Skip backend integration tests when cli flag isn't passed #527
Conversation
|
@IronPan @neuromage does it make sense to run backend unit tests in Travis rather than Prow? Why wait for 17+ minutes to find out the result of these tests when you can wait for 3-4 minutes? |
|
/lgtm |
|
I am in favor or running the unit tests in travis as well if it's faster.
(Maybe we need to run them in prow as well to get the test grid? Yang can
comment on that part).
…On Wed, Dec 12, 2018 at 3:27 PM Ajay Gopinathan ***@***.***> wrote:
/lgtm
—
You are receiving this because your review was requested.
Reply to this email directly, view it on GitHub
<#527 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AYkFTeLPUeLRU5GyTJrFmg2TSoI2iXbyks5u4ZDmgaJpZM4ZQmuG>
.
|
|
the unit tests are not taking long to finish and show up the result in github. but feel free to move them to travis if you want. |
|
/lgtm |
|
@IronPan this needs a /approve as well. |
yebrahim
changed the title
|
/approve |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: IronPan The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
1 similar comment
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: IronPan The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
|
The unit tests and build image are parallel. UT won't depend on the images being built |
|
Need one more lgtm here after resolving conflicts with master. |
|
/lgtm |
* Numerous fixes and improvements to bulk deploying Kubeflow with v06
Fix bug with getting credentials in bulk_deploy
* Need to support obtaining credentials to talk to K8s API servier using
the pods service accounnt by calling load incluster config.
* With v06 deployments we need to handle the case where the DM API might
not be enabled yet and enable it before checking if KF exists.
* Don't use workload identity because with workload identity we are seeing
performance problems when issuing a lot of requests that hit the gke metadata
server
Improve the ability to continue deploying v06 deployments that failed.
* With v06 applications check if the ingress actually exists and if
not we will rerun deploy.
* Delete istio-ingressgateway service if its not fully deployed to try to recover.
* Add a util method for getting Kubernetes credentials
* When running in a pod there are two cases we need to handle
* using the KSA to authenticate to the K8s APIServer the pod is running
* Using a kubeconfig file to talk to a different cluster.
* Add resource requests to the create_unique_kf_instance jobs to better
handle scheduling of many jobs
* Create code to delete Kubeflow deployments in bulk
* Don't use a CSV file to provide info about what deployments to create
* Uploading CSV files to GCS turns out to be a lot of friction
* instead just take command line arguments which specify the range of
projects to iterate over.
* Fix lint.
Backend unit tests now run by default if you run
go test ./.... If you want integration tests to run as well, set theunitTestsOnlyflag to false:$ go test ./... -args -unitTestsOnly=false./area back-end
/assign @neuromage
This change is