"Notify on repeated failure" option and "Notify on recovered"

[scottillogical]

So, I'm not sure how this would be possible to implement in this concourse resource, but I wanted to get the idea out there: For builds that run on timers and notify slack, it can be really noisy unless you have the ability to turn off repeated failure notifications and you can only alert when the job starts working again "the job is back to normal!"

I have no idea how this would work in a course resource, as it requires persistence.

[jhunt]

Yeah, we would definitely need some sort of external memory, like a redis kvstore or a git resource to track history. This is a great idea, but perhaps it requires an addon?

[alexanelli]

just wondering if you've looked into this any further @scottschulthess

[jhunt]

Specifically, should we spin up a separate project (custom API + BOSH release) to handle this "trap"-like behavior?

[jhunt]

I've gotten more and more external requests for this type of functionality in lots of messaging systems, and I've found a need for it more and more myself. However, I don't think it's a good fit for the way this resource is architected, so it needs to be a new project.

[attack]

I was able to replicate this requested functionality using this project and semver-resource. I can extract to a gist if there is interest.

[scottillogical]

interested!