Clusterctl

From ISPWiki

Contents 1 Functions 2 Operating modes 3 Choosing nodes to start a command 3.1 Examples 4 Special commands 4.1 Managing the roles 5 Result 6 Examples

Functions

The utility can be used for:

1. executing a command on multiple nodes.
2. managing the node roles.

Please note! You must have root privileges to run the commands.

Operating modes

There are 3 operating modes: specified commands will run on all the nodes either in parallel (in this case the result will output as long as they are processed), or successively with time delay. A required mode is set by the --runmode key:

--runmode parallel     Run in parallel 
--runmode 0            Run successively without time delay (default mode) 
--runmode N            Run with N sec delay

Choosing nodes to start a command

In order to choose a node on which you wish to start a command, you can provide either a role or the exact node:

--role HTTP            Start on all the nodes that have the HTTP role. You just need to provide the role's name.
--role all             Select all roles (by default).

You can provide the exact node.

--node 12.34.56.78     Run on node 12.34.56.78 
--node LOCAL           Run on the local node 
--node ALL             Run on all nodes (by default) 

Examples

clusterctl --role MASTER

Use all the node that have the MASTER role

clusterctl --node LOCAL

Use a local node

clusterctl --node LOCAL --role MASTER

Use a local node, only if it has the MASTER role

clusterctl --node ALL --role MASTER

See clusterctl --role MASTER

Special commands

The following special commands are available:

my_roles          Show all the roles for the node, on which the command was run. All the local IP-addresses will be located. 
role_executors    Show all the IP-addresses that have the specified role. 
bind_local_ip     Bind the IP-address of all the local node web-servers to the "bind_interface" interface that is specified by the configuration file parameter.

Managing the roles

There is a number of special commands that can be used for managing the role start scripts:

start             Run the start script of the specified role and/or node
stop              Run the stop scripts of the specified role and/or node 
start_scripts     Show all start scripts of the specified role and/or node
stop_scripts      Show all stop scripts of the specified role and/or node 

For example:

clusterctl --role HTTP --node LOCAL start

Run the HTTP role scripts on the local node (if any)

clusterctl --role HTTP start

Run the HTTP role scripts on all the nodes where it is present

clusterctl --role HTTP --runmode parallel start

Run the HTTP role scripts on all the nodes where this role is present in the parallel mode.

Result

The command execution results from different servers are separated by the following lines:

#clusterctl -- 127.0.0.1 ------------------

Everything that the program logged in STDOUT and STDERR will be locally displayed in STDOUT and STDERR. The status of the command execution is specified in a separate line as follows:

#clusterctl Exit with code: 127

Or:

#clusterctl Exit with signal: 9

Or:

#clusterctl Stopped with signal: 5

Examples

Run "uname -a" on all the servers:

clusterctl --role all "uname -a"

First run "uname -a" on all the servers, then run "date":

clusterctl --role all "uname -a" "date"

Monitor the Apache processes on all the worker HTTP:

clusterctl --role HTTP "ps auxw | grep apache"

Find out the roles of the current node:

clusterctl --role all my_roles

Find out the nodes that process the HTTP role

clusterctl --role HTTP role_executors

Bind the local IP-addresses:

clusterctl bind_local_ip

Use an alternative configuration file:

clusterctl --conf /other/path/oproxy.conf ...

Help:

clusterctl --help

Version:

clusterctl --version

Examples on how to manage the role start scripts are described in the corresponding section given above.