Table of Contents
List of Examples
The Common Reusable SHell (CRaSH) deploys in a Java runtime and provides interactions with the JVM. Commands are written in Groovy and can be developped at runtime making the extension of the shell very easy with fast development cycle.
Table of Contents
CRaSH provides has various ways to be started, it can even be easily embedded.
CRaSH can use a standard web archive to be deployed in a web container. The war file is used for its packaging capabilities and triggering the CRaSH life cycle start/stop. In this mode CRaSH has two packaging available:
A core war file found under deploy/core/crash.war provides the base CRaSH functionnalities.
A gatein war file found under deploy/gatein/crash.war provides additional Java Content Repository (JCR) features but deploys only in a GateIn server (Tomcat or JBoss). It extends the core packaging and adds
JCR browsing and interactions
SCP support for JCR import and export
You have to copy the crash.war in the appropriate server, regardless of the packaging used.
The standalone mode allows you to run CRaSH from the command line directly. It provides the same functionality as the war deployment but does not require a web container as it runs its own virtual machine. The directory crash directory in the application contains the standalone distribution.
The bin directory /crash/bin can be added to the system path, it contains the crash.sh script that will start the standalone mode, for instance you can set it up this way:
> export PATH=/.../crash/bin:$PATH > crash.sh ______ .~ ~. |`````````, .'. ..'''' | | | |'''|''''' .''```. .'' |_________| | | `. .' `. ..' | | `.______.' | `. .' `. ....'' | | 1.0.0-cr2-SNAPSHOT Follow and support the project on http://vietj.github.com/crash Welcome to jerry + ! It is Thu Apr 12 21:19:35 CEST 2012 now
Let's review quickly what you can find in standalone crash:
The bin directory contains the crash.sh script and the standalone crash jar file
The conf directory contains the JVM logging configuration logging.properties
The cmd directory contains the commands that will be available in crash by default it contains a few example commands
The lib directory contains the various libraries used by crash, you should place additional jar files there
The attach mode allows you to attach CRaSH to a JVM located on the same host with the attach API provided by the Hotspot JVM. It works thanks to the standalone mode, the main difference is when you run the command line you can specify a process id of a JVM and CRaSH will hook into the targetted JVM, let's see quickly an example of how to use it
> jps 3165 RemoteMavenServer 20650 Test 20651 Jps > crash.sh 20650 ______ .~ ~. |`````````, .'. ..'''' | | | |'''|''''' .''```. .'' |_________| | | `. .' `. ..' | | `.______.' | `. .' `. ....'' | | 1.0.0-cr2-SNAPSHOT Follow and support the project on http://vietj.github.com/crash Welcome to jerry + ! It is Thu Apr 12 22:09:23 CEST 2012 now %
In this example we will attach crash to the Test JVM. We obtained the Test JVM PID thanks to the jps command that belongs to the Java Platform. During this mode the commands are executed in the target JVM.
You need to connect using telnet, SSH to use the shell, there is a third special mode using the JVM input and output.
Telnet connection is done on port 5000:
(! 520)-> telnet localhost 5000 Trying ::1... Connected to localhost. Escape character is '^]'. ______ .~ ~. |`````````, .'. ..'''' | | | |'''|''''' .''```. .'' |_________| | | `. .' `. ..' | | `.______.' | `. .' `. ....'' | | 1.0.0-cr2 Follow and support the project on http://vietj.github.com/crash Welcome to julien.local + ! It is Fri Dec 03 16:20:40 CET 2010 now
The bye command disconnect from the shell.
SSH connection is done on port 2000 with the password crash :
juliens-macbook-pro:~ julien$ ssh -p 2000 -l root localhost root@localhost's password: CRaSH 1.0.0-cr2 (http://vietj.github.com/crash) Welcome to juliens-macbook-pro.local! It is Fri Jan 08 21:12:53 CET 2010 now. %
The bye command disconnect from the shell.
Line edition: the current line can be edited via left and right arrow keys
History: the key up and key down enable history browsing
Quoting: simple quotes or double quotes allow to insert blanks in command options and arguments, for instance "old boy" or 'old boy'. One quote style can quote another, like "ol' boy".
