-
Story
-
Resolution: Unresolved
-
Major
-
None
-
None
-
None
-
8
-
False
-
None
-
False
-
OCPSTRAT-1389 - On Cluster Layering: Phase 3 (GA)
-
-
-
MCO Sprint 259, MCO Sprint 260, MCO Sprint 261, MCO Sprint 262
-
0
-
0.000
Currently, we are using bare pod objects for doing our image builds. While this works, it makes adding retry logic and other things much more difficult since we will have to implement this logic. Instead, we should use Kubernetes Jobs objects.
Jobs have built-in mechanisms for retrying, exponential backoff, concurrency controls, etc. This frees us from having to implement complicated retry logic for build failures beyond our control such as pod evictions, etc.
Done When:
- BuildController uses Kubernetes Jobs instead of bare pods to perform builds.
- All tests have been updated.
- incorporates
-
OCPBUGS-32809 When OCB is enabled, MCO is stuck if a MC is deleted
- New
-
OCPBUGS-21647 When OCB is enabled, a node is drained before completing the image build
- Closed
- links to