This is the RHQ 4.6 release. It was released on January 31st, 2013
RHQ 4.6.0 no longer has a GUI-based installer. Even if you are familiar with installing previous versions of RHQ, be sure to see Running the Installer for updated installation and upgrade instructions.
If upgrading from RHQ 4.2 (4.3 and later are not affected) you must first make a manual change to your database. Have a look at the first note on the RHQ 4.4 release notes
RHQ 4.6 is recommended with Java7, and also supports Java6.
This is a significant infrastructure change. RHQ has migrated from AS 4.2.3 and now runs on AS7. There are many benefits associated with this update and several changes of note.
HornetQ replaces the dated JBoss Messaging implementation. There is one behavioral change for HA RHQ installations. Messages sent to a server must be delivered by that server. This means only that when a server goes down, any undelivered messages will not be delivered until that server returns to service. In the prior system messages could be delivered by other running HA nodes. In practical terms this will have little to no discernible effect.
The embedded agent is no longer functional with the RHQ Server now running inside JBossAS7. Its existence was of minimal benefit and the work to move it to the new platform was better directed elsewhere. Therefore, the embedded agent functionality has been completely removed.
Further enhancments
Alert definitions for resources, groups and resource types can now be created, updates and deleted. THIS MAY CHANGE
Endpoints introduced in 4.5.1 to create platforms and resources have been deprecated - replacements are provided
BZ 896537: The Dashboard view now orders Dashboard Tabs alphabetically, guaranteeing order and allowing users to "land" on a non-default dashboard.
A new Cassandra plugin compatible with Cassandra 1.1.2 is part of this release. The plugin is still experimental but supports monitoring and configuration for features exposed via Cassandra's JMX interface. Additional support for some important internal configuration settings (Eg. gc_grace_seconds) is also available. The plugin has enhanced support for taking and restoring snapshots.
The plugin is not yet compatible with Cassandra 1.2.x. Since Cassandra 1.2.x has a lot of improvements and changes, a decision about how to support it will be taken for the next RHQ release.
The JBoss AS 7 plugin has been updated to use commons-httpclient instead of JDK HttpUrlConnection for uploads to the server management interface. It will prevent OutOfMemoryErrors to occur in the agent as HttpUrlConnection caches payloads in memory (BZ 887320)
JAX-WS based Webservices have been removed. The REST interface, CLI and custom Java clients supersede the WS API.
Method org.rhq.core.util.exec.ProcessExecutor#redirectStreams has been deprecated. Callers should switch to using a new method org.rhq.core.util.exec.ProcessExecutor#redirectAllStreams method.
REST-API Endpoints introduced in 4.5.1 to create platforms and resources have been deprecated - replacements are provided
Method org.rhq.core.util.exec.ProcessExecutor#redirectStreams now returns void. This change restores the behavior of this method as it existed in RHQ 4.4.0 and prior. Clients that used the return type of the method introduced in RHQ 4.5.x and all the new code should use a new method called redirectAllStreams. The method redirectStreams has been deprecated as of RHQ 4.6.0 and will be removed in some future release.
Method org.rhq.core.domain.configuration.definition.PropertyDefinitionMap#getPropertyDefinitions() now returns a java.util.Map. This is to restore the behavior as it existed in RHQ 4.4.0 and prior.
Static fields org.rhq.core.domain.criteria.MeasurementScheduleCriteria#SORT_FIELD_* were removed. A few sort specifiers that were actually in error and if used would have produced a downstream error.
Number of arguments in the constructor of the class org.rhq.core.domain.alert.composite.AlertConditionAvailabilityCategoryComposite changed. While this class is available, it is not used directly in any of our remote APIs. The previous constructor signature was missing an argument required for correct behavior. See BZ 888927.
All the remote EJBs and handlers associated with the webservices were removed (WebServicesRemote, WebServicesTestRemote, RemoteWsInvocationHandler, ServerVersion).
The GWT part of the UI has partially been translated into German, Portuguese, Japanese, Chinese, Czech, Russian and Korean. The language should be automatically selected depending on your browser settings. You can explicitly access other translations by appending a locale specifier in the URL. For example to select the German translation you would append ?locale=de to the base URL, e.g. http://localhost:7080/coregui/?locale=de.
Supported locales are:
zh for Chinese
cs for Czech
de for German
ja for Japanese
pt for Portuguese
ru for Russian
ko for Korean
Please ping us if you want to help translating the UI to your language. Translations are done via the translations project on GitHub, which also has some instructions on how to start.
Please report all bugs you find in Bugzilla. If you find a bug that has been recorded in the above list, please leave a comment on them especially if this needs special steps to reproduce.
Please consult Bugzilla with a target release of RHQ 4.6.0 for a list of resolved issues
You can download the release here
Heiko W. Rupp <hwr@redhat.com> (52):
Remove unused webdav source code.
Add a script that creates a confluence-version of the git shortlog between two tags. Includes a mapping file for committers and their multiple accounts they commit from.
Remove to-be-removed webservices from the build.
Removed plugins
removed plugins
Exclude plugins dir
Revert "Exclude plugins dir"
Revert "Removed plugins"
Revert "removed plugins"
Follow up with the changes to SenderResult()
Make the rest docs generator understand the new @ApiClass annotation.
Document the data classes of the REST api.
Add Thomas as developer and mark Ian as retired
BZ 861239 - add a flush() so that getting the live availability works.
Print the DB-url and user that is used to run the tests to make debugging easier.
Add some excluded metrics, as they may indeed return null.
Make message more clear in case of failure.
Add more excludes for the mem pools
BZ 880136 - disable the Connections available trait by default. (cherry-picked from c38b8eb, BZ 859417)
BZ 880214 - better detect the location of backing files for deployed content.
Get rid of outdated <source>1.5 and <target>1.5 settings.
Get rid of the never really working JBossOSGI plugin - especially as the JBossOSGi project is long dead anyway.
Get the REST-API to work again. * Login module is now called 'Database' and no longer 'DatabaseServer' * Application is relative to /rest , so only put the /1 part in here * Use the servlet api version 3 descriptor and trim non-needed entries. * Change the way the cache is obtained. Fix some javadoc.
Add support for searching of resources, groups and group definitions by name via ?q=
Refactor the REST API to drop the /1/ in the URI. * move to EJB 3.1 local SLSB * separate the examples into their own war file
BZ 885169 - explicitly prepend /rest/ as some browsers may misunderstand relative links.
BZ 861491 - Choose a better display name and description for the 'promote' operation. (cherry-picked from e4203a2, BZ 860665)
Add tests for the REST-Api endpoints and fix some issues in the API.
Add tests for the operations part of the rest-api
BZ 886850 Make some followup changes to the Exception-semantics change in that BZ
Platforms must not be set to disabled according to jshaugn, so catch and test this case.
First cut at testing the Alert-part of the REST-api and of supporting the creation of definitions.
Support CRUD on Conditions and Notifications
Further work on the AlertDefinition part of the Rest-api: - Support for alert dampening. - Split alert handler and definition handler. - Allow to create alert definitions for groups and resource types as well.
Update russian translations from the translation project.
A bit of cleanup. Make sure double resource creation works, allow to submit metrics with metric name instead of schedule id.
Get jar-less plugins going again.
Forgot this in my last commit wrt the rest-api changes.
Improve documentation and get the processor to work again. Generate links from parameters to domain classes, upgrade the docbook plugin.
Improve representation of types.
Do not assume everything is a jar - we also support jar-less plugins.
Use a real UUID for resource creation. Deprecate the endpoints for platform and resource creation.
BZ 906366 If no priority is given for an alert definition, assume LOW and don't bomb.
BZ 905210 retrieval of alertCount failed for requested xml content type
BZ 907029 - Make the reports available at /rest/ as well.
BZ 906155 text/plain is not supported, only json,xml,html
Improve documentation and harden the alert definition part of the REST api
Itests are now itests-2
Improvements in the Events and Group area of the REST-api
Follow the recent changes in availability handling and add another test.
Add a get group by id case with valid id
Add getAlerts4Definition, extend testing
Hudson <jboss-qa-internal@redhat.com> (2):
development RHQ_4.6.0-SNAPSHOT
tag RHQ_4_6_0
Jay Shaughnessy <jshaughn@redhat.com> (222):
BZ 828428 - org.rhq.enterprise.server.cloud.instance.CacheConsistencyManagerBean ERROR ORA-01795: maximum number of expressions in a list is 1000] This time the problem is 1000 or more agents on a server (with status <> 0). Batch the updates to avoid the Oracle limitation.
Add new scripting api module to eclipse source path
BZ 846375 - Resource list views should have name field as initial sort and limit ancestry to client side sorting only] - Apply default initial sort on Name field for resource list views (using ResourceSearchView) - note, updated constructor chaining to be more efficient while adding default sort behavior. Removed an unused constructor. - Disable server-side sorting on Ancestry since sorting on the encoded value is not useful to the end user. Allow client-side sorting when all rows are fetched, this gives the desired alpha sort on the display value.
Eclipse .classpath updates
Fix compile issue in hardware plugin
Merge branch 'master' into rhq-on-as7
Remove WebService support from RHQ.
Fix issue in eclipse I18N launcher
First pass at Hibernate upgrade. We will try and leverage the Hibernate4 modules provided with AS7. - Changes to remove use of deprecated features that were removed in H4 - new mechanism for createBlob - new ElementCollection annotation - new mechanism to work with Connection - Dependency changes to new jar structure - Removal of legacy annotations and jpa deps for compile issues - Also leverage AS7 provided antlr jar, removing transitive deps pulling in multiple jars/versions
Merge branch 'master' into rhq-on-as7
A few more legacy hibernate3 deps removed.
Don't play with the rhq.ear deployment marker files. When building from the top we get an unwanted .dodeploy which messes up the initial server startup. It worked well when building the invidual module, but now the dev will likely need to manually rebuild or redeploy the ear.
The default MBean server for AS7 is the platform MBeanServer, not the "JBoss" MBeanServer. - Undeprecate CoreServer and CoreServerMBean and provide it with the new, proper MBeanServer.
POM Dependency work - move opencsv version to root pom as it is used in multiple child poms - this also fixes issue in CLI pom where opencsv.version had been accidentally removed - add infinispan.version in root pom to be provided AS7 module version - change resteasy version in root pom to match provided AS7 module version
Eclipse, add Resource filter on dev-container
Update more dep versions to those provided with AS7 - comment out container-lib module from build - call out in the root pom which deps are using AS7 provided versions and which are not.
Migrate RhqCache from TreeCache to Inifinispan
Move EAR dependency declarations to jboss-deployment-structure.xml.
Update Resource JNDI names to required AS7 formats
Fix dep issue in misc (perftest) plugin
trivial - Fix jdoc typos
Update reference for security domain. (the domain itself will move to standalone-full.xml)
Add REST security domain def to the installer.
