Brutex/cavis
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
cavis/deeplearning4j/deeplearning4j-nlp-parent/deeplearning4j-nlp/src/main/resources/org/deeplearning4j/ehcache.xml

712 lines
35 KiB
XML
Raw Normal View History

Eclipse Migration Initial Commit
2019-06-06 15:21:15 +03:00
<!--~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
~ Copyright (c) 2015-2018 Skymind, Inc.
~
~ This program and the accompanying materials are made available under the
~ terms of the Apache License, Version 2.0 which is available at
~ https://www.apache.org/licenses/LICENSE-2.0.
~
~ Unless required by applicable law or agreed to in writing, software
~ distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
~ WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
~ License for the specific language governing permissions and limitations
~ under the License.
~
~ SPDX-License-Identifier: Apache-2.0
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~-->
<!--
CacheManager Configuration
==========================
An ehcache.xml corresponds to a single CacheManager.
See instructions below or the ehcache schema (ehcache.xsd) on how to configure.
System property tokens can be specified in this file which are replaced when the configuration
is loaded. For example multicastGroupPort=${multicastGroupPort} can be replaced with the
System property either from an environment variable or a system property specified with a
command line switch such as -DmulticastGroupPort=4446. Another example, useful for Terracotta
server based deployments is <terracottaConfig url="${serverAndPort}"/ and specify a command line
switch of -Dserver36:9510
The attributes of <ehcache> are:
* name - an optional name for the CacheManager. The name is optional and primarily used
for documentation or to distinguish Terracotta clustered cache state. With Terracotta
clustered caches, a combination of CacheManager name and cache name uniquely identify a
particular cache store in the Terracotta clustered memory.
* updateCheck - an optional boolean flag specifying whether this CacheManager should check
for new versions of Ehcache over the Internet. If not specified, updateCheck="true".
* dynamicConfig - an optional setting that can be used to disable dynamic configuration of caches
associated with this CacheManager. By default this is set to true - i.e. dynamic configuration
is enabled. Dynamically configurable caches can have their TTI, TTL and maximum disk and
in-memory capacity changed at runtime through the cache's configuration object.
* monitoring - an optional setting that determines whether the CacheManager should
automatically register the SampledCacheMBean with the system MBean server.
Currently, this monitoring is only useful when using Terracotta clustering and using the
Terracotta Developer Console. With the "autodetect" value, the presence of Terracotta clustering
will be detected and monitoring, via the Developer Console, will be enabled. Other allowed values
are "on" and "off". The default is "autodetect". This setting does not perform any function when
used with JMX monitors.
* maxBytesLocalHeap - optional setting that constraints the memory usage of the Caches managed by the CacheManager
to use at most the specified number of bytes of the local VM's heap.
* maxBytesLocalOffHeap - optional setting that constraints the offHeap usage of the Caches managed by the CacheManager
to use at most the specified number of bytes of the local VM's offHeap memory.
* maxBytesLocalDisk - optional setting that constraints the disk usage of the Caches managed by the CacheManager
to use at most the specified number of bytes of the local disk.
These settings let you define "resource pools", caches will share. For instance setting maxBytesLocalHeap to 100M, will result in
all caches sharing 100 MegaBytes of ram. The CacheManager will balance these 100 MB across all caches based on their respective usage
patterns. You can allocate a precise amount of bytes to a particular cache by setting the appropriate maxBytes* attribute for that cache.
That amount will be subtracted from the CacheManager pools, so that if a cache a specified 30M requirement, the other caches will share
the remaining 70M.
Also, specifying a maxBytesLocalOffHeap at the CacheManager level will result in overflowToOffHeap to be true by default. If you don't want
a specific cache to overflow to off heap, you'll have to set overflowToOffHeap="false" explicitly
Here is an example of CacheManager level resource tuning, which will use up to 400M of heap and 2G of offHeap:
<ehcache xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:noNamespaceSchemaLocation="ehcache.xsd"
updateCheck="true" monitoring="autodetect"
dynamicConfig="true" maxBytesLocalHeap="400M" maxBytesLocalOffHeap="2G">
-->
<ehcache xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:noNamespaceSchemaLocation="http://ehcache.org/ehcache.xsd" updateCheck="true" monitoring="autodetect" dynamicConfig="true">
<!--
DiskStore configuration
=======================
The diskStore element is optional. To turn off disk store path creation, comment out the diskStore
element below.
