Edge Cluster Operations¶

Edge Cluster Limitations¶

Currently it is not possible to connect to VMs running in edge clusters through the normal Sunstone mechanisms to access VM console, such as VNC. Other mechanisms that connect to the VM’s IP, such as SSH and RDP, work as in other VMs running in local clusters.

Creating a Cluster¶

Check the steps here. You can also use the command oneprovision create:

oneprovision create /tmp/provision.yml
ID: 0

Managing the Edge Cluster¶

Listing¶

The list command lists all provisions.

Parameters:

Parameter	Description	Mandatory
`--csv`	Show output as CSV	NO
`--json`	Show output as JSON	NO
`--done`	Show provisions in DONE state	NO

oneprovision list
ID NAME                      CLUSTERS HOSTS VNETS DATASTORES STAT
 0 aws-cluster                      1     1     1          2 RUNNING

Showing¶

The show command lists all provisioned objects of the particular provision.

Parameters:

Parameter	Description	Mandatory
`provision ID`	Valid provision ID	YES
`--csv`	Show output as CSV	NO
`--json`	Show output as JSON	NO
`--xml`	Show output as XML	NO

Examples:

oneprovision show 0
PROVISION 0 INFORMATION
ID        : 0
NAME      : aws-cluster
STATE     : RUNNING
PROVIDER  : aws

Provision Infrastructure Resources

CLUSTERS
100: aws-cluster

DATASTORES
100: aws-cluster-image
101: aws-cluster-system

HOSTS
0: 54.166.142.123
1: 34.234.201.245

NETWORKS
0: aws-cluster-public

Provision Resource Resources

VNTEMPLATES
0: aws-cluster-private

Note

The Terraform state is stored inside the provision information, so the user doesn’t need to manage it directly.

Configuring¶

Warning

It’s important to understand that the (re)configuration can happen only on physical Hosts that aren’t actively used (e.g., no virtual machines running on the Host) and with the operating system/services configuration untouched since the last (re)configuration. It’s not possible to (re)configure the Host with a manually modified OS/services configuration. Also, it’s not possible to fix a seriously broken Host. Such a situation needs to be handled manually by an experienced systems administrator.

The configure command offlines the OpenNebula Hosts (making them unavailable to users) and triggers the deployment configuration phase. If the provision was already successfully configured before, the argument --force needs to be used. After successful configuration, the OpenNebula Hosts are re-enabled.

Parameters:

Parameter	Description	Mandatory
`provision ID`	Valid provision ID	YES
`--force`	Force reconfiguration	NO

Examples:

oneprovision configure 0 -d
ERROR: Hosts are already configured
oneprovision configure 0 -d --force
2018-11-27 12:43:31 INFO  : Checking working SSH connection
2018-11-27 12:43:34 INFO  : Configuring hosts

Adding more hosts¶

Note

You can only add more hosts to a provision in RUNNING state.

The host add will deploy and configure new hosts in the provision.

Parameters:

Parameter	Description	Mandatory	Default
`provision ID`	Valid provision ID	YES	NA
`--amount amount`	Number of hosts to add	NO	1
`--hostnames h1,h2`	Hostnames to add, used in onpremise provision	NO	NA
`--host-params`	For cluster with Ceph, use to pass the Ceph group to host, one of the: `ceph_group=osd,mon` `ceph_group=osd` `ceph_group=clients`	NO	NA

Examples:

oneprovision host add 0 -d
2018-11-27 12:43:31 INFO  : Deploying
2018-11-27 12:43:31 INFO  : Monitoring hosts
2018-11-27 12:43:31 INFO  : Checking working SSH connection
2018-11-27 12:43:34 INFO  : Configuring hosts

oneprovision host add 0 -d --hostnames '10.0.0.110,10.0.0.111'
2018-11-27 12:43:31 INFO  : Deploying
2018-11-27 12:43:31 INFO  : Monitoring hosts
2018-11-27 12:43:31 INFO  : Checking working SSH connection
2018-11-27 12:43:34 INFO  : Configuring hosts