Add RHQ Queue creation support for standalone-fulle.xml
More configuration generation in the installer - add Infinispan cache config - rename JMSJbossASClient to MessagingJBossASClient to follow subsystem naming. Also, clean this up a bit. - Support more of the config options we currently use/need.
Delete unused @Management interfaces. The anotation is obsolete.
Remove use of obsolete @Management and @Service annotations and migrate ServerPluginService to the service archive.
re-enable container-lib in the build. although, comment out most of the deps until proven that they are needed.
Added TODO to perhaps remove the ApplicationException annotation.
Move Server Jat REST doc generation into profile activated with -Ddocgen. Cuts build time by about 40%.
Should have been @Column, the join column is specified in the CollectionTable annotation.
Now that we require at least java6 for builds, and possibly java7, remove special handling for java5 and java6 in the poms.
Bring container-lib back to life and fix ordering
Don't create rhq.ear.dodeploy if rhq.ear.skipdeploy exists. Note, due to maven limitations this only works if dev-container has the default name and location.
Delete any .dodeploy when we know we want .skipdeploy
Don't throw a RuntimeException in getPlugin() as return info for a successfult call. Exceptions can affect transactioning. Just return null.
Fix jdcoc for getPlugin
Move installer's DMR client management stuff into a reusable common module that can be consumed by various components, like server jar. Start using the new DMR client to register RHQ user login modules
New GroovyMonkey script to make target dirs derived since the old monkey script is no longer usable in eclipse.
Fix this dumb .dodeploy thing so that an appserver dev build ensures any rhq.ear.dodeploy file is removed.
Now (theoretically) deploys the ldap related login modules if ldap is configured. May need a little more work to handle the situation when ldap information is updated. I would think in that case we may need to remove and redeploy the security domain.
Hibernate4 is more strict (or perhaps just more complete) in its handling of SequenceGenerator allocationSize. This value must match the increment on the underlying sequence.
Call ModelControllerClient.close() when done with it.
H4 is more strict about the use of identification variables (aka alias) in queries. Make sure we define them and use them if necesssary. This was a particular problem in ServerPlugin queries.
Move server-comm-configuration.xml to the service sar top level so it can be found bu the comm service on the classpath.
The java:app JNDI name isn't applicable to JNDI lookups outside the app, like from our Quartz jobs. The java:global/rhq works as it provides the app namespace.
Tweak custom login module deploy utility to only create the security-domain if it does not already exist.
Remove the JMS_* tables, they are not relevant to HornetQ.
- Get Hibernate Test Page to render. - translation is not yet working. - Get Entity browser working - Get Admin Control page rendering - Rename *TestBeanLocal interfaces to *TestLocal to conform to our naming conventions such that JNDI lookups work as expected
Play with fire, get burned. Fix usage of internal hibernate methods.
Make sure the installer, startupBean and serverManager.getidentity() are all using the same approach for determining the default server name, otherwise duplicate entries end up in the rhq_server table, which can cause problems.
Merge branch 'master' into rhq-on-as7
Merge branch 'rhq-on-as7' of ssh://git.fedorahosted.org/git/rhq/rhq into rhq-on-as7
Bump up the version on our new common module, it didn't get done by the merge since it doesn't exist in master.
Add the new script-modules section as a downloadable-resource
Update script updated in obsolete container module
-Remove the obsolete "container" module, it is replaced with "appserver". -Remove the obsolete "installer-war" module, it is replaced with "installer".
Make sure CLI now has jboss-logging jar
Fix issues in the RemoteSafeInvoker for remote clients - fix JNDI lookups given new JNDI naming for AS7 - this required sending the in fqcn of the remote from the client, which we previously did not do. Remote clients will need to run on compatible CLI/version. - do a better job of validating whether we are invoking a valid remote method
Eclipse generated a change but I'm not sure why.
Fix an issue in getMeasurementScheduleCompositesByContext due to Hibernate4 changes (which require some more strict handling on JPQL). - Also, remove some unnecessary and misused constants in in MeasurementScheduleCriteria
BZ 814305 - Dealing with singletons in inventory] Fix an issue with the previous fix to handle the case where the parent did not actually have to worry about singleton children. Also, defer calling super's onInit until the subclass is done with it's work.
BZ 814305 - Dealing with singletons in inventory] Fix an issue with the previous fix to handle the case where the parent did not actually have to worry about singleton children. Also, defer calling super's onInit until the subclass is done with it's work.
Start migrating RHQ Server support to AS7 plugin/platform - Identify the AS7 process as an RHQ Server and tweak the resource name in discovery - Have the installer create a default rhqadmin mgmt user so that the RHQ Server resource will allow for an immediate connection with the plugin - Update the runs-inside elements in the rhq-server plugin desc to use the AS7 plugin *** The service are not yet integrated properly
alphabetize provided dep versions for easy visual scanning
Working towards AS7 based RHQ Server plugin and MBean management approach for AS7 plugin.
Clean up rhq-server plugin - change "jmx mbeans" server to "RHQ Server Subsystems" service - package jmx plugin and ems jars in the plugin to get around classloader issue where runs-inside plugin does not have same plugin dependency as the injecting plugin. - init management host,port,user,password plugin config props from the parent AS server. (port gets parent port + 9) - Add code to only add "RHQ Server Subsystems" child resource if the parent AS server is the RHQ Server. - Change resource key from "JBossAS7JMX" to "RHQServerSubsystems"
Clean up rhq-server plugin - remove some commented code - fix the default port to be the factory standard
remove dead src dir from .classpath
Replace the old domain AbstractEJB3test infrastructure (embedded jboss) with a new, Arquillian-based, approach. Lots of updates. Some test classes not yet passing due to issues with JNDI lookups in @BeforeMethod.
remove redundant/unnecessary properties
Implement new approach for @BeforeMethod seeing that Arquillian with testng calls this method twice, once in container and once out of container. Ignore the out of container invocation. Do this by having AbstractEJB3Test impl the @BeforeMethod and let subclasses declare simple (non-annotated) overrides for beforeMethod (and afterMethod)
- Add an inContainer() method to be used by test classes that need their own @BeforeXXX annotated methods if, for example, they need to apply attributes, like groups. - Add an executeInTransaction() method that allows commit vs. rollback when complete. - Update the affected test classes.
use the determines datasource
update postgres ver for Eclipse
BZ 865167 -Resource with AvailabilityType of NULL is being returned to UI resulting in ResourceDatasource.copyValues throwing org_rhq_core_domain_resource_Resource_currentAvailability is null] I believe this is the cause of the failure. In certain cases a legacy db may not have an rhq_resource_avail record for a resource. It is now eagerly created but in the past it was possible to have resource without the record. The old codepath for initializing resource avail for these resources had an issue.
update eclipse to arquillian 1.0.2.Final
Domain EJB Jar tests - Change deployment to be an EAR instead of a JAR because it can package the 3rd party dependent jars in its /lib dir. - Use the shrinkwrap maven resolver stuff to help package the 3rd party jars - minor - refactor the package for TransactionCallback to put it in the correct place for stuff used by multiple test classes. - Fix ResourceTest to properly clean up its test resources. - Fix ServerPluginTest to properly look up the DS - Simplify OSGiVersionComparatorTest to not use junit - Workaround Arquillian-as7 protocol bug - Add ejb-jar.xml to the domain jar deployment so it's a valid ear module
We are packaging the test classes and the actual domain jar in separate jars, and the former is a library and the later an ejb jar. So don't try and call protected methods just because we're in the same package.
SUREFIRE-836 is fixed in 2.12.1, upgrading to see if it helps me at all with some current issues.
Example of Arquillian LoadableExtension.
more .classpath stuff for new deps
Just get things to build after removal of embeddable ejb3 and new AbstractEJB3Test based on Arquillian. Not close to running, lots of band-aids and placeholders...
IN PROGRESS: Server JAR Test Migration to Arquillian - moving tests to new server/itests-2 module - not yet in full build - not yet working
One test passing but unprepare stuff not finishing smoothly. Still, progress...
First test class passing (RepoManagerBeanTest) - Clean up and add comments to AbstractEJB3Test - Convert tests to use executeInTransaction (although this isn't required but it's cleaner). - Tweak domain version of AbstractEJB3Test to avoid Class.forName, which is susceptible to NoClassDefFound issues
Get more tests running by altering BeforeXXXX/AfterXXX stuff as needed.
Protect toString() impls from detached entities
- Migrate drift tests and AlertSender tests from server/itests to server/itests-2. This merges them with the other former server jar unit tests that are now arquillian based integration tests. And consolidates itests based on AbstractEJB3Test. - Convert AlertSender test structure to one big test to get around the current @BeforeClass and shared instance var issues with Arquillian and the testng lifdecycle. - Add SystemProperties section to standalone.xml, making project.version property available to test code. - Add support for prepareTestAgents in the new infra - add support for BeforeMethod overrides offering up the Method param
make sure @AfterMethod runs always
More restructured tests now passing - Add utility in AbstractEJB3Test for persisting objects that need to be shared between tests. Since testng basically "new"s the test class on each test. - Add utility in AbstractEJB3Test for registering a custom service.
More test classes working - fixed bug in prepareForTestAgents - simplified other prepare/unprepare methods
More test classes passing - migrate JndiAccessTest from itests to itests-2 - fix issue in RhqManager - NOT TOTALLY PASSING, need Lukas - fix jndi lookup for getEntityManagerFacade - add more required test resources to teast deployment
More test classes passing - Add methods to register custom plugin scanner service - Add more test resource files to deployment -
- More test classes passing - SynchronizationManagerBeanTest running but not passing. BZ 874696 created - Fixed bug in ResourceGroupManagerBean
The rest of the test classes now updated and passing - Added a ton more test resources - Added a try-catch around afterMethod to report cleanup failures - remove preparePluginScannerService(), this service is already deployed by default in the rhq ear. Leave only a way to override with a customized service.
Reintroduce server/jar to the source tree. Server JAR tests now contain only unit tests, defined as non-container based tests, and not integration tests, defined as tests that extend AbstractEJB3Test (and require the AS7 container).
Update the PERF tests for new env but they are not working. Adding these to BZ 874696
Arquillian extension (from https://gist.github.com/3975179) to provide a brute force way of running multiple test classes against a single deployment. Seems to work for domain but still looking at issues in server itests.
Second pass, running all tests, issues with test interactions, etc - ensure all overridden before/afterMethods are declared 'protected', not 'public' because public methods are implicit @Test methods if the class is annotated with @Test. - Exclude the tests (via the surefire config in the POM) we know are not passing and will need to be looked at after we merge. - Fix test interaction issues for the tests extending AbstractDriftServerTest - remove use of InitDB annotation, which is not conducive to test interaction between test classes - Fix a few more places where we assumed test execution was done in the standard file hierarchy (for example, assumoing target dir, etc) We now use java.io.tmpdir for any file system interaction. - Add in a few more missing test resources - Figured out how to add test-jar deps into the deployment. See rhq-core-client-api in AbstractEJB3Test. - -
beef up error logging by adding exception message
More eclipse classpath updates
Fix a compile error in this plugin. Is this right?
