cf push Availability During Pivotal Application Service Upgrades

Page last updated:

This topic describes what you can expect regarding the availability of cf push during upgrades of Pivotal Application Service (PAS).

Overview

Availability of cf push during PAS upgrades varies from release to release. There are various considerations, such as Cloud Controller database (CCDB) migrations or the number of VMs in use, that can impact cf push availability. However, cf push is mostly available for the duration of an upgrade.

Impact on Single VMs vs. High-Availability Infrastructure

Having a single VM in use, such as a WebDAV, a HAProxy, a Gorouter, a UAA, or a Cloud Controller, impacts whether cf push is unavailable during an upgrade.

If you have scaled out your application to achieve high availability and are not using WebDAV, cf push should be available for the entire duration of the upgrade. However, upgrades to certain versions of PAS sometimes require a CCDB schema or data migration, which may cause cf push to be unavailable while Cloud Controllers are rolling during the upgrade.

Availability During a BBR Backup

The Cloud Controller API is taken down during the pre-backup-lock stage of a BBR backup and put back up again during the post-backup-unlock stage. As a result, cf push becomes unavailable during that time. However, the backup that takes place between those stages is very short, no longer than a few minutes. The bulk of the BBR backup operation happens after the post-backup-unlock stage, so the Cloud Controller API and cf push are available for most of the duration of a BBR backup.

The Uptimer tool can help you measure cf push availability during an upgrade.

Create a pull request or raise an issue on the source for this page in GitHub