[JBoss JIRA] (ISPN-9849) Server should allocate less when looking up the cache
by Tristan Tarrant (Jira)
[ https://issues.jboss.org/browse/ISPN-9849?page=com.atlassian.jira.plugin.... ]
Tristan Tarrant updated ISPN-9849:
----------------------------------
Fix Version/s: 10.0.0.Beta1
(was: 10.0.0.Alpha3)
> Server should allocate less when looking up the cache
> -----------------------------------------------------
>
> Key: ISPN-9849
> URL: https://issues.jboss.org/browse/ISPN-9849
> Project: Infinispan
> Issue Type: Enhancement
> Components: Server
> Affects Versions: 10.0.0.Alpha2, 9.4.5.Final
> Reporter: Dan Berindei
> Assignee: Dan Berindei
> Priority: Major
> Labels: performance
> Fix For: 10.0.0.Beta1
>
>
> * A composed String key is allocated to look up a decorated cache
> * Another composed String key is allocated to look up the {{CacheInfo}} object for that decorated cache. We could keep this map and remove the decorated caches map.
> * Reads decorate the cache with {{SKIP_STATISTICS}}, allocating a {{Flag[]}} and a {{DecoratedCache}} per request. I'd like to remove this wrapping, even if updating the statistics slows things down a bit, because without statistics there's no way to find the hit ratio of the cache.
--
This message was sent by Atlassian Jira
(v7.12.1#712002)
5 years, 5 months
[JBoss JIRA] (ISPN-9850) State transfer enabled attribute is overridden by store fetch-state attribute
by Tristan Tarrant (Jira)
[ https://issues.jboss.org/browse/ISPN-9850?page=com.atlassian.jira.plugin.... ]
Tristan Tarrant updated ISPN-9850:
----------------------------------
Fix Version/s: 10.0.0.Beta1
(was: 10.0.0.Alpha3)
> State transfer enabled attribute is overridden by store fetch-state attribute
> -----------------------------------------------------------------------------
>
> Key: ISPN-9850
> URL: https://issues.jboss.org/browse/ISPN-9850
> Project: Infinispan
> Issue Type: Bug
> Components: Configuration, Core
> Affects Versions: 10.0.0.Alpha2, 9.4.5.Final
> Reporter: Dan Berindei
> Assignee: Dan Berindei
> Priority: Major
> Fix For: 10.0.0.Beta1
>
>
> The decision whether to request state is made based on {{(configuration.clustering().stateTransfer().fetchInMemoryState() || configuration.persistence().fetchPersistentState())}}. In the XML {{fetchInMemoryState}} is {{<state-transfer enabled=X/>}}, {{fetchPersistentState}} is {{<store fetch-state=X/>}}.
> We should rename {{fetchInMemoryState}} to match the XML and to indicate that it enables/disables all state transfer, and stores should not be allowed to fetch state if state transfer was disabled in the state transfer configuration.
--
This message was sent by Atlassian Jira
(v7.12.1#712002)
5 years, 5 months
[JBoss JIRA] (ISPN-9882) JdbcStringBasedStore leaks threads
by Tristan Tarrant (Jira)
[ https://issues.jboss.org/browse/ISPN-9882?page=com.atlassian.jira.plugin.... ]
Tristan Tarrant updated ISPN-9882:
----------------------------------
Fix Version/s: 10.0.0.Beta1
(was: 10.0.0.Alpha3)
> JdbcStringBasedStore leaks threads
> ----------------------------------
>
> Key: ISPN-9882
> URL: https://issues.jboss.org/browse/ISPN-9882
> Project: Infinispan
> Issue Type: Bug
> Components: Core
> Affects Versions: 10.0.0.Alpha2, 9.4.5.Final
> Reporter: Dan Berindei
> Assignee: Dan Berindei
> Priority: Major
> Fix For: 10.0.0.Beta1
>
>
> When stopping, {{JdbcStringBasedStore}} only stops the connection factory if it implements {{ManagedConnectionFactory}}. {{PooledConnectionFactory}} doesn't implement it, so Agroal connection pools are not closed and leak threads.
