How to check for build failure within a running build?

[Philip Colmer]

I'm working on a Bamboo plan that interacts with a Service Desk ticket. The idea is that Bamboo does some long-running work and then updates the ticket with the results.

The bit I'm trying to figure out at the moment is how to update the SD ticket if something goes wrong with the Bamboo build. I've added a script to the plan under the "Final tasks" section with the intention that the script would query Bamboo to find out the status of the running build and, if something has gone wrong, update the SD ticket with that information.

I'm querying /rest/api/latest/result/<plan key>/<build number> and getting the data back from Bamboo but "state" and "buildState" are both set to "Unknown".

I could implement a solution by creating a marker file at the start of the plan and deleting it after a successful run of the long-running script then have the "final task" script check for the existence of the marker file but it seems a bit inelegant.

Is there any method for me to retrieve the information I'm looking for?

[Daniel Santos · Accepted solution]

Hi @Philip Colmer

Thank you for clearly stating the problem, it helps a lot who answers.

The perfect solution for you would be using the Pre-Post Build Command Runner plugin. Unfortunately, It is not available for Bamboo versions after 6.7.3. It is an Atlassian Labs plugin and therefore has no support nor updates guaranteed, but there is a request to update it in our issue tracker [BAM-20454] Update Pre-post Build Command Runner for Bamboo 6.8+.

I'm querying /rest/api/latest/result/<plan key>/<build number> and getting the data back from Bamboo but "state" and "buildState" are both set to "Unknown".

The result "Unknown" is expected for the plan assuming the build is still running, which is the case when you call the REST API from your final stage. The requests that would give you some data would be to JOBs.

Assuming your last plan stage is a final stage and you want to check if the build succeeded, you just need to make sure that all jobs in the prior stage succeeded.

I was running some tests and this solution worked for me:

job1=`curl -u<USER>:<PASSWORD> -H 'Accept: application/json' \
           -X GET "<BAMBOO_URL>/rest/api/latest/result/<PROJ-PLAN-JOB1>-KEY-${bamboo.buildNumber}?buildstate" \
           | jq -r '.buildState'`

job2=`curl -u<USER>:<PASSWORD> -H 'Accept: application/json' \
           -X GET "<BAMBOO_URL>/rest/api/latest/result/<PROJ-PLAN-JOB2>-${bamboo.buildNumber}?buildstate" \
           | jq -r '.buildState'`

if [ $job1 == 'Successful' ] && [ $job2 == 'Successful' ]
then
     echo '****Successful****'
else
     echo '****Failure****'
fi

Please let me know if that helps you to move forward.

[Philip Colmer]

Hi Daniel

That works perfectly. Thanks for the suggestion.

Regards

Philip

[benjaminramos1]

Would this work for plan branches?

[Daniel Santos]

It should, but you need to make sure the proper job is used.
For the main plan it will be something like mentioned before:

<PROJ-PLAN-JOB1>

For a plan branch it should be like

<PROJ-PLAN0-JOB1>

each plan branch will have a different PROJ-PLAN"X" name.

You should be able to use a variable (from  Bamboo Variables)to build the job key.

I hope it helps.

[Adrian Raszkiewicz]

Hey! 

I am having trouble accessing the ?buildState field. It doesn't look like the API URI actually holds it? 

Not sure if this is still up to date, but trying to do something very similar and struggling to find an approach.