GridGain Developers Hub

GridGain 8.7.38 Release Notes

GridGain 8.7.38 provides access to a number of improvements and bug fixes.

New Features

Extended C++ Client and Java Server Interoperability

A new method Cache::InvokeJava("key", "java_task_name", args) was added for C++ thick client. With this method, you can Invoke Java entry processor by its name.

Cache<int64_t, int64_t> cache = node.GetOrCreateCache<int64_t, int64_t>("MyCache");
cache.Put(15, 40);
std::string res = cache.InvokeJava<std::string, int64_t>(15, "my.package.StringifyValue");
assert(res == "40");

Additional Option to Preserve Snapshots on Cleanup

You can now specify the number of snapshots to keep during snapshot cleanup with the -keep=N option, where N is the number of full snapshots to keep. This ensures that a certain number of snapshots are kept for backup. The option does not affect incremental snapshots.

You set this property in the snapshot management tool:

snapshot-utility.sh delete -keep=5
snapshot-utility.bat delete -keep=5

Changes in Behavior

IGNITE_THRESHOLD_WAL_ARCHIVE_SIZE_PERCENTAGE was Replaced

The IGNITE_THRESHOLD_WAL_ARCHIVE_SIZE_PERCENTAGE system property is now replaced by DataStorageConfiguration#minWalArchiveSize. The default value is set to half of default value of DataStorageConfiguration#maxWalArchiveSize property.

You can change the value in this way:

new IgniteConfiguration().setDataStorageConfiguration(
new DataStorageConfiguration().setMinWalArchiveSize(256 * 1024 * 1024)
);

Transaction States are Always Logged

To improve recovery process, WAL logging is now enabled for each transaction by default. This way GridGain can restore transactional states after a node crash.

This behavior can lead to reduced performance on lower-end servers or servers under high load. If your setup is affected, switch to LOG_ONLY or BACKGROUND WAL mode to reduce the strain.

Historical Rebalance May be Switched to Full Rebalance

Due to changes in WAL segment behavior, older segments may be automatically released when maximum WAL archive size is reached. If there is no segment for historical rebalance, GridGain automatically switches to full rebalance.

Improvements and Fixed Issues

Community Edition Changes

Issue ID Category Description

GG-33512

Platforms & Thin Clients

C++ thick client: Added API that allows calling Invoke with Java entry processor by name: Cache.InvokeJava().

GG-33431

Platforms & Thin Clients

C++ client: Fixed a rare issue with handling of READ_WANTED and WRITE_WANTED SSL messages.

GG-33394

Cluster Affinity and Baseline Topology

Fixed unexpected rebalance on an already rebalanced cluster.

GG-33383

Cluster Storage Engine

Added affinity assignment logging when assignment is not ideal.

GG-33381

Cluster Storage Engine

Fixed a memory leak when caches are created and deleted.

GG-33378

Cluster Storage Engine

Improved transaction latency during checkpoint.

GG-33374

Cluster SQL Engine

Fixed possible inconsistency of new indexes after node restart.

GG-33364

Cluster Storage Engine

Fixed a heap memory leak that happened during topology changes.

GG-33360

Platforms & Thin Clients

.NET: Fixed string and array cache keys handling when partition awareness is enabled.

GG-33359

Platforms & Thin Clients

.NET: Removed the requirement to have bash installed on Linux and macOS systems.

GG-33354

Cluster Communication

Decreased latency on communication messages.

GG-33351

Cluster SQL Engine

Fixed an issue when extra results could be returned if query parallelism was greater than 1.

GG-33134

Cluster Metrics & Monitoring

Fixed CPU load and memory metrics displaying negative values on Java 11.

Enterprise Edition Changes

Issue ID Category Description

GG-33473

Cluster Rolling Upgrade

Fixed thin clients' session serialization in compatibility scenario.

GG-33436

Cluster Storage Engine

System property IGNITE_PDS_WAL_REBALANCE_THRESHOLD has been deprecated. Use distributed property history.rebalance.threshold instead.

GG-33426

Cluster Deployment

Jetty dependency upgraded to version 9.4.42.v20210604.

GG-33414

Cluster Storage Engine

Fixed a transactional inconsistency when a transaction rollback happens after a node failure.

GG-33409

Cluster Rolling Upgrade

Fixed security issues with thin client’s certificate serialization during update from GridGain 8.7.32 and earlier.

GG-33397

Control Center Agent

Added z/OS-specific SSL parameters to startup scripts.

Installation and Upgrade Information

See the Rolling Upgrades page for information about how to perform automated upgrades and for details about version compatibility.

Below is a list of versions that are compatible with the current version. You can rolling-upgrade from any of those. Compatibility with other versions is not guaranteed. If you are on a version that is not listed, contact GridGain for information on upgrade options.

