sfdx force:apex:test:report returns nothing for an AsyncApexJobId with any Aborted ApexClassIds

[Pyronewbic]

Summary

Occasionally, on the platform, we run into issues with particular testMethods for an ApexClass running into loops. Therefore, we abort them (via an sfdx plugin we've created) - but any AsyncApexJobId that has Aborted ApexClassIds returns nothing when the report command is run for the AsyncApexJobId.

Steps To Reproduce:

Here's what a snippet of our code looks like:

private async runAllTests(): Promise<AnyJson>{

    ApexTestRunCommand.id = 'force:apex:test:run' 
    let testRun : any
    if(!this.flags.classnames){
      testRun = await ApexTestRunCommand.run(['-u',this.flags.targetusername,'--json','-l',this.flags.level])
    }else{
      this.ux.log(`Running for Specific TestClassNames!`)
      testRun = await ApexTestRunCommand.run(['-u',this.flags.targetusername,'--json','-n',this.flags.classnames])
    }
    let allTests : any = await this.testRun(testRun.testRunId,0)

    return allTests
  }

private async testRun(testRunId: string, ms:number): Promise<AnyJson>{

    let time = this.flags.retrytime+ms
    let skippedIds = []
    let flag = true

    DataRecordUpdateCommand.id = 'force:apex:data:update'
    DataRecordCreateCommand.id = 'force:apex:data:create'
  
    do {
        let res : any = await this.conn.query(`select Id,ApexClassId,MethodName,RunTime,ApexClass.Name FROM ApexTestResult where AsyncApexJobId='${testRunId}' and RunTime>${time} and ApexTestResult.ApexClassId in (select ApexTestQueueItem.ApexClassId from ApexTestQueueItem where Status='Processing' and ParentJobId='${testRunId}')`)
        
        if(res.records.length>0){
            for(const record of res.records){
              if(skippedIds.includes[record.ApexClassId]){
                this.ux.log(`ApexClassId ${record.ApexClassId} for ${record.ApexClass.Name} has already been Aborted, and records on ApexTestResult have been updated!`)
              }else{
                this.ux.log(`Aborting ApexClassId ${record.ApexClassId} since ${record.ApexClass.Name}.${record.MethodName} has crossed it's Threshold of ${time} milliseconds!`)
                await DataRecordUpdateCommand.run(['-s','ApexTestQueueItem','-w',`ParentJobId=${testRunId} ApexClassId=${record.ApexClassId}`,'-v','status=Aborted'])
                skippedIds.push(record.ApexClassId)
                let resp = await this.conn.query(`select Id,ApexClass.Name,MethodName from ApexTestResult where AsyncApexJobId='${testRunId}' and ApexClassId='${record.ApexClassId}'`)
                let rec : any
                for(rec of resp.records){
                  this.ux.log(`Updating ${rec.ApexClass.Name}.${rec.MethodName} with Id ${rec.Id} as Fail with an error Message!`)
                  await DataRecordUpdateCommand.run(['-s','ApexTestResult','-i',rec.Id,'-v',`Outcome=Fail Message='Class Was Looping!'`])
                }
                this.ux.log('\n') 
              }
            }
        }
        
        await new Promise(r => setTimeout(r, 30000))
        
        let resp: any = await this.conn.query(`select Status FROM ApexTestRunResult where AsyncApexJobId='${testRunId}'`)
        if(resp.records[0].Status=='Completed'){
          this.ux.log(`Running All Tests for id ${testRunId} is done!`)
          flag = false
        }else if(resp.records[0].Status=='Aborted'){
          throw new SfdxError(`This Particular ApexTestRun has been Aborted!`)
        }
    } while (flag)

    let testResults: any = await ApexTestReportCommand.run(['-w', this.TIMEOUT, '-i', testRunId, '-u', this.flags.targetusername , '--json'])
    return testResults
  }

Expected result

The ApexTestReportCommand should've returned results in json, considering that the test run was Completed.

Actual result

The result returned is { status: 0 }

Additional information

I've dug through the code (to salesforce-alm/dist/lib/apex/apexTestApi.js) and noticed this on line 671 under the 'checkProgress' method

if (status.Status === 'Completed' || status.Status === 'Failed'){
    this.completed[status.Id] = status;
  }
  else {
    allCompleted = false;
 }

and have updated it to be

if (status.Status === 'Completed' || status.Status === 'Failed' || status.Status === 'Aborted') {...}

It looks like it returns the test results successfully now - I know this was intended to be by design, but consider this scenario.

SFDX CLI Version: sfdx-cli/7.42.1 win32-x64 node-v8.11.3

SFDX plugin Version: salesforcedx 47.15.0 (core)

OS and version: Windows 10, Version 1809 (OS Build 17763.973)

[clairebianchi]

@Pyronewbic I have notified the team that owns this command and they are taking a look.

[smaddox-sf]

Hi @Pyronewbic - Thanks for raising this. It is working as designed today, but this is an interesting use case. We will look this as a potential enhancement to consider when we do upcoming work in this area.

[iowillhoit]

Hey @Pyronewbic, I am grooming some old Github Issues. This one is a little tricky to track down – a lot of migrations, ownership changes, and features have happened since it was opened. A quick scan of the [migrated] code looks like this might be fixed here.

If this is still not working as you'd expect when using the latest version of sf (currently 2.13.9), could you please open a new Github Issue with updated information. Thank you!

[Pyronewbic]

Hi there, @iowillhoit - it's been a while since I've worked with sfdx, but I've upgraded deps to the latest version, and the changes made help my parallel apex class runs (with aborted ones if they cross a threshold time), return the required output. Thanks for pointing it out!