7.0.1 Release Notes

When creating a new template the JDBC cache store configuration is not copied over, and must instead be manually entered.

This is a known issue in JBoss Data Grid 7.0.1, and no workaround exists at this time.

When executing the JMX operation ClusterCacheStats.resetStatistics() none of the cluster statistics are reset.

This is a known issue in JBoss Data Grid 7.0.1, and to workaround this issue execute Statistics.resetStatistics() on each server individually.

When executing container or cache based commands through the CLI in domain mode the following message is displayed:

Container does not exists

This is a known issue in JBoss Data Grid 7.0.1, and no workaround exists at this time.

Unlike the C++ and Java Hot Rod clients, the C# client does not include asynchronous versions of cache operations at this time.

This is a known issue in JBoss Data Grid 7.0.1, and a workaround exists by using the Task API to wrap the cache operation. For instance, the following example demonstrates performing an asynchronous Put(key, value):

await Task.Run(() => cache.Put(k,v))

When attempting to insert values into a cache using the Spark connector while compatibility mode is enabled the request will fail with a ClassCastException.

This is a known issue in JBoss Data Grid 7.0.1, and to workaround this issue disable compatibility mode.

It is not possible to use scripts that create a Stream in JavaScript when the cluster is in DIST mode. Any attempts to execute these scripts will result in a NotSerializableException, as the lambdas fail when attempting to be serialized.

This is a known issue in JBoss Data Grid 7.0.1, and to workaround this issue it is recommended to manually iterator over data using an Iterator, or to execute lambdas after the data has been transferred from the script to the originator node.

When using CDI annotations with the CDI 1.1 API, such as the version found within JBoss EAP 7, a java.lang.RuntimeException is thrown; this exception is thrown as the cache events used by JBoss Data Grid 7.0 include content that has been removed by CDI 1.1.

This is a known issue in JBoss Data Grid 7.0.1. This issue is resolved in the CDI 1.2 API; to workaround this issue attach the listeners manually, as outlined in the following steps:

When updating one server group's transport configuration through the Administration Console and accepting the prompt to restart the server group remains running. Instead, a second server group is restarted in its place.

This is a known issue in JBoss Data Grid 7.0.1, and no workaround exists at this time.

When attempting to use the JGroups ENCRYPT protocol a ArrayIndexOutOfBoundsException is thrown.

This is a known issue in JBoss Data Grid 7.0.1, and to workaround this issue use the SYM_ENCRYPT or ASYM_ENCRYPT protocols, as described in the JBoss Data Grid Administration and Configuration Guide.

Cross-Site replication with multiple site masters may not work properly when configured asynchronously.

Data can be stored inconsistently because the replication relies on the delivery order of the updates to the remote site to be consistent, but with multiple site masters, it's possible an update for the same key will go through a different route than the update before it. This violates the FIFO order and leads to inconsistent data between sites.

This is related to an issue in JGroups, which JBoss Data Grid uses. The issue won't be fixed in the version of JGroups that JDG 7.0.1 uses. The issue may be fixed in a future 7.1.x release of JDG, which uses a different version of JGroups.

This is a known issue in JBoss Data Grid 7.0.1, and to workaround this issue set max_site_master=1 in RELAY2.

Previously, cross-site replication did not work if security was enabled for the cache.

With this version of JBoss Data Grid cross-site replication works when security is enabled for the cache.

Previously, Red Hat JBoss Data Grid came with a bin/init.d directory which contained the following files:

infinispan-server.conf

infinispan-server-lsb.sh

infinispan-server.sh

jboss-eap.conf

jboss-eap-rhel.sh

The Infinispan files did not work and the JBoss EAP files were not applicable to JBoss Data Grid. As such working files were added and the files which didn't apply have been removed. The following files are now available:

jboss-datagrid.conf

jboss-datagrid-rhel.sh

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, rolling upgrades did not work consistently from Red Hat JBoss Data Grid 6.6.2 through 7.1.

With this version of JBoss Data Grid rolling upgrades now allow seamless data migration with no downtime from version 6.6.2 through 7.1.

The CDI Quickstart Guide demonstrates injection of Infinispan caches into a web application using CDI. Previously, the guide described bundling the JBoss Data Grid EAP modules with the application. However, this is not recommended if the application is deployed inside EAP. The quickstart now specifies the JDG EAP modules be installed directly on the server. If the application is deployed outside of the server the JDG libraries can be still bundled with the application.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, Red Hat JBoss Data Grid API documentation, generated by javadoc, was incomplete. The API docs were updated to include leveldb, objectfilter, and hotrod client packages. Additionally, several internal configuration APIs were removed from the documentation. Many aesthetic changes were also made to make the docs easier to use.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, when attempting to use an eviction listener with the LIRS eviction strategy a NullPointerException was thrown.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, JBoss Data Grid could leak objects from Netty, a network framework used by JDG.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, when attempting to implement security on the REST interface in JBoss Data Grid 7.0.0 the settings were ignored, allowing clients to access this interface without providing the required credentials.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, if a JBoss Data Grid 6 HotRod client running on Java 7 was configured to use SSL to connect to a JDG 7 server running on Java 8 the client couldn't establish the connection sporadically and was very slow. If a JDG 7 client was used with Java 8 the client worked as expected. The exception on the server and client respectively were as follows:

