fix(aws-batch): job lifecycle, scheduling policies, job definition type check and job state-change events

docs/coverage/README.md

@@ -33,7 +33,7 @@ code does not implement. Machine-readable: [`coverage.json`](./coverage.json).
 | `backup` | [Backup](./aws/backup.md) | - | - | - | 25 |
 | `backupdr` | - | - | [BackupDR](./gcp/backupdr.md) | - | 6 |
 | `bastion` | - | [Bastion](./azure/bastion.md) | - | - | 4 |
-| `batch` | [Batch](./aws/batch.md) | - | - | - | 14 |
+| `batch` | [Batch](./aws/batch.md) | - | - | - | 24 |
 | `batch-azure` | - | [Batch](./azure/batch.md) | - | - | 17 |
 | `bedrock` | [Bedrock](./aws/bedrock.md) | - | - | - | 65 |
 | `bedrockagent` | [BedrockAgent](./aws/bedrockagent.md) | - | - | - | 32 |

docs/coverage/aws/README.md

@@ -15,7 +15,7 @@ Services cloudemu emulates for AWS, by native name. Back to the [cross-provider
 | [AppSync](./appsync.md) | `appsync` | 17 |
 | [Athena](./athena.md) | `athena` | 26 |
 | [Backup](./backup.md) | `backup` | 25 |
-| [Batch](./batch.md) | `batch` | 14 |
+| [Batch](./batch.md) | `batch` | 24 |
 | [Bedrock](./bedrock.md) | `bedrock` | 65 |
 | [BedrockAgent](./bedrockagent.md) | `bedrockagent` | 32 |
 | [BedrockAgentRuntime](./bedrockagentruntime.md) | `bedrockagentruntime` | 3 |

docs/coverage/aws/batch.md

@@ -3,25 +3,41 @@
 
 AWS's `batch` service · portable interface `driver.Batch` · [AWS index](./README.md)
 
-## Operations (14)
+## Operations (24)
 
 | Operation | Description |
 | --- | --- |
+| `CancelJob` |  |
 | `CreateComputeEnvironment` | Compute environments. |
 | `CreateJobQueue` | Job queues. |
+| `CreateSchedulingPolicy` | Scheduling policies. |
 | `DeleteComputeEnvironment` |  |
 | `DeleteJobQueue` |  |
+| `DeleteSchedulingPolicy` |  |
 | `DeregisterJobDefinition` |  |
 | `DescribeComputeEnvironments` |  |
 | `DescribeJobDefinitions` |  |
 | `DescribeJobQueues` |  |
+| `DescribeJobs` |  |
+| `DescribeSchedulingPolicies` |  |
+| `ListJobs` |  |
+| `ListSchedulingPolicies` |  |
 | `ListTagsForResource` |  |
 | `RegisterJobDefinition` | Job definitions. |
-| `TagResource` | Tags (by ARN), shared across all three resource kinds. |
+| `SubmitJob` | Jobs. |
+| `TagResource` | Tags (by ARN), shared across every taggable resource kind. |
+| `TerminateJob` |  |
 | `UntagResource` |  |
 | `UpdateComputeEnvironment` |  |
 | `UpdateJobQueue` |  |
+| `UpdateSchedulingPolicy` |  |
 
