(aws-ecs): `shouldUseCircuitBreaker` warning is emitted for DAEMON scheduling-strategy services, where the deployment circuit breaker doesn't apply

[shorn]

Describe the bug

The @aws-cdk/aws-ecs:shouldUseCircuitBreaker warning ("Enable the 'circuitBreaker' property to trigger a quicker deployment failure...") is emitted for an Ec2Service created with daemon: true (DAEMON scheduling strategy).

The ECS deployment circuit breaker is not applicable to daemon services: its failure threshold is defined as 0.5 × desiredCount (min 3, max 200), and a daemon service has no desired count (it runs one task per instance). The circuit breaker is documented for rolling-update replica services. So this warning is a false positive — following its advice (setting circuitBreaker) is meaningless for a daemon service, and the only way to make it go away is to acknowledgeWarning(...) advice that doesn't apply.

Regression Issue

• Select this option if this issue appears to be a regression.

Last Known Working CDK Library Version

No response

Expected Behavior

No shouldUseCircuitBreaker warning is emitted for services using the DAEMON scheduling strategy.

Current Behavior

The warning is emitted on synth for daemon services:

[Warning at /Stack/DaemonService] Enable the 'circuitBreaker' property to trigger a quicker deployment failure if tasks are failing to come start (without this setting deployments may take up to 3 hours to fail). [ack: @aws-cdk/aws-ecs:shouldUseCircuitBreaker]

Reproduction Steps

const vpc = new ec2.Vpc(this, 'Vpc');
const cluster = new ecs.Cluster(this, 'Cluster', { vpc });
cluster.addCapacity('Cap', { instanceType: new ec2.InstanceType('t3.micro') });

const taskDef = new ecs.Ec2TaskDefinition(this, 'TaskDef');
taskDef.addContainer('c', {
  image: ecs.ContainerImage.fromRegistry('amazon/amazon-ecs-sample'),
  memoryReservationMiB: 128,
});

new ecs.Ec2Service(this, 'DaemonService', {
  cluster,
  taskDefinition: taskDef,
  daemon: true, // DAEMON scheduling strategy
  // no circuitBreaker -> warning is emitted even though it cannot apply
});

Run cdk synth and observe the warning on DaemonService.

Possible Solution

The warning is emitted in BaseService (aws-ecs/lib/base/base-service.ts), guarded only by !props.circuitBreaker && this.isEcsDeploymentController — with no checkfor the scheduling strategy. The daemon/scheduling-strategy information is only set in the Ec2Service subclass, so BaseService never accounts for it.

Suggested fix: suppress shouldUseCircuitBreaker when the service uses the DAEMON scheduling strategy. CDK already special-cases daemon services elsewhere (it validatesdesiredCount, placement strategies, and AZ-rebalancing against daemon mode, and has a dedicated @aws-cdk/aws-ecs:minHealthyPercentDaemon warning — #31705), so thisis an inconsistency rather than intended behaviour. Related signal-vs-noise precedent for an unsuppressable circuit-breaker-adjacent warning: #10059.

The circuit breaker requires the ECS (rolling-update) deployment controller, which daemon services do use — so isEcsDeploymentController is true and the warning fires — but the breaker itself is desired-count based and isn't meaningful for daemon scheduling.

Additional Information/Context

No response

AWS CDK Library version (aws-cdk-lib)

"aws-cdk-lib": "2.258.0"

AWS CDK CLI version

"aws-cdk": "2.1126.0"

Node.js Version

v26.2.0

OS

Windows 10

Language

TypeScript

Language Version

Typescript 5.8.3

Other information

No response

[pahud]

Hi @shorn,

This is a good catch, and a really thorough write-up — your root-cause analysis is exactly right.

Current vs Expected:

                    Ec2Service({ daemon: true })  // schedulingStrategy = DAEMON
                                   │
            ┌──────────────────────┴──────────────────────┐
            ▼                                              ▼
      CURRENT ❌                                     EXPECTED ✅
┌──────────────────────────────┐          ┌──────────────────────────────┐
│ guard:                       │          │ guard:                       │
│  !circuitBreaker &&          │          │  !circuitBreaker &&          │
│  isEcsDeploymentController   │          │  isEcsDeploymentController && │
│  (no daemon check)           │          │  !isDaemon                   │
└──────────────┬───────────────┘          └──────────────┬───────────────┘
               ▼                                          ▼
┌──────────────────────────────┐          ┌──────────────────────────────┐
│ ⚠ warning fires (false +ve)  │          │ no warning — circuit breaker  │
│ only silenced by ack()       │          │ is desiredCount-based, N/A to │
│                              │          │ daemon services               │
└──────────────────────────────┘          └──────────────────────────────┘

From my view this is a genuine false positive. The warning in BaseService is gated only on !props.circuitBreaker && this.isEcsDeploymentController, with no scheduling-strategy check — and as you noted, DAEMON is set in the Ec2Service subclass, so BaseService never accounts for it here.

Worth setting expectations on scope: this is purely an advisory annotation, so it has no effect on the synthesized template or deployment — the only impact is the misleading warning itself (and acknowledgeWarning() being the only way to quiet it). So I'd treat it as a low-severity correctness fix rather than anything affecting running services.

In my opinion the fix is consistent with how CDK already special-cases daemon mode elsewhere (the dedicated minHealthyPercentDaemon warning, desiredCount validation, etc.): thread the daemon/scheduling-strategy signal into BaseServiceProps and suppress this warning when the strategy is DAEMON. It's a small, contained change — if you'd like to put up a PR it'd be very welcome, and thanks again for the detailed diagnosis and repro.

[gingeekrishna]

Hi @pahud ,

I've submitted a fix for this in PR #38140.

The change is a one-line guard in BaseService (base-service.ts):

// Before
if (!props.circuitBreaker && this.isEcsDeploymentController) {

// After
if (!props.circuitBreaker && this.isEcsDeploymentController && additionalProps?.schedulingStrategy !== 'DAEMON') {

Ec2Service already passes schedulingStrategy: 'DAEMON' in additionalProps when daemon: true, so no API surface needed to be added. The fix is consistent with how CDK already special-cases daemon mode for desiredCount validation, minHealthyPercent, and placement strategy checks.

A unit test is included confirming that no shouldUseCircuitBreaker warning is emitted for daemon services, while existing tests confirm replica services are unaffected.

[gingeekrishna]

Opened PR #38250 to fix this.

Root cause: The shouldUseCircuitBreaker warning fires in BaseService whenever !props.circuitBreaker && this.isEcsDeploymentController, with no check for the scheduling strategy. Both Ec2Service and ExternalService already pass schedulingStrategy: props.daemon ? 'DAEMON' : ... as part of additionalProps — so the fix is a single guard: additionalProps.schedulingStrategy !== 'DAEMON'.

Fix: One-line condition change in base-service.ts + a regression test confirming daemon services no longer emit the false-positive warning.