container-service-extension
container-service-extension copied to clipboard
Enterprise CSE clusters drop "vdc" designation after being resized
After some additional testing stemming from issue #377 , I've found another issue when attempting to resize an Enterprise CSE cluster post provisioning.
Environment: vCD 9.7 PKS 1.4.0 NSX-T 2.4.1
Logged in as the org admin for my cse-enterprise-org
, which is enabled for Enterprise CSE cluster creation and list the available OvDCs and clusters:
$ vcd login director.example.com cse-enterprise-org cse-ent-admin -iw
$ vcd cse ovdc list
name org k8s_provider
------------ ------------------ --------------
cse-ent-ovdc cse-enterprise-org ent-pks
$ vcd cse cluster list
(nothing returned)
No clusters have been provisioned in the OvDC, so I'll provision one and list it after the create is finished:
$ vcd cse cluster create scale-test
property value
--------------------------- ----------------------
kubernetes_master_host scale-test.example.com
kubernetes_master_ips In Progress
kubernetes_master_port 8443
kubernetes_worker_instances 2
last_action CREATE
last_action_description Creating cluster
last_action_state in progress
name scale-test
worker_haproxy_ip_addresses
$ vcd cse cluster list
k8s_provider name org_name status vdc
-------------- ---------- ------------------ ---------------- ------------
ent-pks scale-test cse-enterprise-org create succeeded cse-ent-ovdc
Cluster is available and tagged with the cse-enterprise-org
org and the cse-ent-ovdc
virutal datacenter as expected. However, because of issue #377 , I need to resize my cluster to add a 3rd node as the cluster did not provision according to the PKS plan definition:
$ vcd cse cluster resize --nodes=3 scale-test
property value
----------- -----------
name scale-test
task_status in progress
When I go to view the status, no clusters are shown:
$ vcd cse cluster list
(nothing shown)
I'll log into the vcd as the administrator
user and list clusters:
$ vcd login director.example.com system administrator -iw
$ vcd cse cluster list
k8s_provider name org_name status vdc
-------------- ---------------- ------------ ------------------ ---------------
native ubuntu-cluster-1 cse-native-1 POWERED_ON cse-native-ovdc
ent-pks scale-test update in progress
The scale-test
cluster is shown but with no org
or vdc
designation. This means my users that need to access information about this cluster in the cse-enterprise-org
are no longer able to see the cluster after it has been resized.
Please let me know if you require any additional logs/information.
This is a known bug in PKS. Resize operations wipe out compute_profile metadata off the cluster. As a result CSE never gets the name of the compute profile on subsequent GET calls. Without this information CSE won't be able to determine on which org vdc the cluster is residing and hence no info on console either.
@rocknes This was fixed in PKS 1.4.1 release. But the change for that is the CSE needs to use new PKS api end point to resize the cluster. I don't know if changes were made by CSE for the same