Summary: Known issues in GigaSpaces 6.0.1 (build 1903). Overview | Core | Configuration, Jini Services, and Tools | .NET

Overview

Below is a list of known issues in GigaSpaces 6.0.1 (build 1903).

See the list of new features and fixed issues in GigaSpaces 6.0.1.

Core

• GS-2837 – Read operations with a template with attributes and class hierarchy fails in secured spaces.
• GS-2825 – When writing an Entry into the space which extends MetaDataEntry, contains a field whose type is a a non-static inner class of the Entry class, and the field's class is not defined in the space classpath nor in a codebase, a StreamCorruptedException is thrown (instead of a ClassNotFoundException).
• GS-2763 – When deploying a fault-tolerant partitioned space and fail-back is enabled in the partitioned-sync2backup cluster schema, active-election might cause a split brain scenario (two primary spaces are displayed in the Space Browser).
• GS-2999 – read()/take() using a SQLQuery snapshot on a POJO return an ExternalEntry instead of a POJO. Workaround: Use the ExternalEntry.getObject() method to get the POJO:

  ExternalEntry snapEE = space.snapshot( mySqlQuery);
  ExternalEntry resultEE = (ExternalEntry)space.read( snapEE, null, 0);
  Object myPojo = resultEE.getObject(space);

• GS-2949 – After a partial update on the master space, the local cache is fully updated. As a result, the following read operations from the local cache return wrong results.
• GS-3085 – When clearing a large quantity of Entries in a sync-replicated cluster, a Java Heap OutOfMemory exception is thrown in the local space (holding the UID list). This causes some of the other spaces in the replication group to complete the clear very slowly (sometimes minutes), and some of the spaces not to clear at all. When attempting to clear Entries from another space, an UndeclaredThrowableException is thrown.
• GS-3118 – SQL queries might return a wrong result when the query contains routing index and the partitioned schema is used. The above happens only when the query contains all of the Entry fields, due to an incorrect ordering of the fields in the load-balancing mechanism.
• GS-2453 – Failover and recovery problems in Service Grid and primary election.
  Workaround: Minimize the election failure detection timeout.

Configuration, Jini Services, and Tools

• GS-2914 – Starting more than one space in a single JVM with different lookup group names or Locators causes these to be appended.
  
  Workaround: Only for the groups (the locators will be solved in next release) Set a properties file or object with your Jini groups XPATH. Since the XPATH always overwrites any system property, you can set it and (even in the same JVM) the space(es) container is advertised according to the XPATH values only.
  
  The -Dcom.gs.jini_lus.groups and -Dcom.gs.jini_lus.locators system properties are relevant mainly for the embedded Lookup Service and Mahalo services. Since these are disabled in most cases, then the only issue is with space advertisements.
  
  You can run these 3 spaces embedded in the same VM and pass 3 different properties for each SpaceFinder call, or via a properties file, for example:

  Properties prop1 = new Properties();
  prop1.setProperty("com.j_spaces.core.container.directory_services.jini_lus.groups", "1");
  IJSpace space1 = (IJSpace)SpaceFinder.find("/./space1?groups=1", prop1 );
  
  Properties prop2 = new Properties();
  prop2.setProperty("com.j_spaces.core.container.directory_services.jini_lus.groups", "2");
  IJSpace space2 = (IJSpace)SpaceFinder.find("/./space2?groups=2" , props2);
  
  Properties prop3 = new Properties();
  prop3.setProperty("com.j_spaces.core.container.directory_services.jini_lus.groups", "3");
  IJSpace space3 = (IJSpace)SpaceFinder.find("/./space3?groups=3", props3 );

  You can see that each space is advertised using its specific Jini group only.

• GS-2765 – A NullPointerException is thrown after renaming JSpaces.jar.
• GS-3067 – Unicast locators get appended when multiple spaces are started in the same JVM. This does not allow isolation and partitioning of discovery across the Lookup Service.

GigaSpaces Management Center

• GS-3092 – The GigaSpaces Management Center throws a field serialization exception when a load-balancer is used. A "used" cluster proxy is returned. Workaround: Use a non-clustered proxy, or don't use a load-balancing policy.

.NET

• GS-3000 – LocalCache does not work with clustered spaces.
• GS-2980 – When writing with the UpdateOnly modifier in a remote space, the Id field is changed to 0. Workaround: Use the Update method instead.
• GS-2939 – In a remote space, write with the UpdateOnly modifier, LocalCache, and versionid is not increased when updated in the space.
• GS-2895 – When performing read with a template that does not return Entries, the operation waits more than the specified timeout (100 - 600 milliseconds more). This occurs in embedded, remote, and partitioned spaces.