 ## Not in scope
 
-_Not documented yet. See the [emulator boundary](../../../README.md) for cloudemu-wide non-goals._
+- Jobs run no workload: a job always succeeds (container `exitCode` 0) once its dependencies succeed. By default it is `SUCCEEDED` as soon as it is submitted; with `--async-settle` it spends about one second each in `SUBMITTED`, `RUNNABLE` and `STARTING`, then five seconds in `RUNNING` (and stays `PENDING` while a dependency is unfinished). `attempts` is always empty, and retry strategies and timeouts are stored and echoed but never applied.
+- `CancelJob` on a `PENDING` job takes effect at once. Real Batch waits for the job's dependencies to finish first.
+- Array jobs: children are derived from the parent, not stored one by one. `SEQUENTIAL` and `N_TO_N` dependencies are honoured. An array parent reports `PENDING` while any child is unfinished. Multi-node parallel jobs run as a single job: `ListJobs` with `multiNodeJobId` returns an empty list.
+- Fair-share scheduling policies are stored, validated and bound to queues, but they never change the order in which jobs run. Jobs never stay `RUNNABLE` because no capacity is free, so "Batch Job Queue Blocked" events are never sent.
+- "Batch Job State Change" events go to the EventBridge default bus on the serve scheduler's tick (and on any Batch job API call), one per status the job passed through; none is sent for the initial submission. In library mode, nothing ticks, so events are published when a job API call sees the change or when `Tick` is called. A job that fails reports only `FAILED`, without the earlier steps. Array children send no events of their own. Real Batch sends no state-change events for compute environments or job queues, so the emulator sends none either.
+- Error messages are written by the emulator. The exception type (`ClientException`, HTTP 400) is the documented one.
+- `containerOverrides`, `ecsPropertiesOverride`, `eksPropertiesOverride` and `nodeOverrides` are merged into the job detail `DescribeJobs` returns (environment variables by name, resource requirements by type, other fields replaced). Overrides are not checked against compute-environment platform rules (for example `nodeOverrides` on Fargate), and multi-node jobs report no per-node `nodeDetails`.

docs/coverage/coverage.json

@@ -2655,6 +2655,9 @@
     "service": "batch",
     "interface": "Batch",
     "operations": [
+      {
+        "name": "CancelJob"
+      },
       {
         "name": "CreateComputeEnvironment",
         "doc": "Compute environments."
@@ -2663,12 +2666,19 @@
         "name": "CreateJobQueue",
         "doc": "Job queues."
       },
+      {
+        "name": "CreateSchedulingPolicy",
+        "doc": "Scheduling policies."
+      },
       {
         "name": "DeleteComputeEnvironment"
       },
       {
         "name": "DeleteJobQueue"
       },
+      {
+        "name": "DeleteSchedulingPolicy"
+      },
       {
         "name": "DeregisterJobDefinition"
       },
@@ -2681,16 +2691,35 @@
       {
         "name": "DescribeJobQueues"
       },
+      {
+        "name": "DescribeJobs"
+      },
+      {
+        "name": "DescribeSchedulingPolicies"
+      },
+      {
+        "name": "ListJobs"
+      },
+      {
+        "name": "ListSchedulingPolicies"
+      },
       {
         "name": "ListTagsForResource"
       },
       {
         "name": "RegisterJobDefinition",
         "doc": "Job definitions."
       },
+      {
+        "name": "SubmitJob",
+        "doc": "Jobs."
+      },
       {
         "name": "TagResource",
-        "doc": "Tags (by ARN), shared across all three resource kinds."
+        "doc": "Tags (by ARN), shared across every taggable resource kind."
+      },
+      {
+        "name": "TerminateJob"
       },
       {
         "name": "UntagResource"
@@ -2700,6 +2729,9 @@
       },
       {
         "name": "UpdateJobQueue"
+      },
+      {
+        "name": "UpdateSchedulingPolicy"
       }
     ],
     "providers": {

docs/coverage/nongoals/batch.md

@@ -0,0 +1,7 @@
+- Jobs run no workload: a job always succeeds (container `exitCode` 0) once its dependencies succeed. By default it is `SUCCEEDED` as soon as it is submitted; with `--async-settle` it spends about one second each in `SUBMITTED`, `RUNNABLE` and `STARTING`, then five seconds in `RUNNING` (and stays `PENDING` while a dependency is unfinished). `attempts` is always empty, and retry strategies and timeouts are stored and echoed but never applied.
+- `CancelJob` on a `PENDING` job takes effect at once. Real Batch waits for the job's dependencies to finish first.
+- Array jobs: children are derived from the parent, not stored one by one. `SEQUENTIAL` and `N_TO_N` dependencies are honoured. An array parent reports `PENDING` while any child is unfinished. Multi-node parallel jobs run as a single job: `ListJobs` with `multiNodeJobId` returns an empty list.
+- Fair-share scheduling policies are stored, validated and bound to queues, but they never change the order in which jobs run. Jobs never stay `RUNNABLE` because no capacity is free, so "Batch Job Queue Blocked" events are never sent.
+- "Batch Job State Change" events go to the EventBridge default bus on the serve scheduler's tick (and on any Batch job API call), one per status the job passed through; none is sent for the initial submission. In library mode, nothing ticks, so events are published when a job API call sees the change or when `Tick` is called. A job that fails reports only `FAILED`, without the earlier steps. Array children send no events of their own. Real Batch sends no state-change events for compute environments or job queues, so the emulator sends none either.
+- Error messages are written by the emulator. The exception type (`ClientException`, HTTP 400) is the documented one.
+- `containerOverrides`, `ecsPropertiesOverride`, `eksPropertiesOverride` and `nodeOverrides` are merged into the job detail `DescribeJobs` returns (environment variables by name, resource requirements by type, other fields replaced). Overrides are not checked against compute-environment platform rules (for example `nodeOverrides` on Fargate), and multi-node jobs report no per-node `nodeDetails`.

