.plug(DelayedFailHealthCheck.of(Duration.ofMinutes(1), Duration.ofMinutes(5)));
// or
return DelayedFailHealthCheck.of(actual, Duration.ofMinutes(1), Duration.ofMinutes(5)))

DurationHealthCheck

Measures the duration a supplier or uni needs to complete and reports the result as a check result in milliseconds. To reduce load on the underlying system, it should be combined with a UniSoftCache.

Example:

var check = NagiosCheck.named("queue size")
    .performance()             // export as performance data
    .warningIf().above(5000)   // warning if > 5 seconds
    .criticalIf().above(10000) // critical if > 10 seconds
    .build();

return DurationHealthCheck.measure(check, () -> ...);

StartupOnlyHealthCheck

A health check that will always return OK, after it has returned OK once. Reduces load if an application has multiple start-up checks, where each is expected to pass once the application is initialized.

UniSoftCache

A soft cache that will re-evaluate the supplier after a configurable amount of time, to avoid overloading the underlying system with health checks.

If the underlying check does not complete within a given timeout, it will return the last known result to ensure that the health check remains responsive.

Examples:

Uni<MyValue> actual = ...;
return actual.plug(UniSoftCache.build().waitFor(5).cacheFor(60).initially(myDefault));
return UniSoftCache.build()
    .cacheFor(Duration.ofMinutes(5))
    .deferredValue(myDefault, () -> ...);
Uni<NagiosCheckResult> actual = ...;
return actual.plug(UniSoftCache.build().initiallyTimeout("My Check"));

Extension Configuration Reference

Configuration property fixed at build time - All other configuration properties are overridable at runtime

Configuration property

Type

Default

Root path

Environment variable: QUARKUS_NAGIOS_ROOT_PATH

string

nagios

Default group

Environment variable: QUARKUS_NAGIOS_DEFAULT_GROUP

string

/well/or/group/

Management enabled

Environment variable: QUARKUS_NAGIOS_MANAGEMENT_ENABLED

boolean

true