Completion: an advanced completion system is available
The help command will display the list of known commands by the shell.
[/]% help % help Try one of these commands with the -h or --help switch: cd changes the current node commit saves changes consume collects a set of nodes cp copy a node to another env display the term env exportworkspace Export a workspace on the file system (experimental) fail Fails help provides basic help importworkspace Import a workspace from the file system (experimental) invoke Invoke a static method log logging commands ls list the content of a node man format and display the on-line manual pages mixin mixin commands mv move a node node node commands produce produce a set of nodes pwd print the current node path rm remove one or several node or a property rollback rollback changes select execute a JCR sql query setperm modify the security permissions of a JCR node sleep sleep for some time thread vm thread commands version versioning commands wait Invoke a static method ws workspace commands xpath execute a JCR xpath query
The basic CRaSH usage is like any shell, you just type a command with its options and arguments. However it is possible to compose commands and create powerful combinations.
Any command help can be displayed by using the -h argument:
% ls -h usage: ls [-h | --help] [-h | --help] [-d | --depth] path [-h | --help] command usage [-h | --help] command usage [-d | --depth] Print depth path the path of the node content to list
In addition of that, commands can have a complete manual that can be displayed thanks to the man command:
% man ls
NAME
ls - list the content of a node
SYNOPSIS
ls [-h | --help] [-h | --help] [-d | --depth] [-d | --depth] path
DESCRIPTION
The ls command displays the content of a node. By default it lists the content of the current node, however it also
accepts a path argument that can be absolute or relative.
[/]% ls
/
+-properties
| +-jcr:primaryType: nt:unstructured
| +-jcr:mixinTypes: [exo:owneable,exo:privilegeable]
| +-exo:owner: '__system'
| +-exo:permissions: [any read,*:/platform/administrators read,*:/platform/administrators add_node,*:/platform/administrators set_property,*:/platform/administrators remove]
+-children
| +-/workspace
| +-/contents
| +-/Users
| +-/gadgets
| +-/folder
PARAMETERS
[-h | --help]
Provides command usage
[-h | --help]
Provides command usage
[-d | --depth]
Print depth
path
the path of the node content to list
A CRaSH command is able to consume and produce a stream of object, allowing complex interactions between commands where they can exchange stream of compatible objets. Most of the time, JCR nodes are the objects exchanged by the commands but any command is free to produce or consume any type.
By default a command that does not support this feature does not consumer or produce anything. Such commands usually inherits from the org.crsh.command.ClassCommand class that does not care about it. If you look at this class you will see it extends the the org.crsh.command.BaseCommand.
More advanced commands inherits from org.crsh.command.BaseCommand class that specifies two generic types <C> and <P>:
<C> is the type of the object that the command consumes
<P> is the type of the object that the command produces
The command composition provides two operators:
The pipe operator | allows to stream a command output stream to a command input stream
The distribution operator + allows to distribute an input stream to several commands and to combine the output stream of several commands into a single stream.
Example 1.2. Update the security of all nt:unstructed nodes
% select * from nt:unstructured | setperm -i any -a read + setperm -i any -a write
NAME
sleep - sleep for some time
SYNOPSIS
sleep [-h | --help] time
PARAMETERS
[-h | --help]
Provides command usage
time
Sleep time in seconds
NAME
man - format and display the on-line manual pages
SYNOPSIS
man [-h | --help] command
PARAMETERS
[-h | --help]
Provides command usage
command
the command
NAME
log add - create one or several loggers
SYNOPSIS
log [-h | --help] add ... name
PARAMETERS
[-h | --help]
Provides command usage
... name
The name of the logger
NAME
log set - configures the level of one of several loggers
SYNOPSIS
log [-h | --help] set [-l | --level] [-p | --plugin] ... name
DESCRIPTION
The set command sets the level of a logger. One or several logger names can be specified as arguments
and the -l option specify the level among the trace, debug, info, warn and error levels. When no level is
specified, the level is cleared and the level will be inherited from its ancestors.
% logset -l trace foo
% logset foo
The logger name can be omitted and instead stream of logger can be consumed as it is a <Logger,Void> command.