BZ 834019 -Measurements not collected for agents with a large number of scheduled measurements] Revisit the solution for this BZ and improve by adding more randomization to the reschedule times. - update test accordingly - add some debug and trace level debugging around late collection handling.
trivial - enhance logging slightly
Fix the test exclusions
Fix the DB cleanup, it was leaving all kinds of crap around.
Fix mock code to use AS7 JNDI naming
BZ 877179 -revert changes to plugin api signature for PropertyDefinitionMap.getPropertyDefinitions()] While maintaining the fix done for Bug 786416, restore the public API that was broken in the original fix work. Also, maintain as much of the API introduced in that fix work just in case it is now in use.
Allow the tests to run against postgres or oracle. For oracle you need -Poracle.
BZ 878224 -Updated alert defs may not fire in an HA environment] When setting the server status dirty to notify the need for global condition cache refresh, update all servers. The global condition cache is supposed to be replicated across HA servers. Otherwise, different servers will have different condition sets generating unexpected results.
Merge branch 'master' into rhq-on-as7
Merge issue resulting in javassist.version property being defined twice. Jacoco seems ok with dialing back to the version provided by AS7, so settling on that version.
Replace hardcoded DS creads with our standard datasource property values.
Fix various pom warnings that maven was complaining about
remove the obsolete enterprise/server/itests module
replace server/itests with server/itests-2 in server modules
We can't use the SecureidentityLoginModule in our tests because we need to use the values set for rhq.test.ds.*, and the password will likely not be encrypted. For example, the Jenkins settings.xml files will not have encrypted DB passwords. Trying IdentityLoginModule in the hopes that that is same as SecureIdentityLoginModule but skipping the encryption step.
Still trying to get the unsecured DS connection right...
Add some logging to help figure out jenkins failure.
make sure surefire has the necessary db env vars since test execution may be forked and lose the context of the original process.
I have a feeling that we're not getting the translations we expect due to the "default-value" syntax here. These properties should be set anyway, so eliminate the defaults.
Merge branch 'master' into rhq-on-as7
Add a description to config prop that community member had to figure out looking at code.
Fix version of javax mail to be the version provided by AS7.
Make a note that this src branch should go away after the jndi asccess tests have been fixed. The new home is in server/itests-2.
The metadata update integration tests were not actually running. We had false postives because the BeforeMethod was throwing an exception. When this happens Arquillian (1.0.2) swallows it and passes the test (not good). Made several changes to these test classes to get them running although there may be test interaction issues still. Checking in now to see how things go in a full jenkins run.
Add support (hopefully) for EAP 5.2 in as5.itest profile.
More work on the metadata tests. - better protect against test interaction issues by using explicit "AfterClass" stand-in tests for each subclass. - work on PluginManagerBeanTest such that it registers plugins to be enabled before trying to disable them. - Not passing, maybe a real problem?
trivial - remove a debugging sleep stmt
Server i-test work to address many test interaction issues ( ** There are still some test failures ** ) - thin rhq.ear by removing the services SAR. The standard services actually just get in the way, tests are responsible for prepare/unprepare of any services they need and should not depend on the production services, which may perform undesired work anyway. - Add AbstractEJB3Test.preparePluginScannerService(), which will prepare a default test deployment scanner, and make use of it in more classes that need that service deployed, and were previously depending on the production deployment scanner. - Apply more consistent temp directory usage by settling on java.io.tmpdir/rhq/<class.simplename> as the convention. Add AbstractEJB3Test.getTempDir() and use it ubiquitously. This consolidates the test files under and "rhq" root dir and also identofies them by creating class name. - DatabaseAndFilePluginDeploymentTest - make a non-test method private, not public so it doesn't execute as a test - remove the obsolete OldAbstractEJB3Test class
Variety of work to eliminate itest interaction issues (more work still in progress) - Improve the cleanup in AbstractDriftServerTest. The cleanup here was very aggressive, wiping pretty much all of the drift tables, including any data still in use by other test classes. Added more easily identifiable naming for the test drift entities to allow for more targeted cleanup. - Add itest.debug support in itests-2 (from mazz's branch work) - Use more unique (avoid clashing) and identifiable (trace back to creator) plugin names in test plugin xml files.
- override AbstractEJB3Test.afterMethod() as opposed to an explicitly annotated method
trivial, comment re: potentially unnecessary call
Fix test failure, ensure that existing plugins are scanned and registered prior to easch test.
Adding in the missing server plugin service prepare/unprepare.
First really clean run of server integration tests (itests-2) - Fix another issue in AbstractDriftServerTest cleanup code - Fix additional cleanup issues where we needed to deploy ytest services - uninventoty requires test agents (comm) service - system config requires (drift) server plugin service - Fix some JDoc in AbstractEJB3Test
Disable the indi-access integration test. This test class relied on the legacy AbstractEJB3Test but used it only in a superfluous way. As opposed to migrating this to itests-2 to use the new AbstractEJB3test Lukas is working in a branch on a version that does not need it at all. We'll re-enable when that branch is merged in. The current version is causing some build issues due to obsolete pom refs.
Remove unnecessary AS711 unzip, the integration tests have moved to server/itests-
Moving the db-upgrade jenkins job to 2.4.0 as 2.3.1 is no longer supported by db-upgrade.xml.
Augeas plugin coding error I'm not sure what effect this change may have, but the code as written was clearly wrong and so I changed it.
API Checker detected a change in this constant, no reason not to make this an ObjectName at decl-time, I think.
API checker work for latest release, post rhq-on-as7 merge - Add latest intentional API changes files - Resolve some maven complaints - Add server/jar includes to check goal in pom
Domain itest changes - remove the -ds.xml files and perform datasource decl in the standalone.xml (via the pom), analogous now to the way it is done in the server itests. - beef up the error handling in AbstractEJB3Test _before/_afterMethods to more readily flag exceptions thrown in overrides. - similarly beef up domain test Before/After methods that don't leverage the overrides - remove the SetBigStringTryClob=true connection prop for oracle, trying to set it was causing us to hit https://issues.jboss.org/browse/AS7-4673 and it seems no longer necessary with oracle11 JDBC driver (http://docs.oracle.com/cd/E11882_01/java.112/e16548/oralob.htm#CHDGEIHF). - fix some domain tests that were manhandling the entitymanager
Oracle testing - Add no-tx-separate=pools=true for ora XA datasource to avoid connection issues - remove SetBigStringTryClob conn prop for oracle because we shouldn't need it with the Ora11 driver and trying to set hits an AS bug - apply necessary changes to installer as well as test code - fix issues with aggressive and sometimes non-functioning cleanup code in MeasurementDataManagerBeanTest - Add oracle driver jar to test deployment to support dbunit Also - Add support for recovery-plugin class for postgres to avoid mem leak in 7.1.1.Final. (note - the actual use of that hook will be in a subsequent commit)
installer work - remove debugging system.out.println - protect paths with quotes in rhq-installer.bat - reduce dev installer retry from 5 to 3 if the remote admin endpoint in AS isn't ready - not sure, this retry seems to do some wonky things on win
remove recovery plugin from postgres XA test datasource, not needed since the mem-leak issue is not relevant in a test run and the declaration of this workaround class is problematic.
Replace more instances of straight division with approach that maintains precision.
Fix an issue where the domain and server itests needed one of -Ppostgres or -Poracle to be defined. Now it will default to postgres unless -Pitest.oracle is set.
Add ability to specify Double comparison tolerances in our matcher test utils. Hopefully this will handle the minor roundoff errors that are sinking some some tests. When we move to cassandra we may be able to remove these tolerances from the tests, given more precision in our math.
Make sure oracle driver dependency is in place for -Ddbreset
Add test dep to ensure the oracle driver can be found when testing against oracle.
BZ 884338 (not a fix) - Agent's availability report is ignored due to bogus stale resource error] I failed to reproduce this issue but: - Add some better logging and some attempted repair code for this situation - Add a couple more tests
Avoid the unzipping if it is already unzipped. It takes time for unzip to actually do all of the file compares just to know it doesn't need to add anything.
Try forcing ojdbc driver resolution usinbg the local repo since it can't always be found in profile-defined repositories.
BZ 888927 -Availability duration conditions limited to one alert definition per resource] Add the hooks to further qualify AvailDuration checkCondition quartz job trigger names with the alertDefId. This will allow avail duration conditions on any number of alert defs for the same resource.
Add API diff clirr exclusion
LDAP support in RHQ-on-AS7 - fix up the client API to support: - removal/replacement of the existing security domain for RHQ logins - allow for multiple login-modules for a single security domain - fix up the CustomJaasDeploymentService to play with the new API to be able to add or remove the ldap support based on whether it is enabled.
In dev mode we try and run the installer automatically for the dev, if it hasn't been run already. We do this in the "background" so we use the same window (the /B option). Change it to suppress output (> nul) so it doesn't interfere with the actual server startup. And change it to suppress input (< nul) so it doesn't hang waiting for "Terminate Batch Job (Y/N)" when it completes.
fix merge issue
BZ 886850 - 4.6 on as7: different rollback semantic for Exceptions] It seems AS 4.2.3 did not wrap RuntimeExceptions as EJBExceptions as per the spec. Neither did it rollback a supported transaction. Furthermore, RHQ code was written assuming the bad semantics were correct. AS7 corrects the handling and therefore the RHQ code is now in error in certain places.
Fix a cleanup issue in the MetadataBeanTest base class. It made the mistake of using an instance variable to persist information across multiple test executions. Arquillian with testng effectively "new"s the test class for each test, so instance variables are initialized each time, no information can be carried over from test to test in that fashion. Changes to persist the necessary data to disk between test runs, using the write/read/delete utils in AbstractEJB3Test.
We need to encode the ldap server bind pw in the AS standalone.xml config. And then decode the value when binding to the server and validating users.
Fixed LdapGroupManagerBeanTest to properly update all of the configured ldap settings and tests are now passing using the mock context.
Fix method supplying the LoginModule control flag string. Prior impl was weak and sensitive to locales.
Update the postgres driver version because it contains a change that may solve a memory leak issue in AS7 DS handling (see https://issues.jboss.org/browse/AS7-4975)
Update version of windows java wrapper from 3.2.0 to 3.3.1 to solve issues with exit code handling, which prevented proper restart of the rhq server after an rhq-installer --reconfig command, or any other non-success exit code situation.
Upping the JSW version for the agent to 3.3.1 to be the same as the recently updated server.
