Details
-
Enhancement
-
Resolution: Won't Do
-
Major
-
None
-
None
-
None
-
Not Started
-
Not Started
-
Not Started
-
Not Started
-
Not Started
-
Not Started
Description
I propose that we stop deploying the "backend-cron" pod in 3scale, and replace the functionality with other means.
Main benefits of such a change would be:
- few pods, less resources used
- appearance of few jobs, simpler solution
- slight simplification of Operator/template
This pod was responsible for:
- flushing stats - disabled. Could be re-instated if someone added stats export again.
- retry of failed jobs.
Solution #1 (simplest?)
- replace with an k8s/openshift cron-job to retry failed jobs
Solution #2 (more changes...)
- rework retried jobs, maybe sidekiq or similar?
Can you (dortiz-1 amartine@redhat.com msorianod) discuss this and summarize here?