The following set the level warn on all the available loggers:
% log ls | log set -l warn
PARAMETERS
[-h | --help]
Provides command usage
[-l | --level]
The logger level to assign among {trace, debug, info, warn, error}
[-p | --plugin]
Force the plugin implementation to use
... name
The name of the logger
NAME
log send - send a message to a logger
SYNOPSIS
log [-h | --help] send [-m | --message] [-l | --level] name
DESCRIPTION
The send command log one or several loggers with a specified message. For instance the following impersonates
the javax.management.mbeanserver class and send a message on its own logger.
#% log send -m hello javax.management.mbeanserver
Send is a <Logger, Void> command, it can log messages to consumed log objects:
% log ls | log send -m hello -l warn
PARAMETERS
[-h | --help]
Provides command usage
[-m | --message]
The message to log
[-l | --level]
The logger level to assign among {trace, debug, info, warn, error}
name
The name of the logger
NAME
log info - display info about a logger
SYNOPSIS
log [-h | --help] info ... name
DESCRIPTION
The loginfo command displays information about one or several loggers.
% loginfo javax.management.modelmbean
javax.management.modelmbean<INFO>
The loginfo command is a <Logger,Void> command and it can consumed logger produced by the logls command:
% logls -f javax.* | loginfo
javax.management.mbeanserver<INFO>
javax.management.modelmbean<INFO>
PARAMETERS
[-h | --help]
Provides command usage
... name
The name of the logger
NAME
log ls - list the available loggers
SYNOPSIS
log [-h | --help] ls [-f | --filter]
DESCRIPTION
The logls command list all the available loggers., for instance:
% logls
org.apache.catalina.core.ContainerBase.[Catalina].[localhost].[/].[default]
org.apache.catalina.core.ContainerBase.[Catalina].[localhost].[/eXoGadgetServer].[concat]
org.apache.catalina.core.ContainerBase.[Catalina].[localhost].[/dashboard].[jsp]
...
The -f switch provides filtering with a Java regular expression
% logls -f javax.*
javax.management.mbeanserver
javax.management.modelmbean
The logls command is a <Void,Logger> command, therefore any logger produced can be consumed.
PARAMETERS
[-h | --help]
Provides command usage
[-f | --filter]
A regular expressions used to filter the loggers
NAME
thread stop - stop vm threads
SYNOPSIS
thread [-h | --help] stop
DESCRIPTION
Stop a VM thread, this method cannot be called as is and should be used with a pipe to consume a list of threads.
PARAMETERS
[-h | --help]
Provides command usage
NAME
thread interrupt - interrupt vm threads
SYNOPSIS
thread [-h | --help] interrupt
DESCRIPTION
Interrup a VM thread, this method cannot be called as is and should be used with a pipe to consume a list of threads.
PARAMETERS
[-h | --help]
Provides command usage
NAME
thread ls - list the vm threads
SYNOPSIS
thread [-h | --help] ls [-n | --name] [-f | --filter] [-s | --state]
PARAMETERS
[-h | --help]
Provides command usage
[-n | --name]
Retain the thread with the specified name
[-f | --filter]
Filter the threads with a regular expression on their name
[-s | --state]
Filter the threads by their status (new,runnable,blocked,waiting,timed_waiting,terminated)
NAME
system gc - call garbage collector
SYNOPSIS
system [-h | --help] gc
PARAMETERS
[-h | --help]
Provides command usage
NAME
system propls - list the vm system properties
SYNOPSIS
system [-h | --help] propls
PARAMETERS
[-h | --help]
Provides command usage
NAME
system propset - set a system property
SYNOPSIS
system [-h | --help] propset name value
PARAMETERS
[-h | --help]
Provides command usage
name
The name of the property
value
The value of the property
NAME
system propget - get a system property
SYNOPSIS
system [-h | --help] propget name
PARAMETERS
[-h | --help]
Provides command usage
name
The name of the property
NAME
system proprm - remove a system property
SYNOPSIS
system [-h | --help] proprm name
PARAMETERS
[-h | --help]
Provides command usage
name
The name of the property
NAME
system freemem - show free memory
SYNOPSIS
system [-h | --help] freemem [-u | --unit] [-d | --decimal]
PARAMETERS
[-h | --help]
Provides command usage
