org.forgerock / org.forgerock.opendj

06 Nov, 2013

1 commit

  • 49010f5e   OPENDJ-1053 Improve logging of replication load balancing and fail-over in order… ... Browse Dir » 
    … to diagnose the causes of these events



Functional tests found a ConcurrentModificationException with r9753 (see at the end of the commit message):
The code was removing elements from a Collection while it was iterating on the exact same Collection.



ReplicationBroker.java:
In inner class LocalEvaluation:
- Renamed "filteredRSs" to "accepted".
- Added getAcceptedRSInfos() to allow iterating on them without triggering ConcurrentModificationException.
In rejectAll*(), called getAcceptedRSInfos().
In getCSN(), renamed parameters.

Fixed ConcurrentModificationException.

ComputeBestServerTest.java:
Added a test2ServersUpToDateAnd1EvenMoreUpToDate() to exhibit the ConcurrentModificationException.
Extracted method newServerState() and used it in all tests.
In all test methods but one, do not update the ServerState with unused CSNs because they make the tests harder to read. Did not change test3Servers() because I believe it is good to keep one test testing the unneded CSNs do not impact the test results.




java.util.ConcurrentModificationException
	at java.util.HashMap$HashIterator.nextEntry(HashMap.java:793)
	at java.util.HashMap$ValueIterator.next(HashMap.java:822)
	at org.opends.server.replication.service.ReplicationBroker.rejectAllWithRSOnDifferentVMThanDS(ReplicationBroker.java:1908)
	at org.opends.server.replication.service.ReplicationBroker.filterServersOnSameHost(ReplicationBroker.java:1878)
	at org.opends.server.replication.service.ReplicationBroker.computeBestReplicationServer(ReplicationBroker.java:1600)
	at org.opends.server.replication.service.ReplicationBroker.connectAsDataServer(ReplicationBroker.java:820)
	at org.opends.server.replication.service.ReplicationBroker.connect(ReplicationBroker.java:691)
	at org.opends.server.replication.service.ReplicationBroker.start(ReplicationBroker.java:221)
	at org.opends.server.replication.service.ReplicationDomain.enableService(ReplicationDomain.java:3024)
	at org.opends.server.replication.service.ReplicationDomain.changeConfig(ReplicationDomain.java:3054)
	at org.opends.server.replication.plugin.LDAPReplicationDomain.applyConfigurationChange(LDAPReplicationDomain.java:4089)
	at org.opends.server.replication.plugin.LDAPReplicationDomain.applyConfigurationChange(LDAPReplicationDomain.java:96)
	at org.opends.server.admin.server.ServerManagedObjectChangeListenerAdaptor.applyConfigurationChange(ServerManagedObjectChangeListenerAdaptor.java:74)
	at org.opends.server.admin.server.ConfigChangeListenerAdaptor.applyConfigurationChange(ConfigChangeListenerAdaptor.java:342)
	at org.opends.server.extensions.ConfigFileHandler.replaceEntry(ConfigFileHandler.java:1610)
	at org.opends.server.workflowelement.localbackend.LocalBackendModifyOperation.processModify(LocalBackendModifyOperation.java:599)
	at org.opends.server.workflowelement.localbackend.LocalBackendModifyOperation.processLocalModify(LocalBackendModifyOperation.java:299)
	at org.opends.server.workflowelement.localbackend.LocalBackendWorkflowElement.execute(LocalBackendWorkflowElement.java:690)
	at org.opends.server.core.WorkflowImpl.execute(WorkflowImpl.java:197)
	at org.opends.server.core.WorkflowTopologyNode.execute(WorkflowTopologyNode.java:100)
	at org.opends.server.core.ModifyOperationBasis.run(ModifyOperationBasis.java:451)
	at org.opends.server.core.SynchronousStrategy.enqueueRequest(SynchronousStrategy.java:49)
	at org.opends.server.protocols.ldap.LDAPClientConnection.addOperationInProgress(LDAPClientConnection.java:1276)
	at org.opends.server.protocols.ldap.LDAPClientConnection.processModifyRequest(LDAPClientConnection.java:2268)
	at org.opends.server.protocols.ldap.LDAPClientConnection.processLDAPMessage(LDAPClientConnection.java:1745)
	at org.opends.server.protocols.ldap.LDAPRequestHandler.run(LDAPRequestHandler.java:194)

git-svn-id: https://svn.forgerock.org/opendj/trunk@9784 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     

04 Nov, 2013

2 commits

  • aed0a3a7   OPENDJ-1190 (CR-2523) Under rare circumstances the DS replication recovery threa… ... Browse Dir » 
    …d (RSUpdater) can spin


This change is linked to the misbehaving RSUpdater thread which (wrongly):
- could be started multiple times
- would not shutdown willingly when the server is shutting down
- would try to look for replay operations in the future



LDAPReplicationDomain.java:
In buildAndPublishMissingChanges(), only exit the loop when the currentStartCSN is newer than now (instead of the incorrect currentStartCSN + 10s).

