Openshift 4 Upgrades
Announcing maintenance/version upgrades
There is currently no formal process for announcing maintenance/upgrades for the Centos CI Openshift 4 infrastructure. Until this gets fleshed out fully, proceed with the following process. Openshift 4 upgrades aim to be seamless and non service affecting, end users should hopefully suffer nothing more than temporary slowness in cluster responsiveness.
- Roll updates out to staging first. Roll updates out to production a day or two later, if everything is still working as expected.
- Announce via email list:
ci-users@centos.org, and via IRC:centos-ci@irc.libera.chatthat a maintenance period or upgrade will be taking place and when. - Ensure that the time allotted for maintenance/upgrade is adequate to complete with some leeway for encountering issues. eg 2 hours
- Once complete, or if the maintenance/upgrade is cancelled, inform via email list
ci-users@centos.org, and via IRC on#centos-ci@irc.libera.chatonce more.
Steps
- Silence alerting to prevent oncall engineers getting pinged when services/nodes restart during the upgrade.
- Login with Admin access account and go in
Administrationsection, thenCluster Settings. - Ensure that the channel is set to
stable-major.minorand that theUpdate nowbutton is visible. - If you are updating to a new minor version, you must first update the channel to
fast-major.minorversion which matches the minor version you wish to update to. Follow the latest information on [2] for updating between minor/major versions, as there might be information such as backing up important resources prior to an update. - Now click update now, and choose the latest
major.minor.patchversion. - Keep an eye on the status of the upgrade to ensure it does not stall/fail half way through.
- Perform suite of tests (functional/integration) to ensure the cluster is still working as expected.
- Unsilence monitoring/alerting once the cluster has completed with the upgrade.
Resources