--
This message was sent by Atlassian Jira
(v7.12.1#712002)
5 years, 5 months
[JBoss JIRA] (ISPN-9863) Testsuite leaks threads
by Tristan Tarrant (Jira)
[ https://issues.jboss.org/browse/ISPN-9863?page=com.atlassian.jira.plugin.... ]
Tristan Tarrant updated ISPN-9863:
----------------------------------
Fix Version/s: 10.0.0.Beta1
(was: 10.0.0.Alpha3)
> Testsuite leaks threads
> -----------------------
>
> Key: ISPN-9863
> URL: https://issues.jboss.org/browse/ISPN-9863
> Project: Infinispan
> Issue Type: Bug
> Components: Test Suite - Server
> Affects Versions: 10.0.0.Alpha2, 9.4.5.Final
> Reporter: Dan Berindei
> Assignee: Dan Berindei
> Priority: Major
> Labels: testsuite_stability
> Fix For: 10.0.0.Beta1
>
>
> Some hotrod-client tests do not stop their {{RemoteCacheManager}} instances. This was fine before ISPN-8619, but the Netty-based client starts event loop threads and we need to stop them.
--
This message was sent by Atlassian Jira
(v7.12.1#712002)
5 years, 5 months
[JBoss JIRA] (ISPN-9886) Use UFC_NB and MFC_NB in default embedded configurations
by Tristan Tarrant (Jira)
[ https://issues.jboss.org/browse/ISPN-9886?page=com.atlassian.jira.plugin.... ]
Tristan Tarrant updated ISPN-9886:
----------------------------------
Fix Version/s: 10.0.0.Beta1
(was: 10.0.0.Alpha3)
> Use UFC_NB and MFC_NB in default embedded configurations
> --------------------------------------------------------
>
> Key: ISPN-9886
> URL: https://issues.jboss.org/browse/ISPN-9886
> Project: Infinispan
> Issue Type: Task
> Components: Configuration
> Affects Versions: 10.0.0.Alpha2, 9.4.5.Final
> Reporter: Dan Berindei
> Assignee: Dan Berindei
> Priority: Major
> Fix For: 10.0.0.Beta1, 9.4.6.Final
>
>
> Once properly configured, UFC_NB and MFC_NB are just as fast as their blocking siblings (ISPN-9828). We should use the non-blocking variants everywhere to simplify configuration.
> We should include UFC_NB even in the TCP configuration, to prevent async messages from overwhelming the receiver (JGRP-2301).
--
This message was sent by Atlassian Jira
(v7.12.1#712002)
5 years, 5 months
[JBoss JIRA] (ISPN-9890) Query doesn't stop properly after index data loss
by Tristan Tarrant (Jira)
[ https://issues.jboss.org/browse/ISPN-9890?page=com.atlassian.jira.plugin.... ]
Tristan Tarrant updated ISPN-9890:
----------------------------------
Fix Version/s: 10.0.0.Beta1
(was: 10.0.0.Alpha3)
> Query doesn't stop properly after index data loss
> -------------------------------------------------
>
> Key: ISPN-9890
> URL: https://issues.jboss.org/browse/ISPN-9890
> Project: Infinispan
> Issue Type: Bug
> Components: Embedded Querying
> Affects Versions: 10.0.0.Alpha2, 9.4.5.Final
> Reporter: Dan Berindei
> Priority: Major
> Fix For: 10.0.0.Beta1
>
>
> When index cache entries are lost (e.g. because a test didn't override {{clearContent()}} to clear the indexes properly before stopping the manager), the query module throws an exception during stop:
> {noformat}
> 18:41:36,669 ERROR (testng-Test) [ComponentRegistry] ISPN000538: Error stopping module org.infinispan.query.impl.LifecycleManager:___defaultcache
> java.lang.AssertionError: file _0.cfe does not exist; files=[]
> at org.apache.lucene.index.IndexWriter.filesExist(IndexWriter.java:4338) ~[lucene-core-5.5.5.jar:5.5.5 b3441673c21c83762035dc21d3827ad16aa17b68 - sarowe - 2017-10-20 08:57:09]
