knownasilya
commented
8 years ago Sounds like you are talking about this: https://github.com/Strider-CD/strider-simple-runner/blob/master/lib/index.js#L300
With how the keeper works, it seems that it should be setup for every branch, or it should be refactored to handle other branches. https://github.com/FrozenRidge/dirkeeper
From further inspection, it could be possible to just specify a different folder for each branch..
When deploying new builds sometimes it is beneficial to keep job data. As an example when you run a test deployment on test server you want to keep old data files so application keep running even though new builds are deployed. This can be mitigated by increasing count in keeper on jobs init dirs.
However another issue with this is when there are multiple branches one active branch would starve count in keeper, so branch deployment that you want to keep might also be deleted. Here is an example:
Project with multiple branches: master newFeature anotherFeature development
If count is 4, and if there were one deployment for newFeature, and anotherFeature, having 4 builds in development would destroy builds that you want to keep.
While development branch having new commits, it is very likely that you are testing newFeature branch and committing there as well.
Expected working of keeper is to be branch aware and keep 'count' old builds for each branch as users wish.