COMMAND LINE INTERFACE (CLI) INTRODUCTION

1. Terminology

Node: A VM in the cluster

NodeGroup: A group of VMs with same functionality and spec

Cluster: A group of NodeGroups working together

Distro: A Hadoop distribution version

2. Usage

CLI supports shell mode, command line mode, and execution of script file. After compile, you can find the jar file under cli/target directory.

• Shell mode: java -jar serengeti-cli-*.jar. It supports tab key based command hint and completion. It supports history by up/down arrows.
• Command line mode: java -jar serengeti-cli-*.jar "command1;command2..."
• Execution of script file: in shell mode or command line mode, execute "script --file scriptFileName". The shell history file named cli.history will help to generate the script file.

3. Command Syntax

CLI commands are built on spring shell(https://github.com/SpringSource/spring-shell), so its syntax follows spring shell command syntax below:

Command ::= serengeti <command-category> <command-name> <required-command-key-values>* [<optional-command-key-values>]*

required-command-key-values ::= <command-key-value-fullsize>

optional-command-key-values ::= <command-key-value-fullsize>

required-command-key-values-fullsize ::= <command-key-value-fullsize>

optional-command-key-value-fullsize ::= <command-key-value-fullsize>

command-key-value-fullsize ::= '--'<command-key-full> [<value>]

- Command-category is the type of object that the command will operate on.

- Command-name is the operation to do.

- Required-command-key-values are parameters that required for the operation.

- Optional-command-key-values are parameters that are optional for the operation.

4. Command Help

• Shell level Help: user can get shell level help by using "help" in command line mode or help command in shell mode. A list of objects will be displayed with operations and short descriptions.
• Command category level help: to get help for a particular class of object, user can enter help followed by an object. CLI will display all operations defined for the object.
• Command level help: to get help for a particular operation, user can enter help followed by an object and an operation. This will display all parameters.

