Working with multiple clusters

So for a while, I have had a very backward way of accessing multiple clusters: I would set the kubeconfig environment variable, or change the default file. If I had bothered to learn the first thing about contexts, I could have avoided the confusion of keeping track of multiple files.

When a cluster is created, we often get a basic config file to access the cluster. I had often looked at these as a black box of access. Here is an example below from my rancher cluster:

apiVersion: v1
clusters:
- cluster:
    certificate-authority-data: REDACTED
    server: https://rke1:6443
  name: default
contexts:
- context:
    cluster: default
    user: default
  name: default
current-context: default
kind: Config
preferences: {}
users:
- name: default
  user:
    client-certificate-data: REDACTED
    client-key-data: REDACTED

Thanks to the official documentation (RTFM folks) I think it has finally clicked. We have lists of 3 different object types in the above config:
– Cluster: the connection to the cluster (contains a CA and endpoint)
– User: Identified with the client cert data and key data
– Context: Ties the above together (also namespaces if we want)

Contexts allow me to have multiple configurations and switch between them using the kubectl config use-context command. My goal is to have a connection to both my openshift cluster, and my rancher cluster. So I combined (and renamed some elements) the configuration:

apiVersion: v1
clusters:
- cluster:
    insecure-skip-tls-verify: true
    server: https://api.oc1.lab.local:6443
  name: api-oc1-lab-local:6443
- cluster:
    certificate-authority-data: REDACTED
    server: https://rke1:6443
  name: rancher
contexts:
- context:
    cluster: api-oc1-lab-local:6443
    namespace: default
    user: kube:admin/api-oc1-lab-local:6443
  name: default/api-oc1-lab-local:6443/kube:admin
- context:
    cluster: rancher
    user: rancherdefault
  name: rancher
current-context: rancher
kind: Config
preferences: {}
users:
- name: kube:admin/api-oc1-lab-local:6443
  user:
    token: REDACTED
- name: rancherdefault
  user:
    client-certificate-data: REDACTED
    client-key-data: REDACTED

If we understand a little YAML, we can easily combine the files. Now it is simple to switch between my clusters:

kubectl config get-contexts
CURRENT   NAME                                        CLUSTER                  AUTHINFO                            NAMESPACE
          default/api-oc1-lab-local:6443/kube:admin   api-oc1-lab-local:6443   kube:admin/api-oc1-lab-local:6443   default
*         rancher                                     rancher                  rancherdefault
kubectl config use-context default/api-oc1-lab-local:6443/kube:admin
Switched to context "default/api-oc1-lab-local:6443/kube:admin".

Installing Portworx on Openshift

Today I decided to see about installing Portworx on Openshift with the goal of being able to move applications there from my production RKE2 cluster. I previously installed openshift using the Installer provisioned infrastructure (rebuilding this will be a post for another day). It is a basic cluster with 3 control nodes and 3 worker nodes.

Of course, I need to have a workstation with Openshift Client installed to interact with the cluster. I will admit that I am about as dumb as a post when it comes to openshift, but we all have to start somewhere! Log in to the openshift cluster and make sure kubectl works:

oc login --token=****** --server=https://api.oc1.lab.local:6443

kubectl get nodes

NAME                     STATUS   ROLES    AGE   VERSION
oc1-g7nvr-master-0       Ready    master   17d   v1.23.5+3afdacb
oc1-g7nvr-master-1       Ready    master   17d   v1.23.5+3afdacb
oc1-g7nvr-master-2       Ready    master   17d   v1.23.5+3afdacb
oc1-g7nvr-worker-27vkp   Ready    worker   17d   v1.23.5+3afdacb
oc1-g7nvr-worker-2rt6s   Ready    worker   17d   v1.23.5+3afdacb
oc1-g7nvr-worker-cwxdm   Ready    worker   17d   v1.23.5+3afdacb

Next, I went over to px central to create a spec. One important note! Unlike installing Portworx on other distros, openshift needs you to install the portworx operator using the Openshift Operator Hub. Being lazy, I used the console:

I was a little curious about the version (v2.11 is the current version of portworx as of this writing). What you are seeing here is the version of the operator that gets installed. This will allow the use of the StorageCluster object. Without installing the operator (and just blindly clicking links in the spec generator) will generate the following when we go to install Portworx:

error: resource mapping not found for name: "px-cluster-f51bdd65-f8d1-4782-965f-2f9504024d5c" namespace: "kube-system" from "px-operator-install.yaml": no matches for kind "StorageCluster" in version "core.libopenstorage.org/v1"

