Database.Cassandra.Basic

Low-level functionality for working with Cassandra at the most basic level.

Connection

type CPool = Pool Cassandra

A round-robin pool of cassandra connections

type Server = (HostName, Int)

A (ServerName, Port) tuple

defServer :: Server

A localhost server with default configuration

defServers :: [Server]

A single localhost server with default configuration

type KeySpace = String

createCassandraPool

Arguments

:: [Server]

List of servers to connect to

-> Int

Number of stripes to maintain

-> Int

Max connections per stripe

-> NominalDiffTime

Kill each connection after this many seconds

-> KeySpace

Each pool operates on a single KeySpace

-> IO CPool

Create a pool of connections to a cluster of Cassandra boxes

Each box in the cluster will get up to n connections. The pool will send queries in round-robin fashion to balance load on each box in the cluster.

MonadCassandra Typeclass

class MonadIO m => MonadCassandra m where

All Cassy operations are designed to run inside MonadCassandra context.

We provide a default concrete Cas datatype, but you can simply make your own application monads an instance of MonadCassandra for conveniently using all operations of this package.

Please keep in mind that all Cassandra operations may raise CassandraExceptions at any point in time.

Methods

getCassandraPool :: m CPool

Instances

MonadIO m => MonadCassandra (ReaderT CPool m)

type Cas a = ReaderT CPool IO a

runCas :: CPool -> Cas a -> IO a

Main running function when using the ad-hoc Cas monad. Just write your cassandra actions within the Cas monad and supply them with a CPool to execute.

transCas :: MonadCassandra m => Cas a -> m (IO a)

Unwrap a Cassandra action and return an IO continuation that can then be run in a pure IO context.

This is useful when you design all your functions in a generic form with MonadCassandra m constraints and then one day need to feed your function to a utility that can only run in an IO context. This function is then your friendly utility for extracting an IO action.

mapCassandra :: (Traversable t, MonadCassandra m) => t (Cas b) -> m (t b)

Run a list of cassandra computations in parallel using the async library

Cassandra Operations

getCol

Arguments

:: (MonadCassandra m, CasType k)

=> ColumnFamily

-> ByteString

Row key

-> k

Column/SuperColumn key; see CasType for what it can be. Use ByteString in the simple case.

-> ConsistencyLevel

Read quorum

-> m (Maybe Column)

Get a single key-column value.

get

Arguments

:: MonadCassandra m

=> ColumnFamily

in ColumnFamily

-> ByteString

Row key to get

-> Selector

Slice columns with selector

-> ConsistencyLevel

-> m [Column]

An arbitrary get operation - slice with Selector

getMulti

Arguments

:: MonadCassandra m

=> ColumnFamily

-> KeySelector

A selection of rows to fetch in one hit

-> Selector

Subject to column selector conditions

-> ConsistencyLevel

-> m (Map ByteString Row)

A Map from Row keys to Rows is returned

Do multiple gets in one DB hit

insert

Arguments

:: MonadCassandra m

=> ColumnFamily

-> ByteString

Row key

-> ConsistencyLevel

-> [Column]

best way to make these columns is through packCol

-> m ()

Insert an entire row into the db.

This will do as many round-trips as necessary to insert the full row. Please keep in mind that each column and each column of each super-column is sent to the server one by one.

 insert "testCF" "row1" ONE [packCol ("column key", "some column content")]

delete

Arguments

:: MonadCassandra m

=> ColumnFamily

In ColumnFamily

-> Key

Key to be deleted

-> Selector

Columns to be deleted

-> ConsistencyLevel

-> m ()

Delete an entire row, specific columns or a specific sub-set of columns within a SuperColumn.

Retrying Queries

retryCas

Arguments

:: (MonadCatch m, MonadIO m)

=> RetryPolicy

For default settings, just use def

-> m a

Action to perform

-> m a

retrying with direct cassandra support. Server-related failures will be retried.

UnavailableException, TimedOutException and SchemaDisagreementException will be automatically retried.

casRetryH :: Monad m => Int -> Handler m Bool

Exception handler that returns True for errors that may be resolved after a retry. So they are good candidates for retrying queries.

networkRetryH :: Monad m => Int -> Handler m Bool

IOExceptions should be retried

Filtering

data Selector

A column selector/filter statement for queries.

