What’s new in Rancher v1.6.11?

Let’s explore the new software and features in the latest release of
Rancher, which you can use by running
the rancher/server:v1.6.11{.markup–code .markup–p-code} image. Check
our documentation on Installing Rancher
Server if you need help running Rancher.

NOTE: Keep in mind that Rancher v1.6.11 is tagged as latest,
which means it’s not ready for production use. Our current, stable
version recommended for use in production is Rancher v1.6.10.

Kubernetes

Kubernetes 1.8.3

We’ve added Kubernetes 1.8.3 to this release. For all the information on
new features and improvements in Kubernetes 1.8, see the release blog
post. This update also allows you to run
Docker 1.12.6, 1.13.1, and 17.03.2, as Kubernetes 1.8
supports these versions.

Azure Cloud Provider Support

In our Kubernetes 1.8.3 catalog item, it is possible to configure Azure
as a cloud
provider for Kubernetes. This means you can:

• Use Azure Load Balancer by using type: LoadBalancer{.markup–code
  .markup–p-code}
• Use Azure Blob/Managed Disks by
  specifying kind: StorageClass{.markup–code
  .markup–p-code} and kind: PersistentVolumeClaim{.markup–code
  .markup–p-code}
• Mount Azure File share by using azureFile{.markup–code
  .markup–p-code} inside a volumes{.markup–code
  .markup–p-code} definition.

Ability to Change Subnet for Kubernetes

In previous configurations, the Kubernetes subnet was hardcoded
to 10.43.0.0/16{.markup–code .markup–p-code}. Now parameterized, you
can configure all the parameters to your liking.

Rancher/Cattle

Docker

The latest stable version, Docker 17.09, is supported when using
Cattle orchestration.

Draining Connections on Load-Balanced Services

Are you getting HTTP Bad Gateway (502){.markup–code
.markup–p-code} errors when upgrading a load-balanced service? Rancher
v1.6.11 comes with a load balancer that can drain connections before
stopping the container. Inside the load-balanced service definition
in rancher-compose.yml{.markup–code .markup–p-code}, it is now
possible to specify drain_timeout_ms{.markup–code .markup–p-code} as
an amount of milliseconds to wait before stopping the container. Check
out the Rancher
Docs for more information.

Clustered MySQL Support

We’ve had a large number of requests to be able to use Rancher with a
clustered MySQL setup which requires a primary key on all tables, like
Percona XtraDB
with pxc_strict_mode = ENFORCING or MASTER{.markup–code
.markup–p-code} or some cloud solutions. By
adding --db-strict-enforcing{.markup–code .markup–p-code} to
the docker run{.markup–code .markup–p-code} command, you can now use
those clustered databases as database for Rancher.

Prune Services No Longer Specified in Rancher CLI

The Rancher CLI v0.6.5 now compares the specified services in the
current stack to those in the new stack. If a service is no longer
specified in the new stack, it will be removed when you
use rancher up --prune{.markup–code .markup–p-code}.

Infrastructure Services Updates & Fixes

Here’s a quick mention of some great improvements to infrastructure
services:

• GitHub issue
  9957 — We’ve applied CPU restrictions to the
  metadata container, which should help in the events of scaling up.
  Before, the load on the systems could get really high which could
  impact the services running on it.
• GitHub issue
  9971 — This is a huge improvement on the IPsec side.
  Tunnels that were interrupted would not recover after a certain
  period of time, but in this release we included retry logic to
  re-establish those connections.

Final Note

Hopefully, you all enjoy the new, improved stuff on which the team has
been working hard. This post includes a subset; read the full release
notes to see what else is new and improved. You can also
subscribe to the Rancher Forums announcements so you get a notification
every time there is a new release. For Kubernetes users, be aware of
this bug that
will restart all containers when you upgrade to version 1.8. If you run
into any issues, you can ask a question on the Rancher
Forums,
jump on our Rancher Users
Slack channel, or file an issue on
GitHub.