Again, I chose to let Portworx automatically provision vmdks for this installation (I was less than excited about cracking open the black box of the OpenShift worker nodes).

kubectl apply -f px-vsphere-secret.yaml
secret/px-vsphere-secret created

kubectl apply -f px-install.yaml
storagecluster.core.libopenstorage.org/px-cluster-f51bdd65-f8d1-4782-965f-2f9504024d5c created
kubectl -n kube-system get pods

NAME                                                    READY   STATUS    RESTARTS   AGE
autopilot-7958599dfc-kw7v6                              1/1     Running   0          8m19s
portworx-api-6mwpl                                      1/1     Running   0          8m19s
portworx-api-c2r2p                                      1/1     Running   0          8m19s
portworx-api-hm6hr                                      1/1     Running   0          8m19s
portworx-kvdb-4wh62                                     1/1     Running   0          2m27s
portworx-kvdb-922hq                                     1/1     Running   0          111s
portworx-kvdb-r9g2f                                     1/1     Running   0          2m20s
prometheus-px-prometheus-0                              2/2     Running   0          7m54s
px-cluster-f51bdd65-f8d1-4782-965f-2f9504024d5c-4h4rr   2/2     Running   0          8m18s
px-cluster-f51bdd65-f8d1-4782-965f-2f9504024d5c-5dxx6   2/2     Running   0          8m18s
px-cluster-f51bdd65-f8d1-4782-965f-2f9504024d5c-szh8m   2/2     Running   0          8m18s
px-csi-ext-5f85c7ddfd-j7hfc                             4/4     Running   0          8m18s
px-csi-ext-5f85c7ddfd-qj58x                             4/4     Running   0          8m18s
px-csi-ext-5f85c7ddfd-xs6wn                             4/4     Running   0          8m18s
px-prometheus-operator-67dfbfc467-lz52j                 1/1     Running   0          8m19s
stork-6d6dcfc98c-7nzh4                                  1/1     Running   0          8m20s
stork-6d6dcfc98c-lqv4c                                  1/1     Running   0          8m20s
stork-6d6dcfc98c-mcjck                                  1/1     Running   0          8m20s
stork-scheduler-55f5ccd6df-5ks6w                        1/1     Running   0          8m20s
stork-scheduler-55f5ccd6df-6kkqd                        1/1     Running   0          8m20s
stork-scheduler-55f5ccd6df-vls9l                        1/1     Running   0          8m20s

Success!

We can also get the pxctl status. In this case, I would like to run the command directly from the pod, so I will create an alias using the worst bit of bash hacking known to mankind (any help would be appreciated):

alias pxctl="kubectl exec $(kubectl get pods -n kube-system | awk '/px-cluster/ {print $1}' | head -n 1) -n kube-system -- /opt/pwx/bin/pxctl"
pxctl status
Status: PX is operational
Telemetry: Disabled or Unhealthy
Metering: Disabled or Unhealthy
License: Trial (expires in 31 days)
Node ID: f3c9991f-9cdb-43c7-9d39-36aa388c5695
        IP: 10.0.1.211
        Local Storage Pool: 1 pool
        POOL    IO_PRIORITY     RAID_LEVEL      USABLE  USED    STATUS  ZONE    REGION
        0       HIGH            raid0           42 GiB  2.4 GiB Online  default default
        Local Storage Devices: 1 device
        Device  Path            Media Type              Size            Last-Scan
        0:1     /dev/sdb        STORAGE_MEDIUM_MAGNETIC 42 GiB          27 Jul 22 20:25 UTC
        total                   -                       42 GiB
        Cache Devices:
         * No cache devices
        Kvdb Device:
        Device Path     Size
        /dev/sdc        32 GiB
         * Internal kvdb on this node is using this dedicated kvdb device to store its data.
Cluster Summary
        Cluster ID: px-cluster-f51bdd65-f8d1-4782-965f-2f9504024d5c
        Cluster UUID: 73368237-8d36-4c23-ab88-47a3002d13cf
        Scheduler: kubernetes
        Nodes: 3 node(s) with storage (3 online)
        IP              ID                                      SchedulerNodeName       Auth            StorageNode     Used    Capacity        Status  StorageStatus        Version         Kernel                          OS
        10.0.1.211      f3c9991f-9cdb-43c7-9d39-36aa388c5695    oc1-g7nvr-worker-2rt6s  Disabled        Yes             2.4 GiB 42 GiB          Online  Up (This node)       2.11.1-3a5f406  4.18.0-305.49.1.el8_4.x86_64    Red Hat Enterprise Linux CoreOS 410.84.202206212304-0 (Ootpa)
        10.0.1.210      cfb2be04-9291-4222-8df6-17b308497af8    oc1-g7nvr-worker-cwxdm  Disabled        Yes             2.4 GiB 42 GiB          Online  Up  2.11.1-3a5f406   4.18.0-305.49.1.el8_4.x86_64    Red Hat Enterprise Linux CoreOS 410.84.202206212304-0 (Ootpa)
        10.0.1.213      5a6d2c8b-a295-4fb2-a831-c90f525011e8    oc1-g7nvr-worker-27vkp  Disabled        Yes             2.4 GiB 42 GiB          Online  Up  2.11.1-3a5f406   4.18.0-305.49.1.el8_4.x86_64    Red Hat Enterprise Linux CoreOS 410.84.202206212304-0 (Ootpa)
Global Storage Pool
        Total Used      :  7.1 GiB
        Total Capacity  :  126 GiB

