Skip to content

Latest commit

 

History

History
119 lines (92 loc) · 6.75 KB

README.md

File metadata and controls

119 lines (92 loc) · 6.75 KB
Raw

3 steps to Contiv Cluster Management

If you are trying cluster manager with baremetal hosts or a personal VM setup follow this link to setup the hosts. After that you can manage the cluster as described in the step 3. below.

To try with a built-in vagrant based VM environment continue reading.

0. Ensure correct dependencies are installed

  • vagrant 1.7.3 or higher
  • virtualbox 5.0 or higher
  • ansible 2.0 or higher

1. checkout and build the code

cd $GOPATH/src/github.com/contiv/
git clone https://github.com/contiv/cluster.git

2. launch three vagrant nodes.

Note: If you look at the project's Vagrantfile, you will notice that all the vagrant nodes (except for the first node) boot up with stock centos7.2 OS and a serf agent running. serf is used as the node discovery service. This is intentional to meet the goal of limiting the amount of services that user needs to setup to start bringing up a cluster and hence making management easier.

cd cluster/
CONTIV_NODES=3 make demo-cluster

3. login to the first node to manage the cluster

Note: The first node is slightly special in a way that it is booted up with two additional services viz. collins and clusterm. collins is used as the node lifecycle management and event logging service. clusterm is the cluster manager daemon. clusterctl utility is provided to exercise cluster manager provided REST endpoint.

CONTIV_NODES=3 vagrant ssh cluster-node1

Provision additional nodes for discovery

clusterctl discover <host-ip(s)>

Cluster Manager uses Serf as a discovery service for node health monitoring and for cluster bootstrapping. Use discover command to include additional nodes in the discovery service. The <host-ip> should be an IP address from a management network only used by infra services such as serf, etcd, swarm, etc..

Note:

clusterctl discover 192.168.2.11 192.168.2.12 --extra-vars='{"env" : {}, "control_interface": "eth1" }'
  • The command above will provision the other two vms (viz. cluster-node2 and cluster-node3) in the vagrant setup for serf based discovery. Once it is run, the discovered hosts will appear in clusterctl nodes get output in a few minutes.
  • the clusterctl discover command expects env and control_interface ansible variables to be specified. This can be achieved by using the --extra-vars flag as shown above or by setting them at global level, if applicable. For more information on other available variables, also checkout discovery section of ansible vars

Get list of discovered nodes

clusterctl nodes get

And info for a single node can be fetched by using clusterctl node get <node-name>.

Commission a node

clusterctl node commission <node-name> --host-group=<service-master|service-worker>

Commissioning a node involves pushing the configuration and starting infra services on that node using ansible based configuration management. The services that are configured depend on the mandatory parameter --host-group. Checkout the service-master and service-worker host-groups in ansible/site.yml to learn more about the services that are configured. To quickly check if commissioning a node worked, you can run etcdctl member list on the node. It shall list all the commissioned members in the list.

Note:

  • certain ansible variables need to be set for provisioning a node. The list of mandatory and other useful variables is provided in ansible_vars.md. The variables need to be passed as a quoted JSON string in node commission command using the --extra-vars flag.
clusterctl node commission node1 --extra-vars='{"env" : {}, "control_interface": "eth1", "netplugin_if": "eth2" }' --host-group "service-master"
  • a common set of variables (like environment) can be set just once as global variables. This eliminates the need to specify the common variables for every commission command.

Decommission a node

clusterctl node decommission <node-name>

Decommissioning a node involves stopping and cleaning the configuration for infra services on that node using ansible based configuration management.

Update a node

clusterctl node update <node-name>

Updating a node involves updating the configuration for infra services on that node using ansible based configuration management. Other use-cases for updating a node include installing newer versions of infra services or changing the host-group of the node like changing a node from worker to master and vice-versa.

Note:

clusterctl node update node1 --extra-vars='{"env" : {}, "control_interface": "eth1", "netplugin_if": "eth2" }' --host-group "service-worker"
  • similar to commission command, the --extra-vars flag can be used with the update command to specify ansible variables needed for provisioning the node.
  • to change the host-group of a node, the --host-group flag is used. If this flag is not specified then node's configuration is updated with the last set host-group.

Set/Get global variables

clusterctl global set --extra-vars=<vars>

A common set of variables (like environment, scheduler-provider and so on) can be set just once using the --extra-vars flag with clusterctl global set command.

Note:

  • The variables need to be passed as a quoted JSON string using the --extra-vars flag.
clusterctl global set --extra-vars='{"env" : {"http_proxy": "my.proxy.url"}, "scheduler_provider": "ucp-swarm"}'
  • The variables set at global level are merged with the variables specified at the node level, with the latter taking precedence in case of an overlap/conflict.
  • The list of useful variables is provided in ansible_vars.md.

Get provisioning job status

clusterctl job get <active|last>

Common cluster management workflows like commission, decommission and so on involve running an ansible playbook. Each such run per workflow is referred to as a job. You can see the status of an ongoing (active) or last run job using this command.

Managing multiple nodes

clusterctl nodes commission <space separated node-name(s)>
clusterctl nodes decommission <space separated node-name(s)>
clusterctl nodes update <space separated node-name(s)>

The worflow to commission, decommission or update all or a subset of nodes can be performed by using clusterctl nodes subcommands. Please refer the documentation of individual commands above for details.

##Want to learn more? Read the design spec and/or see the remaining/upcoming features in github issues page