[-u | --unit]
The unit of the memory space size {(B)yte, (O)ctet, (M)egaOctet, (G)igaOctet}
[-d | --decimal]
The number of decimal (default 0)
NAME
system totalmem - show total memory
SYNOPSIS
system [-h | --help] totalmem [-u | --unit] [-d | --decimal]
PARAMETERS
[-h | --help]
Provides command usage
[-u | --unit]
The unit of the memory space size {(B)yte, (O)ctet, (M)egaOctet, (G)igaOctet}
[-d | --decimal]
The number of decimal (default 0)
NAME
jdbc close - close the current connection
SYNOPSIS
jdbc [-h | --help] close
PARAMETERS
[-h | --help]
Provides command usage
NAME
jdbc query - execute SQL query
SYNOPSIS
jdbc [-h | --help] query ... sqlQuery
PARAMETERS
[-h | --help]
Provides command usage
... sqlQuery
The query
NAME
jdbc open - open connection from datasource
SYNOPSIS
jdbc [-h | --help] open datasource
PARAMETERS
[-h | --help]
Provides command usage
datasource
The datasource
NAME
jdbc connect - connect to database with JDBC connection string
SYNOPSIS
jdbc [-h | --help] connect connectionString
PARAMETERS
[-h | --help]
Provides command usage
connectionString
The connection string
Table of Contents
The CRaSH JCR extension allow to connect and interract with Java Content Repository implementations.
NAME
repo info - show info about the current repository
SYNOPSIS
repo [-h | --help] info
DESCRIPTION
The info command print the descriptor of the current repository.
PARAMETERS
[-h | --help]
Provides command usage
NAME
repo ls - list the available repository plugins
SYNOPSIS
repo [-h | --help] ls
DESCRIPTION
The ls command print the available repository plugins.
PARAMETERS
[-h | --help]
Provides command usage
NAME
repo use - changes the current repository
SYNOPSIS
repo [-h | --help] use parameters
DESCRIPTION
The use command changes the current repository used by for JCR commands. The command accepts a set of properties
as main command argument that will be used to select a repository:
% repo use parameterName=parameterValue;nextParameterName=nextParameterValue
The parameters is specific to JCR plugin implementations, more details can be found thanks to the ls command.
PARAMETERS
[-h | --help]
Provides command usage
parameters
The parameters used to instantiate the repository to be used in this session
NAME
ws login - login to a workspace
SYNOPSIS
ws [-h | --help] login [-u | --username] [-p | --password] [-c | --container] workspaceName
DESCRIPTION
This command login to a JCR workspace and establish a session with the repository.
When you are connected the shell maintain a JCR session and allows you to interact with the session in a shell
oriented fashion. The repository name must be specified and optionally you can specify a user name and password to
have more privileges.
Before performing a login operation, a repository must be first selected with the repo command, for instance:
% repo use container=portal
Once a repository is obtained the login operation can be done:
% ws login portal-system
Connected to workspace portal-system
% ws login -u root -p gtn portal-system
Connected to workspace portal-system
PARAMETERS
[-h | --help]
Provides command usage
[-u | --username]
The user name
[-p | --password]
The user password
[-c | --container]
The portal container name (eXo JCR specific)
workspaceName
The name of the workspace to connect to
NAME
ws logout - logout from a workspace
SYNOPSIS
ws [-h | --help] logout
DESCRIPTION
This command logout from the currently connected JCR workspace
PARAMETERS
[-h | --help]
Provides command usage
NAME
cd - changes the current node
SYNOPSIS
cd [-h | --help] path
DESCRIPTION
The cd command changes the current node path. The command used with no argument changes to the root
node. A relative or absolute path argument can be provided to specify a new current node path.
[/]% cd /gadgets
[/gadgets]% cd /gadgets
[/gadgets]% cd
[/]%
PARAMETERS
[-h | --help]
Provides command usage
path
The new path that will change the current node navigation
NAME
pwd - print the current node path
SYNOPSIS
pwd [-h | --help]
DESCRIPTION
The pwd command prints the current node path, the current node is produced by this command.
