* Added support for limits and requests of CPU and memory. Examples:
1. kn service create mysvc --image dev.local/ns/image:latest --request-cpu 100m --requests-memory 64Mi
2. kn service create mysvc --image dev.local/ns/image:latest --limits-cpu 1000m --limits-memory 1024Mi
You can also include and/or omit the requests and limits. So the following should also work:
kn service create mysvc --image dev.local/ns/image:latest --request-cpu 1000m --limits-memory 1024Mi
* Addressed comments about maintaining existing requests and limits and added more tests
* TestServiceCreateWithRequests
* TestServiceCreateWithLimits
* TestServiceCreateRequestsLimitsCPU
* TestServiceCreateRequestsLimitsMemory
* TestServiceCreateRequestsLimitsCPUMemory
* updating dependencies and formatting tests code
* Support --all-namespaces option for service and revision list
`kn {service, revision} list` does not have the all-namespaces
options. Although it is possible to list them by `kn service list -n
""`(empty string), it is difficult to find such secret trick.
Hence, This patch adds `--all-namespaces` option.
* Add all-namespace option to KnParams
* Add AddNamespaceFlags function
* Add GetNamespace func
* Update help message
* Use GetNamespace util in service create cmmand
* Stub in an integration test
This just calls `kn service list` and verifies it doesn't have a
non-zero exit status.
* Build the client in pre_integration_tests
* Need to explicitly build the `kn` client
As opposed to just building the entire repo. That didn't actually
stick a `kn` binary in the root.
* Address code review feedback
Build the client in e2e-tests.sh instead of presubmit-tests.sh, fix a
couple copy/paste places that referred to eventing, remove unnecessary
knative_teardown function.
* Don't attempt to build kn twice
* Turns out pre_integration_tests() is the easiest way to build kn
We don't have any other good hooks at the moment to do something in
`e2e-tests.sh` that runs only once before the test cluster gets setup.
* Update test-infra to use the new cluster_setup function
This updates the version of test-infra to use the new cluster_setup
function added in https://github.com/knative/test-infra/pull/667
It simplifies our e2e setup logic into one place instead of being
split over presubmit-tests.sh and e2e-tests.sh
* Update vendored test-infra as well
* We have to explicitly tell CI to use the vendored modules
* Be a bit more descriptive in the stubbed out integration test
When building with go 1.11.5 I see a hash mismatch for knative/serving.
The algorithm for computing the hash has changed in recent go releases.
To create this PR I ran:
- `rm go.sum`
- `go test ./pkg/...`
- `go build ./cmd/kn`
- `go mod tidy`
All developers should update their version of go and clean their module
cache to flush downloaded modules with an invalid hash.
* Refactor commands to allow insertion of fakes for testing
* Pin client-go to version that matches signatures other libs use
* Add tests for service list
* actually add test file
* Review comments: Fatal instead of Error/return, Split instead of read to newline
* Get revision and service listing skeletons in.
This adds the kn program, with basic commands for service and revision listing
in.
kn service list
kn revision list
They use the genericclioptions library from kubernetes to support jsonpath,
yaml, and json output. The default is to just list the names of the objects, for
now, until we add in some of the kubectl-based libraries for creating tables in
another commit.
This is deliberately bare-bones; I am looking to get the basic skeletons of how
this repository is laid out down.
No tests yet. Soon.
* Move commands to pkg, so cmd only contains minimalist main.go
* RunE instead of Run. The defaults on Cobra command generation are weird.
* Oops forgot to change types of stuff
* Do not panic on bad config file
* Make commands no longer global
* Make configuration initialization not static in the module, but rather triggered by main.go init
dr.max