Configure it if you have disk persistence enabled for any cache or if you use
unclustered indexed search.
If it is not configured, and a cache is created which requires a disk store, a warning will be
issued and java.io.tmpdir will automatically be used.
diskStore has only one attribute - "path". It is the path to the directory where
any required disk files will be created.
If the path is one of the following Java System Property it is replaced by its value in the
running VM. For backward compatibility these should be specified without being enclosed in the ${token}
replacement syntax.
The following properties are translated:
* user.home - User's home directory
* user.dir - User's current working directory
* java.io.tmpdir - Default temp file path
* ehcache.disk.store.dir - A system property you would normally specify on the command line
e.g. java -Dehcache.disk.store.dir=/u01/myapp/diskdir ...
Subdirectories can be specified below the property e.g. java.io.tmpdir/one
-->
<diskStore path="."/>
<!--
TransactionManagerLookup configuration
======================================
This class is used by ehcache to lookup the JTA TransactionManager use in the application
using an XA enabled ehcache. If no class is specified then DefaultTransactionManagerLookup
will find the TransactionManager in the following order
*GenericJNDI (i.e. jboss, where the property jndiName controls the name of the
TransactionManager object to look up)
*Bitronix
*Atomikos
You can provide you own lookup class that implements the
net.sf.ehcache.transaction.manager.TransactionManagerLookup interface.
-->
<transactionManagerLookup class="net.sf.ehcache.transaction.manager.DefaultTransactionManagerLookup" properties="jndiName=java:/TransactionManager" propertySeparator=";"/>
<!--
CacheManagerEventListener
=========================
Specifies a CacheManagerEventListenerFactory which is notified when Caches are added
or removed from the CacheManager.
The attributes of CacheManagerEventListenerFactory are:
* class - a fully qualified factory class name
* properties - comma separated properties having meaning only to the factory.
Sets the fully qualified class name to be registered as the CacheManager event listener.
The events include:
* adding a Cache
* removing a Cache
Callbacks to listener methods are synchronous and unsynchronized. It is the responsibility
of the implementer to safely handle the potential performance and thread safety issues
depending on what their listener is doing.
If no class is specified, no listener is created. There is no default.
-->
<cacheManagerEventListenerFactory class="" properties=""/>
<!--
TerracottaConfig
========================
(Enable for Terracotta clustered operation)
Note: You need to install and run one or more Terracotta servers to use Terracotta clustering.
See http://www.terracotta.org/web/display/orgsite/Download.
Specifies a TerracottaConfig which will be used to configure the Terracotta
runtime for this CacheManager.
Configuration can be specified in two main ways: by reference to a source of
configuration or by use of an embedded Terracotta configuration file.
To specify a reference to a source (or sources) of configuration, use the url
attribute. The url attribute must contain a comma-separated list of:
* path to Terracotta configuration file (usually named tc-config.xml)
* URL to Terracotta configuration file
* <server host>:<port> of running Terracotta Server instance
Simplest example for pointing to a Terracotta server on this machine:
<terracottaConfig url="localhost:9510"/>
This element has two attributes "rejoin" and "wanEnabledTSA", which can take values of either "true" or "false":
<terracottaConfig rejoin="true" wanEnabledTSA="true" url="localhost:9510" />
By default, these attributes are false.
Without rejoin, if the Terracotta Server is restarted the client cannot connect back to the
server. When enabled, this allows the client to connect to the new cluster without the
need to restart the node.
When wanEnabledTSA is true, the client will wait for the WAN Orchestrator to provide the
list of WAN enabled caches. Once the Orchestrator is up and running then the client will proceed
to create the clustered data structures.
