Conversation
As of the update, ansible does not default to the platform-python. This causes errors with the SELinux commands as the module they require is in python3.6, not 3.8, which ansible defaults to. Using platform Linux as it is is python3.6 solves this issue.
This task fails on the new version of ansible on the epel repo as the use of url now refers to the grafana_url, not the data source as it did previously. Making the url explicit with ds_url solves this problem.
|
I've been annoyed that we're stuck on Ansible 2.9 for so long so the fact that 5 is now available in EPEL is very welcome news. Given how much has changed in Ansible in the meantime, I'm surprised that this is all the changes that are needed. To give some more confidence, would you be able to also edit the CI config file ( to perhaps I expect there are things that will not be compatible. Furthermore, the mgmt node bootstrap from Terraform (https://github.com/clusterinthecloud/terraform/blob/master/common-files/bootstrap.sh.tpl) runs Ansible before having installed EPEL so it will currently still be getting Ansible 2.9 (whereas the compute nodes to install EPEL before Ansible so will get the new version - https://github.com/clusterinthecloud/ansible/blob/6/roles/packer/templates/prepare_ansible.sh.j2). We might also need to install |
Epel has now released version 5.4 of Ansible within stable with this there are a few things that broke these are the fixes.