HTTPS clone URL
Subversion checkout URL
Command Line tools for Administering the NorthScale Membase Server
Nothing to show
Fetching latest commit...
Cannot retrieve the latest commit at this time.
|Failed to load latest commit information.|
Membase Command Line Tool Set Specification High Level Description There will be a command line tool set that exists that allows users to have the same access to functions such as: * Starting and stopping of the server * The dynamic cluster membership changes of the NorthScale Membase Server * Query/fetch a given key * Joining a host:port * Listing nodes * Replication information * Set a given key/value Requirements: * Not only support the Northscale Membase Server, but could also support the Northscale Memcached Server since both of these products have a well-defined streaming REST (Representational State Transfer) interface. * Be able to point to any server node in the cluster to be able to manage the entire Membase cluster. * Be extensible. To add new functions or commands to the tool, there needs to be a way to easy add new functions with the least amount of changes to the overall tool as well as existing tools Mid-level Discription These tools will have some primary components: * Use the chain of responsibility design pattern. A good example of this can be found at: http://www.linuxtopia.org/online_books/programming_books/c++_practical_programming/c++_practical_programming_247.html Although this example is in C++, the basic concept will be possible to be implemented in Python. * These tools will be written in Python, using a framework which works best for extensibility * Top level source file: - parses user input to obtain * operation name * command line flags specific to that command of the desired operation. - has dictionary of functions keyed by the name which will be a "dispatch hash" * Source file with the implementation of functions for sending REST requests to the specified server and processing the JSON response. - PycURL library will provide the HTTP client functionality - Parsing of JSON into objects will be processed by the json module. The ouput of this processing can be used by the command line to display the JSON results of the REST command in a printable format http://pycurl.sourceforge.net/ http://docs.python.org/library/json.html For instance: * membase.py - contains parsing of command, dictionary of functions keyed by command string to dispatch appropriate method. * membase_cli_list_servers.py - contains the function that implements the servers listing functionality * membase_cli_$operation - other operations, as they are implemented * membase_cli_rest_client.py - contains the functions for : - converting json replies from the REST interface to something that can be printed out by the program for the given command - HTTP client interface for sending REST commands and recieving the JSON replies Operations/Functions: The following are operations that each function source file will implement: command as run by user REST command ---------------------------------------------------------- membase listservers GET /pools/default membase join POST /node/controller/doJoinCluster -c cluster (IP:port, host:port) -u user -p password membase eject POST /node/controller/ejectNode -c cluster (IP:port, host:port) -u user -p password membase listpools GET /pools/default -s servers [server,] membase listbuckets GET /pools/default/buckets/default -p pool [pool,] membase flush GET /pools/default/bucket/default/controller/doFlush membase stats GET /pools/default/stats?stat=$stat&period=$period -p pools [pool,] -s servers [server,] -b buckets membase bucket POST /pools/$pool/buckets (pool != default) -c create -n bucketname name=$bucketname -g get -p pool -d delete -f flush [bucket,] -s cachesize cacheSize=$cachesize membase log -g get -e entry text membase version membase Low-level Discription Chain of responsibility/dispatcher One crude idea is a big switch() statement to call the appropriate command or function. But the article above brings to light a more flexible and extensible model: using a Python dictionary of functions, keyed by the operation for whatever functions exist and to simple call the appropriate function/method based on the command (hash key, value being the function/method), without having to use a switch or interpolation of the command string into a numeric value tested by a switch statement. Each command implementation will exist in its own source file, and whatever happens to be imported into the top-level program, that command should simply run as it is keyed from the function dictionary. Networking Layer using REST The source file membase_cli_rest_client.py will contain the functions for sending REST requests to the server using PyCURL. Imported by each source file implementing a particular command, the programs that implment each REST action will have access to these functions/methods JSON parsing of REST replies membase_cli_rest_client.py will also contain the functions/methods for parsing the REST JSON replies into data objects that can be displayed to the user. Command line parsing The top level source file, membase.py, will contain the functionality for parsing user input to the main command line program.