-
Bug
-
Resolution: Unresolved
-
Normal
-
None
-
4.14
-
No
-
1
-
OSDOCS Sprint 243, OSDOCS Sprint 244, OSDOCS Sprint 245, OSDOCS Sprint 246, OSDOCS Sprint 247, OSDOCS Sprint 248, OSDOCS Sprint 249, OSDOCS Sprint 250, OSDOCS Sprint 251
-
9
-
False
-
-
N/A
-
Release Note Not Required
We have an issue after migrating to the OVN-Kubernetes network with OpenShift on OpenStack - OCPBUGS-16910.
The minimal requirements of the worker nodes in openstack are 2 vCPUs and 8 GB RAM. [0] [1]
Increased the flavor size to match the master nodes requirements: 4 vCPUs and 16 GB RAM, and the issue didn't reproduce - the worker nodes are not overloaded during the conformance tests anymore. [2]
Discussed in a Slack thread, and decided we might state in the docs that it is recommended to use the flavor that matches the master nodes when migrating to the OVN-Kubernetes network. [3]
We have a similar statement in https://github.com/shiftstack/installer/blob/master/docs/user/openstack/README.md#worker-nodes:
However, if you are experiencing Out Of Memory issues, or your installs are timing out, try increasing the size of your flavor to match the master nodes: 4 vCPUs and 16 GB RAM.
Note that using the workers minimal requirements, all the functionality works, but the workers can get overloaded when running many tests in parallel.
[0] https://github.com/shiftstack/installer/blob/master/docs/user/openstack/README.md#worker-nodes
[1] https://docs.openshift.com/container-platform/4.13/installing/installing_openstack/installing-openstack-installer-custom.html#installation-osp-compute-machines_installing-openstack-installer-custom
[2] https://issues.redhat.com/browse/OCPBUGS-16910?focusedId=22871729&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-22871729
[3] https://redhat-internal.slack.com/archives/CH98TDJUD/p1693205778786629