5. Command List

command-category	command-name	Parameters	descriptions
connect		Mandatory --host <Serengeti server host>:8080 Options --username <serengeti user name> --password <serengeti password>	Function Connect a Serengeti server. Remark The Serengeti host with optional port number, e.g. hostname:port . Connect command requires username and password interactively. If connect failed, the other Serengeti command is not allowed to execute.
resource pool	add	Mandatory --name <resource pool name> --vcrp < VC rp name > --vccluster < vsphere cluster name > Options --username	Function Add a new resource pool. Remark User must add a resource pool before create a cluster.
	list	Mandatory None Options --name <resource pool name> --detail <flag to show node information>	Function Show resource pool information.
	delete	Mandatory --name <resource pool name> Options None	Function Delete an unused resource pool. Remark User cannot delete a resource pool which is in use.
datastore	add	Mandatory --name <data store name> --spec <datastore name(s) in the vsphere> Options --type <specify the type for storage: SHARED or LOCAL>	Function Add new datastore(s) . Remark When the spec parameter include some datastore names, users need to use "," to separate them. Users may also specify multiple data stores by a wildcard.
	list	Mandatory None Options --name <data store name> --detail < flag to show datastore detail information >	Function Show datastore information.
	delete	Mandatory --name <data store name> Options None	Function Delete an unused datastore .
network	add	Mandatory --name <network name> --portGroup<vsphere port group name> Options Combination 1 -- dhcp < mark as dhcp type> Combination 2 -- ip <ip range information> -- dns< first dns information > -- secondDNS <second dns information> -- gateway <gateway information> -- mask<mask information>	Function Add a network to Serengeti. Remark Users must enter either ip range or dhcp, but not both. Example network add --name ipNetwork --ip 192.168.1.1-100,192.168.1.256-300 --portGroup pg1 --dns 202.112.0.1 --gateway 192.168.1.255 --mask 255.255.255.1 network add --name dhcpNetwork --dhcp
	list	Mandatory None Options --name <network name> --detail <flag to show network detail information >	Function Show network information.
	delete	Mandatory --name <network name> Options None	Function Delete an unused network. Remark User cannot delete a network which is in use.
cluster	create	Mandatory --name <cluster name> Options --distro < hadoop distro name> --specFile <spec file pathname> --rpNames < resource pools for the cluster > --dsNames < datastores for the cluster> --networkName <network name> --resume < flag to resume cluster creation > --topology < Specify which topology type will be used for rack awareness: HVE, RACK_AS_RACK, or HOST_AS_RACK. > --skipConfigValidation < flag to skip cluster configuration validation > --yes < flag to answer 'yes' to all Y/N questions >	Function Create a Hadoop cluster Remark The specFile is defined in json format. Example cluster create --name testCluster
	config	Mandatory --name <cluster name> --specFile <spec file pathname> Options --skipConfigValidation < flag to skip cluster configuration validation > --yes < flag to answer 'yes' to all Y/N questions >	Function Config an existing cluster. Remark The specFile is defined in json format.
	delete	Mandatory --name <cluster name> Options None	Function Delete an unused cluster.
	resize	Mandatory --name <cluster name> --nodeGroup < node group name> --instanceNum <instance number> Options None	Function Change the number of nodes in a node group. Parameter Definition instance number ::= <final number> Remark The instanceNum should be larger than the existing instance numbers in the node group. Example cluster resize --name testCluster --nodeGroup slave -- instanceNum 10
	start	Mandatory --name <cluster name> Options --nodeGroupName <node group name> --nodeName <node name>	Function Start VMs in a cluster.
	stop	Mandatory --name <cluster name> Options --nodeGroupName <node group name> --nodeName <node name>	Function Stop VMs in a cluster.
	limit	Mandatory --name <cluster name>: Name of the Hadoop cluster in Serengeti --activeComputeNodeNum <active node number>: Number of active compute nodes for the specified Hadoop cluster or node group within that cluster. Options --nodeGroupName <node group name>: Name of a node group in the specified Hadoop cluster in Serengeti (only supports node groups with task tracker role)	Function Enable or disable provisioned compute nodes in the specified Hadoop cluster or node group in Serengeti to reach the limit specified by activeComputeNodeNum. Compute nodes are re-commissioned and powered-on or decommissioned and powered-off to reach the specified number of active compute nodes.
	unlimit	Mandatory --name <cluster name>: Name of the Hadoop cluster in Serengeti Options --nodeGroupName <node group name>: Name of a node group in the specified Hadoop cluster in Serengeti (only supports node groups with task tracker role)	Function Remove cluster limit restriction for provisioned compute nodes in the specified Hadoop cluster or node group in Serengeti. All compute nodes in specified cluster or node group are all re-commissioned and powered-on.
	list	Mandatory None Options --name <cluster name> --detail <flag to show node information>	Function Show cluster information. Note: with this option specified, Serengeti will query from vCenter server to get latest node status. That operation may take some time, for example, longer than 7 seconds for each cluster. Please be patient.
	export	Mandatory --spec <falg to export cluster specification> --name <cluster name> Options --output <output file name>	Function Export cluster specification.
	target	Mandatory None Options --name <cluster name> --info <flag to show target information>	Function Set or query target cluster to run hadoop commands.
distro	list	Mandatory None Options --name <distro name> --detail < flag to show distro detail information>	Function Show distro information.
topology	list	Mandatory None Options None	Function Show the rack->hosts mapping topology.
	upload	Mandatory --fileName <topology file name> Options --yes <answer 'yes' to all Y/N questions.>	Function Upload a rack-->hosts mapping topology file.
disconnect			Function Disconnect the Serengeti server.

## 6. Hadoop Commands From CLI 0.6.0, we integrated impala(https://github.com/SpringSource/impala) hadoop hdfs, map/reduce, pig, and hive commands into CLI. You need to use "cluster target" command to set hdfs or jobtracker url before launching hdfs or map/reduce commands. More details can be found from CLI help.