Re-running os-net-config nukes resolv.conf
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
tripleo |
Triaged
|
High
|
Unassigned |
Bug Description
Observerd in RDO CI jobs fs001 and fs039:
When overcloud nodes are deployed with '--network-config', i.e networking is configured as part of baremetal node provisioning. Then network config is then re-applied by the deploy-steps playbook. Result in empty resolv.conf[1]
When not running node provisioning with '--network-config' the resolv.conf[2] is ok.
Failing jobs have a backup config[3] in /etc/os-net-config, both configs have the correct dns_servers.
[1] ...
- https:/
- https:/
[2] ...
- https:/
- https:/
Changed in tripleo: | |
milestone: | xena-2 → xena-3 |
Looks like the actual problem is that different templates are used for the two os-net-config runs.