Closed tdddblog closed 3 years ago
🤪 Unstable integration & delivery is failing because of wrong JDK version. It is trying to compile with JDK 1.8, but JDK 11 is required to build registry manager and harvest.
Fixed JDK issue in Unstable integration & delivery
, but now it fails with
CRITICAL:pds.roundup.util:💥 Process with command line ['git', 'pull', 'origin', 'master'] failed with status 128
CRITICAL:pds.roundup.util:🪵 Stdout = «»
CRITICAL:pds.roundup.util:📚 Stderr = «From https://github.com/NASA-PDS/pds-registry-mgr-elastic
* branch master -> FETCH_HEAD
* [new branch] master -> origin/master
fatal: refusing to merge unrelated histories
@tdddblog I imagine this actually going to fail for all repos because it is trying to work off of master by default and there is no way to specify the current branch in roundup. Sigh. Will try to look later today or this weekend unless @nutjob4life is blocked on the API and can take a look 😎
You can have multiple workflow files with different triggering events. For example, suppose a repository had:
stable-cicd.yaml
unstable-cicd.yaml
integration-only.yaml
And integration-only.yaml
contained:
on:
push:
branches-ignore:
- 'main'
- 'master'
jobs:
integration-test:
name: ∫ Integration Test
runs-on: ubuntu-latest
steps:
-
name: 💳 Check it out
uses: actions/checkout@v2
with:
lfs: true
token: ${{secrets.ADMIN_GITHUB_TOKEN}}
fetch-depth: 0
# The `ref` is not necessary; it defaults to the branch that triggered the workflow
-
name: ∲ Run integration test
run: |
integ-o-matic --run-parallel 4 --module blah
The above ↑ would be run on a push to any branch but master
(or main
as it should be called these days; master
is passé and potentially offensive). It'll do the integration test only on pushes to non-main branches.
The Roundup could be used here in place of integ-o-matic
or whatever but we'd need to configure it to do a different kind of assembly because the standard assemblies also include creating releases.
Right now it supports three kinds of assemblies:
Only the last two can actually be invoked, but the framework is there to support n kinds of assemblies.
EDIT: after typing this comment I see you do already have an extra .yaml
file in this commit! The key, then, is branches-ignore
.
thanks @nutjob4life . i was hoping to be able to fix that integration issue easily across all repos by replacing the previous with **
, but of course it cannot be that easy :-) .
@nutjob4life actually, thinking about this some more, part of the reason we want to use the unstable_cicd on branches is so that we could potentially tag a SNAPSHOT a tester could use to test with for a PR. not sure if that makes sense, but that was part of the thought process here. it looks like we reference master directly here and here where I think if we just referenced the current branch (or passed it in via the Github action, the part where it breaks would not be an issue
See https://github.com/NASA-PDS/pds-registry-app/issues/154