Pivotal GemFire 9.1 Release Notes
Note: General support includes security vulnerability resolutions and critical bug fixes in all supported minor versions, while other maintenance is applied only to the latest supported minor release.
Pivotal GemFire 9.1 is based on Apache Geode version 1.2.0. Pivotal GemFire 9.1 and Apache Geode share a common set of documentation. Therefore, references to both GemFire and Apache Geode appear throughout this documentation; consider them equivalents.
A fully-integrated Lucene text search using indexes. See Apache Lucene Integration.
Improved the resiliency of servers during SSL handshake.
PartitionResolverenables the colocation of related data without writing and deploying a custom partition resolver.
There is a
gfsh importoption to invoke callbacks such as CacheListeners when importing a region snapshot file.
The following previously-deprecated classes have been removed:
- some overloads of
- some overloads of
You can download Pivotal GemFire 9.1 from the Pivotal GemFire product download page.
For details on how to install Pivotal GemFire 9.1, see Installing Pivotal GemFire.
To upgrade from an earlier version of GemFire to the current version, see Upgrading Pivotal GemFire.
Note that an extra restart of servers within clusters that are upgraded to GemFire version 9.1.1 or
a more recent version from a version prior to version 9.1.1 will be needed once the clients have
been upgraded, if the
SecurityManager interface has been implemented.
The GemFire 9.1.0 release was based on the Apache Geode 1.2.0 release. Here is a list of all issues resolved for Geode 1.2.0.
The following issues have been resolved in GemFire 9.1.1. Ticket numbers of the form GEODE‑NNNN can be inspected at the Apache Software Foundation site for Geode issues.
GEODE-2018, GEM-1069: Modified the session caching mechanism to register instantiators earlier in the startup process.
GEODE-3030: Set possibleDuplicate=true for all bucket events after failover.
GEODE-3043, GEM-1449: Modified the Membership Manager to eliminate addition of a “surprise member” when the member is already in the cluster.
GEODE-3052: Eliminated a potential “split brain” condition resulting from restarting locators in rapid succession.
GEODE-3097: Enabled gfsh with SSL over HTTP.
GEODE-3101: Improved the release of local locks held in a client when a transaction fails on the server.
GEODE-3147: Fixed a communication buffer issue for threads sending transaction synchronization messages.
GEODE-3193: Deferred pid file removal to a later stage of the shutdown sequence to improve diagnosis and recovery.
GEODE-3206, GEM-1594: Ensured that
gfsh lucene search results are listed in order of relevancy.
GEODE-3217, GEODE-3410: Reimplemented
gfsh query as a single step command.
GEODE-3286: Fixed a condition in which improperly handled closed connections could lead to an
GEODE-3291, GEM-1592: Added the soundex analyzer’s jar to support Lucene searches.
GEODE-3249: Increased authorization requirements for internal client/server messages.
GEODE-3314: Fixed a re-use conflict in the handling of remote and local locking tokens.
GEODE-3315: Fixed a deserialization issue with client queues that could result in a
GEODE-3379, GEM-1517: Improved the coordination of transaction commits following data rebalancing.
GEODE-3445: Added a
gfsh connect option to skip SSL validation.
GEODE-3513: For generic app servers, improved the cleanup of native sessions in proxy mode.
GEODE-3519: Implemented additional locking on invalidate and remove operations initiated by clients.
GEM-1256: Addressed a Java-level deadlock affecting function execution and some other membership tasks.
GEM-1492: Improved the handling of JTA exception and the subsequent release of local locks on a client node.
GEM-1576: Fixed one-way SSL commit between client and server.
GEM-1645: Added documentation for automated rebalancing.
The following known issues affect GemFire 9.1:
GEM-868/GEODE-2180: Reconfiguration of the cache when autoreconnecting does not work with the
default settings for
use-cluster-configuration if the cache is configured through API calls.
Either disable autoreconnect or disable use of the cluster configuration service.