With MongoDb as job repository, restarting Abrupt Shutdown does not work as expected #4943
Description
Bug description
With MongoDb as job repository, restarting Abrupt Shutdown does not work.
Environment
spring-boot-starter-batch 3.5.3 / MongoDB 8.0.12
Steps to reproduce
When a job is stopped properly with
jobOperator.stop(execId);
Here is the result in db :
// collection: content_video.BATCH_JOB_EXECUTION
{
"_id": "689231387c9236ffa8bc83d8",
"jobExecutionId": "42",
"jobInstanceId": "33",
"jobParameters": {
"run.id": {
"value": "fixed",
"type": "java.lang.String",
"identifying": true
}
},
"stepExecutions": [
{
"stepExecutionId": "40",
"jobExecutionId": "42",
"name": "ytVaoSyncStep",
"status": "STOPPED",
"readCount": "6000",
"writeCount": "6000",
"commitCount": "6",
"rollbackCount": "0",
"readSkipCount": "0",
"processSkipCount": "0",
"writeSkipCount": "0",
"filterCount": "0",
"startTime": "2025-08-05T16:28:40.824Z",
"createTime": "2025-08-05T16:28:40.714Z",
"endTime": "2025-08-05T16:29:38.126Z",
"lastUpdated": "2025-08-05T16:29:38.156Z",
"executionContext": {
"map": {
"batch.taskletType": "org.springframework.batch.core.step.item.ChunkOrientedTasklet",
"batch.executed": true,
"batch.version": "5.2.2",
"batch.stepType": "org.springframework.batch.core.step.tasklet.TaskletStep",
"yt.sync.channel.index": 5,
"yt.sync.video.index": 35
},
"dirty": true
},
"exitStatus": {
"exitCode": "STOPPED",
"exitDescription": "org.springframework.batch.core.JobInterruptedException"
},
"terminateOnly": true
}
],
"status": "STOPPED",
"startTime": "2025-08-05T16:28:40.533Z",
"createTime": "2025-08-05T16:28:40.357Z",
"endTime": "2025-08-05T16:29:38.280Z",
"lastUpdated": "2025-08-05T16:29:38.280Z",
"exitStatus": {
"exitCode": "STOPPED",
"exitDescription": "org.springframework.batch.core.JobInterruptedException"
},
"executionContext": {
"map": {
"batch.version": "5.2.2"
},
"dirty": true
},
"_class": "org.springframework.batch.core.repository.persistence.JobExecution"
}
// collection: content_video.BATCH_STEP_EXECUTION
{
"_id": "689231387c9236ffa8bc83d9",
"stepExecutionId": "40",
"jobExecutionId": "42",
"name": "ytVaoSyncStep",
"status": "STOPPED",
"readCount": "6000",
"writeCount": "6000",
"commitCount": "6",
"rollbackCount": "0",
"readSkipCount": "0",
"processSkipCount": "0",
"writeSkipCount": "0",
"filterCount": "0",
"startTime": "2025-08-05T16:28:40.824Z",
"createTime": "2025-08-05T16:28:40.714Z",
"endTime": "2025-08-05T16:29:38.126Z",
"lastUpdated": "2025-08-05T16:29:38.156Z",
"executionContext": {
"map": {
"batch.version": "5.2.2",
"batch.taskletType": "org.springframework.batch.core.step.item.ChunkOrientedTasklet",
"batch.stepType": "org.springframework.batch.core.step.tasklet.TaskletStep",
"yt.sync.channel.index": 5,
"yt.sync.video.index": 35
},
"dirty": true
},
"exitStatus": {
"exitCode": "STOPPED",
"exitDescription": "org.springframework.batch.core.JobInterruptedException"
},
"terminateOnly": true,
"_class": "org.springframework.batch.core.repository.persistence.StepExecution"
}
As you can see, BATCH_JOB_EXECUTION contains the stepExecutions.
When restarting the job with the same parameters, it works as excepted and the job restart with the right context.
But when the job is stopped abruptly (server shutdown), spring doc recommend the set the status to FAILED manually : https://spring.io/blog/2021/01/27/spring-batch-on-kubernetes-efficient-batch-processing-at-scale#4-gracefulabrupt-shutdown-implication
So we update the status from STARTED to FAILED
// collection: BATCH_JOB_EXECUTION
{
"_id": "68930f7d06fc21b2e19f5b64",
"jobExecutionId": "45",
"jobInstanceId": "34",
"jobParameters": {
"runId": {
"value": "1",
"type": "java.lang.Long",
"identifying": true
}
},
"status": "FAILED",
"startTime": "2025-08-06T08:17:01.498Z",
"createTime": "2025-08-06T08:17:01.363Z",
"lastUpdated": "2025-08-06T08:17:01.499Z",
"exitStatus": {
"exitCode": "FAILED",
"exitDescription": "Marked STOPPED after abrupt shutdown"
},
"executionContext": {
"map": {
"batch.version": "5.2.2"
},
"dirty": true
},
"_class": "org.springframework.batch.core.repository.persistence.JobExecution",
"endTime": "2025-08-06T08:18:14.865Z",
"stepExecutions": []
}
// collection: BATCH_STEP_EXECUTION
{
"_id": "68930f8206fc21b2e19f5b65",
"stepExecutionId": "43",
"jobExecutionId": "45",
"name": "ytVaoSyncStep",
"status": "FAILED",
"readCount": "4000",
"writeCount": "4000",
"commitCount": "4",
"rollbackCount": "0",
"readSkipCount": "0",
"processSkipCount": "0",
"writeSkipCount": "0",
"filterCount": "0",
"startTime": "2025-08-06T08:17:06.268Z",
"createTime": "2025-08-06T08:17:06.169Z",
"lastUpdated": "2025-08-06T08:17:50.327Z",
"executionContext": {
"map": {
"batch.version": "5.2.2",
"batch.taskletType": "org.springframework.batch.core.step.item.ChunkOrientedTasklet",
"batch.stepType": "org.springframework.batch.core.step.tasklet.TaskletStep",
"yt.sync.channel.index": 3,
"yt.sync.video.index": 444
},
"dirty": true
},
"exitStatus": {
"exitCode": "FAILED",
"exitDescription": ""
},
"terminateOnly": false,
"_class": "org.springframework.batch.core.repository.persistence.StepExecution"
}
But the BATCH_JOB_EXECUTION.stepExecutions stays empty.
When restarting, the method org.springframework.batch.core.repository.dao.MongoStepExecutionDao#getLastStepExecution get the stepExecution from jobExecution.getStepExecutions() which is empty and causes problems.
I think, like the jdbc version it should fetch the stepExecution from the collection BATCH_STEP_EXECUTION and not from BATCH_JOB_EXECUTION.stepExecutions