ERROR [org.infinispan.server.hotrod.CacheDecodeContext] (HotRodServerWorker-4-2) ISPN005003: Exception reported: org.infinispan.server.hotrod.UnknownVersionException: Unknown version:-96

TcpTransport [TRACE] Could not connect to server: /192.168.1.1:11222 java.net.ConnectException: Connection refused: connect

A new configuration property was added to set the TLS level. Clients running on Java 7 connecting to servers over SSL running Java 8 should force TLS 1.2 since Java 7 does not enable it by default. The property should be set as follows:

infinispan.client.hotrod.ssl_protocol=TLSv1.2

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, the clear of a map wouldn't work if passivation was enabled. The size would never be equal to 0. If passivation was set to false the clear worked as expected and the result of the cache.size() method on the map would be 0.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, when enabling security with JBoss Data Grid with an LDAP backend using common-role-name-mapper for authorization, the role name failed to extract when its attribute in the distinguished name was cn instead of CN.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, when using a JBoss Data Grid server and any Java client, if there was a connection failure ClientListener would stop working. This could be reproduced with tcpkill. After killing the connection and putting a key, and then restoring the connection, ClientCacheEntryModifiedEvent is expected but does not occur.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, when AtomicMaps were used with write skew and simple versioning, lifespan limits from the config file were ignored and instead entries had an infinite lifespan. The configured default value was used if any of the following were done:

Disabled write skew checks

Used a normal put (not atomic maps)

Did not start transactions first

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, the endpoint subsystem didn't use the expression resolver to evaluate resource attributes. This could cause JBoss Data Grid clients to be unable to resolve the system property external-host.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, when auto commit in JBoss Data Grid was set to false on clustered caches, expired events were not generated.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, in JBoss Data Grid, DMR operation register-proto-schemas could fail with a NullPointerException if the proto file had syntax errors. This could also cause the Administration console to not show an error message when using an incorrect schema.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, JBoss Data Grid used SimpleDateFormat in org.infinispan.rest.Server. Under load this could cause an exception. The affected code now uses equivalent thread safe libraries instead of SimpleDateFormat.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, during shutdown an exception could be thrown by TransactionTable if the cluster view changed at the same time.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, the expiration metadata was not applied to a newly created entry that was read in the same transaction as a null. Changes in JBoss Data Grid which simplified entry wrapping addressed this issue.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, a NullPointerException could occur in BoundedEquivalentConcurrentHashMapV8.java when using LIRSEvictionPolicy under high load.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, a NullPointerException could occur in CacheNotifierImpl.notifyCacheEntriesEvicted when using LIRSEvictionPolicy.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, in the HotRod client, the SSLContext was created with the hardcoded TLS protocol. This could only be overridden programmatically by providing a custom instance of SSLContext. With this update a property was added to change the protocol from the default SSLContext via properties files. The new property is infinispan.client.hotrod.ssl_protocol

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, TcpTransport passed through the provided buffer regardless of its size. If a large buffer was passed in this could lead to excessive direct memory being allocated in JDK classes and pooled in thread-local caches. TcpTransport now uses fixed size buffers to avoid this issue.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, the HotRod .NET client included dozens of unnecessary or redundant files in the lib and bin folders. With this update unnecessary or redundant files are not included.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, if an application with module dependencies was started with JBoss Data Grid, it could show WARN messages like the following:

WARN [org.jboss.as.weld] JBAS016017: Using deployment classloader to load proxy classes for module org.infinispan.jcache:jdg-7.1. Package-private access will not work. To fix this the module should declare dependencies on [org.jboss.weld.core, org.javassist]

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, due to a query caching issue JBoss Data Grid could produce an error under certain conditions.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, if protobuf encoded and indexed items were stored in a replicated cache, when state-transfer started on a node joining the cluster, the transfer triggered exceptions. This update ensures the __protobuf_metadata cache is started and state transfer is complete before a cache using protobuf gets started.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, JBoss Data Grid did not properly enforce security when using the REST interface. This was fixed in JDG-931. The changes required updates to the XML configuration and this has been documented in the Developer Guide and the Administration and Configuration Guide

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, JBoss Data Grid 7.1 did not exist. Now that it does and its default configuration settings have been verified to perform better than 7.0.0 default settings, they have been backported to 7.0.1.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.

Previously, when examining the port for containers in the Administration Console the port-offset is not considered, causing multiple containers to show at the same port. After refactoring the codebase this issue no longer occurs.

This issue is resolved as of Red Hat JBoss Data Grid 7.0.1.