Upgrade Arquillian BOM from 1.0.2.Final ro 1.0.3.Final. This seems to work in my local tests and supposedly fixes an issue requiring a workaround for jenkins [1]. So, take out the workaround in AbstractEJB3Test on the hopes that Jenkins will run.
Arquillian 1.0.3 bom and its packaged update to the shrinkwrap resolver did not seem to solve the "settings.xml repositories" problem as hoped. So, restoring the workaround.
Fix issue with the jndi names used for the report exports.
Try to make the Welcome message portlet cleaner and more useful by removing useless canned text and offering a few basic links.
Don't let dashboards come back in random order. Order alphabetically by name.\ This also provides a poor-mans's way of starting with a favorite dash, or ordering/grouping in a certain fashion.
Initial conversion to Shrinkwrap 2 API
Shrinkwrap-resolver 2.0.0-alpha-7 is released and should contain fixes for our reported issues. Update version and retest.
Add back in the test-jar dep that does not get pulled automatically. The new version of shrinkwrap did not change the behavior. I don't know if it's a bug or not, maybe not.
trivial code tweak while debugging tests
Updates to move to Shrinkwrap 2 API and get associated SW fixes. - change poms to use the shrinkwrap-resolver-depchain pom dep in order to suck in the necessary jars - other minor 2.0 API tweaks
Let arquillian/shrinkwrap dependency mgmt be handled by the root pom, because there is specific ordering in place to use newer shrinkwrap resolver than what the arquillian-bom pulls in.
BZ 797331 - RHQ discovers network adapters that don't exist] It is OK to DISABLE some NetworkAdapters at component start. But do this once-per-start check in a thread because for DOWN adapters this could be a slow call, also the disable() method requires a server round trip. Together this can seemingly hang agent startup (actually plugin container startup) as component starts are done sequentially.
BZ 901700 - RHQ Server resource being named as vanilla standalone AS7 on Windows] Add logic to make the distinction on windows. Since the .bat files don't use the JAVA_OPTS var as does linux, a different approach was needed.
Update Eclipse .classpath to latest shrinkwrap resolver
trivial, remove commented code
Try to fix issues in AS5 itests due to shrinkwrap upgrade
BZ 902406 - large number of services cause inventory report to not get processed] Break up the single-transaction approach of DiscoveryBossBean.mergeInventoryReport() into a transaction for each resource being merged. This should prevent a long-running transaction time dependent on the inventory report size.
BZ 902406 - large number of services cause inventory report to not get processed] Since we're in the act of merging into inventory, disable the default InventoryStatus filter on the Resource criteria searches.
BZ 902406 - large number of services cause inventory report to not get processed] - Clarify behavior of ResourceManager.createResource in the jdoc. - Fix testcode depending on false assumption of cascase persist
Rearrange things to allow for batching resource merge based on the rhq.server.discovery.merge.batch.size sysproperty. The default is 200 per transaction. Also, a variety of other optimizations to reduce the pull of unnecessary lazy data as well as the number of SLSB calls needing to establish their own transaction.
trivial - fix log message
Change batch looping strategy. Simplify range handling and perhaps help free memory as we chcunk through the batches.
Perf enhancement to mergeInventoryReport, replace Criteria calls with local named Query use because it ended up just being an ID search with no optional data.
Added some more test code to try and repro a type removal issue. It didn't force the issue but it's still useful test code.
Change some test logging from info to debug
More mergeInventoryReport speed - add map cache of attached parent entities to avoid redundant DB fetches - replace named query fetch with simple em.find where possible - trivial, rename some private methods that were named inappropriately
Even more MergeInventoryReport speed - avoid slsb call overhead by performing relevant query locally - remove unnecessary em.find - convert some em.find calls to em.getReference
Hopefully final tweeks to MergeInventoryReport
More test interaction work for the server itests. Many tests need a work area on disk. The test area being used was typically ${java.io.tmpdir}/rhq/testClassName. There were two issues with this: - Different test jobs running concurrently would end up using the same work directory. This is something that may have been happening on Jenkins. - An mvn clean had no effect on anything hanging around in those directories, so if test cleanup was poor it may have left legacy data that could possibly impact future test runs.
Fix test failure, when I beefed up type removal checking I neglected to notice a dependent test that also needed updating.
get rid of hardcoded project version, an oversight from long back
Trivial - some jdoc improvements
BZ 906500 - quartz throwing NoClassDefFoundError: oracle/sql/BLOB] I believe this is fixed. I added an optional module dependency on the Oracle driver, for the server jar. Our datasources had no problem, they are defined correctly to find/use the oracle driver, but since Quartz wants to perform a test directly against a driver class, we had to ensure the driver jar was accessible to the classloader.
BZ 907407 - RHQ master build fails starting - Failed to load module: deployment.rhq.ear.rhq-enterprise-server-ejb3.jar:main] Server JAR has a dependency on the OJDBC driver because Quartz sometimes needs to load a BLOB class from the driver. Postgres installs will not have (and don't need) the Oracle driver, so we declared the dependency "optional" in jboss-deployment-structure.xml. But "optional" just means AS7 will ignore the fact that the module may not exist at all. Our build and install process is simplified by always creating the oracle driver module structure. But "optional" will not ignore the fact that the module directory is empty (no jar file). So, this commit ensures an empty/mock OJDBC driver jar will be placed in the module directory if the driver is not available at build time.
BZ 907407 - RHQ master build fails starting - Failed to load module: deployment.rhq.ear.rhq-enterprise-server-ejb3] A fix to the original fix
Fix a bug in the ResourceType pojo to ResourceType entity conversion code. Criteria String matching is fuzzy (i.e. substring matching) by default. This needs to have the Criteria.setStrict(true) added to make it exact.
Make sure Criteria.setStrict(true) is set when using a Criteria search where exact matches are expected and substrings could return unwanted results.
Changes to Plugin and ResourceType deletion - refactor PluginManagerBean.deletePlugins for scalability and perf by avoiding an umbrella transaction and allowing for deferred resource removal. - refactor ResourceMetadataManagerBean.removeObsoleteTypes to utilize the same approach as plugin deletion for type removal. - fix bug in ResourceManagerBean.uninventoryResources() that did not utilize a new transaction for each resource uninventory. - Various perf enhancements. - avoid unnecessary and timely authz checks when uninventory is being performed by overlord/inventoryManager - use overlord a little more efficiently in places (but still take into consideration its session timeout limit). - Add a little defensive coding to AsyncResourceDeleteJob to log less if the doomed resource has already been removed. This can happen when the quartz job runs at the same time as a runtime plugin update that removes types.
Fix issue with SLSB invocation.
Scalability work to chunk processing of availability reports. Start outside of a transaction and execute one transaction per chunk.
Add Criteria.getSingleResult() convenience method.
Simplify availability logic by removing the legacy assumption that an availability check may not report avail changes to the server. Previously the caller of the avail check could optionally send the report. Now avail checks will always forward the resulting report to the server (although empty reports will not actually be sent). The legacy behavior was likely due to the old always-perform-a-full-scan approach. The new behavior gives us a scalability enhancement. We no longer have to follow up on-demand avail checks (like the 'avail' prompt command or agent operation) with full reports, which can be costly for agents with large inventory.
Fix a transactioning issue with agent backfill. It was possible for the Agent to be marked backfilled in one transaction yet have the actual avail changes fail in a second transaction, leaving the dead agent and its children with UP avail and no longer eligible for backfill due to the flag being set.
To maintain the test dep ordering, which is sometimes important, and definitel is important here, move all of the test deps to the new test.dependencies profile.
Fix test given slight change in avail behavior. In commit c516ad09fcebcedfd1a505d6bfd434997ac6642d we stopped kicking off an immediate avail scan when new resources were discovered, simplifying the code, and approach, to just wait for the next scheduled scan, which likely occurs nearly as fast. From the test's perspective this meant one resource, the platform, had its avail change at a slightly different time.
Fix issues in mergeAvailabilityReport "repair code". This is code that attempts to fix problems in the Availability records, if they somehow get corrupted. In general, it doesn't execute. But if and when it did, it wasn't repairing things as well as it could. This makes improvements.
Fix test to reflect the improved behavior when mergeAvailabilityReport performs Availability "repair".
Inventory sync no longer schedules avail check runs in response to unknown/modified resources, instead just waiting for the next scheduled run, which is soon given only 30s intervals.
Fix resource bulk delete when the resource is a member of groups, on oracle. The issue is that a native Query is used that returns resource group ids and since Oracle represents integers as Number, it actually returns a BigDecimal, not an Integer, for the Ids. Deal with it in the generic way we deal with DB type differences.
BZ 912871 - Perf: resource configuration update taking a lot of backend resources] Scalability work for resource config update. We already added concurrency limits that should throttle how many config updates the server gets hit with at one time but we also want to see if we can relieve the intense banging on the db and speed up the update time for each resource.
Add another test for the work done in Bug 912871.
This should allow us to run Arquillian's spawned AS7 instances (for domain and server jar itests) on non-default ports. This can be helpful in environments where you want to run multiple AS7 instances at the same time.
Fix tests so they run on windows as well.
Don't access the resource to delete transactionally and hold it while trying to then uninventory it via slsb calls. This could cause a deadlock.
Jirka Kremser <jkremser@redhat.com> (98):
The plugin descriptor has been restructured, new metrics have been added
Process scan strings now math the hadoop nodes
Metrics subsystem is now collecting the data via JMX
Metric definitions for server of type JobTracker
Added metric definitions for TaskTracker node; typos
Adding the class infrastructure for performing the operations
3 simple operations on NameNode have been added
New set of operations: start, stop for all node types; list job queue for job tracker
change of plugin.displayName (hadoopPlugin -> Hadoop)
Support for percentage metrics from interval (0,100); new operations: rebalance
New operation – kill job
BZ 858282 - Globally uncaught exception when creating duplicate role] If rhq_role_name_key db constraint is violated, more informative message is returned to the client. Instead of "Globally uncaught exception..." with db implementation details. The same thing for user creation.
BZ 857423 - [EAP 5.1.2]inconsistent result of shut down operation when using JMX and shutdown.sh script and server is already stopped.] Check for return code was added.
BZ 860818 - operation schedule doesn't show proper "repeat every" value] The "seconds" unit type was used no matter what actual units were selected. Now when the operation is scheduled, the "repeat every" value is persisted to DB in millis and when displaying the details about the op., the millis are converted to human readable form.
BZ 857462 - RFE: Disable 'Purge' button on destination level when no deployment is LIVE] If at least one deployment is LIVE the button is enabled, otherwise it is disabled. The call is made, when constructing the UI components.
