Red Hat Bugzilla – Bug 1622256
Master api does not start due to slow image pull
Last modified: 2018-10-11 03:26:27 EDT
Description of problem: Upgrading from openshift 3.10 to 3.11 forces registry.access.redhat.com users to migrate to registry.redhat.io. During upgrades, we update static pod definitions to create api service from new registry. We do not attempt to pre-pull or ensure image is present on the host before attempting to start the api service. Timeout of ~120 seconds (120 retries, 1 second delay) might not be enough considering the size of the image. How reproducible: Depends on network/registry speed, but should happen with some regularity. Steps to Reproduce: 1. Upgrade openshift 3.10 -> 3.11 using upgrade_control_plane.yml Actual results: Timeout waiting for api to start, due primarily to image now being present on host and taking ~4 minutes to download. Expected results: We should pre-pull the image across all masters before attempting to upgrade.
PR Created in master: https://github.com/openshift/openshift-ansible/pull/9779
Version:openshift-ansible-3.11.0-0.25.0.git.0.7497e69.el7.noarch Checked that pr9779 merged. Upgrade succeed with control plane images pulled in advance.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2018:2652