(In Progress)
- Include table indexes in
KeyspaceMetadata.export_as_string()
Jan 29, 2014
- Prevent leak of Scheduler thread (even with proper shutdown)
- Correctly handle ignored hosts, which are common with the DCAwareRoundRobinPolicy
- Hold strong reference to prepared statement while executing it to avoid garbage collection
- Add NullHandler logging handler to the cassandra package to avoid warnings about there being no configured logger
- Fix bad handling of nodes that have been removed from the cluster
- Properly escape string types within cql collections
- Handle setting the same keyspace twice in a row
- Avoid race condition during schema agreement checks that could result in schema update queries returning before all nodes had seen the change
- Preserve millisecond-level precision in datetimes when performing inserts with simple (non-prepared) statements
- Properly defunct connections when libev reports an error by setting errno instead of simply logging the error
- Fix endless hanging of some requests when using the libev reactor
- Always start a reconnection process when we fail to connect to a newly bootstrapped node
- Generators map to CQL lists, not key sequences
- Always defunct connections when an internal operation fails
- Correctly break from handle_write() if nothing was sent (asyncore reactor only)
- Avoid potential double-erroring of callbacks when a connection becomes defunct
- Add default query timeout to
Session - Add timeout parameter to
Session.execute() - Add
WhiteListRoundRobinPolicyas a load balancing policy option - Support for consistency level
LOCAL_ONE - Make the backoff for fetching traces exponentially increasing and configurable
- Raise Exception if
TokenAwarePolicyis used against a cluster using theMurmur3Partitionerif the murmur3 C extension has not been compiled - Add encoder mapping for
OrderedDict - Use timeouts on all control connection queries
- Benchmark improvements, including command line options and eay multithreading support
- Reduced lock contention when using the asyncore reactor
- Warn when non-datetimes are used for 'timestamp' column values in prepared statements
- Add requirements.txt and test-requirements.txt
- TravisCI integration for running unit tests against Python 2.6, Python 2.7, and PyPy
Nov 12, 2013
This release makes many stability improvements, especially around prepared statements and node failure handling. In particular, several cases where a request would never be completed (and as a result, leave the application hanging) have been resolved.
- Add timeout kwarg to
ResponseFuture.result() - Create connection pools to all hosts in parallel when initializing new Sesssions.
- Properly set exception on ResponseFuture when a query fails against all hosts
- Improved cleanup and reconnection efforts when reconnection fails on a node that has recently come up
- Use correct consistency level when retrying failed operations against a different host. (An invalid consistency level was being used, causing the retry to fail.)
- Better error messages for failed
Session.prepare()opertaions - Prepare new statements against all hosts in parallel (formerly sequential)
- Fix failure to save the new current keyspace on connections. (This could cause problems for prepared statements and lead to extra operations to continuously re-set the keyspace.)
- Avoid sharing
LoadBalancingPoliciesacrossClusterinstances. (When a secondClusterwas connected, it effectively mark nodes down for the firstCluster.) - Better handling of failures during the re-preparation sequence for unrecognized prepared statements
- Throttle trashing of underutilized connections to avoid trashing newly created connections
- Fix race condition which could result in trashed connections being closed before the last operations had completed
- Avoid preparing statements on the event loop thread (which could lead to deadlock)
- Correctly mark up non-contact point nodes discovered by the control connection. (This lead to prepared statements not being prepared against those hosts, generating extra traffic later when the statements were executed and unrecognized.)
- Correctly handle large messages through libev
- Add timeout to schema agreement check queries
- More complete (and less contended) locking around manipulation of the pending message deque for libev connections
- Prepare statements in batches of 10. (When many prepared statements are in use, this allows the driver to start utilizing nodes that were restarted more quickly.)
- Better debug logging around connection management
- Don't retain unreferenced prepared statements in the local cache. (If many different prepared statements were created, this would increase memory usage and greatly increase the amount of time required to begin utilizing a node that was added or marked up.)
Oct 22, 2013
- Use lazy string formatting when logging
- Avoid several deadlock scenarios, especially when nodes go down
- Avoid trashing newly created connections due to insufficient traffic
- Gracefully handle un-handled Exceptions when erroring callbacks
- Node state listeners (which are called when a node is added, removed, goes down, or comes up) should now be registered through Cluster.register_listener() instead of through a host's HealthMonitor (which has been removed)
Oct 10, 2013
- SSL support
- Avoid KeyError when building replica map for NetworkTopologyStrategy
- Work around python bug which causes deadlock when a thread imports the utf8 module
- Handle no blist library, which is not compatible with pypy
- Avoid deadlock triggered by a keyspace being set on a connection (which may happen automatically for new connections)
- Switch packaging from Distribute to setuptools, improved C extension support
- Use PEP 386 compliant beta and post-release versions
Sep 24, 2013
- Handle new blob syntax in Cassandra 2.0 by accepting bytearray objects for blob values
- Add cql_version kwarg to Cluster.__init__
- Fix KeyError when building token map with NetworkTopologyStrategy keyspaces (this prevented a Cluster from successfully connecting at all).
- Don't lose default consitency level from parent PreparedStatement when creating BoundStatements
Sep 20, 2013
- Support for LZ4 compression (Cassandra 2.0+)
- Token-aware routing will now utilize all replicas for a query instead of just the first replica
- Fix libev include path for CentOS
- Fix varint packing of the value 0
- Correctly pack unicode values
- Don't attempt to return failed connections to the pool when a final result is set
- Fix bad iteration of connection credentials
- Use blist's orderedset for set collections and OrderedDict for map collections so that Cassandra's ordering is preserved
- Fix connection failure on Windows due to unavailability of inet_pton and inet_ntop. (Note that IPv6 inet_address values are still not supported on Windows.)
- Boolean constants shouldn't be surrounded by single quotes
- Avoid a potential loss of precision on float constants due to string formatting
- Actually utilize non-standard ports set on Cluster objects
- Fix export of schema as a set of CQL queries
- Use cStringIO for connection buffer for better performance
- Add __repr__ method for Statement classes
- Raise InvalidTypeParameterError when parameters of the wrong type are used with statements
- Make all tests compatible with Python 2.6
- Add 1s timeout for opening new connections
Aug 19, 2013
- Fix pip packaging
Aug 16, 2013
Initial release