> at org.apache.lucene.index.IndexWriter.startCommit(IndexWriter.java:4409) ~[lucene-core-5.5.5.jar:5.5.5 b3441673c21c83762035dc21d3827ad16aa17b68 - sarowe - 2017-10-20 08:57:09]
> at org.apache.lucene.index.IndexWriter.prepareCommitInternal(IndexWriter.java:2865) ~[lucene-core-5.5.5.jar:5.5.5 b3441673c21c83762035dc21d3827ad16aa17b68 - sarowe - 2017-10-20 08:57:09]
> at org.apache.lucene.index.IndexWriter.commitInternal(IndexWriter.java:2970) ~[lucene-core-5.5.5.jar:5.5.5 b3441673c21c83762035dc21d3827ad16aa17b68 - sarowe - 2017-10-20 08:57:09]
> at org.apache.lucene.index.IndexWriter.shutdown(IndexWriter.java:1082) ~[lucene-core-5.5.5.jar:5.5.5 b3441673c21c83762035dc21d3827ad16aa17b68 - sarowe - 2017-10-20 08:57:09]
> at org.apache.lucene.index.IndexWriter.close(IndexWriter.java:1125) ~[lucene-core-5.5.5.jar:5.5.5 b3441673c21c83762035dc21d3827ad16aa17b68 - sarowe - 2017-10-20 08:57:09]
> at org.hibernate.search.backend.impl.lucene.IndexWriterHolder.closeIndexWriter(IndexWriterHolder.java:173) ~[hibernate-search-engine-5.10.3.Final.jar:5.10.3.Final]
> at org.hibernate.search.backend.impl.lucene.AbstractWorkspaceImpl.closeIndexWriter(AbstractWorkspaceImpl.java:109) ~[hibernate-search-engine-5.10.3.Final.jar:5.10.3.Final]
> at org.hibernate.search.backend.impl.lucene.AbstractWorkspaceImpl.shutDownNow(AbstractWorkspaceImpl.java:104) ~[hibernate-search-engine-5.10.3.Final.jar:5.10.3.Final]
> at org.hibernate.search.backend.impl.lucene.LuceneBackendResources.shutdown(LuceneBackendResources.java:103) ~[hibernate-search-engine-5.10.3.Final.jar:5.10.3.Final]
> at org.hibernate.search.backend.impl.lucene.WorkspaceHolder.close(WorkspaceHolder.java:68) ~[hibernate-search-engine-5.10.3.Final.jar:5.10.3.Final]
> at org.hibernate.search.util.impl.Closer.push(Closer.java:92) ~[hibernate-search-engine-5.10.3.Final.jar:5.10.3.Final]
> at org.hibernate.search.util.impl.Closer.push(Closer.java:107) ~[hibernate-search-engine-5.10.3.Final.jar:5.10.3.Final]
> at org.hibernate.search.indexes.spi.DirectoryBasedIndexManager.destroy(DirectoryBasedIndexManager.java:76) ~[hibernate-search-engine-5.10.3.Final.jar:5.10.3.Final]
> at org.hibernate.search.indexes.impl.IndexManagerGroupHolder.close(IndexManagerGroupHolder.java:80) ~[hibernate-search-engine-5.10.3.Final.jar:5.10.3.Final]
> at org.hibernate.search.util.impl.Closer.push(Closer.java:92) ~[hibernate-search-engine-5.10.3.Final.jar:5.10.3.Final]
> at org.hibernate.search.util.impl.Closer.pushAll(Closer.java:120) ~[hibernate-search-engine-5.10.3.Final.jar:5.10.3.Final]
> at org.hibernate.search.indexes.impl.IndexManagerHolder.stop(IndexManagerHolder.java:186) ~[hibernate-search-engine-5.10.3.Final.jar:5.10.3.Final]
> at org.hibernate.search.util.impl.Closer.push(Closer.java:92) ~[hibernate-search-engine-5.10.3.Final.jar:5.10.3.Final]
> at org.hibernate.search.util.impl.Closer.push(Closer.java:107) ~[hibernate-search-engine-5.10.3.Final.jar:5.10.3.Final]
> at org.hibernate.search.engine.impl.ImmutableSearchFactory.close(ImmutableSearchFactory.java:268) ~[hibernate-search-engine-5.10.3.Final.jar:5.10.3.Final]
> at org.hibernate.search.engine.impl.MutableSearchFactory.close(MutableSearchFactory.java:133) ~[hibernate-search-engine-5.10.3.Final.jar:5.10.3.Final]
> at org.infinispan.query.impl.LifecycleManager.cacheStopping(LifecycleManager.java:411) ~[classes/:?]
