Clear salt-keys in baremetal cicd deployments
When a BM lab is re-deployed, cfg01 is bootstrapped first.
But all BM and VCP nodes from previous deployment are still active,
and register their salt-minions once the new salt-master is up.
To prevent deployment errors, before re-deploying MAAS nodes:
- clear all the salt keys and re-register the cfg01 node
Minor fix to VCP nodes:
- add ens2 'role: single_dhcp' to get the node accessible
from salt-master after reboot
Change-Id: I4f511c32ad40a3c27235abc7cb8330bcc7cb6a50
diff --git a/tcp_tests/templates/cookied-cicd-bm-os-contrail40-maas/salt.yaml b/tcp_tests/templates/cookied-cicd-bm-os-contrail40-maas/salt.yaml
index 4ea0035..a7b3677 100644
--- a/tcp_tests/templates/cookied-cicd-bm-os-contrail40-maas/salt.yaml
+++ b/tcp_tests/templates/cookied-cicd-bm-os-contrail40-maas/salt.yaml
@@ -48,6 +48,17 @@
retry: {count: 1, delay: 5}
skip_fail: false
+- description: Remove all the salt-minions and re-register the cfg01 minion
+ cmd: |
+ salt-key -y -D;
+ salt-call test.ping
+ sleep 5
+ # Check that the cfg01 is registered
+ salt-key | grep cfg01
+ node_name: {{ HOSTNAME_CFG01 }}
+ retry: {count: 1, delay: 5}
+ skip_fail: false
+
- description: provision the automatically commissioned physical nodes through MAAS
cmd: |
salt-call maas.deploy_machines;