HistoricalCsnOrderingTest.java:
Code cleanup.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9767 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     
  • 9796b35c   OPENDJ-1190 (CR-2523) Under rare circumstances the DS replication recovery threa… ... Browse Dir » 
    …d (RSUpdater) can spin


This change is linked to the misbehaving RSUpdater thread which (wrongly):
- could be started multiple times
- would not shutdown willingly when the server is shutting down
- would try to look for replay operations in the future



LDAPReplicationDomain.java:
Added AtomicReference for RSUpdater thread.
Used DirectoryThread.isShutdownInitiated() for ServerStateFlush and RSUpdater threads.
In RSUpdater, added shutdown field to pass it down to buildAndPublishMissingChanges() + overrode initiateShutdown() to set the shutdown field to true.
In shutdown(), called initiateShutdown() for RSUpdater and RSUpdater threads.
In sessionInitiated(), only start the RSUpdaterThread if it does not already exist.
In buildAndPublishMissingChanges(), added early exits in case of shutdown.
Added now() method.

HistoricalCsnOrderingTest.java:
Consequence of the change to LDAPReplicationDomain.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9764 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     

29 Oct, 2013

2 commits

  • 79d7cb2c   Fixing failing tests in continuous integration. ... Browse Dir » 
    
AssuredReplicationServerTest.java
Fixed lost configuration lost in r9746.

FakeReplicationDomain.java, ReplicationDomainTest.java:
Added back configuration accidentally removed in r9746.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9759 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     
  • 2c86336c   StateMachineTest.java: ... Browse Dir » 
    In createReplicationBroker(), added back configuration accidentally removed in r9746

git-svn-id: https://svn.forgerock.org/opendj/trunk@9758 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     

28 Oct, 2013

2 commits

  • dcc0d28e   AssuredReplicationServerTest.java: ... Browse Dir » 
    Used Assertions to have more descriptive assertion failures.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9754 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     
  • ec480c97   OPENDJ-1053 (CR-2520) Improve logging of replication load balancing and fail-ove… ... Browse Dir » 
    …r in order to diagnose the causes of these events


This change records how each replication server is evaluated by computeBestReplicationServer().
It also logs why a DS switches from one RS to another RS.


ReplicationBroker.java:
Added static classes RSEvaluations and LocalEvaluation.
Moved keepBest() method to RSEvaluations.
In filterServersWithAllLocalDSChanges(), reworked the code to be more understandable + Added method getCSN() and rejectAllWithRSIsLaterThanBestRS(), rejectAllWithRSOnDifferentVMThanDS().
In computeBestServerWhenConnected(), extracted method isServerOverloadingRS().

replication.properties:
Added tons of messages.

ComputeBestServerTest.java:
Moved this class to package org.opends.server.replication.service to match the package of the tested code.
Consequence of the changes to ReplicationBroker.
Added methods containsOnly(), getDescription(), getEval1() and getEval2().
Reformatted the code a bit.
Removed @throws from javadocs.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9753 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     

23 Oct, 2013

1 commit

  • ac9a8c6f   LDAPReplicationDomain.java: ... Browse Dir » 
    Replaced instance fields with directly storing and using the ReplicationDomainCfg object.

ReplicationBroker.java:
Replaced instance fields with directly storing and using the ReplicationDomainCfg object.
Removed start(Set<String>), setChangeTimeHeartbeatInterval() (now useless), and setGroupId() (never used).
Added getBaseDN(), getReplicationServerUrls(), getGroupId().
In several methods, used Map.entrySet() instead of Map.keySet() with Map.get(Object)  + renamed replicationServerInfo to rsInfo.

replication.properties:
Removed now unused message.


