Skip to content

Latest commit

 

History

History
132 lines (86 loc) · 6.44 KB

README.md

File metadata and controls

132 lines (86 loc) · 6.44 KB
Raw

Overview

The scripts here will deploy an OpenShift 4 cluster with a Bastion host. Provisioning will likely take on order of ~60 minutes to complete.

Usage

Create Cluster

  1. Ensure that you have a ../secret.yml in the parent directory
  2. Ensure that you have cp my_vars.sample.yml my_vars.yml and you have edited 'my_vars.yml'
  3. Ensure that AGNOSTICD_HOME environment variable is set
  4. Run: create_ocp4_workshop.sh
  5. Wait ... ~60 minutes

Destroy Cluster

  1. Ensure that you have a ../secret.yml in the parent directory
  2. Ensure that you have cp my_vars.sample.yml my_vars.yml and you have edited 'my_vars.yml'
  3. Ensure that AGNOSTICD_HOME environment variable is set
  4. Run: delete_ocp4_workshop.sh
  5. Wait ... ~5-10 minutes
    • If something goes wrong, manual cleanup is difficult. The rough steps are later in this document.

Tips

Example: Running oc or kubectl commands with provided 'KUBECONFIG', need to export KUBECONFIG

You will find a kubeconfig file has been copied to your host in the output_dir specified in my_vars.yml

  • Example, my output_dir is /tmp/agnostic_jwm0702ocp4a and in there is a file ocp4-workshop_jwm0702ocp4a_kubeconfig

      $ export KUBECONFIG=/tmp/agnostic_jwm0702ocp4a/ocp4-workshop_jwm0702ocp4a_kubeconfig
  $ oc whoami
  system:admin

Obtaining kubeadmin password

Look in the directory configured by output_dir specified in my_vars.yml, in that directory will be a file *_kubeadmin-password it will contain the password for 'kubeadmin'

    $ cat /tmp/agnostic_jwm0702ocp4a/ocp4-workshop_jwm0702ocp4a_kubeadmin-password 
    d34yS-vY429C-7fj2-CmEa9

Log into the web console

The console address is computed as: https://console-openshift-console.apps.cluster-${GUID}.${guid}.${subdomain_base_suffix}/

  • For example in my_vars.yml I have

      guid:                   jwm0702ocp4a
  subdomain_base_suffix:  .mg.dog8code.com

  • So my console is at: https://console-openshift-console.apps.cluster-jwm0702ocp4a.jwm0702ocp4a.mg.dog8code.com/

  • I can also determine this by looking at the routes in the openshift-console namespace on the cluster

      $ export KUBECONFIG=/tmp/agnostic_jwm0702ocp4a/ocp4-workshop_jwm0702ocp4a_kubeconfig
  $ oc get routes -n openshift-console
  NAME        HOST/PORT       PATH   SERVICES PORT    TERMINATION          WILDCARD
  console     console-openshift-console.apps.cluster-jwm0702ocp4a.jwm0702ocp4a.mg.dog8code.com            console     https   reencrypt/Redirect   None
  downloads   downloads-openshift-console.apps.cluster-jwm0702ocp4a.jwm0702ocp4a.mg.dog8code.com          downloads   http    edge                 None

  • I log in with credentials of:

    • Username: 'kubeadmin'
    • Password: From ${output_dir}/ocp4-workshop_${guid}_kubeadmin-password

SSH

OCP 4 environments are not typically ones you will SSH into, yet for this case the install is happening on the 'clientvm', so you may find it useful to ssh to the clientvm

  • To find the 'clientvm' look at your output_dir from my_vars.yml, then find the file ending in *_ssh_conf in that directory.

    • Example, mine is /tmp/agnostic_jwm0702ocp4a/ocp4-workshop_jwm0702ocp4a_ssh_conf

        $ cat /tmp/agnostic_jwm0702ocp4a/ocp4-workshop_jwm0702ocp4a_ssh_conf
  ##### BEGIN ADDED BASTION PROXY HOST clientvm.jwm0702ocp4a.internal ocp4-workshop-jwm0702ocp4a ######
  Host clientvm.jwm0702ocp4a.internal clientvm
      Hostname ec2-34-209-31-72.us-west-2.compute.amazonaws.com
      IdentityFile ~/.ssh/libra.pem
      IdentitiesOnly yes
      User ec2-user
      ControlMaster auto
      ControlPath /tmp/jwm0702ocp4a-%r-%h-%p
      ControlPersist 5m
      StrictHostKeyChecking no
      ConnectTimeout 60
      ConnectionAttempts 10
      UserKnownHostsFile /tmp/agnostic_jwm0702ocp4a/ocp4-workshop_jwm0702ocp4a_ssh_known_hosts
  ##### END ADDED BASTION PROXY HOST clientvm.jwm0702ocp4a.internal ocp4-workshop-jwm0702ocp4a ######

    • SSH into the clientvm

        $ ssh -i ~/.ssh/libra.pem ec2-user@ec2-34-209-31-72.us-west-2.compute.amazonaws.com
  [ec2-user@clientvm 0 ~]$ cd cluster-jwm0702ocp4a/
  [ec2-user@clientvm 0 ~/cluster-jwm0702ocp4a]$ ls
  auth/  metadata.json  terraform.aws.auto.tfvars  terraform.tfstate  terraform.tfvars  tls/
  $ openshift-install destroy cluster

Troubleshooting

Manual cleanup if something goes wrong from delete_ocp4_workshop.sh

  1. Attempt to run "openshift-install destroy cluster" from the clientvm that deployed the cluster

    • Follow steps above in 'Tips' to ssh into your clientvm

        $ ssh -i ~/.ssh/libra.pem ec2-user@ec2-34-209-31-72.us-west-2.compute.amazonaws.com
  
  [ec2-user@clientvm 0 ~]$ cd cluster-jwm0702ocp4a/

  [ec2-user@clientvm 0 ~/cluster-jwm0702ocp4a]$ ls
  auth/  metadata.json  terraform.aws.auto.tfvars  terraform.tfstate  terraform.tfvars  tls/

  $ openshift-install destroy cluster

  2. After you have cleaned up the terraform provisioned resources and only after, then you can tear down the clientvm by deleting the associated CloudFormation template

    • Remember...the terraform state of the ocp 4 cluster is stored locally on the clientvm .... don't delete the clientvm until you have attempt to run openshift-install destroy cluster

Error converting YAML to JSON. Could not find expected key ‘Install Config’

This is most likely because of a bad custom configuration file generated for cluster installation. In my case, it was because I did not put ocp4_pull_secret (pull-secret) within single inverted commas.

Playbook hangs on openshift-installer create phase

Make sure that the cluster creation is complete. To ensure that it is complete, login to clientvm host. There would be a directory created at /home/ec2-user/cluster-<GUID> for your cluster installation. Tail .openshift_installer.log file in that directory. Check whether the cluster creation step succeded. If not, wait till it completes. You can also ensure whether the openshift-installer process exited. If all of this looks okay to you, then your playbook just hung on shell completion or a bad network connection. You can follow steps in this issue.

htpasswd replaces the default kubeadmin user

Use opentlc-mgr as user and r3dh4t1! as password