-
Notifications
You must be signed in to change notification settings - Fork 132
GKE: Enable integration tests #214
GKE: Enable integration tests #214
Conversation
sameersbn
commented
Nov 5, 2018
•
edited
Loading
edited
- Configure CI setup for testing on GKE clusters
- Configure DNS suffix for GKE tests
- Clean up disk leakage
- Debug monitoring test failures in GKE
57658b1
to
44b7461
Compare
3344ff1
to
489c6ee
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
(lgtm so far, with the query regarding dnsSuffix in monitoring_test. Hooking the dns zone up shouldn't be hard at this point - I'm doing that part now for demos)
Jenkinsfile
Outdated
@@ -224,15 +242,15 @@ spec: | |||
// $AZURE_SUBSCRIPTION_ID, $AZURE_TENANT_ID. | |||
withCredentials([azureServicePrincipal('jenkins-bkpr-owner-sp')]) { | |||
def resourceGroup = 'jenkins-bkpr-rg' | |||
def clusterName = ("${env.BRANCH_NAME}".take(8) + "-${env.BUILD_NUMBER}-${platform}-" + UUID.randomUUID().toString().take(5)).replaceAll(/[^a-zA-Z0-9-]/, '-').toLowerCase() | |||
def clusterName = ("${env.BRANCH_NAME}".take(8) + "-${env.BUILD_NUMBER}-" + UUID.randomUUID().toString().take(5) + "-${platform}").replaceAll(/[^a-zA-Z0-9-]/, '-').toLowerCase() |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We need more uniqueness than platform+BUILD_NUMBER? This expression is getting a bit silly - I think if we're going to include a random component, then lets just drop all the non-random bits and rely on tags/labels to report meta-info.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The truncation removes the uniqueness. This change essentially is to make sure the uniqueness does not get removed as a result of truncation. As a result I can reliably query leftover gce disks using a filter and remove them.
99af496
to
7de4193
Compare
@anguslees please take a look. |
a430a75
to
2b64d72
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Nice work!
2b64d72
to
3fbcb8f
Compare
Signed-off-by: Sameer Naik <[email protected]>
Signed-off-by: Sameer Naik <[email protected]>
when the cluster is delete synchronously the dns zone gets cleaned up and the subsequent call to delete the zone executes successfully.
95e2237
to
cf5c023
Compare
* GKE: Enable integration tests Signed-off-by: Sameer Naik <[email protected]> * fix GKE integration test setup Signed-off-by: Sameer Naik <[email protected]> * let gke select the latest available k8s patch release * define `waitForRollout` utility function * drop `+k8s` string from platform name * insert the random string before the platform substring * wait for deployments in `kube-system` namespace to rollout * fix command to delete dns zone * remove `--async` flag from GKE cluster deletion when the cluster is delete synchronously the dns zone gets cleaned up and the subsequent call to delete the zone executes successfully. * fix gke disk leakage * break multiline command block * purge zone records before deleting the zone * cleanup `gcloud container clusters create` command * remove double dashes from clusterName * add utility functions to update glue records * setup glue records for gke tests * remove `--dns-suffix` requirement for monitoring tests * specify machine-type in gke cluster create command * let kubeprod create the dns zone * nameServers as a ArrayList to insertGlueRecords (cherry picked from commit bda226f) Signed-off-by: Sameer Naik <[email protected]>