Example using a path to Terracotta configuration file:
<terracottaConfig url="/app/config/tc-config.xml"/>
Example using a URL to a Terracotta configuration file:
<terracottaConfig url="http://internal/ehcache/app/tc-config.xml"/>
Example using multiple Terracotta server instance URLs (for fault tolerance):
<terracottaConfig url="host1:9510,host2:9510,host3:9510"/>
To embed a Terracotta configuration file within the ehcache configuration, simply
place a normal Terracotta XML config within the <terracottaConfig> element.
Example:
<terracottaConfig>
<tc-config>
<servers>
<server host="server1" name="s1"/>
<server host="server2" name="s2"/>
</servers>
<clients>
<logs>app/logs-%i</logs>
</clients>
</tc-config>
</terracottaConfig>
For more information on the Terracotta configuration, see the Terracotta documentation.
-->
<terracottaConfig url="localhost:9510"/>
<!--
Cache configuration
===================
The following attributes are required.
name:
Sets the name of the cache. This is used to identify the cache. It must be unique.
maxEntriesLocalHeap:
Sets the maximum number of objects that will be created in memory. 0 = no limit.
In practice no limit means Integer.MAX_SIZE (2147483647) unless the cache is distributed
with a Terracotta server in which case it is limited by resources.
maxEntriesLocalDisk:
Sets the maximum number of objects that will be maintained in the DiskStore
The default value is zero, meaning unlimited.
eternal:
Sets whether elements are eternal. If eternal, timeouts are ignored and the
element is never expired.
The following attributes and elements are optional.
maxEntriesInCache:
This feature is applicable only to Terracotta distributed caches.
Sets the maximum number of entries that can be stored in the cluster. 0 = no limit.
Note that clustered cache will still perform eviction if resource usage requires it.
This property can be modified dynamically while the cache is operating.
overflowToOffHeap:
(boolean) This feature is available only in enterprise versions of Ehcache.
When set to true, enables the cache to utilize off-heap memory
storage to improve performance. Off-heap memory is not subject to Java
GC. The default value is false.
maxBytesLocalHeap:
Defines how many bytes the cache may use from the VM's heap. If a CacheManager
maxBytesLocalHeap has been defined, this Cache's specified amount will be
subtracted from the CacheManager. Other caches will share the remainder.
This attribute's values are given as <number>k|K|m|M|g|G for
kilobytes (k|K), megabytes (m|M), or gigabytes (g|G).
For example, maxBytesLocalHeap="2g" allots 2 gigabytes of heap memory.
If you specify a maxBytesLocalHeap, you can't use the maxEntriesLocalHeap attribute.
maxEntriesLocalHeap can't be used if a CacheManager maxBytesLocalHeap is set.
Elements put into the cache will be measured in size using net.sf.ehcache.pool.sizeof.SizeOf
If you wish to ignore some part of the object graph, see net.sf.ehcache.pool.sizeof.annotations.IgnoreSizeOf
maxBytesLocalOffHeap:
This feature is available only in enterprise versions of Ehcache.
Sets the amount of off-heap memory this cache can use, and will reserve.
This setting will set overflowToOffHeap to true. Set explicitly to false to disable overflow behavior.
Note that it is recommended to set maxEntriesLocalHeap to at least 100 elements
when using an off-heap store, otherwise performance will be seriously degraded,
and a warning will be logged.
The minimum amount that can be allocated is 128MB. There is no maximum.
maxBytesLocalDisk:
As for maxBytesLocalHeap, but specifies the limit of disk storage this cache will ever use.
timeToIdleSeconds:
Sets the time to idle for an element before it expires.
i.e. The maximum amount of time between accesses before an element expires
Is only used if the element is not eternal.
Optional attribute. A value of 0 means that an Element can idle for infinity.
The default value is 0.
timeToLiveSeconds:
Sets the time to live for an element before it expires.
i.e. The maximum time between creation time and when an element expires.
