You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository was archived by the owner on Oct 8, 2025. It is now read-only.
We had an issue where a missing Ansible host (due to a temporary API outage at AWS) caused ce-deploy to run but skip playbooks - which is just a warning and not a failure, so we get a zero return code from Ansible - thus the track file was incorrectly updated:
Executing a deploy will need to be dependent on the host check succeeding and hostname will need to be a variable like it is in ce-provision, we'll have to stop coding the hostname/group directly into the playbook for the deploy.
A bit of refactoring to do, but something like this should work nicely to stop a false-positive build caused by a host being skipped.
We had an issue where a missing Ansible host (due to a temporary API outage at AWS) caused ce-deploy to run but skip playbooks - which is just a warning and not a failure, so we get a zero return code from Ansible - thus the track file was incorrectly updated:
We need Ansible to fail when the host is missing to avoid this. I have tested this playbook, which will do the job:
We'll have to pass this
_ce_deploy_ansible_hostin to a play as an extra var and knit a call to this playbook intobuild.shsomehow:Executing a deploy will need to be dependent on the host check succeeding and hostname will need to be a variable like it is in ce-provision, we'll have to stop coding the hostname/group directly into the playbook for the deploy.
A bit of refactoring to do, but something like this should work nicely to stop a false-positive build caused by a host being skipped.