[/gadgets]% pwd
/gadgets
PARAMETERS
[-h | --help]
Provides command usage
NAME
ls - list the content of a node
SYNOPSIS
ls [-h | --help] [-d | --depth] path
DESCRIPTION
The ls command displays the content of a node. By default it lists the content of the current node, however it also
accepts a path argument that can be absolute or relative.
[/]% ls
/
+-properties
| +-jcr:primaryType: nt:unstructured
| +-jcr:mixinTypes: [exo:owneable,exo:privilegeable]
| +-exo:owner: '__system'
| +-exo:permissions: [any read,*:/platform/administrators read,*:/platform/administrators add_node,*:/platform/administrators set_property,*:/platform/administrators remove]
+-children
| +-/workspace
| +-/contents
| +-/Users
| +-/gadgets
| +-/folder
PARAMETERS
[-h | --help]
Provides command usage
[-d | --depth]
The depth of the printed tree
path
The path of the node content to list
NAME
cp - copy a node to another
SYNOPSIS
cp [-h | --help] source target
DESCRIPTION
The cp command copies a node to a target location in the JCR tree.
[/registry]% cp foo bar
PARAMETERS
[-h | --help]
Provides command usage
source
The path of the source node to copy
target
The path of the target node to be copied
NAME
mv - move a node
SYNOPSIS
mv [-h | --help] source target
DESCRIPTION
The mv command can move a node to a target location in the JCR tree. It can be used also to rename a node. The mv
command is a <Node,Node> command consuming a stream of node to move them and producing nodes that were moved.
[/registry]% mv Registry Registry2
PARAMETERS
[-h | --help]
Provides command usage
source
The path of the source node to move, absolute or relative
target
The destination path absolute or relative
NAME
rm - remove one or several node or a property
SYNOPSIS
rm [-h | --help] ... paths
DESCRIPTION
The rm command removes a node or property specified by its path either absolute or relative. This operation
is executed against the JCR session, meaning that it will not be effective until it is commited to the JCR server.
[/]% rm foo
Node /foo removed
It is possible to specify several nodes.
[/]% rm foo bar
Node /foo /bar removed
rm is a <Node,Void> command removing all the consumed nodes.
PARAMETERS
[-h | --help]
Provides command usage
... paths
The paths of the node to remove
NAME
node add - creates one or several nodes
SYNOPSIS
node [-h | --help] add [-t | --type] ... paths
DESCRIPTION
The addnode command creates one or several nodes. The command takes at least one node as argument, but it can
take more. Each path can be either absolute or relative, relative path creates nodes relative to the current node.
By default the node type is the default repository node type, but the option -t can be used to specify another one.
[/registry]% addnode foo
Node /foo created
[/registry]% addnode -t nt:file bar juu
Node /bar /juu created
The addnode command is a <Void,Node> command that produces all the nodes that were created.
PARAMETERS
[-h | --help]
Provides command usage
[-t | --type]
The name of the primary node type to create.
... paths
The paths of the new node to be created, the paths can either be absolute or relative.
NAME
node set - set a property on the current node
SYNOPSIS
node [-h | --help] set [-t | --type] propertyName propertyValue
DESCRIPTION
The set command updates the property of a node.
Create or destroy property foo with the value bar on the root node:
[/]% set foo bar
Property created
Update the existing foo property:
[/]% set foo juu
When a property is created and does not have a property descriptor that constraint its type, you can specify it
with the -t option
[/]% set -t LONG long_property 3
Remove a property
[/]% set foo
set is a <Node,Void> command updating the property of the consumed node stream.
