GKE now includes managed support for Stackdriver. Installing or updating Stackdriver support happens automatically when you select a GKE version for your cluster and a support option. There are three support options:
- Legacy Stackdriver support (GA).
- Stackdriver Kubernetes Monitoring support (Beta).
- No support for Stackdriver.
All three options are provided by all GKE versions presently available for new clusters and for updates to existing clusters. For installation instructions, see Installing Stackdriver support.
Which Stackdriver option should I use?
You can use the default Legacy Stackdriver, or select the Beta release of Stackdriver Kubernetes Monitoring. Both options are provided by all GKE versions presently approved for new GKE clusters and for updates to existing clusters. For installation instructions, see Installing Stackdriver support. Here are some issues to conisider:
Legacy Stackdriver remains as the default and preferred option through GKE versions 1.10, 1.11, and 1.12. It is the most stable but it isn't getting any improvements and lacks features present in Stackdriver Kubernetes Monitoring.
Only Legacy Stackdriver lets you disable Stackdriver Logging while still using Stackdriver Monitoring.
There is value in getting started with Stackdriver Kubernetes Monitoring as soon as possible, since it will become the preferred version of Stackdriver.
If you are using Legacy Stackdriver, then you will have to manually change your Stackdriver monitoring and logging configurations in order to update your existing clusters to Stackdriver Kubernetes Monitoring. See Migrating to Stackdriver Kubernetes Monitoring.
About Stackdriver Kubernetes Monitoring
Stackdriver Kubernetes Monitoring, a replacement for Legacy Stackdriver, is designed to monitor GKE clusters. It manages Stackdriver services together, including both Monitoring and Logging. It also features a Stackdriver Kubernetes Monitoring Console that provides a dashboard customized to GKE clusters:
- In the summary pane, you can view a cluster's key metrics, such as CPU utilization, memory utilization, and the number of open incidents.
- You can view clusters by their infrastructure, workloads, or services.
- You can inspect namespaces, nodes, workloads, services, pods, and containers.
- For pods and containers, you can view metrics as a function of time and view log entries using the Logs Viewer.
When you create or upgrade a cluster, you can choose Stackdriver Kubernetes Monitoring in place of Legacy Stackdriver. For more information, see Installing Stackdriver Monitoring.
Selecting your Stackdriver support for GKE
If you are using Legacy Stackdriver, then you must switch to Stackdriver Kubernetes Monitoring before support for Legacy Stackdriver is removed.
Both versions send chargeable metric data and log data to Stackdriver; see the Pricing page for details.
Legacy Stackdriver how-to guides
Stackdriver Monitoring: Describes how to set up Stackdriver Monitoring for your GKE clusters.
Stackdriver Logging: Describes how to set up Stackdriver Logging for your GKE clusters.
Stackdriver Kubernetes Monitoring how-to guides
Installing Stackdriver Kubernetes Monitoring: Describes how to enable Stackdriver Kubernetes Monitoring in new and existing clusters.
Migrating to Stackdriver Kubernetes Monitoring: Describes the differences between Legacy Stackdriver and Stackdriver Kubernetes Monitoring and what you must change in your Stackdriver configurations as part of updating your Stackdriver support.
Observing your system: Describes the Stackdriver Kubernetes Monitoring dashboard interface, how to organize cluster information, and how to view alerting details.
Release guide: Lists the progression of product capabilities and known issues during the Stackdriver Kubernetes Monitoring beta period.