after deployment trigger result variables

[Prem Chudzinski [extensi]]

Hi,

I need to pass a variable deployment.environmentName back to my plan but that doesn’t happen.

This is my workflow: build -> deployment to multiple environments -> after deployment trigger BuildPlan to run integration tests

On the screenshot attached to the marketplace listing I can see some deploymentResult Keys and variables?

https://marketplace.atlassian.com/plugins/com.atlassianlab.bamboo.plugins.bamboo-after-deployment-trigger-plugin/server/overview

How do I pass them to a triggered plan? 

Cheers

Przemyslaw

[rsperafico · Accepted solution]

Hello @Prem Chudzinski [extensi],

You can retrieve the deployment.environmentName variable by calling it in a Script task as example:

echo "bamboo.deployment.environmentName" ${bamboo.deployment.environmentName}

So in your example:

build -> deployment to multiple environments -> after deployment trigger BuildPlan to run integration tests

"build", lets say PROJ-PLAN
"deployment to multiple environments", lets say QA
"after deployment trigger BuildPlan", lets say PROJ-TESTS

In PROJ-TESTS-JOB1, you could add a Script task with the above mentioned, resulting on:

build 02-Jan-2018 19:21:45 bamboo.deployment.environmentName QA

Kind regards,

Rafael

[Prem Chudzinski [extensi]]

That seems to be the solution,

but now I've got a new issue with the after-deployment-trigger

Here's the scenario

1. PROJ-PLAN1

2. Deployment to QA1 and QA2

3. PROJ-PLAN2 (integration tests) using After Deployment Trigger

At this point I assume the QA1 and QA2 deployments should trigger Proj-PLAN2 twice. I use here 

bamboo.deployment.environmentName

To determine against which environment my integration tests should be executed.

What is happening?

Proj-PLAN2 is triggered once.

Sometimes by QA1 sometimes by QA2

In the trigger configuration I've selected Triggering Deployment to "Any" but also tried to create two triggers for QA1 and QA2.

Same result.

Seems like a bug to me.

What do you think?

[Prem Chudzinski [extensi]]

In the logs I can see

2018-01-03 10:15:42,761 INFO [10-BAM::PlanExec:pool-17-thread-3] [ChainExecutionManagerImpl] Build Proj-PLAN2-JOB1-66 has been dispatched

Only once, and should be triggered twice.

[rsperafico]

Hello @Prem Chudzinski [extensi],

Could you please go to Bamboo administration >> Overview >> Plans >> Concurrent builds and Enable it, setting the Default number of concurrent builds allowed to 2

Save the configuration and re-do the testing.

Bamboo's concurrent builds feature allows you to build a plan concurrently on several agents. You might find this useful if a plan is likely to be triggered again before the current build completes.

For more information -  Configuring concurrent builds

Kind regards,

Rafael

[Prem Chudzinski [extensi]]

Hi @rsperafico,

Yes that helped but that seems to be very strange behaviour.

I'm looking into the plugin code and API

nonBlockingPlanExecutionService.tryToStart(plan, triggeringAction);

https://docs.atlassian.com/atlassian-bamboo/6.2.3/com/atlassian/bamboo/plan/NonBlockingPlanExecutionService.html#tryToStart-com.atlassian.bamboo.trigger.Triggerable-com.atlassian.bamboo.build.BuildDetectionAction-

In other words, the method should be used for cases where build absolutely, positively needs to trigger (unless encountering internal error or hitting concurrent build limit)

I do understand now that it was hitting the concurrent build limit. But no errors or logs were printed out to the end user.

I'd expect all triggered builds to be queued.

Will this method put the execution into queue?

https://docs.atlassian.com/atlassian-bamboo/6.2.3/com/atlassian/bamboo/plan/PlanExecutionManager.html#start-com.atlassian.bamboo.plan.cache.ImmutableChain-com.atlassian.bamboo.build.BuildDetectionAction-com.atlassian.bamboo.util.AcquisitionPolicy-

[rsperafico]

Hello @Prem Chudzinski [extensi],

Thank you for raising the issue on https://bitbucket.org/atlassianlabs/bamboo-after-deployment-trigger/issues/3/multiple-environments-not-triggering-a

And regarding to your inquire, "start" would do the job.

Kind regards,

Rafael

[Prem Chudzinski [extensi]]

Just tried it start doesn't work.

[rsperafico]

Hello @Prem Chudzinski [extensi],

You should be looking at com.atlassian.bamboo.plan.NonBlockingPlanExecutionService::tryToStart in the same way https://bitbucket.org/atlassianlabs/bamboo-after-deployment-trigger/src/db69092655fd6a20500c3217b64e45301f495871/src/main/java/com/atlassianlab/bamboo/plugins/afterdeploymenttrigger/DeploymentResultListener.java does.

Kind regards,

Rafael

[Prem Chudzinski [extensi]]

@rsperafico

I've modified the plugin itself to use PlanExecutionManager instead of NonBlockingPlanExecutionService.

I just think my use case is not supported by bamboo.

I'd need an API that would enforce bamboo to queue next job regardless of the concurrent build limit.

My use case is pretty common.

Build->Multiple Deployments->Integration tests against each environment using a single Plan. 

Additionally it should be possible to push some custom variables from the deployment result to the triggered Plan.

At the moment I' have to use the default 

bamboo.deployment.environmentName

to determine which env has triggered my integration tests.

I need to pass more variables like host name for example.

[Artur Grigoryan]

hi. Were you able to pass more variables?

[Bill Rebey]

Note that the answer provided by @rsperafico above:

echo "bamboo.deployment.environmentName" ${bamboo.deployment.environmentName}

Is for use in shell scripts that are run by the bamboo agent.

For use within a program, the actual environment variables have the dots are replaced with underscores, such as this (C# .Net example):

strTargetEnv = Environment.GetEnvironmentVariable("bamboo_deployment_environmentName");