Note: Stack stable-v7-1.0.6, which was released on August 3rd, has been removed due to an issue with how the backend built the release. As such, the EY Platform will automatically skip to use stable-v7-1.0.7.
If you are using stable-v7-1.0.6, it is recommended to upgrade to stable-v7-1.0.7. Please feel free to reach out to Support for any questions, or if assistance is needed.
Release Version: stable-v7-1.0.7
Action: You apply the following changes the next time you click the Upgrade button for your Engine Yard Stack V7 environment.
- Created new (default) Puma stack (PR: #92)
- Created new Puma Legacy stack, which uses the legacy way of launching Puma (PR: #96)
- Fixed HTTP2 Nginx issue (PR: #95)
- Fixed issue of chef not running during reboot causing instances to fail (PR: #93)
- Added support for PostgreSQL Maintenance which adds a scheduled job (PR: #91)
Changes to Puma
We have now split Puma into Puma Legacy and Puma. Puma Legacy is for any version prior to Puma 5.1 and doesn't utilise Cluster mode. The new version of the standard Puma stack requires the `sd_notify` gem to be bundled prior to deployment, uses systemd to start the process, and leverages Cluster mode.
To keep the legacy Puma setup
If you're already using Puma and wish to keep the old method, of using Monit and the deploy script, please contact Support and we can alter the environment for you with no downtime.
To use the new Puma setup
This may cause a few minutes of downtime whilst the upgrades are applied and the Puma workers started up again.
- Update your Gemfile and Gemfile.lock after bundling `sd_notify`
- Deploy your application as normal
- SSH into each application instance and stop the current Puma workers using `sudo monit stop -g <appname>` Use `sudo monit summary` if unsure of the correct name
- Click Upgrade at the top of the Environment page in the Engine Yard UI and proceed with the upgrades
- The resulting Chef run should start Puma up again and the application become accessible again
Article is closed for comments.