Action Required Service Updates

Increased Availability with HA Masters in the Kubernetes Service: Actions You Must Take

Share this post:

Increased Availability with HA Masters in the IBM Cloud Kubernetes Service

Your apps and clusters continue to be accessible, even if you do not migrate to HA Masters. If you want increased availability, you must complete the steps described in this blog before these dates (listed by Kubernetes version):

  • Version 1.10: December 4, 2018
  • Version 1.11: November 27, 2018

What are HA masters?

IBM Cloud Kubernetes Service will soon increase your cluster availability by adding replicas of the Kubernetes master with fix pack version 1.11.3_1531, released on November 1, 2018. After you update, your clusters will have three Kubernetes master replicas set up with a highly available (HA) configuration.

IBM deploys each master on separate physical hosts. Furthermore, if your cluster is in a multizone capable region, the master replicas are spread across the zones. This HA configuration increases the overall service availability. For example, with HA masters, you won’t experience downtime during a cluster update and can continue making updates to your Kubernetes deployments and services.

Complete HA master documentation.

How do I get HA masters?

For 1.11 clusters: HA masters are available with the Kubernetes master fix pack version 1.11.3_1531. New version 1.11 clusters that you create have HA masters by default. You can update your existing cluster to this version or later to get HA masters. Initially, 1.11 masters will not be automatically updated to the latest 1.11 fix pack with HA master configuration. Unlike regular master updates, we are temporarily disabling automatic updates for 1.11 to give you time to assess the impact and complete the update steps, if necessary. Failure to make the necessary changes by these dates may break apps within your clusters.

For 1.10 clusters: You can update your existing cluster to 1.11 now that the 1.11 HA master fix pack is available. Additionally, a Kubernetes master fix pack version 1.10.8_1530 to update to HA master configuration has been released on November 7, 2018. As with 1.11 clusters, IBM will provide you a period of time in which automatic updates of masters are disabled so that you can follow the update steps.

For 1.9 clusters: You can update your existing cluster to 1.10 or 1.11 now that the HA master fix packs are available. Clusters running 1.9 will not get a Kubernetes master fix pack to update to HA master configuration based on the end of support now that 1.12 is available.

Future versions: Future cluster versions, such as 1.12, will have HA masters by default. In addition, all future Kubernetes master fix packs for 1.10 and 1.11 will only support HA masters.
Note: If you are planning to update your 1.9 or 1.10 clusters to 1.11 or 1.12, in addition to HA masters, you also get performance benefits by moving from Docker to containerd as the container runtime. For update steps related to containerd, see the docs.

What’s the timeline for these changes?

For version 1.11 clusters:

  • November 1, 2018: HA master update available, but not automatically applied to existing clusters.
  • November 1, 2018 – November 26, 2018: Your time to assess impact and update to Kubernetes master fix pack 1.11.3_1531 or later for HA masters.
  • November 27, 2018: HA master updates for 1.11 clusters are automatically applied. Please ensure that you have taken all of the necessary update steps before this date.

For version 1.10 clusters:

  • November 7, 2018: HA master update available, but not automatically applied to existing clusters.
  • November 7, 2018 – December 3, 2018: Your time to assess impact and update to Kubernetes master fix pack 1.10.8_1530 or later for HA masters.
  • December 4, 2018: HA master updates for 1.10 clusters are automatically applied. Please ensure that you have taken all of the necessary update steps before this date.

How do I know if my cluster might be impacted? What update actions do I take?

Many of the HA master configuration changes relate to improving the availability of your cluster networking. As such, you need to take update steps under the following circumstances:

Contact us

If you have any concerns or need help with the HA master update, open an IBM Cloud Support case. Be sure to include your region, cluster ID, and mention HA master update in the case.

For general questions, engage our team via Slack by registering here and join the discussion in the #general channel on our public IBM Cloud Kubernetes Service Slack.

 

Program Director, Offering Management, IBM Kubernetes Service & IBM Container Registry

More Action Required Service Updates stories
March 5, 2019

Deprecating IBM Workload Scheduler on IBM Cloud

We are announcing the deprecation of the IBM Workload Scheduler on IBM Cloud service on March 31, 2019. We are also offering the option to migrate to IBM Workload Automation on Cloud for all customers who are currently using IBM Workload Scheduler.

Continue reading

February 19, 2019

IBM Analytics Engine: Changes to Cluster Credential Access

To make your cluster more secure and to prevent malicious conduct, IBM Analytics Engine will now follow the security best practices and only return the cluster credentials via the reset password API when requested by the user.

Continue reading

January 8, 2019

Sydney? Version 9.1.1? IBM MQ MFT? IBM MQ on Cloud Welcomes You with Open Arms

IBM MQ is now available in Sydney, Australia. Additionally, you can enjoy MQ 9.1.1 in the managed service by upgrading your queue managers manually or waiting until the automatic upgrade on February 6, 2019.

Continue reading