Recovering from an Ops Manager and PAS Upgrade Failure
Page last updated:
This topic explains how to use BOSH Backup and Restore (BBR) to restore Ops Manager and Pivotal Application Service (PAS) back to a healthy state after a failed upgrade between PAS minor versions.
Note: BBR can only restore across the same PAS minor versions. To verify which version of PAS was used to create the backup artifact, see metadata/cf-VERSION.yml
in the backup artifact.
Evaluate Your Deployment
You can restore Ops Manager and PAS to a healthy state after a failed upgrade between PAS minor versions if your backups, BOSH Director, and PAS deployment meet certain conditions. Follow the procedures in this section to confirm that your deployment can be restored by BBR.
Review Your Backup
Before you perform a restore, you must:
Run the following
bbr
commands from your Ops Manager VM:Back up your deployments by running:
bbr deployment --deployment DEPLOYMENT-NAME backup
Where
DEPLOYMENT-NAME
is the name of your PAS deployment.Back up your BOSH Director by running:
bbr director backup
The above
bbr
commands produce the following backup artifacts:DEPLOYMENT_NAME_TIMESTAMP/
DIRECTOR_IP_TIMESTAMP/
installation.zip
Export your Ops Manager settings. For more information, see Settings Page in Using the Ops Manager Interface.
Validate Your Environment for Restore
Before you can restore after a failed upgrade between PAS minor versions, you mustvconfirm that your BOSH Director and PAS deployment are healthy.
vConfirm That Your BOSH Director Is Healthy
To check that the BOSH Director is in a healthy state for a bbr restore
, you should SSH to the Ops Manager VM and ensure that all jobs are running.
To ensure that all jobs are running:
SSH into your BOSH Director. For more information, see Log In to the BOSH Director VM with SSH in Advanced Troubleshooting with the BOSH CLI.
View all deployment job instances by running:
monit summary
Monit should report that all jobs are
running
.Ensure that the BOSH API is responsive by running the following command:
bosh configs
If this command runs successfully, then the BOSH Director API is in a healthy and responsive state and can be restored by BBR.
If your BOSH Director is unhealthy, see Restoring an Unhealthy BOSH Director.
Confirm That Your PAS Deployment Is Healthy
To confirm that PAS is in a healthy state:
Run:
bosh -d PAS-DEPLOYMENT-NAME instances
Where
PAS-DEPLOYMENT-NAME
is the name of your PAS BOSH deployment.
If your PAS deployment is healthy, then the output states that all of your instances are in therunning
state. If all instances are running, PAS can be restored by BBR.
If your PAS deployment is unhealthy, see Restoring an Unhealthy PAS Deployment.
Restore Your Deployment
After you confirm that your BOSH Director is healthy with its data intact, and all instances in your PAS deployment are responsive, you can restore PAS with BBR.
Restoring PAS
If your BOSH Director is healthy, you can restore across the same PAS minor version by following the procedures in Step 14: Restore PAS in Restoring Pivotal Platform from Backup with BBR.
Recover to a Healthy State
If either the BOSH Director or PAS are not in a healthy state, they must be recovered before you can run bbr restore
. If you need guidance on the following subsections, contact Pivotal Support.
Restoring an Unhealthy BOSH Director
If the BOSH Director is not in a healthy state, VMware recommends following the full restore workflow to regain your Ops Manager, BOSH Director, and BOSH deployments. For more information, see Restore Workflow in Restoring Pivotal Platform from Backup with BBR.
Restoring an Unhealthy PAS Deployment
If your BOSH Director is healthy, you can restore your previous BOSH Director state by following the procedure in Step 8: Restore the BOSH Director in Restoring Pivotal Platform from Backup with BBR.
After your BOSH Director is restored, you can then restore your deployments by following the procedure in Step 14: Restore PAS in Restoring Pivotal Platform from Backup with BBR.