Hello, Thanks for looking into this. I think the problem you found on the haproxy config for the Ceph Dashboard is related to [1], where is given to the operator the flexibility of moving the ceph dashboard haproxy frontend vip on a network != ctlplane when no CephStorageDashboard composable network is defined [2]. On Mon, Sep 6, 2021 at 5:42 PM wodel youchi <wodel.youchi@gmail.com> wrote:

Hi,

I deployed Openstack Train using TripleO, I added ceph dashboard template to my deployment. But it is not reachable via the provisioning network, it is reachable via the storage network. Is this the default behavior? Did I miss something?

I modified the haproxy configuration to point the dashboard to the ctplane vip address and it worked.

How do I make this work from the deployment?

You can do that by overriding the CephDashboardNetwork parameter, which means you can add to your existing deployed *CephDashboardNetwork:ctlplane * By doing this, during the next stack update the haproxy config that points to the ctlplane network should be maintained.

Could an update of the overcloud, reset the configuration of the haproxy to

the default?

Exactly, a stack update will result in applying the default values defined in your tht. For this reason, you should pass CephDashboardNetwork:ctlplane to your overcloud deploy.

Thanks in advance. Regards.

I started a patch series [3] to make sure the Ceph Dashboard network uses the right networks, maintaining the ability to change it overriding the parameter described above. Thank you!! [1] https://review.opendev.org/c/openstack/tripleo-ansible/+/801757 [2] https://docs.openstack.org/project-deploy-guide/tripleo-docs/latest/features... [3] https://review.opendev.org/q/topic:%22ceph_dashboard_network%22+(status:open...) -- Francesco Pantano GPG KEY: F41BD75C