Detecting failures

[matteosuppo]

If I provide a missing image the deployment is sent to kubernetes, but it fails. Is there a way to detect that it failed?

[Ulexus]

Perhaps a modification of the script to subsequently execute kubectl rollout status --watch on the deployment would work. I do this in my own scripts... but I don't recall if it provides machine-readable return codes when failures occur.

[Ulexus]

With a quick check: yes, it appears a non-zero status code is returned when the rollout fails.

[so0k]

This looks like an awesome addition, @MacTynow can review

Internally we've moved away from directly updating deployments in kubernetes and instead use Helm for release management (giving us powerful rollbacks - which include rollback of all resources related to a deployment such as configmaps / secrets / ... )

more details here - http://tech.honestbee.com/articles/devops/2017-07/drone-helm-repository

[Ulexus]

I can understand that; I just find Helm to be too bureaucratic for my tastes.

[mmaryo]

hello any update ?