PARAMETERS
[-h | --help]
Provides command usage
[-t | --type]
The property type to use when it cannot be inferred
propertyName
The name of the property to alter
propertyValue
The new value of the property
NAME
node import - imports a node from an nt file
SYNOPSIS
node [-h | --help] import source target
DESCRIPTION
Imports a node from an nt:file node located in the workspace:
[/]% importnode /gadgets.xml /
Node imported
PARAMETERS
[-h | --help]
Provides command usage
source
The path of the imported nt:file node
target
The path of the parent imported node
NAME
node export - export a node to an nt file
SYNOPSIS
node [-h | --help] export source target
DESCRIPTION
Exports a node as an nt file in the same workspace:
[/]% node export gadgets /gadgets.xml
The node has been exported
PARAMETERS
[-h | --help]
Provides command usage
source
The path of the exported node
target
The path of the exported nt:file node
NAME
mixin add - add a mixin to one or several nodes
SYNOPSIS
mixin [-h | --help] add mixin ... paths
DESCRIPTION
The add command addds a mixin to one or several nodes, this command is a <Node,Void> command, and can
add a mixin from an incoming node stream, for instance:
[/]% select * from mynode | mixin add mix:versionable
PARAMETERS
[-h | --help]
Provides command usage
mixin
the mixin name to add
... paths
the paths of the node receiving the mixin
NAME
mixin remove - removes a mixin from one or several nodes
SYNOPSIS
mixin [-h | --help] remove mixin ... paths
DESCRIPTION
The remove command removes a mixin from one or several nodes, this command is a <Node,Void> command, and can
remove a mixin from an incoming node stream, for instance:
[/]% select * from mynode | mixin remove mix:versionable
PARAMETERS
[-h | --help]
Provides command usage
mixin
the mixin name to remove
... paths
the paths of the node receiving the mixin
NAME
select - execute a JCR sql query
SYNOPSIS
select [-h | --help] [-o | --offset] [-l | --limit] [-a | --all] ... query
DESCRIPTION
Queries in SQL format are possible via the ##select## command. You can write a query with the same syntax defined
by the specification and add options to control the number of results returned. By default the number of nodes is limited
to 5 results:
[/]% select * from nt:base
The query matched 1114 nodes
+-/
| +-properties
| | +-jcr:primaryType: nt:unstructured
| | +-jcr:mixinTypes: [exo:owneable,exo:privilegeable]
| | +-exo:owner: '__system'
| | +-exo:permissions: [any read,*:/platform/administrators read,*:/platform/administrators add_node,*:/platform/administratorsset_property,*:/platform/administrators remove]
+-/workspace
| +-properties
| | +-jcr:primaryType: mop:workspace
| | +-jcr:uuid: 'a69f226ec0a80002007ca83e5845cdac'
...
Display 20 nodes from the offset 10:
[/]% select * from nt:base -o 10 -l 20
The query matched 1114 nodes
...
It is possible also to remove the limit of displayed nodes with the -a option (you should use this option with care) :
[/]% select * from nt:base -a
The query matched 1114 nodes
...
select is a <Void,Node> command producing all the matched nodes.
PARAMETERS
[-h | --help]
Provides command usage
[-o | --offset]
The offset of the first node to display
[-l | --limit]
The number of nodes displayed, by default this value is equals to 5
[-a | --all]
Display all the results by ignoring the limit argument, this should be used with care for large result set
... query
The query, as is
NAME
xpath - execute a JCR xpath query
SYNOPSIS
xpath [-h | --help] [-o | --offset] [-l | --limit] [-a | --all] query
DESCRIPTION
Executes a JCR query with the xpath dialect, by default results are limited to 5.All results matched by the query are produced by this command.
PARAMETERS
[-h | --help]
Provides command usage
[-o | --offset]
The offset of the first node to display
[-l | --limit]
The number of nodes displayed, by default this value is equals to 5
[-a | --all]
Display all the results by ignoring the limit argument, this should be used with care for large result set
query
The query
NAME
commit - saves changes
SYNOPSIS
commit [-h | --help] path
DESCRIPTION
Saves the changes done to the current session. A node can be provided to save the state of the
this nodes and its descendants only.
PARAMETERS
[-h | --help]
Provides command usage
path
The path of the node to commit
NAME
rollback - rollback changes
SYNOPSIS
rollback [-h | --help] path
DESCRIPTION
Rollbacks the changes of the current session. A node can be provided to rollback the state of the
this nodes and its descendants only.
