Refactored the code to switch the distribution source of the roles
to new collections for the roles that have already been migrated into
collections.
Updated:
- requirements
- role invocations
- documentation
- release notes
- minor whitespace fixes
Change-Id: I8dbb88aa2061b3afcab07caeb30f45b3a5c939af
There are a few places where we have a play with one task that uses the
group_by module to generate groups that are used in a subsequent play.
These plays do not need to gather facts for their target hosts.
In particular, the kolla-openstack.yml playbook should really be able to
operate without access to remote hosts.
This change also adds changed_when: false to all group_by tasks, since
they show as changed but haven't really made any changes outside of
Ansible.
TrivialFix
Change-Id: I18f67eda4e48058f3f402b9d0e692d6eeb02855f
This extends the physical network configuration in Kayobe to configure
Cumulus physical switches using the nclu Ansible module.
Change-Id: I960027ead301c5793a0ada1959a23549a71bdbfb
Adds support for a '--disable-discovery' argument to the 'kayobe
physical network configure' command. This can be used to configure the
physical network after discovery of bare metal compute nodes is
complete, to return the network to a normal state. The interface
configuration to be applied is configured via
'switch_interface_config_disable_discovery'.
Essentially, this is the mirror of the '--enable-discovery' argument.
Change-Id: I40426a8fd22e51e8ea3350013a614dfbcf2afccf
Story: 2004048
Task: 27054
In Ansible, playbook group variables take precedence over inventory
group variables. Typically in a kayobe-config we might specify the
switch_type variable for a group containing switches using inventory
group_vars. This seemed to work in Ansible 2.3, but since Ansible 2.4
the default value of None in kayobe's playbook group_vars takes
precedence. This behaviour is as documented, not sure why it worked
previously.
This change removes the default value of None set in the playbook
group_vars.
Change-Id: I042ad8d115cdc0dd57a116e1be9d397ef4d216ae
Story: 2002903
Task: 22870
Mellanox switches running MLNX-OS cannot be managed using the Ansible onyx*
modules, as the OS is too old. This change uses the stackhpc.mellanox-switch
Ansible role to configure these devices using the Ansible expect module.
Change-Id: I86f654c95be05e3a82d6015e4c89093127553408
Story: 2002687
Task: 22512
PowerConnect switches do not support the dellos Ansible modules, since
they are based on DellOS version 4. This role uses the expect module to
perform simple switch configuration.
Change-Id: Icb59aaff75f9faf089423610fecb06598f75e35e
Story: 2002106
Task: 19784
Allow the physical network interface configuration to be limited to a subset of
interfaces, either by interface name or switch interface description. This is done
via:
kayobe physical network configure --interface-limit interface1,interface2
or
kayobe physical network configure --interface-description-limit host1,host2
Fixes: #25
As of ansible 2.3, it is no longer possible to use delegate_to with a host other
than localhost. Use of a jump host can be enabled via setting ProxyCommand in
ansible_ssh_common_args as for other hosts.
After discovery Neutron needs to own the control of ports for the baremetal compute
nodes. We separate the interface configuration and add an --enable-discovery flag
to kayobe physical network configure
Jakub Darmach