Open waciumawanjohi opened 2 years ago
Just to +1 this, this causes some unexpected behavior at the workload level too if a supplychain is stamping out a runnable because the workload will report that it is Ready
before the runnable tasks have run because the runnable itself is reporting that it is Ready
.
Just to clarify, the desired behaviour for the Ready
condition on Runnable
would be as follows:
Ready: True
-> Latest thing passed
Ready: False
-> Latest thing failed
Ready: Unknown
-> Something Running
Current state of the code:
Based on the current behavior, Outputs should be read as "LatestSuccessfulOutputs" -
We can absolutely implement @jwntrs spec, I'll use this policy:
Open questions about Runnable Status:
Additional:
Did #876 and #900 fix this? Check and resolve @squeedee
Bug description:
When a Runnable is submitted, it will not display output until the created object has a condition
Ready
with a statusTrue
. But while waiting for a healthy child object, the Runnable itself will still have a statusReady
with statusTrue
.Steps to reproduce:
Submit these objects:
Expected behavior:
k get -o yaml runnable cm
returns a Runnable with conditionReady
== FalseActual behavior:
k get -o yaml runnable cm
returns a Runnable with conditionReady
== TrueAdditional context:
See branch
runnable-status-bug
for a test that exercises this.TODO