Document URL: https://access.redhat.com/documentation/en-us/openshift_container_platform/3.7/pdf/installation_and_configuration/OpenShift_Container_Platform-3.7-Installation_and_Configuration-en-US.pdf Section Number and Name: Advanced Installation section: Configuring Schedulability on Masters https://docs.openshift.com/container-platform/3.7/install_config/install/advanced_install.html#marking-masters-as-unschedulable-nodes Describe the issue: I have a 2 node openshift 3.7 cluster which became unresponsive because I was running etcd on the same node as my (ML) machine learning application. My ML application starved etcd of memory and cpu caused etcd to become "dead" as returned from "systemctl status etcd" Suggestions for improvement: Instead of this warning which is currently in the installation guide: "You can manually set a master host to schedulable during installation using the openshift_schedulable=true host variable, though this is not recommended in production environments:" State that it is "IMPORTANT* not to run etcd and other master services on the same node as user pods. Additional information:
Tracking this in https://github.com/openshift/openshift-docs/pull/9293
Thank you Diane, Can you please look at the changes in PR https://github.com/openshift/openshift-docs/pull/9293. These changes are for OCP 3.3 till OCP 3.7 as we already removed that note from OCP 3.9 docs.
No response form reporter, closing this now.