internal/settle/settle.go

@@ -83,6 +83,16 @@ const (
 	DefaultRestoreExpeditedSettle = 1 * time.Second
 	DefaultRestoreStandardSettle  = 3 * time.Second
 	DefaultRestoreBulkSettle      = 5 * time.Second
+
+	// DefaultBatchJobPhase is how long an AWS Batch job dwells in SUBMITTED,
+	// RUNNABLE and STARTING before moving on; RUNNING lasts
+	// DefaultBatchJobRunDwell.
+	DefaultBatchJobPhase = 1 * time.Second
+
+	// DefaultBatchJobRunDwell is how long an AWS Batch job stays RUNNING. It is
+	// longer than the other phases so a real CLI or SDK client can observe a
+	// running job and call TerminateJob on it before it finishes.
+	DefaultBatchJobRunDwell = 5 * time.Second
 )
 
 // Window is a read-time overlay describing a resource still settling into its

providers/aws/aws.go

@@ -598,6 +598,7 @@ func wireLifecycleEvents(p *Provider) {
 	p.Glue.SetEventPublisher(p.EventBridge)
 	p.CloudWatch.SetEventPublisher(p.EventBridge)
 	p.SageMaker.SetEventPublisher(p.EventBridge)
+	p.Batch.SetEventPublisher(p.EventBridge)
 }
 
 // awsDrivers assembles the resource-discovery driver set from the provider's
@@ -639,7 +640,7 @@ func awsDrivers(p *Provider) *resourcediscovery.Drivers {
 func (p *Provider) Tickables() []config.Tickable {
 	// SQS and EFS run before CloudWatch so the alarms evaluated in the same
 	// tick see the queue gauges and AWS/EFS metrics they just published.
-	return []config.Tickable{p.SQS, p.EFS, p.CloudWatch, p.SSM, p.CloudFormation}
+	return []config.Tickable{p.SQS, p.EFS, p.CloudWatch, p.SSM, p.CloudFormation, p.Batch}
 }
 
 // Close tears down any real engines wired into the provider via

providers/aws/batch/batch.go

@@ -1,7 +1,10 @@
 // Package batch provides an in-memory mock implementation of AWS Batch: the
-// compute-environment, job-queue, and job-definition control plane. Resources
-// provision synchronously (VALID/ACTIVE immediately), ARNs and ECS cluster ARNs
-// are synthesized, and job-definition revisions increment on re-register.
+// compute-environment, job-queue, job-definition and scheduling-policy control
+// plane plus the job lifecycle. Resources provision synchronously
+// (VALID/ACTIVE immediately), ARNs and ECS cluster ARNs are synthesized, and
+// job-definition revisions increment on re-register. Jobs run no workload: a
+// submitted job walks SUBMITTED -> RUNNABLE -> STARTING -> RUNNING ->
+// SUCCEEDED on the configured clock (instantly unless AsyncSettle is on).
 package batch
 
 import (
@@ -11,6 +14,7 @@ import (
 	"sync"
 
 	"github.com/stackshy/cloudemu/v2/config"
+	"github.com/stackshy/cloudemu/v2/internal/awsevents"
 	"github.com/stackshy/cloudemu/v2/internal/idgen"
 	"github.com/stackshy/cloudemu/v2/internal/memstore"
 	"github.com/stackshy/cloudemu/v2/services/batch/driver"
@@ -24,6 +28,8 @@ const (
 	kindComputeEnvironment = "compute-environment"
 	kindJobQueue           = "job-queue"
 	kindJobDefinition      = "job-definition"
+	kindSchedulingPolicy   = "scheduling-policy"
+	kindJob                = "job"
 )
 