Is only used if the element is not eternal.
Optional attribute. A value of 0 means that and Element can live for infinity.
The default value is 0.
diskExpiryThreadIntervalSeconds:
The number of seconds between runs of the disk expiry thread. The default value
is 120 seconds.
diskSpoolBufferSizeMB:
This is the size to allocate the DiskStore for a spool buffer. Writes are made
to this area and then asynchronously written to disk. The default size is 30MB.
Each spool buffer is used only by its cache. If you get OutOfMemory errors consider
lowering this value. To improve DiskStore performance consider increasing it. Trace level
logging in the DiskStore will show if put back ups are occurring.
clearOnFlush:
whether the MemoryStore should be cleared when flush() is called on the cache.
By default, this is true i.e. the MemoryStore is cleared.
memoryStoreEvictionPolicy:
Policy would be enforced upon reaching the maxEntriesLocalHeap limit. Default
policy is Least Recently Used (specified as LRU). Other policies available -
First In First Out (specified as FIFO) and Less Frequently Used
(specified as LFU)
copyOnRead:
Whether an Element is copied when being read from a cache.
By default this is false.
copyOnWrite:
Whether an Element is copied when being added to the cache.
By default this is false.
Cache persistence is configured through the persistence sub-element. The attributes of the
persistence element are:
strategy:
Configures the type of persistence provided by the configured cache. This must be one of the
following values:
* localRestartable - Enables the RestartStore and copies all cache entries (on-heap and/or off-heap)
to disk. This option provides fast restartability with fault tolerant cache persistence on disk.
It is available for Enterprise Ehcache users only.
* localTempSwap - Swaps cache entries (on-heap and/or off-heap) to disk when the cache is full.
"localTempSwap" is not persistent.
* none - Does not persist cache entries.
* distributed - Defers to the <terracotta> configuration for persistence settings. This option
is not applicable for standalone.
synchronousWrites:
When set to true write operations on the cache do not return until after the operations data has been
successfully flushed to the disk storage. This option is only valid when used with the "localRestartable"
strategy, and defaults to false.
The following example configuration shows a cache configured for localTempSwap restartability.
<cache name="persistentCache" maxEntriesLocalHeap="1000">
<persistence strategy="localTempSwap"/>
</cache>
Cache elements can also contain sub elements which take the same format of a factory class
and properties. Defined sub-elements are:
* cacheEventListenerFactory - Enables registration of listeners for cache events, such as
put, remove, update, and expire.
* bootstrapCacheLoaderFactory - Specifies a BootstrapCacheLoader, which is called by a
cache on initialisation to prepopulate itself.
* cacheExtensionFactory - Specifies a CacheExtension, a generic mechanism to tie a class
which holds a reference to a cache to the cache lifecycle.
* cacheExceptionHandlerFactory - Specifies a CacheExceptionHandler, which is called when
cache exceptions occur.
* cacheLoaderFactory - Specifies a CacheLoader, which can be used both asynchronously and
synchronously to load objects into a cache. More than one cacheLoaderFactory element
can be added, in which case the loaders form a chain which are executed in order. If a
loader returns null, the next in chain is called.
* copyStrategy - Specifies a fully qualified class which implements
net.sf.ehcache.store.compound.CopyStrategy. This strategy will be used for copyOnRead
and copyOnWrite in place of the default which is serialization.
Example of cache level resource tuning:
<cache name="memBound" maxBytesLocalHeap="100m" maxBytesLocalOffHeap="4g" maxBytesLocalDisk="200g" />
Cache Event Listeners
+++++++++++++++++++++
All cacheEventListenerFactory elements can take an optional property listenFor that describes
which events will be delivered in a clustered environment. The listenFor attribute has the
following allowed values:
* all - the default is to deliver all local and remote events
* local - deliver only events originating in the current node
* remote - deliver only events originating in other nodes
Example of setting up a logging listener for local cache events:
<cacheEventListenerFactory class="my.company.log.CacheLogger"
listenFor="local" />
Search
++++++
A <cache> can be made searchable by adding a <searchable/> sub-element. By default the keys
and value objects of elements put into the cache will be attributes against which
queries can be expressed.