Deleting¶

The delete command releases the physical resources to the remote provider and deletes the provisioned OpenNebula objects.

oneprovision delete 0 -d
2018-11-27 12:45:21 INFO  : Deleting provision 0
2018-11-27 12:45:21 INFO  : Undeploying hosts
2018-11-27 12:45:23 INFO  : Deleting provision objects

Only provisions with no running VMs or images in the datastores can be easily deleted. You can force oneprovision to terminate VMs running on provisioned Hosts and delete all images in the datastores by using the --cleanup parameter.

Parameters:

Parameter	Description	Mandatory
`provision ID`	Valid provision ID	YES
`--delete-all`	Delete all contained objects (VMs, images)	NO

Examples:

oneprovision delete 0 -d
2018-11-27 13:44:40 INFO  : Deleting provision 0
ERROR: Provision with running VMs can't be deleted

   $ oneprovision delete 0 -d --cleanup
   2018-11-27 13:56:39 INFO  : Deleting provision 0
   2018-11-27 13:56:44 INFO  : Undeploying hosts
   2018-11-27 13:56:51 INFO  : Deleting provision objects

- states

Customization of the Edge Cluster¶

Ansible¶

Ansible is used to configure the Hosts. All the playbooks and roles are located in /usr/share/one/oneprovision/ansible. OpenNebula comes with a set of roles ready to configure the provision, but in case you want to add new roles or modify the existing ones, please check this guide.

Provision Elements¶

You can create multiple elements with a single provision; check this guide for more information.

Networking¶

Warning

This sections does not apply to onprem provider. If you’re using onprem provider please check the Virtual Network Management section.

Adding/Removing Public IPs¶

Note

You can only add more IPs to a provision in RUNNING state.

Note

You can only add more IPs to elastic networks

The ip add will request the new IP in the remote provider.

Parameters:

Parameter	Description	Mandatory	Default
`provision ID`	Valid provision ID	YES	NA
`--amount amount`	Number of IPS to add	NO	1

Examples:

oneprovision ip add 0

Check this to know more about public networking.

Adding Virtual Network¶

Check this to know how you can add more private networks to an existing Edge Cluster.

CLI Commands¶

Validate¶

The validate command checks the provided provision template is correct. It returns exit code 0 if the template is valid.

Host Management¶

Individual Hosts from the provision can be managed by the oneprovision host subcommands.

Cluster Management¶

Individual clusters from the provision can be managed by the oneprovision cluster subcommands.

Datastore Management¶

Individual datastores from the provision can be managed by the oneprovision datastore subcommands.

Virtual Networks Management¶

Individual virtual networks from the provision can be managed by the oneprovision network subcommands.

Images¶

Individual images from the provision can be managed by the oneprovision image subcommands.

Templates¶

Individual VM templates from the provision can be managed by the oneprovision template subcommands.

VNet Templates¶

Individual VNet templates from the provision can be managed by the oneprovision vntemplate subcommands.

Flow Templates¶

Individual Flow templates from the provision can be managed by the oneprovision flowtemplate subcommands.

Logging Modes¶

The oneprovision tool in the default mode returns only minimal requested output (e.g., provision IDs after create), or errors. Operations on the remote providers or the Host configuration are complicated and time-consuming tasks. For better insight and for debugging purposes there are two logging modes available, providing more information on the standard error output.

verbose (--verbose/-d). Only the main steps are logged.
debug (--debug/-D). All internal actions, including generated configurations with sensitive data, are logged.

Running Modes¶

The oneprovision tool is ready to deal with common problems during execution. It’s able to retry some actions or clean up an incomplete provision. Depending on where and how the tool is used, it offers two running modes:

interactive (default). If an unexpected condition appears, the user is asked how to continue.
batch (--batch). It’s expected to be run from scripts. No questions are asked and the tool tries to deal automatically with the problem according to the failover method specified as a command line parameter: