Delete any of cnci_image.img.xz, cnci_image.img or .cnci_ready files
to let ansible rebuild the cnci image.
If the tree files mentioned above already exists. Skip the
image creation phase.
Signed-off-by: Alberto Murillo Silva <alberto.murillo.silva@intel.com>
When the managed nodes are fedora, they require the following
packages to be present:
- python-dnf
- netaddr
- libselinux-python
Fixes#50
Signed-off-by: Alberto Murillo <alberto.murillo.silva@intel.com>
Several bundles got deprecated in clearlinux.
This patch updates cleanup/ciao.yml to remove the bundles
that are actually installed by ciao.
Fixes#74
Signed-off-by: Alberto Murillo <alberto.murillo.silva@intel.com>
the `qemu-nbd` tool it's used to manipulate the cnci image, which
is already a raw image (not qcow as thought), so we can use
raw image manipulation, removing all the need of nbd manipulation
tools.
Signed-off-by: Simental Magana, Marcos <marcos.simental.magana@intel.com>
Using $GOPATH as the value for gopath is not
so easily achievable because the GOPATH value
for the deployment machine does not get copied
automatically to other nodes.
Also setting a fact on one node does not set it
on other nodes.
The easiest way to share this value across the
different ciao roles is to set it as a variable
that can be overriden in groups_vars/all file or
via the cmdline with --extra-vars "gopath=$GOPATH"
Fixes: #57
Signed-off-by: Alberto Murillo <alberto.murillo.silva@intel.com>
docker containers uses /etc/resolv.conf file from the
host machine filtering any localhost reference in it.
If after the filter there are no nameservers entries then
the daemon adds google public dns servers (8.8.8.8 and 8.8.4.4)
This Changes adds a note to remind the administrator to setup a
dns server that the containers can use to resolv the ciao cluster
nodes names.
Fixes#39
Signed-off-by: Alberto Murillo <alberto.murillo.silva@intel.com>
The playbooks were creating a ciaorc to authenticate the cli
with the ciao user.
The ciao user is reserved for ciao services authenticating into keystone
and should not be used in the cli to create instances.
This changes creates admin-openrc with credentials for the admin user
and demo-openrc with credentials for the demo user to be used by ciao-cli.
Fixes#46
Signed-off-by: Alberto Murillo <alberto.murillo.silva@intel.com>
When deploying from an ubuntu or fedora machine to a
ClearLinux node, ansible tries to use sftp which is
disabled in ClearLinux by default.
Enabling scp_if_ssh in ansible.cfg makes anisble use
ssh to copy files to the nodes.
Fixes#54
Signed-off-by: Jesus Ornelas Aguayo <jesus.ornelas.aguayo@intel.com>
- Specify ansible requirements for managed nodes and control machine
- Specify passwordless sudo is required in deployment machine and managed nodes
- Specify which commands have to be run as sudo
Fixes#37
Signed-off-by: Alberto Murillo <alberto.murillo.silva@intel.com>
- Deployment of ciao has changed from installing everyghin in packages
to install keystone and ciao-webui from containers.
- ciao-laucher will detect the operating system and install the required
dependencies
- Support for Ubuntu 16.04 and Fedora 24 has been added.
Signed-off-by: Alberto Murillo <alberto.murillo.silva@intel.com>
profile_tasks prints how much time takes each task to run
The feature is disabled by default but can be enabled by
uncommenting the corresponding line in ansible.cfg
Signed-off-by: Alberto Murillo <alberto.murillo.silva@intel.com>
Ceph FileSystem requires a metadata server.
This functionallity is already provided.
This commit updates the example to install a metadata server.
Signed-off-by: Alberto Murillo <alberto.murillo.silva@intel.com>
- Add an ansible.cfg file to enable pipelining.
Pipelining increases the performance by executing many ansible
modules without actual file transfer.
- Prerun ciao-common
ciao-common role install ciao code and dependencies. By running
this role on all nodes in parallel we can save some time.
- Set strategy: free
By default all hosts will run each task before any host starts the next task
Using the 'free' strategy allows each host to run until the end of the play
as fast as it can.:
Signed-off-by: Alberto Murillo <alberto.murillo.silva@intel.com>
A demo user and a demo project is included in the example
for testing purposes of a ciao cluster.
Signed-off-by: Alberto Murillo <alberto.murillo.silva@intel.com>
If ciao_dev is set to false (default) it will try to install
ciao from the operating system packages which is only supported
in Clear Linux.
In order to install ciao in Ubuntu, ciao_dev must be set to
true to build ciao from github master branch.
This commit adds the variable to the examples to have it handy
for users who wants to deploy ciao in Ubuntu.
Signed-off-by: Alberto Murillo <alberto.murillo.silva@intel.com>
- ciao_dev might not be defined and should be treated as false by default
- mysql_packages is defined for ubuntu and should not cause a failure
in clearlinux when not defined.
Fixes#30
Signed-off-by: Alberto Murillo <alberto.murillo.silva@intel.com>
* ciao: Install ciao from github sources
- Add a new variable ciao_dev to ciao-common
When set to True, it will build ciao from the master branch
and put the binaries in /usr/local/bin
When set to False (default) it will install ciao from the OS
packages
Fixes#21
Signed-off-by: Alberto Murillo <alberto.murillo.silva@intel.com>
* ciao: Add ciao_cleanup.yml playbook (#28)
- This playbook cleanups a ciao cluster
Signed-off-by: Alberto Murillo <alberto.murillo.silva@intel.com>
the [ciao_network] group in `examples/ciao/hosts` is
duplicated. This commits removes duplication of such group.
Signed-off-by: Simental Magana, Marcos <marcos.simental.magana@intel.com>
There were usage of different FQDNs for some variables.
this commit use a single FQDN for better understanding of the setup.
Signed-off-by: Simental Magana, Marcos <marcos.simental.magana@intel.com>
Users might not know what is the function of a variable in the
example playbook.
This change adds a reference for a role documentation where the
user can find more information about the variables below.
Fixes#18
Signed-off-by: Alberto Murillo <alberto.murillo.silva@intel.com>
Rally needs a compute endpoint in order to run.
- Create ciao admin user in openstack by default
- Create ciao compute service in openstack by default
- Create ciao compute endpoint in openstack by default
- Remove optional variables in example playbook to create
the items above.
Fixes#14
Signed-off-by: Alberto Murillo <alberto.murillo.silva@intel.com>
- Update ciao example playbook with newer variables
- Remove glance from the example since it is not yet
supported by ciao.
Signed-off-by: Alberto Murillo <alberto.murillo.silva@intel.com>
- Remove nova_public_interface_name and neutron_public_interface_name
since both are referring to the same thing, the interface where services
will run on that node.
- Add public_interface var with default value of the interface name
with a configured default gateway (ansible_default_ipv4.alias)
- Remove common_interface_detect.yml since it was testing for groups
names that might not be defined when deploying only keystone without
nova and neutron
Alberto Murillo Silva