Hi all As I’m sure you’ve noticed, Travis builds have recently been getting more and more unreliable: either there is a queue of several hours before the build is executed, or in many cases no build is triggered at all. This seems to be due to the limits on Apache’s Travis account. I asked about this in the Apache Infra JIRA, and the answer was as follows:
"Our Travis CI capacity is very limited and we will not be increasing that limit in the foreseeable future. If you are not happy with it, we suggest you switch to GitHub Actions or our Jenkins/BuildBot CI instead.” Given this, I think it would make sense for us to explore alternatives.To try this out, I’ve created a CI build pipeline using Github Actions and raised a PR for it (https://github.com/apache/fineract/pull/1955 <https://github.com/apache/fineract/pull/1955>). This is a “parallel” pipeline - in that I haven’t removed the Travis one. So once this PR is merged, we should get both a Github build as well as a Travis build for each PR. My suggestion would be to merge this and see how the two compare for the next week or so. If Github is more reliable / faster, then we could switch off the Travis one altogether. The only caveat is: I have not been able to test this very comprehensively, as the actions do not run unless the PR is merged. They seem to work OK in my own repository, but I can’t guarantee they will immediately work in the Apache repo. So there may be some false failures initially until we get this build working without issues. However, despite the possible “noise” from these failures, I think this would be a worthwhile exercise. Any concerns? If not, I can merge the actions in tomorrow and then address any issues to get them working asap. Regards Petri