<cache>
<searchable/>
</cache>
An "attribute" of the cache elements can also be defined to be searchable. In the example below
an attribute with the name "age" will be available for use in queries. The value for the "age"
attribute will be computed by calling the method "getAge()" on the value object of each element
in the cache. See net.sf.ehcache.search.attribute.ReflectionAttributeExtractor for the format of
attribute expressions. Attribute values must also conform to the set of types documented in the
net.sf.ehcache.search.attribute.AttributeExtractor interface
<cache>
<searchable>
<searchAttribute name="age" expression="value.getAge()"/>
</searchable>
</cache>
Attributes may also be defined using a JavaBean style. With the following attribute declaration
a public method getAge() will be expected to be found on either the key or value for cache elements
<cache>
<searchable>
<searchAttribute name="age"/>
</searchable>
</cache>
In more complex situations you can create your own attribute extractor by implementing the
AttributeExtractor interface. Providing your extractor class is shown in the following example:
<cache>
<searchable>
<searchAttribute name="age" class="com.example.MyAttributeExtractor"/>
</searchable>
</cache>
Use properties to pass state to your attribute extractor if needed. Your implementation must provide
a public constructor that takes a single java.movingwindow.Properties instance
<cache>
<searchable>
<searchAttribute name="age" class="com.example.MyAttributeExtractor" properties="foo=1,bar=2"/>
</searchable>
</cache>
If you intend to use dynamic attribute extraction (see net.sf.ehcache.Cache.registerDynamicAttributesExtractor) then
you need to enable it as follows:
<cache>
<searchable allowDynamicIndexing="true"/>
</cache>
Cache Exception Handling
++++++++++++++++++++++++
By default, most cache operations will propagate a runtime CacheException on failure. An
interceptor, using a dynamic proxy, may be configured so that a CacheExceptionHandler can
be configured to intercept Exceptions. Errors are not intercepted.
It is configured as per the following example:
<cacheExceptionHandlerFactory class="com.example.ExampleExceptionHandlerFactory"
properties="logLevel=FINE"/>
Caches with ExceptionHandling configured are not of type Cache, but are of type Ehcache only,
and are not available using CacheManager.getCache(), but using CacheManager.getEhcache().
Cache Loader
++++++++++++
A default CacheLoader may be set which loads objects into the cache through asynchronous and
synchronous methods on Cache. This is different to the bootstrap cache loader, which is used
only in distributed caching.
It is configured as per the following example:
<cacheLoaderFactory class="com.example.ExampleCacheLoaderFactory"
properties="type=int,startCounter=10"/>
Element value comparator
++++++++++++++++++++++++
These two cache atomic methods:
removeElement(Element e)
replace(Element old, Element element)
rely on comparison of cached elements value. The default implementation relies on Object.equals()
but that can be changed in case you want to use a different way to compute equality of two elements.
This is configured as per the following example:
<elementValueComparator class="com.company.xyz.MyElementComparator"/>
The MyElementComparator class must implement the is net.sf.ehcache.store.ElementValueComparator
interface. The default implementation is net.sf.ehcache.store.DefaultElementValueComparator.
SizeOf Policy
+++++++++++++
Control how deep the SizeOf engine can go when sizing on-heap elements.
This is configured as per the following example:
<sizeOfPolicy maxDepth="100" maxDepthExceededBehavior="abort"/>
maxDepth controls how many linked objects can be visited before the SizeOf engine takes any action.
maxDepthExceededBehavior specifies what happens when the max depth is exceeded while sizing an object graph.
"continue" makes the SizeOf engine log a warning and continue the sizing. This is the default.
"abort" makes the SizeOf engine abort the sizing, log a warning and mark the cache as not correctly tracking
memory usage. This makes Ehcache.hasAbortedSizeOf() return true when this happens.