ReplicationDomain.java:
Replaced instance fields with directly storing and using the ReplicationDomainCfg object.
Code cleanups:
- Removed useless use of "this." for method calls.
- In processErrorMsg() and processUpdateDone(), collapsed if statements
- Increased vertical density
- Extracted method getGenId()
- In waitForAckIfAssuredEnabled(), used early exit.


DomainFakeCfg.java:
Comments => javadocs
baseDn => baseDN
Removed several unused methods.
Added setWindowSize() + instance member.

*Test*.java
Used the DomainFakeCfg everywhere + SortedSet instead of Set.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9746 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     

17 Oct, 2013

2 commits

  • 3c108d77   Fixed GenerationIdTest.generationIdTest(): test was failing precisely because th… ... Browse Dir » 
    …e directory server's broker was not connected to a replication domain, hence it was impossible to add the changes to the changelogDB and then find them when querying the changelogDB.


GenerationIdTest.java:
Renamed assertConnectedToReplicationDomain() to waitConnectionToReplicationDomain().

LDAPReplicationDomain.java:
Removed uses of StringBuilder, relied on *Operation.toString() to do it for us.

ReplicationBroker.java:
Improved toString().

ReplicationTestCase.java:
Removed a wrong comment.
Used Assertions.assertThat().

ReplicationServerTest.java:
Removed a useless call to DirectoryServer.getSynchronizationProviders().

git-svn-id: https://svn.forgerock.org/opendj/trunk@9736 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     
  • 17bdce54   GenerationIdTest.generationIdTest() test is not fixed, but I have now determined… ... Browse Dir » 
    … that it is failing because the server is not connected to a replication domain, hence it is impossible to add the changes to the changelog.

GenerationIdTest.java:
Added assertConnectedToReplicationDomain() + called it.
Fixed a few logged strings.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9735 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     

16 Oct, 2013

1 commit

  • 1d9f4773   HeartbeatThread.java, CTHeartbeatPublisherThread.java, DirectoryThread.java: ... Browse Dir » 
    Used StaticUtils.stackTraceToSingleLineString(e)) for logging on one line only.

ReplicationBackend.java:
Renamed all *containers to *domains.
Reworked findExportContainers() and findSearchContainers() into only one method: selectReplicationDomains().

ListenerThread.java:
Fixed formatting.

ReplicationServerLoadBalancingTest.java:
Code cleanup.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9729 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     

15 Oct, 2013

3 commits

  • c5cb87ee   GenerationIdTest.java: ... Browse Dir » 
    More code cleanup.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9728 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     
  • 217d78e1   GenerationIdTest.java: ... Browse Dir » 
    Code cleanup.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9727 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     
  • 6df8f774   Fixing UpdateOperationTest.csnGeneratorAdjust() test failure in Continuous Integration. ... Browse Dir » 
    UpdateOperationTest.java:
Added consumeAllMessages(ReplicationBroker).
In csnGeneratorAdjust(), called consumeAllMessages().
Extracted local variables serverId.
Fixed formatting.
Increased vertical density.
Removed a useless call to Thread.sleep().
Removed useless comments.

ReplicationBroker.java:
Used local variables "localSession" to avoid concurrency issues when setSession(null) is called by another thread.

*.java:
Implemented toString().

git-svn-id: https://svn.forgerock.org/opendj/trunk@9725 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     

14 Oct, 2013

1 commit

  • d0449d18   UpdateOperationTest.java: ... Browse Dir » 
    Tried to simplify the code + separated similar looking test data between the different tests to help debug the (not so) random test failure.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9724 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     

11 Oct, 2013

2 commits

  • a7b6a1d1   Trying to fix random failures for HistoricalTest.testRecurringPurgeIn1Run(). ... Browse Dir » 
    
ReplicationTestCase.java:
In executeTask() and waitTaskState() added maxWaitTimeInMillis parameter.

HistoricalTest.java:
In testRecurringPurgeIn1Run(), added a sleep before configuring conflicts purge delay + waited purge for a maximum of 120 seconds (was approximatively 20 seconds before): as long as task max duration.

InitOnLineTest.java, ReplicationServerTest.java:
Consequence of the change to ReplicationTestCase.

