Rename cheaha.yml playbook to cluster.yml

See merge request rc/hpc-factory!116

Rename it because cheaha is not the only cluster we deploy against. So
give it a generic name.

Add a task to template DNS config

See merge request rc/hpc-factory!115

This will allow you to define the variables during runtime whose values
you may want to override. In this case, we want to override the DNS
config related to nameserver and search domain

Mount dirs besides adding them to fstab

See merge request rc/hpc-factory!114

Otherwise, since we run this role in deploy the dirs won't be mounted
and cause unexpected behavior of applications.

Create multi homed proxy

See merge request rc/hpc-factory!111

removed dev_key to move var to gitlab pipeline variable

See merge request rc/hpc-factory!113

Avoids error that the IP has already been assigned while trying multiple
proxy deploys

The instance creation with internal port attached was causing issues
with login because the default gateway was set to internal network
gateway. This will only allow login via headnode but not the host.
This commit will fix that issue by giving a way to define the gw ip
for default route.

We wanted to create a multi-homed instance connected to proxy-net and
internal-net because the nfs-mounts are shared via internal net by the server.
However, the NIC on internal-net wasn’t getting an IP assigned automatically
and this was creating an issue for NFS server to communicate with the clients.
So we created a port and assigned that port during instance creation

Change path to avoid ansible log deletion

See merge request rc/hpc-factory!109

feat: Run clusterhooks at deploy time for proxies

See merge request rc/hpc-factory!108

We are providing hooks into the cluster we need at deploy time rather
than baking them at build time.
This will enable us to build once and deploy across multiple clusters.
Otherwise, we need to build image whenever you need to deploy to a
different cluster.

Feat ci proxy builds

See merge request rc/packer-openstack-hpc-image!103

fix yum install failures by fixing the baseurl via shell provisioners
temporarily until we upgrade the base OS

Modify roles in cheaha.yml playbook to enable a standalone proxy

See merge request rc/packer-openstack-hpc-image!90

Add packer template for proxy build

See merge request rc/packer-openstack-hpc-image!95

Enable TLS only if the ldap_uri var is having https substring in the
value

A proxy node need not be a slurm client so we don't run this role

We changed it to "all" because that's the convention we follow.
When running ansible playbook you need to limit the hosts with -l
so that it doesn't run on all hosts.

Some background on why it was set to default in the first place,
`hosts: default` is used as a placeholder during development
to indicate where actual host groups will later be defined.
You can define specific host groups within the inventory
file with names that are more relevant to your infrastructure,
replacing the "default" placeholder.

We parameterized a few tasks in the cheaha.yml and put the vars in
group_vars/all so that the existing knightly builds won't break.
We added values required for the CoD deploy in group_vars/proxy. Now if
we put the proxy nodes in proxy group they will inherit values meant for
CoD deploy which is not desired in production.
We later realized that the appropriate file is group_vars/knightly and
it gives flexibility to put proxy nodes in knightly group to override
the vars meant for CoD in group_vars/all.

Change the default timeout from 10 secs to 60 secs to accomodate the delay for slurm pkg yum install tasks after running ldap_config  role
Refer #125