Description of problem:
When cluster is deleted, luci does not send request to ricci agents to delete
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Delete an existing cluster via the luci web app interface
2. The cluster is deleted from the DB used by luci
3. The /etc/cluster/cluster.con files are not edited/deleted by the ricci agents
on the cluster nodes.
The /etc/cluster/cluster.con files are not edited/deleted by the ricci agents on
the cluster nodes.
The /etc/cluster/cluster.con files on the cluster nodes should be deleted -
should the ccsd service also be restarted on the cluster nodes.
I'm seeing messages in the debug log on the luci server that look like this:
Oct 30 06:40:40 tng3-5 luci: [auth 1] reported cluster_info =
(Oct30_a,Oct30_a) for tng3-1.lab.msp.redhat.com
The cluster that was deleted was named "Oct30_a" - why is the cluster name
showing up twice in these log messages?
Is this related to: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=212584 ?
re: cluster name showing up twice: the first field is the actual cluster name,
and the second field is the cluster alias.
re: comment #1: i don't think they're related.
just to be clear, when you delete the cluster, are you using the "manage
systems" tab in homebase, or are you using the delete link in the cluster tab?
Good question - when I referred to "deleting the cluster" I meant to say that
this was from the "manage systems" tab in homebase.
ok, removing systems and clusters via the "manage systems" page merely removes
them from Luci management; it won't destroy clusters, so it ought not send any
ricci commands (except unauthenticate, when appropriate).
Thanks Ryan - but, how does a user delete a cluster?