For the next bit of housekeeping, I want to get a kubectl config so I can add this cluster in to PX Backup. Because of the black magic when I used the oc command to log in, I’m going to export the kubecfg with:

apiVersion: v1
clusters:
- cluster:
    insecure-skip-tls-verify: true
    server: https://api.oc1.lab.local:6443
  name: api-oc1-lab-local:6443
contexts:
- context:
    cluster: api-oc1-lab-local:6443
    namespace: default
    user: kube:admin/api-oc1-lab-local:6443
  name: default/api-oc1-lab-local:6443/kube:admin
current-context: default/api-oc1-lab-local:6443/kube:admin
kind: Config
preferences: {}
users:
- name: kube:admin/api-oc1-lab-local:6443
  user:
    token: REDACTED

Notice that the token above is redacted, you will need to add your token from the oc when pasting it to PX Backup

And as promised, the spec I used to install:

# SOURCE: https://install.portworx.com/?operator=true&mc=false&kbver=&b=true&kd=type%3Dthin%2Csize%3D32&vsp=true&vc=vcenter.lab.local&vcp=443&ds=esx2-local3&s=%22type%3Dthin%2Csize%3D42%22&c=px-cluster-f51bdd65-f8d1-4782-965f-2f9504024d5c&osft=true&stork=true&csi=true&mon=true&tel=false&st=k8s&promop=true
kind: StorageCluster
apiVersion: core.libopenstorage.org/v1
metadata:
  name: px-cluster-f51bdd65-f8d1-4782-965f-2f9504024d5c
  namespace: kube-system
  annotations:
    portworx.io/install-source: "https://install.portworx.com/?operator=true&mc=false&kbver=&b=true&kd=type%3Dthin%2Csize%3D32&vsp=true&vc=vcenter.lab.local&vcp=443&ds=esx2-local3&s=%22type%3Dthin%2Csize%3D42%22&c=px-cluster-f51bdd65-f8d1-4782-965f-2f9504024d5c&osft=true&stork=true&csi=true&mon=true&tel=false&st=k8s&promop=true"
    portworx.io/is-openshift: "true"
spec:
  image: portworx/oci-monitor:2.11.1
  imagePullPolicy: Always
  kvdb:
    internal: true
  cloudStorage:
    deviceSpecs:
    - type=thin,size=42
    kvdbDeviceSpec: type=thin,size=32
  secretsProvider: k8s
  stork:
    enabled: true
    args:
      webhook-controller: "true"
  autopilot:
    enabled: true
  csi:
    enabled: true
  monitoring:
    prometheus:
      enabled: true
      exportMetrics: true
  env:
  - name: VSPHERE_INSECURE
    value: "true"
  - name: VSPHERE_USER
    valueFrom:
      secretKeyRef:
        name: px-vsphere-secret
        key: VSPHERE_USER
  - name: VSPHERE_PASSWORD
    valueFrom:
      secretKeyRef:
        name: px-vsphere-secret
        key: VSPHERE_PASSWORD
  - name: VSPHERE_VCENTER
    value: "vcenter.lab.local"
  - name: VSPHERE_VCENTER_PORT
    value: "443"
  - name: VSPHERE_DATASTORE_PREFIX
    value: "esx2-local4"
  - name: VSPHERE_INSTALL_MODE
    value: "shared"