riptano / cassanova Goto Github PK

This implements an obsolete Cassandra protocol in an obsolete version of Python. Do not use!

Cassanova is a fake Cassandra cluster. It runs in a single Python thread and is very lightweight for startup and shutdown- making it particularly useful for automated testing of Cassandra client interfaces and libraries.

Features:

• Passes most of the Cassandra Thrift system tests!
• Can mimic multiple nodes on different network interfaces
• Easy to start and stop
• Implemented as a Twisted Service, so it can share a process and reactor with your own Twisted code

What it does not support (yet?):

• Partitioners other than RandomPartitioner
• Indexes
• Column validators or custom comparators
• TTLs
• Consistency levels (there's only one actual data store between all "nodes")
• Write to disk (everything is lost when the process exits)
• CQL (durr)
• describe_splits (durr)
• Perform efficiently (data is stored according to what was easiest to write, not what would be most efficient for writes or lookups)

Requirements: Python (2.5 or higher), Twisted Matrix (8.0 or higher, I think), Thrift python library

The simplest way to run Cassanova is with the twistd tool. To run in the foreground and log to the console, listening on the default port (12379):

twistd -noy test.tap

To fork and run in the background, leaving a pid file:

twistd -y test.tap --pidfile whatever.pid --logfile cassanova.log

To change the listening port, set the $CASSANOVA_CLUSTER_PORT environment variable.

To run the test suite, have a Cassandra 0.7 source tree handy (I've been working against the 0.7.3 branch, so I know it works there), make sure the nose test coverage tool is installed, and run:

./run_tests.sh ~/path/to/cassandra-tree

All the real code is in cassanova.py. The more important classes are:

• CassanovaInterface: Provides the implementation of each of the supported Thrift calls. Corresponds to a single Thrift connection to the service.
• CassanovaNode: Corresponds to a single (fake) Cassandra node, listening on its own interface. You'll need to have a local network interface for each node you want to run.
• CassanovaService: The central service, which holds the schema and data. Corresponds to a whole cluster. The add_node method is used to instantiate new listening CassanovaNodes, given a unique network interface.

There is short explanation of the way data is stored in memory at the top of cassanova.py.