Description of problem: In /roles/openshift_node/defaults/main.yml there is a setting - { var: SystemKeepFree, val: 20% } that is incorrect, journald doesnt accept percentages as input for config files as specified in https://bugzilla.redhat.com/show_bug.cgi?id=874631 ' This cause log disk filling up because of wrong default configuration settings in openshift-ansible. How reproducible: Always Actual results: Log disk filling up because of wrong default configuration settings in openshift-ansible. Expected results: Good journald settings to be used Additional info: While We wait until https://bugzilla.redhat.com/show_bug.cgi?id=874631 is resolved and therefore the percentage value would be correct, we should ammend the main.yml file with a fixed value like: - { var: SystemKeepFree, val: 1G }
Reading through the referenced bug it seems like the default if not specified is 15% but when specifying a value it must be in absolute units. So I think we should just drop this value as it's not possible to specify a percentage via configuration and it's not possible to come up with a reasonable absolute value otherwise.
Created upstream Pull Request: https://github.com/openshift/openshift-ansible/pull/11187
Thanks Javier!
Fix is available in openshift-ansible-3.11.84-1
Verify this bug with openshift-ansible-3.11.87-1.git.0.a7b07ff.el7.noarch.rpm SystemKeepFree was removed from roles/openshift_node/defaults/main.yml After a fresh installation, no such setting in journald.conf [root@qe-gpei-2master-registry-router-etcd-1 ~]# grep SystemKeepFree /etc/systemd/journald.conf [root@qe-gpei-2master-registry-router-etcd-1 ~]#