Summary: Fixed issues in GigaSpaces 5.2.3 (build 1780). Overview | Configuration, Jini Services, and Tools | Core | POJO and Spring | Service Grid | .NET

Overview

Below is a list of fixed issues since GigaSpaces 5.2.2 (build 1739).

See the release notes, fixed and known issues in previous versions.

Configuration, Jini Services, and Tools

• GS-1289 - Fixed: LookupFinder bug - client sometimes connected to a space with the same name in a different Jini group: This occurred with two spaces that have the same name (different container), each a member of one shared Jini group and one unique one, and only three lookup services up, each registered for only one Jini group of the three (the two unique and the shared). If you used SpaceFinder.find() for the space with the shared group, and then looked for the space with one of the unique groups, while less than a minute passed between the finds - in the second find, you could have received a proxy to the wrong space.
• GS-1318 - Jini service registration lease timeout has been extended. The com.sun.jini.reggie.minMaxServiceLease is now set to 15 minutes in the This is set in the Service Grid config\services\service.config, the config\services\reggie.config and the config\services\mahalo.config files.
• GS-1510 - Fixed: Clustered spaces disappeared (e.g. from the GUI) approximately 5 minutes after the benchmark began writing Entries. Such behavior occurred when a large amount of memory was consumed in the process, causing extensive GC spikes. This caused high CPU usage and distracted the LeaseRenewManager (a long GC/CPU clock caused the LeaseRenewManager to miss the default 4 seconds, or a lease renew attempt which caused a space service un-registering event to be fired).
  
  The solution is to add resources (memory, CPU) or spaces, or to tune the LeaseRenewal maxLeaseDuration and roundTripTime. These two values can be configured using the following system properties:

  //Default value for net.jini.lease.LeaseRenewalManager.roundTripTime default value 4 seconds
  -Dcom.gs.jini.config.roundTripTime=40000
  //Default value for net.jini.lookup.JoinManager.maxLeaseDuration default value 8 seconds.
  -Dcom.gs.jini.config.maxLeaseDuration=80000

• GS-1679 – Fixed: When performing a SpaceFinder.find before the space was loaded, before lookup timeout expired, a FinderException was thrown, even though the space was published and available. A second SpaceFinder.find managed to find the published space.
• GS-1831 – Replicator report threads are now created only for FINEST loggers, saving redundant threads creation.
• You can set any of the gs_logging.properties using a system property. For example, if you want to have a separate logger file-handler URL for each process, set:

  -Djava.util.logging.FileHandler.pattern=../mypathdir/mylogger.log

The same is true for all other properties, like patterns, log levels, etc.

GigaSpaces Browser

• GS-1561 - Fixed: Some spaces did not appear in the Space View after a large cluster was started. The container node was added without the spaces.
• GS-1766 - Fixed: After relocating a space, the relevant space node disappeared from its container in the tree, but still appeared in the Deployment tab.
• GS-2057 – Fixed: Attempting to run a benchmark on a secured space in a cluster threw a SpaceSecurityException (for a working clustered proxy). The same exception also occurred when attempting to inspect an object from the Query view in a secured space.
• GS-2260 – Fixed: A NullPointerException was thrown in the SpaceRuntimePanel. After this, the Classes view was not updated anymore.
• GS-1870 – Fixed: High CPU usage during the GUI Cluster View refresh.
• GS-1939 – The ServiceManagerImpl.FindSpaceTask (inner class) has been removed, since it caused high CPU utilization, and redundant expensive calls to SpaceFinder.find().
• GS-1998 – Fixed: The setenv script was missing a first check if the JAVA_OPTIONS had already been defined externally.
• GS-1898 – Fixed: High CPU usage was discovered during the GUI Cluster View refresh and its lease cancellation.

Core

• GS-1365 - Fixed: A NullPointerException was thrown when using Map local cache with put-first=false.
• GS-1868 – Fixed: When performing writeMultiple of Entries with a lease (but with no UID), the lease didn't expire.
  GS-1902 – Fixed: A NullPointerException was thrown when launching a custom partitioned-sync2backup cluster using RMI.
• GS-2075 – Fixed: The Lease Manager got into an endless loop when 32 classes were used.
• GS-181 – Fixed: An OutOfMemory exception was thrown when performing the clear operation on a large quantity of Entries in a sync-replicated cluster. To solve this, the support of <multi-opers-chunk-size> has been added to the <sync-replication> tag in sync-replicated cluster schemas.
  
  When performing batch operations (writeMultiple, readMultiple, takeMultiple, and updateMultiple), and using the <multi-opers-chunk-size>, the Entries are split up into chunks, according to the tag value. For example: <multi-opers-chunk-size>100</multi-opers-chunk-size> splits up the Entries into 100 chunks.
  
  This element's default value is -1, which means it is not enabled by default, unless you set a certain value as seen in the example above.
  
  Transaction.commit() doesn't support multiple chunks. Commit replicates as one chunk (as in previous versions).

Query Engine

• GS-71 – Fixed: Continues Query – when using SQLQuery with notify, snapshot returned ExternalEntry.
• GS-1687 - Fixed: Wrong query result when using orderBy and rownum for simple queries - instead of ordering the results and then limiting the result set, the data was first limited and then ordered.
• GS-1888 – Fixed: BroadcastNotify didn't work with SQLQuery. The query was not converted to an ExternalEntry, and therefore was not broadcasted.
• GS-2040 – Fixed: The Embedded Query Processor (or any embedded worker) changed data serialization during select query.

POJO and Spring

• GS-1323 - Fixed: When using the gs.xml file in order to map POJOs, the fifo, replicate, and persist flags are not required (these were required in previous versions).
• GS-1977 – Fixed: When GigaSpaces runs in a Spring context, the exported class was occasionally not assignable from the Spring dynamic proxy.

Service Grid

• GS-1532 - Fixed: Partial deployment - when deploying one space partition and multiple instances of a collocated service, starting another GSC, and killing the first GSC, all deployed instances of the collocated service should have been be recreated in a secondary GSC, but actually not all of the instances were recreated.

.NET

• GS-1593 – Fixed: PONO fields of type object could not contain an array of bytes (Byte[]).
• GS-1714 – Fixed: An empty log was created in the .NET client.
• GS-1728 – Fixed: PONOs with an object type field were not read properly from the space.