8.7.2-p15, 8.7.3, 8.7.4, 8.7.5, 8.7.5-p1, 8.7.6, 8.7.6-p1, 8.7.7, 8.7.8, 8.7.9, 8.7.10, 8.7.11, 8.7.12, 8.7.13, 8.7.14, 8.7.15, 8.7.16, 8.7.17, 8.7.18, 8.7.19, 8.7.19-p1, 8.7.20, 8.7.21, 8.7.22, 8.7.23, 8.7.24, 8.7.25, 8.7.26, 8.7.26-p1, 8.7.27, 8.7.27-p1, 8.7.28, 8.7.29, 8.7.29-p1, 8.7.30, 8.7.31, 8.7.32, 8.7.33, 8.7.33-p2, 8.7.34, 8.7.35, 8.7.36, 8.7.37, 8.8.1, 8.8.2, 8.8.3, 8.8.4, 8.8.5, 8.8.6

Known Limitations

Jetty Configuration Incompatibility in GridGain 8.7.21 and Later

If you are upgrading from version 8.7.20 or earlier, consider an incompatibility issue related to Jetty configuration introduced in GridGain 8.7.21.

Your setup may be affected if:

  • You use the ignite-rest-http module (e.g. to connect to GridGain Web Console)

  • You have a custom Jetty configuration that enables SSL for REST

  • Your Jetty configuration uses the org.eclipse.jetty.util.ssl.SslContextFactory class

  • The keystore specified in the Jetty configuration contains both the CA certificate and the private certificate

In this case, after starting a new version, an exception is thrown with an error message similar to the following:

java.lang.IllegalStateException: KeyStores with multiple certificates are not supported on the base class
org.eclipse.jetty.util.ssl.SslContextFactory. (Use org.eclipse.jetty.util.ssl.SslContextFactory$Server
or org.eclipse.jetty.util.ssl.SslContextFactory$Client instead)

To workaround this issue, alter the Jetty configuration to use org.eclipse.jetty.util.ssl.SslContextFactory$Server or org.eclipse.jetty.util.ssl.SslContextFactory$Client. See the configuration example at the Client Certificate Authentication page.

Default rebalanceThreadPoolSize in GridGain 8.7.26 and Later

In GridGain 8.7.26, the default value of the property IgniteConfiguration.rebalanceThreadPoolSize changed from 1 to min(4, number of CPU / 4). It may cause a compatibility issue under the following conditions:

  • When a Rolling Upgrade is performed

  • The upgrade is performed from 8.5.7 version (or earlier) to 8.5.x or from 8.7.3 (or earlier) to 8.7.x

  • The server nodes have at least 8 CPU cores

  • The nodes configuration does not have the property IgniteConfiguration.rebalanceThreadPoolSize, so the default value is used

In this case, an exception is thrown with an error message similar to the following:

сlass org.apache.ignite.IgniteException: Rebalance configuration mismatch (fix configuration or set -DIGNITE_SKIP_CONFIGURATION_CONSISTENCY_CHECK=true system property).
Different values of such parameter may lead to rebalance process instability and hanging.  [rmtNodeId=5fc58fb7-209d-489a-8034-0127a81abed6, locRebalanceThreadPoolSize = 4, rmtRebalanceThreadPoolSize = 1]

To workaround this issue, change the configuration of the server nodes to rebalanceThreadPoolSize=1 so that it matches the previous default configuration. For example:

<bean class="org.apache.ignite.configuration.IgniteConfiguration">
    <property name="rebalanceThreadPoolSize" value="1"/>

    <!-- The rest of the configuration goes here -->
</bean>

Jetty Doesn’t Accept Incorrect Configuration in GridGain 8.7.31 and Later

In GridGain 8.7.31 Jetty was upgraded to 9.4.33. Starting that version, Jetty has more strict validation of the provided configuration files. Before that version, an incorrectly spelled property in the configuration file had no effect. Starting this version, errors in the configuration will lead to an error on start.

Your setup may be affected if:

  • You use the ignite-rest-http module (e.g. to connect to GridGain Web Console)

  • You have a custom Jetty configuration for REST

  • The custom configuration has errors in it

You will need to fix the custom Jetty configuration before upgrading.

ignite.sh No Longer Enables Remote JMX by Default in GridGain 8.7.31 and Later

Starting from 8.7.31 version, GridGain no longer attempts to automatically enable the remote JMX. Default settings are known to cause issues if customized (for example, secure the connection). Also, in most cases, remote JMX is not required since many tools use local JMX connections (not using TCP).

Your setup may be affected if:

  • You start GridGain nodes via ignite.sh script

  • You connect to GridGain nodes' JMX interface remotely over TCP using the default configuration

To continue using remote JMX, you need to manually specify the required JMX settings. Please see the example below. Note that you don’t need remote JMX if you use a local connection, such as connecting JConsole to a GridGain process on the same host.

export JVM_OPTS="-Dcom.sun.management.jmxremote -Dcom.sun.management.jmxremote.port=33333 \
    -Dcom.sun.management.jmxremote.authenticate=false -Dcom.sun.management.jmxremote.ssl=false"

bin/ignite.sh

We Value Your Feedback

Your comments and suggestions are always welcome. You can reach us here: https://gridgain.freshdesk.com/support/login or docs@gridgain.com

Please visit the documentation for more information.