Add latency recommendations between the different kinds of servers, specifically for masters. Reported by: sgarciam https://access.redhat.com/documentation/en-us/openshift_container_platform/4.4/html/installing_on_bare_metal/installing-on-bare-metal#annotations:af79bea4-3e86-4a65-8f06-757e950e1ee2
Hey, can we get an update in the bugzilla about the status of the request?
There are some generic recommendations already in the etcd documentation [1], but something more specific in the OpenShift documentation is needed. [1] https://github.com/etcd-io/etcd/blob/master/Documentation/op-guide/hardware.md
Hi, can we get an update in the bugzilla about the status of the request?
(In reply to oarribas from comment #2) > There are some generic recommendations already in the etcd documentation > [1], but something more specific in the OpenShift documentation is needed. > > > [1] > https://github.com/etcd-io/etcd/blob/master/Documentation/op-guide/hardware. > md Unfortunately there's nothing about latency recommendations in the link, and "Low latency ensures etcd members can communicate fast" seems pretty obvious for what we're expecting here. I hope the asignee can provide some information about the request soon.
In documentation for 4.6 we have an whole chapter talking about remote nodes. We don't have some explicit talk about minimal latency between nodes, however it's pretty clear that we will support this kind of cluster feature as well the pre-requisites for that. https://docs.openshift.com/container-platform/4.6/nodes/edge/nodes-edge-remote-workers.html In a k8s cluster, the communication in control plane stuff is very sensible to latency, however, the nodes themselves can have much more independency and tolerant to some latency and network downtime.
Any feedback for this bugzilla?
I'm currently tracking down some engineering SMEs to help guide this request. I will provide an update as soon as I can. Thanks!
Work in progress. https://github.com/openshift/openshift-docs/pull/26585 covers the disk latency requirements for etcd, but we are working to build out the information around tolerable etcd network peer latency.
Further work in: https://github.com/openshift/openshift-docs/pull/28280
QE review was in the PR
Content is now published: https://access.redhat.com/documentation/en-us/openshift_container_platform/4.6/html/scalability_and_performance/recommended-host-practices#recommended-etcd-practices_