GenerationIdTest.java:
Consequence of the change to ReplicationTestCase.
Extracted method createSetGenerationIdTask().

git-svn-id: https://svn.forgerock.org/opendj/trunk@9719 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     
  • 8fd0a15a   ReplicationBroker.java: ... Browse Dir » 
    In computeInitialServerStatus(), used early exits.
In computeBestServerForWeight(), extracted methods computeBestServerWhenNotConnected() and computeBestServerWhenConnected().
Changed replicationServerUrls from Collection<String> to Set<String>.
Removed useless field initialization to null.
Renamed _publish() to publish() + reduced local variables scope.
In receive(), renamed local variable replicationServerID to previousRsServerID and used this one more rather than the field.
In changeConfig(), used Set.equals().
Changed getReplicationMonitor() to getReplicationMonitorInstanceName().

ReplicationDomain.java:
Consequence of the change to ReplicationBroker.getReplicationMonitor(), inlined getReplicationMonitorInstanceName().
Consequence of the change to ReplicationBroker.replicationServerUrls.

TestCaseUtils.java:
Moved newSet(), newSortedSet(), newList() here from test classes.

*Test.java:
Consequence of the changes to ReplicationBroker.replicationServerUrls.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9712 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     

10 Oct, 2013

8 commits

  • 36d4e2bf   Added some checks to avoid python exception ... Browse Dir » 
    git-svn-id: https://svn.forgerock.org/opendj/trunk@9706 41b1ffd8-f28e-4786-ab96-9950f0a78031
    csovant
     
  • 8ac0a10c   OPENDJ-1116 Introduce abstraction for the changelog DB ... Browse Dir » 
    Changes after review from Matthew Swift.

Renamed CNIndexRecord to ChangeNumberIndexRecord to be consistent with ChangeNumberIndexDB name.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9703 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     
  • 80700011   OPENDJ-1116 Introduce abstraction for the changelog DB ... Browse Dir » 
    
Removed ChangeNumberIndexDB.getRecord(long).


ECLServerHandler.java:
Replaced calls to ChangeNumberIndexDB.getRecord() by calls to ChangeNumberIndexDB.getCursorFrom().
Inlined all the crossDomainStartState local variables.
Added getCursorFrom() to preserve old behaviour from JEChangeNumberIndexDBCursor ctor.


ChangeNumberIndexDB.java:
Removed getRecord().

JEChangeNumberIndexDB.java:
Removed getReadCursor(), getRecord() + some surprising and never used code like lock field, hasLock(), lock() and release().

JEChangeNumberIndexDBCursor.java:
In ctor, do not throw ChangelogException if the changeNumber cannot be found.

JEChangeNumberIndexDBTest.java:
Used JEChangeNumberIndexDB.getCursorFrom() instead of JEChangeNumberIndexDB.getReadCursor().
Renamed a few variables.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9702 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     
  • 304e0d76   ExternalChangeLogTest.java: ... Browse Dir » 
    Added tests on the CSNs returned by changelog and on their order too.
Removed stupid for loops.
Extracted methods readCookie() and commonAssert().
Renamed:
- getCookie() to assertContainsAndReadCookie()
- assertEntries() to assertFourEntries()
- assertDnEquals() to assertDNEquals()
Added getCSNs().

git-svn-id: https://svn.forgerock.org/opendj/trunk@9701 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     
  • c4e0644d   Clean up unit test assertions and hopefully get better exception stack traces. ... Browse Dir » 
    git-svn-id: https://svn.forgerock.org/opendj/trunk@9700 41b1ffd8-f28e-4786-ab96-9950f0a78031
    matthew
     
  • 9d011d7b   OPENDJ-1116 Introduce abstraction for the changelog DB ... Browse Dir » 
    Changes after review from Matthew Swift.

Consolidated ChangeNumberIndexDBCursor and ReplicaDBCursor into a single interface named DBCursor.



DBCursor.java: ADDED
It does not implement Comparable like ReplicaDBCursor. Instead a Comparator has been extracted to JEChangelogDB.CrossReplicaDBCursor .

ChangeNumberIndexDBCursor.java, ReplicaDBCursor.java: DELETED - replaced by DBCursor

*.java:
Used DBCursor instead of removed cursor interfaces.


git-svn-id: https://svn.forgerock.org/opendj/trunk@9699 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     
  • 2eea6297   Stabilizing JEChangeNumberIndexDBTest.testClear(). ... Browse Dir » 
    

Here is the failure scenario for this test (with one CPU):

 t1 (test thread)                  t2 (CNIndexDB purge thread)
 ----------------                  ---------------------------
        |                              |
        |                              v
        |                         JEChangeNumberIndexDB
        |                          run()
        |                          trim()
        |                          clear(null) <- entered, but not completed
        v
ChangeNumberIndexDB
 setPurgeDelay(0);
 addRecord(1)
 addRecord(2)
 addRecord(3)
        |                              |
        |                              +--> executes the clear()
        v
 getFirstRecord() <- BOOM NullPointerException!