Remember that SuperColumns are always fully deserialized, so we don't offer a way to filter columns within a SuperColumn.

Column names and ranges are specified by any type that can be packed into a Cassandra column using the CasType typeclass.

Constructors

All

Return everything in Row

forall a . CasType a => ColNames [a]

Return specific columns or super-columns depending on the ColumnFamily

forall a b . (CasType a, CasType b) => SupNames a [b]

When deleting specific columns in a super column

forall a b . (CasType a, CasType b) => Range

Return a range of columns or super-columns.

Fields

rangeStart :: Maybe a

rangeEnd :: Maybe b

rangeOrder :: Order

rangeLimit :: Int32

Instances

Show Selector

Default Selector

range :: Selector

A default starting point for range Selector. Use this so you don't run into ambiguous type variables when using Nothing.

 range = Range (Nothing :: Maybe ByteString) (Nothing :: Maybe ByteString) Regular 1024

boundless :: Maybe ByteString

data Order

Order in a range query

Constructors

Regular

Reversed

Instances

Show Order

reverseOrder :: Order -> Order

data KeySelector

A Key range selector to use with getMulti.

Constructors

Keys [Key]

Just a list of keys to get

KeyRange KeyRangeType Key Key Int32

A range of keys to get. Remember that RandomPartitioner ranges may not mean much as keys are randomly assigned to nodes.

Instances

Show KeySelector

data KeyRangeType

Encodes the Key vs. Token options in the thrift API.

InclusiveRange ranges are just plain intuitive range queries. WrapAround ranges are also inclusive, but they wrap around the ring.

Constructors

InclusiveRange

WrapAround

Instances

Show KeyRangeType

Exceptions

data CassandraException

Constructors

NotFoundException

InvalidRequestException String

UnavailableException

TimedOutException

AuthenticationException String

AuthorizationException String

SchemaDisagreementException

ConversionException String

OperationNotSupported String

Instances

Eq CassandraException

Data CassandraException

Ord CassandraException

Read CassandraException

Show CassandraException

Typeable CassandraException

Exception CassandraException

Utility

getTime :: IO Int64

Cassandra is VERY sensitive to its timestamp values. As a convention, timestamps are always in microseconds

throwing :: IO (Either CassandraException a) -> IO a

Make exceptions implicit.

wrapException :: IO a -> IO a

Wrap exceptions of the underlying thrift library into the exception types defined here.

Basic Types

type ColumnFamily = String

type Key = ByteString

type ColumnName = ByteString

type Value = ByteString

data Column

A Column is either a single key-value pair or a SuperColumn with an arbitrary number of key-value pairs

Constructors

SuperColumn ColumnName [Column]

Column

Fields

colKey :: ColumnName

colVal :: Value

colTS :: Maybe Int64

Last update timestamp; will be overridden during write/update ops

colTTL :: Maybe Int32

A TTL after which Cassandra will erase the column

Instances

Eq Column

Ord Column

Read Column

Show Column

col :: ByteString -> ByteString -> Column

A short-hand for creating key-value Column values. This is pretty low level; you probably want to use packCol.

packCol :: CasType k => (k, ByteString) -> Column

Pack key-value pair into Column form ready to be written to Cassandra

unpackCol :: CasType k => Column -> (k, Value)

Unpack a Cassandra Column into a more convenient (k,v) form

packKey :: CasType a => a -> ByteString

Pack a column key into binary, ready for submission to Cassandra

type Row = [Column]

A full row is simply a sequence of columns

data ConsistencyLevel

Constructors

ONE

QUORUM

LOCAL_QUORUM

EACH_QUORUM

ALL

ANY

TWO

THREE

Instances

Enum ConsistencyLevel

Eq ConsistencyLevel

Ord ConsistencyLevel

Show ConsistencyLevel

Typeable ConsistencyLevel

Helpers

class CKey a where

A typeclass to enable using any string-like type for row and column keys

Methods

toColKey :: a -> ByteString

fromColKey :: ByteString -> Either String a

Instances

CKey String

CKey ByteString

CKey ByteString

CKey Text

CKey Text

CKey [ByteString]

For easy composite keys, just serialize your data type to a list of bytestrings, we'll concat them and turn them into column keys.

fromColKey' :: CKey a => ByteString -> a

Raise an error if conversion fails

Cassandra Column Key Types

module Database.Cassandra.Pack