Use Rolling Updates

Introduction

Goal

Redeploy a backward-compatible distribution without downtime using the rolling update method.

Background

When redeploying a backward-compatible distribution in a Bloomreach Cloud environment, there are two available methods:

  • Stop-start
  • Rolling update

By default, redeployment is done using the stop-start method which will first stop all existing containers before starting new containers with the new distribution. This is the most reliable option because it avoids potential backward-compatibility problems such as two different versions communicating with the same database.

However, the downside of the start-stop method is a downtime period. When redeploying a backward-compatible distribution, such a downtime period can be avoided by using the rolling update method instead of start-stop. A rolling update will start a new container, waits until it is ready to serve traffic, then stop an old container. This process is repeated until all old containers have been removed.

Please note that because new containers are started when using the rolling update method, session state is lost. This means all active CMS users are logged out and need to log in again.
In general, it is recommened (and required for non-backward compatible distributions) to use the Blue-Green Deployment strategy. This way, downtime is avoided by deploying to an offline environment first, and switching live traffic to it only once it has passed verification tests.

Instructions

  1. Make sure that the new, to-be-deployed, distribution is backward-compatible with the old, currently deployed, distribution.
  2. In Mission Control, select the environment your want to redeploy a distribution in.
  3. Select the Details tab.
  4. Under Distribution file, select Rolling update.
  5. Click on Redeploy a file.

Did you find this page helpful?
How could this documentation serve you better?
On this page
    Did you find this page helpful?
    How could this documentation serve you better?