BZ 864870 - [CLI] it is possible to create Drift Detection Definition without manageme drift permission] Authorization has been done on the SLSB layer; user w/o MANAGE_DRIFT can still read the drift definitions (C,U,D operations are forbidden)
BZ 866461 - It is possible to change the drift detection without MANAGE_DRIFT permission] Button
ablement has been modified accordingly; if user has read access only, in there are "view icons" insdead of "modify icons"
BZ 870533 - Server silent/autoinstall sets http and https ports to sslsocket port value] The code responsible for overriding the "rhq.server.startup.web.http
.port" properties with the port # taken from "rhq.communications.connector.bind-port" has been commented out
BZ 871407 - NPE on auto completion for javascript object] the contextComplete() method now returns -1 instead of throwing NPE
BZ 872185 - Traceability of particular measurements in agent's log] If TRACE log lvl is enabled, each measurement is logged into the agent's log
Merge feature/hadoop-plugin into master
i18n for the new gwt HAAC
skeleton implementation for server view in High Availability Administration Console including the inner class – datasource for the Server entity
new GWT service for CRUD operations with RHQ server
exposing the CloudManagerBean via CLI
Greasemonkey script for FF for typing in the default credentials after pressing Alt + C (can be used also in Chrome with Tampermonkey extension)
BZ 850818 - Globally uncaught exception on clicking OK button while adding CLI Script alert notification to the alert definition] Another possible scenario where NPE was thrown was fixed by this commit.
detail view for server node, also the datasource is now a standalone class
Merge branch 'master' into jkremser/ha-jsf-to-gwt-rewrite
AgentTableView class added including the data fetching via datasource
Detail of server is now fully functional, it also displays the list of connected agents
ServerNodeDatasource has been split into two separate, AgentDetailView added, some refactoring of existing code
hyperlinks across table data and detail information rendered on detail pages are now working
partition event table, datasource and finder for PartitionEvent by criteria have been added
Merge branch 'master' into jkremser/ha-jsf-to-gwt-rewrite
new methods for EventPartition table, UI imporovements for event partition view
i18n for Partition Events
criteria based finding is now fully working for Partition Event entities
Partition Event detail view with the list of agent-server assignments has been added
Affinity Groups view has been added, creating and deleting the groups is working
Affinity Group detail view added; refactor of data sources for entities derived from the Server class (various composites)
Assigning servers into affinity group + new methods in GWT service for this purpose
More work behind assigning the servers into groups; added logic for updating details of existing server and existing affinity group
Buttons were moved to tool strips; button enablement is now refreshing properly even after the modal window is closed by "X" button
Merge branch 'master' into jkremser/ha-jsf-to-gwt-rewrite
adding comment
BZ 886287 - NullPointerException when another exception should have been thrown] quick fix, lazy evaluation of Java should ensure the prevention of throwing NPE
Revert "exposing the CloudManagerBean via CLI"
ServerCriteria and Server finder have been added
AgentCriteria and Agent finder have been added
Assigning the agent to the affinity group; some refactoring (renaming AgentNodeDatasource -> AgentDatasource, AgentNodeDatasourceField -> AgentDatasourceField)
i18n for error messages, exception handling for agent and server views
the rest of l10n for error and other messages, some javadoc
Modal window for assigning the servers/agents into affinity group has now the buttons in the ToolStrip
Deprecation of newly unused methods (were used from portal war before), checking the rights
Old JSF views are not accessible from UI anymore, instead the new GWT views are on their place (without the "GWT" suffix)
AuthZ for CloudManagerBean (adding Subject param to methods)
Merge branch 'master' into jkremser/ha-jsf-to-gwt-rewrite
Overriding getBasePath() methods for allowing double clicking on table record for detail views & some JavaDoc
JavaDoc for SLSBs (describing needed permissions)
ServerCriteria now allows filtering by multiple operation modes, AgentCriteria allows filtering by agenttoken
Refactoring: renaming "CloudManager" to "TopologyManager".
Merge branch 'master' into jkremser/ha-jsf-to-gwt-rewrite
Adding renamed servlet (TopologyGWTService) to the web.xml
Merge branch 'master' into jkremser/ha-jsf-to-gwt-rewrite
BZ 888369 - Time range setting on monitoring table should be more robust] Adding form validators for the format of the times as well as for the start time < end time; if the end time is after the current server time, the end time is set to the current server time. This caused the NaNs in the table before. However, It is still not perfect, because the user with bad time settings will have prefilled values in the start/end time fields. The simple form ("last n hours") on the monitoring tab is working as it should.
Remote endpoint (the connection string) information has been added to the agent detail page. It allows to deduce the protocol, numAcceptThread, maxPoolSize, clientMaxPoolSize, etc.
set initial criteria for member servers in affinity group
BZ 893112 - UI widgets incorrectly hide when creating operation schedule] Instead of using Canvas.setVisibility, methods Canvas.show() and Canvas.hide() are now used
BZ 889108 - Scheduled Operations Portlet does not show operations scheduled on the group The support for operation scheduled on the groups has been added for the portlet. Now, it shows both types of operations sorted by the next exec. time.
quick fix (typo): changing "unitialized" to "uninitialized".
BZ 895664 - Globally uncaught exception on navigating a resource after adding 'Resource Metric Graph' portlet to dashboard] Gflot now doesn't use its own version of jQuery. This bug was caused by the fact that Gflot lazily loads the jQuery to the page, so if user visited the dashboard before the page when Sparkline is used (resource summary tab), the exception was thrown.
BZ 878117 - Empty group is displayed among compatible groups] Added support for recalculating the group's resource type after resource uninventorization.
Trivial: comments added for changes made in commit 091b0903f
BZ 880960 - SNMP Trap notifications sent from JON using incorrect port] typo: default port for snmptrapd has been set to 162
Added itests (regression tests) for BZ 878117
[BZ 845363 - Bundle Deployment - Newly created Resource group which contains \~100 resources is not being <discovered> during the destination creation|]
The Bundle Deployment Wizzard doesn't let the user to create a mixed group or a compatible group with a resource type which can't serve for deployments. The check is done using the ResourceTypeRepository. I18n and l10n added.
Improvements in the "Edit Row" popup window for modyfing the configuration: close button added; buttons moved to the toolstrip
Adding "Last Availability Ping" to the agent detail page in topology section and to the agent detail page in Inventory->Agent tab
refactoring: renaming class "ResourceResourceAgentView" to "ResourceAgentView"
refactoring: renaming class "ResourceResourceAgentView" to "ResourceAgentView"
Schedule an operation using cron expression via JBoss ON CLI] Methods scheduleResourceOperationUsingCron() and scheduleGroupOperationUsingCron() have been added to the OperationManager.
BZ 880819 - RFE: expose GroupDefinitionManager remotely] Subset of methods defined in the local intercace has been exposed via the remote intercase and registered in the RhqManager.
BZ 880819 - RFE: expose GroupDefinitionManager remotely] No need for exposing the "getById()" method. There is a common criteria finder available and other remote managers don't expose this getById() methods either. We should keep the api consistent.
BZ 858282 - Globally uncaught exception when creating duplicate role RoleManager now explicitly checks whether the role with the same name already exists (without trying to create the role first and then catching the exception like it was done before)
BZ 906754 - It is not possible to delete a resource group which is assigned to more than one roles. ConcurrentModificationException was thrown when iterating through the set and removing its elements at the same time. I created a copy of the set.
BZ 907037 - Missing Column in Schedules Table] Schedule ID column added. By default it is a hidden field and if needed, it has to be enabled in the table header menu.
BZ 902823 - Nonsense data returned as backing content of non-content-backed resources The UnsupportedOperationException is thrown when the resource is non-content-backed. I was checking also the usages of the method and it seems to be ok.
Adding the Hadoop plugin to the validate-all-plugins module.
Fixing the build. It was failing because of message "Component class org.rhq.plugins.hadoop.JobJarComponent for resource type Job Jar from plugin hadoop does not support the content management facet but defines package types.". The skeleton implementation has been added.
BZ 907407 - RHQ master build fails starting - Failed to load module: deployment.rhq.ear.rhq-enterprise-server-ejb3] The dummy oracle driver was created by 7zip, now it is plain old .zip renamed to .jar
BZ 893802 - [as7] Restart operation fails to start AS when shutdown hasn't completed when start script is invoked] If the AS is deployed on the same node as the agent, the check whether the process is runing or not is done in a loop once a second. This new check was added after the original (using the DMR api polling) check. In total, the AS has up to 20 second to finish its work and completely shut down.
Setting auto fit for button's width when title consits from more than 15 chars
BZ 893802 [as7] Restart operation fails to start AS when shutdown hasn't completed when start script is invoked] I have added NPE check, because I got NPE in certain circumstances.
BZ 902823 - Nonsense data returned as backing content of non-content-backed resources Instead of throwing the exception, the null is returned. Now it's in compliance w/ the the method's jdoc.
BZ 909157 - Operation parameters are not shown in schedule operation details.] If the method refreshOperationParametersItem() is called when scheduling a new operation, the existent parameters are not forget.
BZ 913121 - Manual import from resource tree's context menu opens the create child child wizard instead] Both import and create option led to the openint the same wizzard. Also some typos in javadoc has been fixed.
Unifying the width of buttons in the toolstrip (removing autoFit by default, autoFit property is set if the title length is greater than 15 characters) and removing unused class SimpleCollapsiblePanel.
John Mazzitelli <mazz@redhat.com> (313):
the start of a GWT installer. this builds and runs for AS7 only (it uses EE6)
this illustrates how we can have a war obtain a co-located ModelController management client without going through a remote connector. the getAppServerVersion method doesn't work correctly yet, but that is only because the ModelNode operation is incorrectly formed. But this is showing the ability to successfully send management API requests to the app server.
illustrates how we can use the co-located management client to obtain simple things like the AS7 release-version string.
show how to get attributes for resources below the root
refactor client stuff so we can more easily move it to common module later
more prototyping - i have code that can deploy the RHQ security domain - which is necessary before we can deploy datasources
It's happening again - I'm building out an installer with no one from UXD walking me away from the ledge. Oh well, here's more of the new installer prototype. I'm sure it will be the foundation of the new installer and people will hate its UI design. At least it doesn't have a red background.
trying to get the database portion of the installer to work. half way there
layout of the system settings
the workflow in the gui is mostly done i think. just a matter of getting the server-side code to install everything that the user asked for.
fix typo
the start of the new container build ant script. still have a ways to go before this produces a functional rhq server on as7
getting the startup scripts to work with AS7.
more container build stuff
get the rhq server on as7 running on windows using the java service wrapper - i tested this successfully on win xp, both in console mode and installed as a service
rename the installer module and the war itself so it doesn't have the annoying -war.war file ending
fix eclipse classpath to use the new installer module
remove the webdav module - it is obsolete, was never in the distro and REST is now the way to obtain this functionality
in all pom.xml files' dev profiles, need to point to the new AS7 deployments folder
fix some poms implement a method to obfuscate a pw using picketbox
add de-obfuscate method
fix the pom so -Pdev works when doing both clean and install and when the appserver is already running with the app already deployed or not
force another connection test if changing advanced view.
opps. forgot to take this out
more on the new installer
gwt installer is getting close to doing dbsetup/dbupgrade
gwt installer now does dbsetup/dbupgrade and sets up the HA server info in the db
more installer stuff
show messages during install
change the eclipse tool to build the new installer, not the old
add contextual help
installer now completely deploys datasources and everything they need (security domain, and child connection properties)
fix datasource
minor cleanup
get auto-installer to work with the new installer
when in dev mode, put the jars in the new ear location
setup the mail service
add log message re: the mail service being configured
workaround as7 bug
move all services that used to be deployed at .xml files to SARs embedded in the EAR this doesn't work yet and need to figure out what to do with mdb service
trying to get our MBean services to be deployed within an AS7 SAR
missed this in test
the JNDI urls are different in AS7
JIRA AS7-5343 forces us to explicitly replace ${x} ourselves
no sense leaving this in the .properties file - we don't want this to be anything other than 1 anyway
put our own JbossAS modules in a separate, top level directory. This will help support running RHQ in a separately installed JBossAS container (like from an RPM install)
get the modules path setup for windows (I need to test this, have to make sure the syntax using ";" is correct)
we need quartz jars in the ear
AS7 doesn't ship with Quartz jars anymore, we put them in the EAR now.