PARAMETERS
[-h | --help]
Provides command usage
path
the path to rollback
NAME
version checkin - checkin a node
SYNOPSIS
version [-h | --help] checkin path
DESCRIPTION
Perform a node checkin
PARAMETERS
[-h | --help]
Provides command usage
path
The node path to checkin
NAME
version checkout - checkout a node
SYNOPSIS
version [-h | --help] checkout path
DESCRIPTION
Perform a node checkout
PARAMETERS
[-h | --help]
Provides command usage
path
The node path to checkout
Secure copy can be used to import or export content. The username/password prompted by the SSH server will be used for authentication against the repository when the import or the export is performed.
The following command will export the node /gadgets in the repository portal-system of the portal container portal:
scp -P 2000 root@localhost:portal:portal-system:/production/app:gadgets gadgets.xml
The node will be exported as app_gadgets.xml.
Note that the portal container name is used for GateIn. If you do omit it, then the root container will be used.
Table of Contents
CRaSH is configured by a set of properties, these properties are defined in a configuration file. In the war file packaging, the configuration file can be found under /WEB-INF/crash/crash.properties file of the archive. Configuration can be overriden by Java Virtual Machine system properties by using the same property name.
CRaSH properties are always prefixed by the crash. value
The key can be changed by replacing the file WEB-INF/sshd/hostkey.pem. Alternatively you can configure the server to use an external file by using the crash.ssh.keypath parameter in the crash.properties. Uncomment the corresponding property and change the path to the key file.
#crash.ssh.keypath=/path/to/the/key/file
The ports of the server are parameterized by the crash.ssh.port and crash.telnet.port parameters in the crash.properties file
# SSH configuration crash.ssh.port=2000
# Telnet configuration crash.telnet.port=5000
to remove the telnet access, remove the jar file in the WEB-INF/lib/crsh.shell.telnet-1.0.0-cr2.jar .
to remove the SSH access, remove the jar file in the WEB-INF/lib/crsh.shell.ssh-1.0.0-cr2.jar .
The /WEB-INF/crash/commands/base/login.groovy file contains two closures that are evaluated each time a message is required
The prompt closure returns the prompt message
The welcome closure returns the welcome message
Those closure can be customized to return different messages.
Authentication is used by the SSH server when a user authenticates. Authentication interface is pluggable and has default implementations. The the section called “Pluggable authentication ” explains how to write a custom authentication plugin, in this section we cover the configuation of the authentication.
The configuration of the authentication plugin is done via property, this is necessary because several plugins can be detected by CRaSH, and the plugin is selected via the property crash.auth that must match the authentication plugin name:
crash.auth=simple
CRaSH comes out of the box with two authentication plugins.
Simple authentication provides a simple username/password authentication configured with the crash.auth.simple.username and crash.auth.simple.password properties:
# Authentication configuration crash.auth=simple crash.auth.simple.username=admin crash.auth.simple.password=admin
Table of Contents
The shell command system is based on the Groovy language and can easily be extended.
Each command has a corresponding Groovy file that contains a command class that will be invoked by the shell. The files are located in the /WEB-INF/crash/commands directory and new files can be added here.
New commands can directly be placed in the commands dirctory however they can also be placed in a sub directory of the command directory, which is useful to group commands of the same kind.
In addition of that there are two special files called login.groovy and logout.groovy that are executed upon login and logout of a user. They are useful to setup and cleanup things related to the current user session.
When the user types a command in the sell, the command line is parsed by the cmdline framework and injected in the command class. Previously the args4j framework was used but this framework does not support natively code completion and could not be extended to support it. The support of command line completion is the main motivation of the development of such a framework. To learn more, the best is to study the existing commands as the framework is quite easy to use, the following features are supported:
Annotation based framework
Provide accurate contextual code completion
Support sub commands (? la git like "git add") for grouping commands of the same kind inside the same class as methods
Advanced support for usage and manual
A simple example, the sleep 1 command pauses the shell for one second, let's briefly study its code:
class sleep extends CRaSHCommand {
@Usage("sleep for some time")
@Command
Object main(@Usage("Sleep time in seconds") @Argument int time) throws ScriptException {
if (time < 0)
throw new ScriptException("Cannot provide negative time value $time");
Thread.sleep(time * 1000);
return null;
}
}The @Usage annotation gives short information about the command itself, another annotation is available for documenting more formally the command: @Man but it is not used in this example.