The SizeOf policy can be configured at the cache manager level (directly under <ehcache>) and at
the cache level (under <cache> or <defaultCache>). The cache policy always overrides the cache manager
one if both are set. This element has no effect on distributed caches.
Transactions
++++++++++++
To enable an ehcache as transactions, set the transactionalMode
transactionalMode="xa" - high performance JTA/XA implementation
transactionalMode="xa_strict" - canonically correct JTA/XA implementation
transactionMode="local" - high performance local transactions involving caches only
transactionalMode="off" - the default, no transactions
If set, all cache operations will need to be done through transactions.
To prevent users keeping references on stored elements and modifying them outside of any transaction's control,
transactions also require the cache to be configured copyOnRead and copyOnWrite.
CacheWriter
++++++++++++
A CacheWriter can be set to write to an underlying resource. Only one CacheWriter can be
configured per cache.
The following is an example of how to configure CacheWriter for write-through:
<cacheWriter writeMode="write-through" notifyListenersOnException="true">
<cacheWriterFactory class="net.sf.ehcache.writer.TestCacheWriterFactory"
properties="type=int,startCounter=10"/>
</cacheWriter>
The following is an example of how to configure CacheWriter for write-behind:
<cacheWriter writeMode="write-behind" minWriteDelay="1" maxWriteDelay="5"
rateLimitPerSecond="5" writeCoalescing="true" writeBatching="true" writeBatchSize="1"
retryAttempts="2" retryAttemptDelaySeconds="1">
<cacheWriterFactory class="net.sf.ehcache.writer.TestCacheWriterFactory"
properties="type=int,startCounter=10"/>
</cacheWriter>
The cacheWriter element has the following attributes:
* writeMode: the write mode, write-through or write-behind
These attributes only apply to write-through mode:
* notifyListenersOnException: Sets whether to notify listeners when an exception occurs on a writer operation.
These attributes only apply to write-behind mode:
* minWriteDelay: Set the minimum number of seconds to wait before writing behind. If set to a value greater than 0,
it permits operations to build up in the queue. This is different from the maximum write delay in that by waiting
a minimum amount of time, work is always being built up. If the minimum write delay is set to zero and the
CacheWriter performs its work very quickly, the overhead of processing the write behind queue items becomes very
noticeable in a cluster since all the operations might be done for individual items instead of for a collection
of them.
* maxWriteDelay: Set the maximum number of seconds to wait before writing behind. If set to a value greater than 0,
it permits operations to build up in the queue to enable effective coalescing and batching optimisations.
* writeBatching: Sets whether to batch write operations. If set to true, writeAll and deleteAll will be called on
the CacheWriter rather than write and delete being called for each key. Resources such as databases can perform
more efficiently if updates are batched, thus reducing load.
* writeBatchSize: Sets the number of operations to include in each batch when writeBatching is enabled. If there are
less entries in the write-behind queue than the batch size, the queue length size is used.
* rateLimitPerSecond: Sets the maximum number of write operations to allow per second when writeBatching is enabled.
* writeCoalescing: Sets whether to use write coalescing. If set to true and multiple operations on the same key are
present in the write-behind queue, only the latest write is done, as the others are redundant.
* retryAttempts: Sets the number of times the operation is retried in the CacheWriter, this happens after the
original operation.
* retryAttemptDelaySeconds: Sets the number of seconds to wait before retrying an failed operation.
Pinning
+++++++
Use this element when data should remain in the cache regardless of resource constraints.
Unexpired entries can never be flushed to a lower tier or be evicted.
This element has a required attribute (store) to specify which data tiers the cache should be pinned to:
* localMemory: Cache data is pinned to the local heap (or off-heap for BigMemory Go and BigMemory Max).
* inCache: Cache data is pinned in the cache, which can be in any tier cache data is stored.
Example:
<pinning store="inCache"/>
Cache Extension
+++++++++++++++
CacheExtensions are a general purpose mechanism to allow generic extensions to a Cache.
CacheExtensions are tied into the Cache lifecycle.