Ensured the CNIndexDB is not trimmed until we explicitly test it:
to do this, ensured the thread is only started when needed by the test.
Also took the occasion to ensure a faster shutdown for the CNIndexDB trimming thread.



JEChangeNumberIndexDB.java:
Ensured we use volatile and AtomicBoolean.
Renamed thread field to trimmingThread.
Extracted method startTrimmingThread().
In trim(), added an AtomicBoolean parameter.
Added methods clear(DN, AtomicBoolean shutdown) and mustShutdown() to ensure faster shutdown.



JEChangelogDB.java:
Called JEChangeNumberIndexDB.startTrimmingThread() after creating the JEChangeNumberIndexDB instance.

JEChangeNumberIndexDBTest.java:
Extracted constants.
In testTrim(), called JEChangeNumberIndexDB.startTrimmingThread().
In testClear(), removed

git-svn-id: https://svn.forgerock.org/opendj/trunk@9698 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     
  • b1cdab9f   *Test*.java: ... Browse Dir » 
    Code cleanup.
Used InternalClientConnection.process*() methods.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9697 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     

09 Oct, 2013

3 commits

  • c972820b   *Test.java: ... Browse Dir » 
    More code cleanups.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9694 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     
  • 9c22c9da   OPENDJ-1116 Introduce abstraction for the changelog DB ... Browse Dir » 
    
ReplicationServerDomain.java:
Renamed getStartState() to getOldestState().

ServerState.java:
In duplicateOnlyOlderThan(), changed parameter from CSN to long.


ReplicationServer.java:
Consequence of the change to ServerState and ReplicationServerDomain.
In getECLChangeNumberLimits(), brought the database empty code to the top of the method: to get something more readable.

ECLServerHandler.java:
Consequence of the change to ServerState and ReplicationServerDomain.
Changed a few names / comments.

JEChangeNumberIndexDB.java
Consequence of the change to ReplicationServerDomain.


ExternalChangeLogTest.java
Consequence of the change to ReplicationServerDomain.
Code cleanup:
- removed gblCSN instance field
- inlined sleep()
- used connection.process*() + inlined runModifyOperation() and runDeleteOperation().
- renamed getReplicationDomainStartState() to getDomainOldestState() + added better asserts

git-svn-id: https://svn.forgerock.org/opendj/trunk@9691 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     
  • 57640b0a   OPENDJ-1116 Introduce abstraction for the changelog DB ... Browse Dir » 
    
JEChangeNumberIndexDBTest.java:
Fixed the build.

ChangeNumberIndexDB.java, JEChangeNumberIndexDB.java:
Improved the javadocs.
Made instance variables volatile.

*.java:
First => Oldest
Last => Newest

git-svn-id: https://svn.forgerock.org/opendj/trunk@9690 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     

08 Oct, 2013

1 commit

  • 542f360c   UpdateOperationTest.java: ... Browse Dir » 
    Big code cleanup to remove as much useless variables and useless code as possible to only leave the important parts of the tests.
There's more to do but I am tired now, I'll see later for the rest.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9684 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     

07 Oct, 2013

2 commits

05 Oct, 2013

2 commits

  • db4b50b3   InitOnLineTest.java: ... Browse Dir » 
    Fixing this test: the main problem was in afterTest() that was trying to remove non leaves entries, which ended up in a failure.
Renamed all the "changelog" to "replServer".
Extracted a few methods.
Removed useless parameters from other methods.
Removed useless methods.
Removed unneeded try / catch / log.
Made better use of assertions and assertj.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9668 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     
  • 92b48125   ReplicationTestCase.java ... Browse Dir » 
    Fixed problem introduced in tests at r9648.