* enable AS7 debug logging when RHQ_SERVER_DEBUG env var is set * use the standalone-full.xml since that has hornet-q and things preconfigured * when using mvn -Pdev mode, set the auto-installer to run
after the installer finishes, make sure it deploys the ear have installer pop up messages if already installed or the install failed
looks like AS no longer supports body-content of none - it is "empty" now.
get MBean services to deploy - add a blank ServiceMBean to workaround JBossAS JIRA AS7-887
get the rest stuff to avoid spitting errors at deploy time
add "clean" option to rhq-server.sh - I'm seeing odd behavior trying to restart after deployments fail on previous starts. clean will remove all tmp and data directories so it starts fresh. it also removes log files
Merge remote-tracking branch 'origin/master' into rhq-on-as7
this legacy tag doesn't exist, AS7 reports it as deployment error, so don't include it
jndi names are different in AS7
looks like AS7 now requires this <interceptors> section in ejb-jar.xml which we didn't have before
comment out obsolete tags
JNDI changes for AS7
we use internal arjuna classes, so we need a dep on it
do not deploy the "client" because it has everything in it - including the RHQApplication which AS7 doesn't like (because the ejb-jar already deploys it)
another thing that should have been removed but was not
work around bug https://issues.jboss.org/browse/AS7-5343 so our SAR can deploy
our dev profile build in server/jar doesn't produce the proper ejb3 jar - its missing our top level files (rhq-server-version.properties is missing, for example, and that is needed by CoreServer MBean)
need this to workaround AS7-5336
set ejb version to 3.0 in the maven plugin config add apache commons loggnig dep so follow AS7 documentation
enable the datasource
fix eclipse tool to delete the plugins from the new as7 location
make sure the default JBoss welcome screen is disabled add the ability to reload the configuraiton at installation end also add API to restart the JVM because I have a feeling at some point, we'll need to use that and not reload
we are going to need hibernate API so add it to the dependencies of the rhq ear
rip out Seam since we don't want to upgrade it when porting the ear to run inside AS7. the only thing "broken" appears to be the content tab's file upload feature. that still needs to be fixed.
remove entitlement service - this is old PoC legacy code that was never implemented
remove Seam
remove the rest for the vestiges of perspectives this also removes the rest of entitlement server plugin
fix file upload
start using the jboss-deployment-structure deployment file, rather than the Dependencies: entry in manifest file because we need some features only the .xml file gives us
opps - build the new installer, not the old
* fix xmlns name and point to 1.1 version of schema * export module dependencies
have installer configure the deployment scanner so it lengthens the deployment timeout
fix core server MBean to use the new AS7 mbeans
fix agent update servlet so it lazily initializes. this way, we don't require the EJBs to be deployed beforehand. only when the first request to obtain the agent binary will the servlet initialize and by then the ejbs will have been deployed
fix the file path to match that of AS7
move StartupServlet to server/jar - it will be converted to an EJB 3.1 singleton startup bean
get StartupServlet to at least compile - but it doesn't do anything since its not an EJB 3.1. singleton EJB yet
add core-util as an explicit dep to installer
fix eclipse setup so it uses jdk1.7
fix many deps so they use AS7 deps and not AS4.2.3
this will no longer be a servlet, so rename it
convert the StartupServlet code to be a singleton startup EJB. None of the actual work has changed; we just moved this from servlet-to-startup EJB
this legacy code used old servlet 2.4 HttlUtils which has been deprecated and removed. since we don't use jsp tags in any pages we care about, just comment it out so this compiles. this legacy code will all be removed at some point in the future
the arjuna stuff needs to be "provided" scope
rest docs generator stuff should be provided scope
exclude some things from the ear lib that AS7 will provide
legacy portal-war doesn't have dep on ejb api anymore. remove usage of ejbexception this code isn't used anyway, just doing this so it can compile
need the ejb annotations like @ApplicationException
no need for the servlet 2.4 dep (we are using EE6 anyway)
rename the MBean interface so it becomes a valid Standard MBean
* add dependencies on the startup bean to the EJBs it needs. * make mbean server lookups get the platform mbean server * add the shutdown listener as a predestroy method on a singleton EJB
fix subject preference cache - making it a ejb singleton also made ShutdownListener a ejb singleton with PreDestroy method
this was blowing up startup. not even sure if this renderer is used anymore, but lazily look up this slsb to avoid the startup error
Merge remote-tracking branch 'origin/master' into as7mastermerge
i don't think we need these - this came over in the master merge, but since we require building on jdk7 right now, we don't need hib annotations on the classpath. webservices have also gone away in the as7 branch.
some minor pom issues
explicitly declare our transaction requirements on our singleton EJBs - the StartupBean should not be given any tx during init
fix eclipse classpath
rip out the latched service stuff - we don't concurrently deploy plugins anyway.
workaround AS7-5530 bug
add tests to the dmr client module
use ejb timer to perform the periodic plugin scan
don't throw NoResultException, just return null
the installer should skip if the ear is currently deploying
work around a deadlock inside AS7 when it tries to deploy our app
add logging
this MBean no longer exists in AS7 and I can't find any place in AS7 that tells me when it started. So I will use the RHQ MBean service that has this information
don't do anything about uninstalling the installer. I think we can leave our new AS7 installer intact. just commenting this out instead of deleting it just in case we do want to resurrect this (its possible we may end up wanting to remove the installer after it runs)
don't do jndi lookups, use injection
fix the EJB timers so they are not persistent
fix the login module so it gets the proper initial context
fix download locations
use the new filesystem layout of AS7
set the ports as defined in rhq-server.properties
clean up test pages and put server name in system settings this commit is broken - i'll be committing fixes shortly
use the new property names
fix control.jsp
forgot to set the server name form item
this looks like it should have been in the test resources - moving to correct spot
move alert email template to the proper place where AS7 can find it for us
fix the agent's libraries
let's stop pretending the ear name can change
in order to workaround https://issues.jboss.org/browse/AS7-5342 we are no longer supporting Properties as directly settable via our sar descriptor. Instead, the properties are in a .properties file now and our MBean will read in the properties and set the overrides that way.
fix agent update servlet
use constants
increment pom version
BZ 860807 ] have the else statement actually do something
Merge branch 'master' of ssh://git.fedorahosted.org/git/rhq/rhq
fixes BZ 860807
have no idea if these tests work anymore, but we know for sure the JNDI names are different in AS7
AS7 doesn't use jnp initial context factory
put this back the way it was
add the jndi context factory builder service, but comment it out because it causes the AS7 server to blow up at startup we need to figure out if we still need this and if so how to refactor it
port the DBInstallationTest tests
add more schemas to test
Merge remote-tracking branch 'origin/master' into rhq-on-as7
remove dup method that I screwed up with the last merge
fix the use-case when you specific -Pdev -Ddbsetup while building the dbutils module
try to resurrect embedded agent. this still doesn't work in AS7 due to JMX deps. I think its best to move the sar's embedded agent into an AS7 module
do not deploy embedded agent. it is not working yet
take out embedded agent properties definitions from installer constants
take agent-sar out of the build
add new as7-jmx plugin
don't bother with embedded agent during startup
fix the lib/ content
stub out the ClassLoaderFacet - we need this so we can point to the client jars at runtime, rather than ship the plugin with a specific client jar for a specific version of AS7
trying to prototype the new rhq-plugin. has a hack in it - we don't want the hack
fold the prototype as7jmx plugin into rhq-server plugin and remove as7jmx plugin
the name of the rhqserver plugin is different
forgot to update the groupId of the rhq server plugin
just add a comment showing the location of a bug with the BZ #
use constant for default port
the old remoting 2.2.2.SP8 stuff needs the old jboss-common jar. the new commons-core has a different impl of org.jboss.util.LRUCachePolicy; the old impl in the old jboss-common had a protected data member m_map that the new impl does not have. the old remoting subclasses LRUCachePolicy and that subclass accesses m_map; so it can't use the new impl; we must keep the old impl.
we don't bundle the connectors in with the agent anymore, no need for this dependency
Merge remote-tracking branch 'origin/master' into as7-master-merge
BZ 864213 ] fix the web resource's ability to collect response time data.
install the rtfilter in the RHQ Server so we can monitor it using the AS7 plugin's web resource's response time metric
BZ 864213 ] fix the web resource's ability to collect response time data.
we don't need the rtfilter stuff in our modules directory because we deploy the rtfilter directly in the EAR
limit what we monitor with rtfilter
the AS7 plugin code forgot to set the log dir in the plugin config. this is necessary if someone started AS7 with a different log dir sysprop (which RHQ itself does when starting its own AS7 server)
this isn't right - you don't map the context. go back to /*
BZ 864619 ] fix the defaults so the AS7 plugin looks at the same place the RtFilter writes the log file
the AS7 plugin code forgot to set the log dir in the plugin config. this is necessary if someone started AS7 with a different log dir sysprop (which RHQ itself does when starting its own AS7 server)
BZ 864619 ] fix the defaults so the AS7 plugin looks at the same place the RtFilter writes the log file
get calltime view to work for both resource and group views in gwt
get the tests to build the test AS7 container for arquillian to use
we can just create initial context with no-arg constructor and it should work now
make sure we assign ids to the antrun-plugin executions
BZ 707349 ] applying community patch from Ken Rumer - see https://bugzilla.redhat.com/show_bug.cgi?id=707349#c13
starting the work to get server/jar to support AS7 testing.