The @Command tags the main(...) method as a command method. Any number of method can be tagged as such, providing a convenient way to pack commands of the same kind. By default the main is a special convention indicating that executing the command should not require to explicitely use the main.
The @Argument annotation describes the command unique argument that is the time to sleep. The same @Usage annotation is used again to describe briefly the argument. Again it could be possible to use the @Man annotation.
Creating a custom is done by implementing a CRaSH plugin that provides an implementation of the AuthenticationPlugin interface, let's study the simple authentication plugin implementation.
The AuthenticationPlugin is the interface to implement to integrate CRaSH with an authentication mechanism:
public interface AuthenticationPlugin {
/**
* Returns the authentication plugin name.
*
* @return the plugin name
*/
String getName();
/**
* Returns true if the user is authentified by its username and password.
*
* @param username the username
* @param password the password
* @return true if authentication succeeded
* @throws Exception any exception that would prevent authentication to happen
*/
boolean authenticate(String username, String password) throws Exception;
}The integration as a CRaSH plugin mandates to extend the class CRaSHPlugin with the generic type AuthenticationPlugin:
public class SimpleAuthenticationPlugin extends
CRaSHPlugin<AuthenticationPlugin> implements
AuthenticationPlugin {
public String getName() {
return "simple";
}
@Override
public AuthenticationPlugin getImplementation() {
return this;
}
...
}The getName() method returns the simple value that matchs the crash.auth configuration property
The getImplementation() method returns the object that implements the AuthenticationPlugin class, this method is implemented from the CRaSHPlugin abstract class, in our case it simply returns this as the plugin and the implementation of AuthentionPlugin are the same class
Now let's study how the plugin retrieves the configuration properties crash.auth.simple.username and crash.auth.simple.password:
public class SimpleAuthenticationPlugin extends
CRaSHPlugin<AuthenticationPlugin> implements
AuthenticationPlugin {
public static final PropertyDescriptor<String> SIMPLE_USERNAME =
PropertyDescriptor.create(
"auth.simple.username",
"admin",
"The username");
public static final PropertyDescriptor<String> SIMPLE_PASSWORD =
PropertyDescriptor.create(
"auth.simple.password",
"admin",
"The password");
@Override
protected Iterable<PropertyDescriptor<?>> createConfigurationCapabilities() {
return Arrays.<PropertyDescriptor<?>>asList(
SIMPLE_USERNAME,
SIMPLE_PASSWORD);
}
private String username;
private String password;
@Override
public void init() {
PluginContext context = getContext();
this.username = context.getProperty(SIMPLE_USERNAME);
this.password = context.getProperty(SIMPLE_PASSWORD);
}
...
}The createConfigurationCapabilities() method returns the constants SIMPLE_USERNAME and SIMPLE_PASSWORD that defines the configuration properties that the plugin uses
The init() method is invoked by CRaSH before the plugin will be used, at this moment, the configuration properties are retrieved from the plugin context with the method getContext() available in the CRaSHPlugin base class
Finally the plugin needs to provide the authenticate() method that implement the authentication logic:
public boolean authenticate(String username, String password)
throws Exception {
return this.username != null &&
this.password != null &&
this.username.equals(username) &&
this.password.equals(password);
}The logic is straightforward with an equality check of the username and password.
Last but not least we must declare our plugin to make it recognized by CRaSH, this is achieved thanks to the java.util.ServiceLoader class. CRaSH uses the ServiceLoader for loading plugins and the loader needs a file to be present in the jar file containing the class under the name META-INF/services/org.crsh.plugin.CRaSHPlugin containing the class name of the plugin:
org.crsh.auth.SimpleAuthenticationPlugin
When all of this is done, the plugin and its service loader descriptor must be package in a jar file and available on the classpath of CRaSH.
You can learn more about the java.util.ServiceLoader by looking at the online javadoc
Table of Contents