 // Mock is an in-memory implementation of AWS Batch. Each resource kind lives in
@@ -33,26 +39,45 @@ const (
 // reads are copy-on-write (mutations build fresh values, never mutating stored
 // ones in place).
 type Mock struct {
-	computeEnvs  *memstore.Store[driver.ComputeEnvironment]
-	jobQueues    *memstore.Store[driver.JobQueue]
-	jobDefs      *memstore.Store[driver.JobDefinition]
-	jobDefMaxRev *memstore.Store[int32]
+	computeEnvs   *memstore.Store[driver.ComputeEnvironment]
+	jobQueues     *memstore.Store[driver.JobQueue]
+	jobDefs       *memstore.Store[driver.JobDefinition]
+	jobDefMaxRev  *memstore.Store[int32]
+	schedPolicies *memstore.Store[driver.SchedulingPolicy]
+	jobs          *memstore.Store[jobRecord]
+	jobSeq        *memstore.Store[int64]
 
 	// registerMu serializes job-definition revision allocation so two concurrent
 	// RegisterJobDefinition calls for the same name can't claim the same revision.
 	registerMu sync.Mutex
 
+	// policyBindMu serializes binding a scheduling policy to a job queue
+	// (CreateJobQueue, UpdateJobQueue) against DeleteSchedulingPolicy, so a
+	// policy can't be deleted between the in-use check and a queue binding it.
+	policyBindMu sync.Mutex
+
+	// jobsMu serializes job submission (sequence allocation), cancel/terminate
+	// decisions and the event-emission bookkeeping on job records.
+	jobsMu sync.Mutex
+
+	// events publishes "Batch Job State Change" events to the EventBridge
+	// default bus once wired by the provider factory.
+	events awsevents.Emitter
+
 	opts *config.Options
 }
 
 // New creates a new AWS Batch mock with the given configuration options.
 func New(opts *config.Options) *Mock {
 	return &Mock{
-		computeEnvs:  memstore.New[driver.ComputeEnvironment](),
-		jobQueues:    memstore.New[driver.JobQueue](),
-		jobDefs:      memstore.New[driver.JobDefinition](),
-		jobDefMaxRev: memstore.New[int32](),
-		opts:         opts,
+		computeEnvs:   memstore.New[driver.ComputeEnvironment](),
+		jobQueues:     memstore.New[driver.JobQueue](),
+		jobDefs:       memstore.New[driver.JobDefinition](),
+		jobDefMaxRev:  memstore.New[int32](),
+		schedPolicies: memstore.New[driver.SchedulingPolicy](),
+		jobs:          memstore.New[jobRecord](),
+		jobSeq:        memstore.New[int64](),
+		opts:          opts,
 	}
 }
 

providers/aws/batch/batch_test.go

@@ -189,8 +189,10 @@ func TestRegisterJobDefinitionRevisionIncrements(t *testing.T) {
 		t.Fatalf("first revision/status wrong: %+v", first)
 	}
 
-	if len(first.PlatformCapabilities) != 1 || first.PlatformCapabilities[0] != "EC2" {
-		t.Fatalf("platformCapabilities default should be [EC2], got %v", first.PlatformCapabilities)
+	// No default is echoed when the caller sent none (it would drift in
+	// Terraform, which reads the list back as-is).
+	if len(first.PlatformCapabilities) != 0 {
+		t.Fatalf("platformCapabilities should be empty when none was sent, got %v", first.PlatformCapabilities)
 	}
 
 	second, err := m.RegisterJobDefinition(ctx, in)

providers/aws/batch/compute_environment.go

@@ -33,6 +33,10 @@ func (m *Mock) CreateComputeEnvironment(
 		return nil, cerrors.New(cerrors.InvalidArgument, "computeEnvironmentName is required")
 	}
 
+	if err := validateTags(in.Tags); err != nil {
+		return nil, err
+	}
+
 	if in.Type != driver.CETypeManaged && in.Type != driver.CETypeUnmanaged {
 		return nil, cerrors.Newf(cerrors.InvalidArgument, "type must be MANAGED or UNMANAGED, got %q", in.Type)
 	}