Reconnect policy package

Package: Pulse.Mqtt.Resilience.Polly

Use this package when reconnect timing should be driven by an existing resilience pipeline instead of the built-in exponential backoff strategy.

Install

dotnet add package Pulse.Mqtt.Resilience.Polly

What it provides

Type	Contract	Purpose
PollyReconnectStrategy	IReconnectStrategy	Runs each connection attempt through a Polly v8 ResiliencePipeline.

The resilient client still owns state transitions, lifecycle hooks, re-subscription, offline queue flush, publish redelivery, and sticky terminal faults. The package only replaces the retry timing loop.

Configure a pipeline

using Polly;
using Polly.Retry;
using Pulse.Mqtt.Resilience.Polly;

var pipeline = new ResiliencePipelineBuilder()
    .AddRetry(new RetryStrategyOptions
    {
        BackoffType = DelayBackoffType.Exponential,
        UseJitter = true,
        MaxRetryAttempts = int.MaxValue,
        Delay = TimeSpan.FromMilliseconds(500),
        MaxDelay = TimeSpan.FromSeconds(30),
    })
    .Build();

Register the strategy:

builder.Services
    .AddPulseMqttClient("devices", configure)
    .UseReconnectStrategy(_ => new PollyReconnectStrategy(pipeline));

Direct construction uses the same option:

new ResilientMqttClientOptions
{
    Connect = new MqttConnectPacket { ClientId = "devices" },
    ReconnectStrategy = new PollyReconnectStrategy(pipeline),
};

Retry classification

PollyReconnectStrategy controls retry timing. IReconnectDecision still classifies whether an MQTT failure is retryable or terminal. Keep authentication and identity failures terminal unless the application has a real token-refresh path:

.UseReconnectDecision(_ => new TokenAwareReconnectDecision(tokens))

If a decision says "do not retry", the client faults instead of feeding the attempt back into the pipeline.

Operational notes

• Avoid unbounded rapid retries. Always include a delay, jitter, or circuit behavior.
• Keep retry policies cancellation-aware; host shutdown should stop promptly.
• Do not hide terminal broker rejections behind forever-retry policies.
• Use diagnostics snapshots and health checks to observe reconnect pressure.

Related docs

• Resilience
• Extending reconnect strategy
• Health checks