Merge remote-tracking branch 'origin/master' into bug/707349
Merge remote-tracking branch 'origin/master' into rhq-on-as7
rip out Seam since we don't want to upgrade it when porting the ear to run inside AS7. the only thing "broken" appears to be the content tab's file upload feature. that still needs to be fixed.
fix file upload
Merge remote-tracking branch 'origin/master' into bug/707349
Merge remote-tracking branch 'origin/master' into rhq-on-as7
support the ability to restart the server in AS7 admin-only mode (this may be useful in the future when supporting the new AS7 proposed patching mechanism)
support adding a JDBC driver that does not support XA
introduce a new "startup-subsystem" extension module for AS7. This currently is not used, but will eventually be the mechanism by which the ear will be deployed.
Merge remote-tracking branch 'origin/master' into bug/707349
BZ 849394 deprecate old API
BZ 849394 ] denote the intentional API change back to the original API
build the startup-subsystem
BZ 865167 ] db upgrade changes to fix avail table data
these must have come over in a merge from master. these should be deleted
add some new capabilities to the client
put some more as7 deps in root pom
the start of the new-new installer.
need this so jenkins can run DBInstallationTest properly
remove the old, now unused, modules/jopr
remove the duplicate dependency for javassist
the start of the new-new installer. the old-new installer (the GWT webapp) is now removed. there is now a new standalone installer. this also fixes the poms so dev mode puts the artifacts in the module where the ear is now deployed
Merge remote-tracking branch 'origin/master' into merge-rhq-on-as7
starting to get the downloads stuff to work
get server client-api module to get the scripts from new correct place
we don't use this anymore, don't bother with directories under rhq.ear in this class
disable secure jndi lookup checks for now
look in the proper places for files inside the ear
be able to attach debugger to the arquillian-launched AS container. you do this by passing in -Ditest.debug on the mvn cmd line this also adds an eclipse launcher to connect to it on the proper port
make sure installer detects if it already ran
look for CLI version info in proper place
Merge remote-tracking branch 'origin/master' into rhq-on-as7
there is no need for this undeployInstaller method anymore
opps..forgot to delete the import
add new remote client SLSB so it can manage the version file in the data directory
name the master PC's data directory name so it better identifies itself as being owned by rhq
all a user to point to an external JBossAS installation via JBOSS_HOME or RHQ_SERVER_JBOSS_HOME
Merge remote-tracking branch 'origin/master' into rhq-on-as7
windows installer script for the new installer
BZ 882280 ] purge old perspective plugin from the DB during db upgrade
BZ 882280 ] purge old perspective plugin from the DB during db upgrade (cherry picked from commit dadcd3fbff9d883c7d6a166ad42687c26c9b5a70)
Merge remote-tracking branch 'origin/master' into rhq-on-as7
Merge remote-tracking branch 'origin/master' into rhq-on-as7
Merge remote-tracking branch 'origin/master' into rhq-on-as7
Merge remote-tracking branch 'origin/master' into rhq-on-as7
Merge remote-tracking branch 'origin/master' into rhq-on-as7
Merge remote-tracking branch 'origin/master' into rhq-on-as7
do not look up coreservice mbean if we already set our plugin dirs (like we will do in our tests)
Merge remote-tracking branch 'origin/master' into rhq-on-as7
trivial - use constants for exit codes
installer supports --servers to list all register servers and their details
fix npe
change option name
abort on unknown option
re-arrange rhq-server.properties to provide example oracle settings and move the settings more likely to be changed towards the top
more rhq-server.properties stuff
ocd kicking in
no need for the separate generate-db-password scripts (which is good because the duplicate alot of script code in the other scripts) we can now generate db passwords from rhq-installer.sh --dbpassword=foo
Merge remote-tracking branch 'origin/master' into rhq-on-as7
we don't need jboss.zip / jboss.unzip.location since we dont' have to prepare AS7 in this module anymore
Merge remote-tracking branch 'origin/master' into rhq-on-as7
Merge remote-tracking branch 'origin/master' into rhq-on-as7
be alittle better at knowing how long we should wait before we can start the comm layer
be alittle better at knowing how long we should wait before we can start the comm layer
Merge remote-tracking branch 'origin/master' into rhq-on-as7
added a commented out section in server/jar pom.
for developers, have the dev builds automatically run the installer for you so you don't have to do it in a separate step.
revert to the last known copy that is correct - stefan, put back your cassandra stuff in here
Merge remote-tracking branch 'origin/master' into rhq-on-as7
check to see if the extension is there to know if the installer already ran, that's more foolproof than checking for the ear being deployed
give alittle luv to our brethern developers running on windows
trivial comment change
Merge remote-tracking branch 'origin/master' into rhq-on-as7
container-lib module went away
fix default location for download files
we dont use this now, but in case we bring it back, i'll fix this now
make the oracle module an optional dep in the installer to support those cases where we can't ship the Oracle JDBC driver jar.
trivial changes to generated props file
fix location where cassandra plugin goes in dev profile mode (though there is no dev profile declaration - this should be fixed also)
put the war in the new ear location
fix windows script so dir names are quoted when starting the vm
only have it try 2 times (as opposed to 5). if the server isn't up in time for our installer to do its thing by then, something is probably wrong anyway this is only for windows developers use (dev mode) so this isn't involving production functionality
this is for unix developers - only retry 3 times - if the installer can't do its thing by then, something is probably wrong anyway. this doesn't involve production functionality - rhq-autonistaller.sh is only for dev mode builds
add https connector - use self-signed cert by default
add a comment to the props file
BZ 887411 ] don't uninventory everything just because we commited some top level server
behavior of tomcat in AS7 appears different than before - change client-auth from want to false. "want" appears to still require certs coming from the agent and I don't know why (like I say, it didn't work like that in Tomcat inside AS 4)
dbsetup stuff uses the old i18nlog still and we need to tell it to not use log4j because otherwise it looks like the log messages aren't getting spit out (log4j isn't configured)
allow for a user to do a dbsetup without having to do a full install. we had some people ask for this, its a rare use-case, but it can be useful. also useful for devs .. rhq-installer.sh --setupdb (or use the shorthand "-b" option)
don't dump the whole stack, just show the error messages
add generic API to remove a resource
trivial change to comment (this line of code doesn't deploy the ear)
rename the methods by removing "Request" - these don't create the requests, they actually do the thing itself. JBossASClient and its subclasses have createXYZRequest static methods, but they only create the ModelNode requests, but they don't submit them for execution. That's the difference - if you see a method called "createXYZRequest", it returns a ModelNode for the caller to manipulate and later submit for execution. If the method does the execution itself, don't append "Request" to its name to avoid confusion.
add code to update SecureIdentity security domain credentials add code to remove web connector skip adding default mgmt user if it already exists add installer --reconfig feature to update AS7 settings based on new rhq-server.properties settings
can't just reload at the end, need to restart - our Mbeans don't go away on just a reload, so we must restart
if an agent is re-registering in order to change its remote endpoint (say, going from non-ssl to ssl), destroy our old client the server may still have hanging around
don't start all agent clients at startup - let us create them lazily during server runtime
need to keep replacing properties in case ${x} replaces with a string that itself has ${y}
we want to support agent->server (ssl)socket transport to allow for uses to circumvent tomcat. this is a rarely used feature but can be useful in certain use-cases
code that circumvents actually doing a reconfigure if the non-expression settings are the same. this helps avoid reconfiguring/restarting the server when nothing really changed
do not allow the reconfigure to occur if the installer hasn't been fully run on the server yet
have the #START SERVER marker appear in the background/service start portion of the scripts so the autoinstaller can be used for both running in console/foreground and background
during a reconfig, wait some time for the server to come up
add a simple method to check that the web subsystem is available
run "installer --reconfig" automatically for the user when starting the server. if we don't do this, we'd require the user to remember to run the installer --reconfig after changing rhq-server.properties, and we all know people won't remember to do that once we are on a version of AS7 that supports expressions in subsystems fully, we can revert this.
we can't run installer --reconfig automatically for the user on windows because for some reason JavaServiceWrapper won't restart AS7 when an exit code 10 is seen, even though we have it configured to do so.
clean up some eclipse tools
fix some messages for avail duration alert condition editor
we want to bump up the default tx timeout to 10 minutes since that is what we used in older versions of RHQ (the out of box AS7 default is 5)
we want to bump up the default tx timeout to 10 minutes since that is what we used in older versions of RHQ (the out of box AS7 default is 5)
have the installer create a org.rhq logger category at the default level of INFO. allow a user to set it in rhq-server.properties (unforunately, this is another AS7 attribute that does not support expressions so you have to run rhq-installer --reconfig to pick up the change after the initial install)
have the installer create a org.rhq logger category at the default level of INFO. allow a user to set it in rhq-server.properties (unforunately, this is another AS7 attribute that does not support expressions so you have to run rhq-installer --reconfig to pick up the change after the initial install)
BZ 902876 ] make sure we are really installing in an AS7 instance that is meant to be the RHQ Server.
Merge remote-tracking branch 'origin/master' into bug/902406
allow perftest plugin to have 6-level hierarchies
Merge remote-tracking branch 'origin/master' into bug/902406
Merge remote-tracking branch 'origin/master' into bug/902406
make sure long-lasting inventory report merges don't step on each other. introduce a inventory report serializer
point to the proper RHQ wiki docs
Merge remote-tracking branch 'origin/master' into bug/902406
provide an agent configuration setting to disable resource types on the agent
don't do this - will NPE when nothign is disabled and we will rebuild this collection here anyway
Merge remote-tracking branch 'origin/master' into bug/902406
Squashed commit of the following:
provide a --quiet option to avail prompt command
append to the installer log so we don't lose any installer log messages when rhq-installer.sh is run multiple times
installer now provides a "--force" option - this lets you re-run the installer even if you didn't set autoinstall.enabled=true and it also lets you attempt another installation if a previous one failed in the middle of its prior attempt.
the name of the perftest sysprops are different now
BZ 912525 ] add new concurrency limit for configuration updates
don't deprecate these methods - criteria is nice, but these specific methods are very efficient and can be used when we need to eek out as much performance as possible. no need to go through the criteria API just to get a single agent using very targeted WHERE conditions.
this should not be here
BZ 913606 ] agent will send a full avail report if the server tells it at connect time that it has been backfilled. the server will also update the last ping time of the agent when the connect comes in.