HeartBeatMsg.java:
Added toString().

git-svn-id: https://svn.forgerock.org/opendj/trunk@9666 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     

04 Oct, 2013

2 commits

  • a7dd0d16   Fix typo and increase wait time when enabling the JMXConnectionHandler to see if… ... Browse Dir » 
    … it helps with CI that runs on single CPU machine.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9665 41b1ffd8-f28e-4786-ab96-9950f0a78031
    ludo
     
  • c5569e14   OPENDJ-1116 Introduce abstraction for the changelog DB ... Browse Dir » 
    Renamed some classes in package src.server.org.opends.server.replication.server.changelog.je:
- DbHandler.java         => JEReplicaDB.java (and tests)
- DraftCNDbHandler.java  => JEChangeNumberIndexDB.java (and tests)
- DraftCNDbIterator.java => JEChangeNumberIndexDBCursor.java

*.java:
Consequence of the renames.
Changed methods/variables/parameter names and comments to reflect the new names. In particular, bye bye to DraftCNDb, DbHandler and welcome to CNIndexDB and ReplicaDB.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9662 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     

03 Oct, 2013

4 commits

  • b4ce9c60   HistoricalTest failure: Code added a task but did not wait for it to complete be… ... Browse Dir » 
    …fore proceeding. Adding a wait should solve the problem.


ReplicationTestCase.java
Added executeTask().
Used simpler InternalConnection.process*() APIs.

HistoricalTest.java:
Used ReplicationTestCase.executeTask().
Inlined local variables.
Extracted method getEntryValue().
Changed signature of publishModify().

GenerationIdTest.java:
Used ReplicationTestCase.executeTask().
Renamed assertNoMessageReceivedBadGenId() to assertNoMessageReceived().

ReplicationServerTest.java
Used ReplicationTestCase.executeTask().


TestCaseUtils.java:
Used simpler InternalConnection.process*() APIs.

UpgradeTestCase.java
Renamed isOutputContainsExpectedMessage() to assertContainsMessage().

*Test.java:
Used assertj.
Let the exceptions go up.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9650 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     
  • efcc0583   ReplicationTestCase.java: ... Browse Dir » 
    Change entryList and configEntryList from LinkedList to Set and renamed them to entriesToCleanup and configEntriesToCleanup.
Extracted method connect() and waitForSpecificMsg() + Changed the return type of the existing waitForSpecificMsg() methods.
In configureReplication(), added 2 String parameters + extracted method addSynchroServerEntry() and addConfigEntry().
Simplified code a lot.

*Test[Case].java:
Consequence of the changes to ReplicationTestCase.
Factorized code with ReplicationTestCase.
Various cleanups:
- used assertj
- used StaticUtils.close()
- made fields private
- used foreach
- used assertEquals() instead of assertTrue()
- removed try/catch/fail test anti pattern.
- let exceptions go up the stack

ProtocolWindowTest.java:
Extracted method searchNbMonitorEntries().

git-svn-id: https://svn.forgerock.org/opendj/trunk@9648 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     
  • 9e77e64c   UpdateComparator.java, UpdateComparatorTest.java: REMOVED ... Browse Dir » 
    This code is never used + UpdateMsg already implements Comparable.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9647 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     
  • afc6c490   Maybe the end of endless pain and suffering on the Continuous Integration front - take 2. ... Browse Dir » 
    In ExternalChangeLogTest.ECLReplicationServerFullTest(), debugging somewhat allows to reproduce the problem. The sequence of events is... interesting.
In ECLAfterChangelogTrim(), the last published DeleteMsg might not even have even been received on the other side when the test ends (i.e. ServerReader.run() might never have called/completed calling Session.receive()), hence clearing the changelogDB cannot clear this yet to be received message!
Looking closely at this test, I cannot see any reason to publish this DeleteMsg, so I removed this code. I also took the opportunity to get rid of what look like a useless 1s sleep.

ExternalChangeLogTest.java:
In ECLReplicationServerFullTest(), removed sending a useless DeleteMsg that changes nothing to the test result + removed a useless 1s sleep (Yay!) + removed the call to clearChangelogDB() in the finally block (already called by a @AfterTest)

git-svn-id: https://svn.forgerock.org/opendj/trunk@9646 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac
     

02 Oct, 2013

1 commit

  • 2db5afe8   Maybe the end of endless pain and suffering on the Continuous Integration front. ... Browse Dir » 
    In ExternalChangeLogTest.ECLReplicationServerFullTest(), it looks like the server does not have the time to trim the last "changetype: delete" from the previous tests method, so let's force the test to clear the changelogDB before it finishes.

ExternalChangeLogTest.java:
In ECLReplicationServerFullTest(), called clearChangelogDB() in the finally block to force cleaning test data.

git-svn-id: https://svn.forgerock.org/opendj/trunk@9645 41b1ffd8-f28e-4786-ab96-9950f0a78031
    JnRouvignac