-
Suggestion
-
Resolution: Unresolved
-
None
-
None
-
6
-
1
-
Problem Definition
Currently, there is no feature in bamboo to perform automatic retry of failed incomplete builds/jobs or rerun the entire build automatically.
Suggested Solution
Introduce a feature to schedule the automatic restart of failed or incomplete jobs. If a plan continues to fail, even after multiple retries of failed or incomplete jobs multiple times, the feature can schedule a complete rerun of the entire build.
Also, has the option to set a number of tries to trigger the failed or incomplete builds/jobs.
Workaround
Currently, this can be implemented by utilizing the scripting/coding.
For example:
Using REST Call for checking build status:
curl --user <user-name>:<password> -X GET http://robhit-proxy.com/rest/api/latest/result/ATP-ATA/latest # Here "ATP-ATA" is the key of the plan
Then using REST Call to trigger a build:
curl --user <user-name>:<password> -X PUT http://robhit-proxy.com/rest/api/latest/queue/ATP-ATA-buildNumber # Here "ATP-ATA" is the key of the plan
[BAM-19006] Automatic Retry for Bamboo
We would also like to see this functionality added to Bamboo along with the ability to re-run failed jobs BEFORE the build is complete. The situation we're running into that's frustrating engineers is this: Let's say we have a build with two long-running jobs. One job fails occasionally and needs to be re-run. The other job succeeds. One cannot re-queue the failed job until AFTER the build completes and fails. This results in wasted cycles because the dev must wait for the time for the first job to pass and the second job (re-run) to pass before they have a successful build result whereas if the job can be re-run while the first (passing) job runs, the increase in time is only the time between the job failure and when it was re-queued.
One use model we have is to make a fix, then do 'Rerun failed/incomplete jobs only'. Often, we have other jobs running, so doing the rerun is not possible without stopping the current build or just waiting. Waiting is tedious, so it would be useful to select 'Rerun only failed/incomplete jobs when current run is finished'.