I'm happy to announce another update to OpenShift Online. As always remember to update to the latest client tools; for most this is as simple as running: gem update rhc

More CPU, Less Throttling, and Greater Performance

We've made some significant changes to performance metrics in this release. In OpenShift's gear model, resources are constrained by cgroups. This allows OpenShift to make smarter decisions about resource management on a node, and to control things like memory/swap and CPU limits.

Prior to today, CPU was throttled in such a way that even if more resources on a node were available, the gear would be limited. This is because there were instances in scaled applications where two gears of the same application would end up on the same node. We decided to make every gear behave similarly so that scaling would be more predictable. This worked well, but had the side effect of causing issues in some very common use cases.

Starting today, users have access to much more CPU for short time periods. This should make things like startup, building, and sudden load spikes a much better experience for people. This CPU bursting will be a welcome change to many. Please let us know how your experiences have changed. If you have questions about scaling please feel free to ask us - openshift@redhat.com

Enter Your Coupon Code

Our marketing and evangelists teams has been hard at work meeting with all sorts of users. If you happen to be one of the lucky people receiving a coupon code, it's now easier than ever to use it. Just look for the coupon field when upgrading to the Silver Plan

REST API Enhancements

We've also been making lots of changes to our backend setup and REST API. Some of these changes include better consistency between the terms "delete" and "destroy". "delete" is now used and "destroy" is no more. Additionally consumers of our REST API can access applications directly by their application ID not just their domain/app name.

Improved JBoss Cartridge Startup

We've updated our JBoss cartridges to better deal with startup and deployment times. Prior to this change JBoss started up and the startup script happily exited even if user applications had not fully deployed. Now our scripts are properly blocking during the deployment phase to provide a better start up experience for JBoss.

More Good News for Cartridge Writers

To our growing community of cartridge writers, we've got some cartridge updates going out in this release as well. First, as a cartridge writer you are now able to subscribe to all environment variables published by other cartridges in your application. To enable this for your custom cartridge, add this snippet to your metadata/manifest.yml:

Subscribes:

  set-env:

    Type: "ENV:*"

    Required: false

In addition to the new environment variables we've added a new upgrade script that is run when a cartridge is upgraded to a new version. It is called after the new cartridge is in place but before setup is run.

upgrade {current software version number} {current cartridge version number} {new cartridge version number}

Quickstart Maintenance Moves Upstream--Make Sure Your Quickstart Isn't Affected

The OpenShift community has taken over maintenance of some of our quickstarts. To avoid duplication we'll be deleting some of our outdated quickstart repositories on GitHub before the next release. Please read the blog post about quickstart housekeeping to find out if your application will be affected and what you need to do to stay up-to-date.

What's Next?

Categories

OpenShift Online

< Back to the blog