CacheExtensions are created using the CacheExtensionFactory which has a
<code>createCacheCacheExtension()</code> method which takes as a parameter a
Cache and properties. It can thus call back into any public method on Cache, including, of
course, the load methods.
Extensions are added as per the following example:
<cacheExtensionFactory class="com.example.FileWatchingCacheRefresherExtensionFactory"
properties="refreshIntervalMillis=18000, loaderTimeout=3000,
flushPeriod=whatever, someOtherProperty=someValue ..."/>
Cache Decorator Factory
+++++++++++++++++++++++
Cache decorators can be configured directly in ehcache.xml. The decorators will be created and added to the CacheManager.
It accepts the name of a concrete class that extends net.sf.ehcache.constructs.CacheDecoratorFactory
The properties will be parsed according to the delimiter (default is comma ',') and passed to the concrete factory's
<code>createDecoratedEhcache(Ehcache cache, Properties properties)</code> method along with the reference to the owning cache.
It is configured as per the following example:
<cacheDecoratorFactory
class="com.company.DecoratedCacheFactory"
properties="property1=true ..." />
Distributed Caching with Terracotta
+++++++++++++++++++++++++++++++++++
Distributed Caches connect to a Terracotta Server Array. They are configured with the <terracotta> sub-element.
The <terracotta> sub-element has the following attributes:
* clustered=true|false - indicates whether this cache should be clustered (distributed) with Terracotta. By
default, if the <terracotta> element is included, clustered=true.
* copyOnRead=true|false - indicates whether cache values are deserialized on every read or if the
materialized cache value can be re-used between get() calls. This setting is useful if a cache
is being shared by callers with disparate classloaders or to prevent local drift if keys/values
are mutated locally without being put back in the cache.
The default is false.
* consistency=strong|eventual - Indicates whether this cache should have strong consistency or eventual
consistency. The default is eventual. See the documentation for the meaning of these terms.
* synchronousWrites=true|false
Synchronous writes (synchronousWrites="true") maximize data safety by blocking the client thread until
the write has been written to the Terracotta Server Array.
This option is only available with consistency=strong. The default is false.
* concurrency - the number of segments that will be used by the map underneath the Terracotta Store.
Its optional and has default value of 0, which means will use default values based on the internal
Map being used underneath the store.
This value cannot be changed programmatically once a cache is initialized.
The <terracotta> sub-element also has a <nonstop> sub-element to allow configuration of cache behaviour if a distributed
cache operation cannot be completed within a set time or in the event of a clusterOffline message. If this element does not appear, nonstop behavior is off.
<nonstop> has the following attributes:
* enabled="true" - defaults to true.
* timeoutMillis - An SLA setting, so that if a cache operation takes longer than the allowed ms, it will timeout.
* searchTimeoutMillis - If a cache search operation in the nonstop mode takes longer than the allowed ms, it will timeout.
* immediateTimeout="true|false" - What to do on receipt of a ClusterOffline event indicating that communications
with the Terracotta Server Array were interrupted.
<nonstop> has one sub-element, <timeoutBehavior> which has the following attribute:
* type="noop|exception|localReads|localReadsAndExceptionOnWrite" - What to do when a timeout has occurred. Exception is the default.
Simplest example to indicate clustering:
<terracotta/>
To indicate the cache should not be clustered (or remove the <terracotta> element altogether):
<terracotta clustered="false"/>
To indicate the cache should be clustered using "eventual" consistency mode for better performance :
<terracotta clustered="true" consistency="eventual"/>
To indicate the cache should be clustered using synchronous-write locking level:
<terracotta clustered="true" synchronousWrites="true"/>
-->
<cache name="cache" eternal="true"
diskPersistent="true"
maxEntriesLocalHeap="1000000"
overflowToDisk="true"
diskSpoolBufferSizeMB="2000"
memoryStoreEvictionPolicy="LFU"
transactionalMode="off">
</cache>
</ehcache>
Reference in New Issue Copy Permalink