Clone
 

mircea markus <mircea.markus@jboss.org> in Infinispan

ISPN-263 Handle cluster partitions

* fixed reviews comments

  1. … 3 more files in changeset.
ISPN-263 Handle cluster partitions

* detail how the key ownership is determined

ISPN-263 Handle cluster partitions

* don't allow partition handling for replicated caches

ISPN-263 Handle cluster partitions

* added user documentation

ISPN-263 Handle cluster partitions

* fixed test suite

ISPN-263 Handle cluster partitions - allow configuration through XML

ISPN-263 Handle cluster partitions - enhanced the test suite with various nodes failures

ISPN-263 Handle cluster partitions

* don't allow for stale data in an resurrected member as this messes the consistency guarantees the partition handling offers

ISPN-4424 Synchronize updating cache entry and making a copy - making the CacheEntry immutable is not enough, the whole CacheEntry hierarchy needs to be made immutable - using clone is simpler that making the hierarchy immutable

ISPN-263 Handle cluster partitions - implemented functionality to determine multiple successive failures

BZ1111221 ISPN-4420 Allow configuring keepalive for hotrod connections - fixed test

ISPN-4420 Allow configuring keepalive for hotrod connections - fixed test

ISPN-4420 Allow configuring keepalive for hotrod connections

  1. … 3 more files in changeset.
BZ1111221 ISPN-4420 Allow configuring keepalive for hotrod connections

  1. … 3 more files in changeset.
updated release name

ISPN-4332 Associating the same transaction with multiple threads is not something JTA spec covers, so we shouldn't expect this behaviour from Geronimo.

ISPN-4332 If we catch an exception during commit, we initiate a rollback (TransactionCoordinator.handleCommitRollback) and consequently signal this unilateral decision to the TM with an XA_HEURRB exception. Narayana throws this to the user as a RollbackException whilst Geronimo throws it as an HeuristicRollbackException, so tests needed to be updated to cover both scenarios.

ISPN-263 Handle cluster partitions

* add logic to handle split brains

* when a split brain happens, the consistent hash of the cache freezes till the nodes that disapeared ressurect

* the cluster enters degraded mode, i.e. only servers data that hasn't been affected by the split brain

  1. … 19 more files in changeset.
ISPN-263 Handle cluster partitions * refactored: moved cache specific logic from ClusterTopologyManagerImpl into the ClusterCacheStatus * This makes it easier to take rebalance related decition at cache level.

fixed compilation issue

remove directories created during tests

ISPN-4170 Upgrade to JGroups 3.5.0.Beta3

    • -1
    • +0
    /core/src/main/resources/jgroups-ec2.xml
    • -1
    • +0
    /core/src/main/resources/jgroups-tcp.xml
    • -1
    • +0
    /core/src/main/resources/jgroups-udp.xml
    • -1
    • +0
    /core/src/test/resources/stacks/tcp.xml
    • -1
    • +0
    /core/src/test/resources/stacks/udp.xml
BZ1059489 ISPN-3957 Preload with async cache store is not efficient

ISPN-3957 Preload with async cache store is not efficient

ISPN-3957 Preload with async cache store is not efficient

ISPN-3852 Add a section describing the configuration for the java hotrod client

ISPN-3784 Exiting because has NOT shut down all the cache managers it has started

Fixed typo: hasSharedCacheLoadOrWriter -> hasSharedCacheLoaderOrWriter

ISPN-3498 Cannot construct org.infinispan.util.KeyValuePair as it does not have a no-args constructor

* an JBossMarshaller is built under the hood for each TestObjectStreamMarshaller instance

* users need to call TestObjectStreamMarshaller.stop explicitly

ISPN-3537 Custom interceptor with Position.LAST set programmatically doesn't work