Closed omus closed 1 year ago
The "pending-timeout" test is failing for me locally with this error being reported on the manager:
nested task error: KubeError: Warning: spec.containers[0].resources.limits[memory]: fractional byte value "1Ei" is invalid, must be an integer
Warning: spec.containers[0].resources.requests[memory]: fractional byte value "1Ei" is invalid, must be an integer
This is odd as it used to work with the tests were originally written. I feel like this is a K8s server version difference which makes me think we should be testing against multiple versions of Kubernetes clusters
Using manusa/actions-setup-minikube@v2.4.0
now reports this error:
Error: Unsupported OS, action only works in Ubuntu 18 or 20
Pulled in changes from #95
Ugh, GitHub actions are now having problems
Merging #96 (1043eb2) into main (426c2bd) will increase coverage by
17.75%
. The diff coverage isn/a
.
@@ Coverage Diff @@
## main #96 +/- ##
===========================================
+ Coverage 49.10% 66.86% +17.75%
===========================================
Files 4 4
Lines 167 172 +5
===========================================
+ Hits 82 115 +33
+ Misses 85 57 -28
see 3 files with indirect coverage changes
:mega: We’re building smart automated test selection to slash your CI/CD build times. Learn more
When running the cluster integration tests I was seeing issues with errors such as resource "NotFound". This turned out to be that the job TTL was cleaning up resources before the tests validated the results. I've since switched to deleting the resources manually to avoid such a scenario. This also as the added benefit of removing resources immediately when tests pass.
Additionally, having users specify
eval $(minikube docker-env)
before running the tests is easy to miss so I added some code to do that for the user.