@InterfaceAudience.Public public interface BufferedMutator extends Closeable
Used to communicate with a single HBase table similar to Table
but meant for
batched, asynchronous puts. Obtain an instance from a Connection
and call
close()
afterwards. Customizations can be applied to the BufferedMutator
via
the BufferedMutatorParams
.
Exception handling with asynchronously via the BufferedMutator.ExceptionListener
.
The default implementation is to throw the exception upon receipt. This behavior can be
overridden with a custom implementation, provided as a parameter with
BufferedMutatorParams.listener(BufferedMutator.ExceptionListener)
.
Map/Reduce jobs are good use cases for using BufferedMutator
. Map/reduce jobs
benefit from batching, but have no natural flush point. BufferedMutator
receives the
puts from the M/R job and will batch puts based on some heuristic, such as the accumulated size
of the puts, and submit batches of puts asynchronously so that the M/R logic can continue
without interruption.
BufferedMutator
can also be used on more exotic circumstances. Map/Reduce batch jobs
will have a single BufferedMutator
per thread. A single BufferedMutator
can
also be effectively used in high volume online systems to batch puts, with the caveat that
extreme circumstances, such as JVM or machine failure, may cause some data loss.
NOTE: This class replaces the functionality that used to be available via
HTable#setAutoFlush(boolean) set to false
.
See also the BufferedMutatorExample
in the hbase-examples module.
ConnectionFactory
,
Connection
Modifier and Type | Interface and Description |
---|---|
static interface |
BufferedMutator.ExceptionListener
Listens for asynchronous exceptions on a
BufferedMutator . |
Modifier and Type | Field and Description |
---|---|
static String |
CLASSNAME_KEY
Key to use setting non-default BufferedMutator implementation in Configuration.
|
static long |
MIN_WRITE_BUFFER_PERIODIC_FLUSH_TIMERTICK_MS
Having the timer tick run more often that once every 100ms is needless and will
probably cause too many timer events firing having a negative impact on performance.
|
Modifier and Type | Method and Description |
---|---|
void |
close()
Performs a
flush() and releases any resources held. |
default void |
disableWriteBufferPeriodicFlush()
Disable periodic flushing of the write buffer.
|
void |
flush()
Executes all the buffered, asynchronous
Mutation operations and waits until they
are done. |
org.apache.hadoop.conf.Configuration |
getConfiguration()
Returns the
Configuration object used by this instance. |
TableName |
getName()
Gets the fully qualified table name instance of the table that this BufferedMutator writes to.
|
default long |
getWriteBufferPeriodicFlushTimeoutMs()
Returns the current periodic flush timeout value in milliseconds.
|
default long |
getWriteBufferPeriodicFlushTimerTickMs()
Returns the current periodic flush timertick interval in milliseconds.
|
long |
getWriteBufferSize()
Returns the maximum size in bytes of the write buffer for this HTable.
|
void |
mutate(List<? extends Mutation> mutations)
Send some
Mutation s to the table. |
void |
mutate(Mutation mutation)
Sends a
Mutation to the table. |
void |
setOperationTimeout(int timeout)
Set operation timeout for this mutator instance
|
void |
setRpcTimeout(int timeout)
Set rpc timeout for this mutator instance
|
default void |
setWriteBufferPeriodicFlush(long timeoutMs)
Sets the maximum time before the buffer is automatically flushed checking once per second.
|
default void |
setWriteBufferPeriodicFlush(long timeoutMs,
long timerTickMs)
Sets the maximum time before the buffer is automatically flushed.
|
static final String CLASSNAME_KEY
static final long MIN_WRITE_BUFFER_PERIODIC_FLUSH_TIMERTICK_MS
TableName getName()
org.apache.hadoop.conf.Configuration getConfiguration()
Configuration
object used by this instance.
The reference returned is not a copy, so any change made to it will affect this instance.
void mutate(Mutation mutation) throws IOException
Mutation
to the table. The mutations will be buffered and sent over the
wire as part of a batch. Currently only supports Put
and Delete
mutations.mutation
- The data to send.IOException
- if a remote or network exception occurs.void mutate(List<? extends Mutation> mutations) throws IOException
Mutation
s to the table. The mutations will be buffered and sent over the
wire as part of a batch. There is no guarantee of sending entire content of mutations
in a single batch; it will be broken up according to the write buffer capacity.mutations
- The data to send.IOException
- if a remote or network exception occurs.void close() throws IOException
flush()
and releases any resources held.close
in interface AutoCloseable
close
in interface Closeable
IOException
- if a remote or network exception occurs.void flush() throws IOException
Mutation
operations and waits until they
are done.IOException
- if a remote or network exception occurs.default void setWriteBufferPeriodicFlush(long timeoutMs)
timeoutMs
- The maximum number of milliseconds how long records may be buffered
before they are flushed. Set to 0 to disable.default void setWriteBufferPeriodicFlush(long timeoutMs, long timerTickMs)
timeoutMs
- The maximum number of milliseconds how long records may be buffered
before they are flushed. Set to 0 to disable.timerTickMs
- The number of milliseconds between each check if the
timeout has been exceeded. Must be 100ms (as defined in
MIN_WRITE_BUFFER_PERIODIC_FLUSH_TIMERTICK_MS
)
or larger to avoid performance problems.default void disableWriteBufferPeriodicFlush()
default long getWriteBufferPeriodicFlushTimeoutMs()
default long getWriteBufferPeriodicFlushTimerTickMs()
long getWriteBufferSize()
The default value comes from the configuration parameter hbase.client.write.buffer
.
void setRpcTimeout(int timeout)
void setOperationTimeout(int timeout)
Copyright © 2007–2019 The Apache Software Foundation. All rights reserved.