Generation 1 has been deprecated and is not recommended for new applications.

Generation 1
Rolling Updates

When a Release is promoted, new processes are gracefully rolled out into production.

If there are errors starting new processes, new processes are not verified as healthy, or the rollout doesn’t complete in 10 minutes, an automatic rollback is performed.

The number of new processes started at a time is configurable so large apps can be fully rolled out in 10 minutes.

Rollouts and rollbacks do not cause any service downtime.

Rollout

A rollout coordinates starting new processes in a way that maintains service uptime and capacity. The basic flow is:

  • Start 1 new process
  • Verify new process is healthy
  • Stop 1 old process
  • Repeat for all processes in the formation

Automatic Rollback

If there are errors starting new processes, new processes are not verified as healthy, or the rollout doesn’t complete in 10 minutes, an automatic rollback is performed. This will result in a rollback or failed state for the app:

$ convox apps info
Name       httpd
Status     rollback

$ convox apps info
Name       httpd
Status     failed

Health Checks

For a rolling update to succeed, certain criteria must be met. For more information, see Health Checks.

Configuring Deployment Parameters

By default, a process must boot and pass the network health check in 3 seconds. (See the Health Check doc for more information about configuring the timeout as well as setting a custom health check port and HTTP path.)

By default, 100% of the processes on the old release stay running and the same number of processes on the new release will attempt to start at once. As new processes are deemed healthy, old ones will be stopped.

If we want to spare no expense for the fastest deployment speed, we can scale a Rack to have enough capacity to run two releases of the app simultaneously. For example if your app needs 5 instances to support 5 web processes that expose ports, running 10 instances will guarantee fast deploys:

$ convox rack scale --count 10

If we can tolerate a temporary decrease in service capacity, we can configure an app to not require 100% of the old processes to stay running. For example we can say that we’re ok stopping 50% of the old processes immediately to free up capacity for new processes:

version: '2'
  services:
    web:
      build: .
      image: convox/rails
      labels:
        - convox.deployment.minimum=50

See the deployment labels doc for more information about setting a deployment minimum.

Most apps will not need to configure any of these settings to gracefully roll out.