> at org.infinispan.factories.ComponentRegistry.preStop(ComponentRegistry.java:200) [classes/:?]
> at org.infinispan.factories.AbstractComponentRegistry.stop(AbstractComponentRegistry.java:373) [classes/:?]
> at org.infinispan.cache.impl.CacheImpl.performImmediateShutdown(CacheImpl.java:1159) [classes/:?]
> at org.infinispan.cache.impl.CacheImpl.stop(CacheImpl.java:1124) [classes/:?]
> at org.infinispan.cache.impl.AbstractDelegatingCache.stop(AbstractDelegatingCache.java:520) [classes/:?]
> at org.infinispan.cache.impl.AbstractDelegatingCache.stop(AbstractDelegatingCache.java:520) [classes/:?]
> at org.infinispan.cache.impl.AbstractDelegatingCache.stop(AbstractDelegatingCache.java:520) [classes/:?]
> at org.infinispan.manager.DefaultCacheManager.terminate(DefaultCacheManager.java:734) [classes/:?]
> at org.infinispan.manager.DefaultCacheManager.stopCaches(DefaultCacheManager.java:786) [classes/:?]
> at org.infinispan.manager.DefaultCacheManager.stop(DefaultCacheManager.java:762) [classes/:?]
> at org.infinispan.test.TestingUtil.killCacheManagers(TestingUtil.java:847) [test-classes/:?]
> at org.infinispan.test.MultipleCacheManagersTest.clearContent(MultipleCacheManagersTest.java:158) [test-classes/:?]
> {noformat}
> When this exception appears, all but the first {{SyncWorkProcessor}} threads are leaked. Interesting enough, in {{ClusteredCachePerfIspnTest}} the exception is only thrown when stopping node B -- node A stops cleanly and all {{SyncWorkProcessor}} threads are stopped.
--
This message was sent by Atlassian Jira
(v7.12.1#712002)
5 years, 5 months
[JBoss JIRA] (ISPN-9897) ClusterExecutor shouldn't rely on JGroupsTransport
by Tristan Tarrant (Jira)
[ https://issues.jboss.org/browse/ISPN-9897?page=com.atlassian.jira.plugin.... ]
Tristan Tarrant updated ISPN-9897:
----------------------------------
Fix Version/s: 10.0.0.Beta1
(was: 10.0.0.Alpha3)
> ClusterExecutor shouldn't rely on JGroupsTransport
> --------------------------------------------------
>
> Key: ISPN-9897
> URL: https://issues.jboss.org/browse/ISPN-9897
> Project: Infinispan
> Issue Type: Bug
> Components: Core
> Reporter: William Burns
> Assignee: William Burns
> Priority: Major
> Fix For: 10.0.0.Beta1
>
>
> Cluster Executor currently uses transport to send messages. However the initial implementation used JGroupsTransport. Either this was a mistake at the time or enough classes have been refactored that this is no longer needed and just the Transport class can be used instead.
--
This message was sent by Atlassian Jira
(v7.12.1#712002)
5 years, 5 months