Support automatic retry of failed builds
It would be great if CircleCI allowed you to configure an automatic retry of a build upon failure. Ideally, you would be able to specify the max number of times you would like it to retry.
Sebastian Lerner it does not solve the issue, which is automatically re-run the failed tests.
Hi folks, we do not have to configure automatically retries, but we do have something that should help some of the scenarios where an automatic retry would be wanted.
Currently in a closed preview/beta: https://circleci.com/docs/rerun-failed-tests-only/
Let me know if there are questions, you can comment on our community forum as well: https://discuss.circleci.com/t/product-launch-re-run-failed-tests-only-circleci-tests-run/47775/4
we need this feature. especially for QA framework executions
We need this
I came here looking for this. Such a thing would be very helpful to resolve an issue with certain steps that interact with external services or steps that have to wait for resources to load that are irreparably flaky. Please consider adding a
max_retriesparameter or something similar!
With GH merge queue this would be great feature to prevent the merge train from being stuck
This would definitely come in handy for situations where the failure is very likely to be transient. For instance, we run a link checker, and that will fail during brief outages of the websites we are linking to.
We have some custom bash logic to retry our test suite when it fails. However, retrying the test suite is _slow_, as you would imagine. I would love greater visibility into transient failures without failing people's builds, so that they can be investigated.
Would also like to see this implemented. Any updates here?
I think what most of us are looking for is essentially being able to automatically execute “Rerun Workflow From Failed” n times when a job fails.
It's easy enough to retry a step with some basic bash logic, however, particularly when running integration tests we occasionally run into transient memory issues that require the job itself to be rerun in a new container. This cannot be achieved via bash or orbs to my best knowledge.
I know that it's possible to retry an entire pipeline by making an API call to CircleCI when: on_fail but that is wasteful when only 1 or 2 jobs are failing in a workflow consisting of dozens of jobs.