BZ 913778 ] fix NPE
John Sanda <jsanda@redhat.com> (34):
BZ 838861 ] Handle empty zip file on java 7
BZ 869015 ] Add more detail logging when handling inventory report with stale types
BZ 845700 ] Do not use SQL LIKE operator when checking if plugin is ready for purge
BZ 845700 ] adding debug logging
BZ 857189 ] change transaction boundaries when installing resource types
Revert BZ 857189 change transaction boundaries when installing resource types"
BZ 857189 ] increase transaction timeout for registerPluginTypes method
BZ 869063 ] do not purge plugin until all resource types have been purged
Removing drift-mongodb plugin from build
BZ 870561 ] Fix data points query to use minvalue and maxvalue columns
deleting test code that was just intended for prototyping
fixing exception in JPQL query in set up/tear down code
adding test for getAggregate method when fetching raw data
Initial commit for Cassandra plugin
Adding operation taking snapshot of an entire keyspace
Add some docs around what versions of cassandra are currently supported.
Adding initial support for ColumnFamily resource type
Adding Keyspace resource type
Initial commit for CassandraUtil
Adding some metrics and resource configuration for StorageService
Adding metrics for ColumnFamily type
Adding metrics for StorageProxy type
Adding Cassandra Server JVM type to monitor Cassandra JVM
Adding start operation for cassandra daemon
Make the shutdown operation for the Cassandra Node type more robust
Adding logic for Windows platforms in Cassandra Node start operation
Updating logic for getAvailability
Adding some traits and resource operation to StorageService type
Adding CommitLog resource type
Adding initial implementation of ConfigurationFacet for ColumnFamilyComponent
Adding repair operation for StorageService type
initial commit for MeasurementOOBManagerBeanTest
adding support for making overlord available via CDI injection
make sure we clean up when downloading jboss distro
Jürgen Hoffmann <jhoffmann@redhat.com> (1):
Improving the way the Dashboard view is displayed (correct row/column alignment).
Larry O'Leary <loleary@redhat.com> (1):
BZ 905183 - [as5 plugin]Unable to collect metric from web application context due to WebApplicationContextComponent not initializing 'clustered' property value
Lukas Krejci <lkrejci@redhat.com> (73):
Making discovery and obtaining the JMX connection work, basics of config properties we are going to support.
formatting
formatting and general cleanup before we start doing anything serious.
more formatting and code cleanup before we do anything serious.
try to read the hadoop home dir from the commandline if it is there, otherwise fall back to the CWD of its process.
Reading of hadoop config files should work now.
start of config update
Pulling the configuration support out of the main component class to make it more readable. Config update should now work, too.
some more config options + initial support for log polling
Renaming stuff, making config props not required
* added events support to the rest of the hadoop resource types * renamed "DataNode" to "Hadoop DataNode" * Tied event polling registration to the avail check because we are only able to find out the name of the log if the hadoop server is running.
Correcting a merge conflict.
bumping the required java version to 6 (just to be consistent with the min reqs of the RHQ agent).
Implementing Hadoop job duration as a calltime metric.
Clear out the pending calltime metrics after reporting them.
More rigorous error handling while processing the log files to prevent a runtime exception cancel the scheduled job for log processing for given resource.
Make the date formats used by the log4j log processor instance fields so that they are not shared by multiple instances that would use those formats from different threads. The DateFormat is not a thread safe class.
BZ 846269 ] - AS5's shutdown operation doesn't require availability to be up.
Support for Job Jars as creatable resources. Job Jars can be submitted (using an operation) to the job tracker, providing a convenient way of running hadoop jobs.
Fixing the child resource creation for resource types that do not have resource configuration.
BZ 859982 ] - Resetting the custom blur event handling before and after a popup dialog is shown in the configuration editor. This prevents unhandled errors when for example entering an entry into the list of maps multiple times.
BZ 846292 ] - The directory component no longer blows up on updating directory entries in other files than the main Apache config file.
Apache plugin integration test compiles again.
Making sure the integration-tests profile can build (still the apache tests don't succeed but that is another problem
).
Make the RhqManager class the sole place to define everything is necessary for defining and making available a new *Manager to the scripting clients.
Fixing the apache integration tests by adding the missing expectations and guarding against custom localhost name.
Fix the PluginContainerTest to not choke on test methods without a plugin container configured.
start of reimplementing the JNDI access tests using Arquillian.
Fixing the JNDI access integration test.
Replacing all the cobertura definitions in all the poms with one definition in the root pom.
Changed the code-coverage provider to JaCoCo. We now generate a separate coverage report for unit tests and integration tests. We also generate one overall report per project. Additionally an overall report including the results of all unit and integration tests is generated in a new top-level module called "code-coverage".
Removed unnecessary property def.
Build everything including all the integration tests in the code-coverage profile.
BZ 803737 ] - Bumping Clirr maven plugin version to 2.5. No need for the patched version to do the API checks anymore.
BZ 875848 ] - Use the fixed version of the password obfuscation as found in Picketbox.
Merge branch 'lkrejci/code-coverage'
Slightly improving logging on apache upgrade tests.
Backup the apache error logs on per-test-method basis.
Make the copied error logs really reflect the test they were collected for. Also bumped up the number and size of the retained logs greatly, because we collect A LOT of data in the debug mode, which is needed until we fix the failures in the CI environment.
Fixing the incorrect assertions in the ApplicationServerComponentTest in AS5 integration tests.
Removing last remnants of cobertura from the poms.
Recovering from the loss of ant as a transitive dep of cobertura in tests.
stuff to help debug the apache integration test failures.
Improvements in Apache integration tests: * improved logging * randomization of used ports * *WithoutSNMP tests don't use SNMP even after upgrade (makes the behavior more understandable)
BZ 879655 ] - Properly deactivate the resource instead of just calling the component's stop() method when forcing reactivation of a started component. This will ensure the PC can't enter an inconsistent state in the case the stop() method fails.
Randomizing ports in the JNDI access test. Maybe this will help the test failures in the integration-tests Jenkins job.
tmp
BZ 876479 ] - Hardcode the values of symbolic time constants in the measurement_utils.js rather than referencing the previous ones during the declaration. That just didn't work, giving all such props a value of NaN.
This reverts commit 2de428af239ee6b4af0af2a439e7e27f65755acb.
Moving the JNDI security infrastructure to the server/jar and updating it to work with AS7.
Removing the obsolete javaURLContextFactory from the server/jar. It was a left-over from a failed attempt at securing the JNDI look up in a saner way than we do now.
Removing the SecurityActions for obtaining a logger instance during class initialization.
Code coverage support for Arquillian-based tests.
Forgot to re-enable the now fixed and working JndiAccessTest in itests-2.
Recovering from the merge of commit 2ce8b7aa063a56840090ed553f47b6e3a0ba7845. The SessionManager should do the security checks, we've got JNDI security working.
Modified the access checking JNDI lookup wrappers to take into account the slightly different behavior needed when there was originally no initial context factory builder or if there originally was some.
Fixed the ExportingInputStream to correctly define the XML namespaces (and thus work under woodstox, which is the XML streaming provider in AS7). Re-enabled the SynchronizationManagerBeanTest.
Made the JndiAccessTest remote integration test work again. For that (among other things) the suite extension previously defined in the core/domain test-jar was moved into a standalone module core/arquillian-integration/suite-extension.
added a .gitignore entry for cachedir of the python script engine generated during some tests.
Ooops, forgot to implement the exception check in the JndiAccessTest.
BZ 880795 ] - The upgraded Augeas library no longer contains the file "libaugeas.so", which causes the default native library lookup mechanisms to fail. Standard ld should be able to find the .so.x.y files, too, but the presence of java.library.path system property seems to be causing this not to kick in during the library loading as done by the JVM.
Define the jna.platform.library.path in the tests, too, so that the correct augeas library gets picked up.
BZ 844475 ] - Don't try to do code completion on native javascript arrays.
Trivial - trailing whitespace removal and adding missing @Override
The upgrade of JAXB to 2.2.4 exposed an issue in the configuration object which would cause the disappearance of the properties of a configuration object if a collection obtained from Configuration#getProperties() was passed to Configuration#setProperties().
Adding Jacoco setup to arquillian config for remote JNDI access integration tests.
Remove the @Override annotation on clone() which makes GWT fail to compile the class.
BZ 906495 ] - The JBoss AS specific bundle functions should work again.
BZ 907558 ] - JavaCommandLine and subclasses now parse the commandline lazily to prevent calling overriden methods in super-class constructor. A couple of other cosmetic improvements like static logger, etc.
Fix the expectations of the JMock-based apache plugin integration tests after the fixes in the inventory report processing that changed the workflow slightly.
The apache integration tests are now resilient against apache servers running on the target machine out of the test control. These will no longer affect the test results.
Fix the JNDI access integration test. Default activation of profiles in Maven just doesn't work the way I expected it to
Adding the API checks to the script bindings.
Mike Thompson <mithomps@redhat.com> (9):
BZ 841045 advisoryInfo-plain.xhtml is missing parameters on the repo-plain.xhtml link] applied patch from jlivings.
BZ 826224 Creation of Compatible Group fails with value too long for type character when a description is entered which is longer then 100 characters] Group names limited now to 100 chars to match DB.
BZ 861127 - Upgrade issue with respect to min/max property attributes] - Needed to upgrade existing nulls to 0 for new min/max columns. (cherry picked from commit 8b08fce535742c6bd7f24d3ab0d8c7d626fe7923)
BZ 861127 - Upgrade issue with respect to min/max property attributes] - Fix the rhq_config_prop_def.max_entries default to 32767 instead of 0 if it is null. And changing signature of setter back to setMax(int) from setMax(Integer) although compile time coercion makes them equivalent. And fix 2 bugs in DBInstallationTest.java that wasn't checking the schema version since RHQ3.3.0 and JON3.0.0. (cherry picked from commit 7d937712636fcad497fec9e89afc3fe84168ad23)
Trivial: Comment on skipped db-upgrade version 2.124
BZ 873880 - CVE-2012-4563 GWT Unknown XSS flaw] Upgrade GWT 2.4.0 to GWT 2.5.0
BZ 873880 Downgrade maven gwt plugin back to 2.4.0 from 2.5.0.] The maven gwt plugin version 2.5.0 doesnt pick up changes when running dev mode (mvn gwt:run). GWT version remains at 2.5.0. Also changed DTD for *.gwt.xml to 2.5.0.
BZ 873880 maven gwt plugin must match gwt version 2.5.0]
BZ 889375 ] Doc type change to eliminate the strange checkbox checking/unchecking behavior. Changing <meta http-equiv=X-UA-Compatible content=IE=9 />
Remy Maucherat <rmaucher@redhat.com> (1):
BZ 865459 ] adding support for adding users for tomcat 7 UserDatabase.
Richard Hensman <richard@onevisionconsulting.co.uk> (2):
Add support for monitoring table spaces to the Oracle plugin
Add support for monitoring Flash Recovery Areas to the Oracle plugin
Simeon Pinder <spinder@redhat.com> (10):
BZ 839552 ] passed in the wrong configuration when overriding behavior for SecurityDomain as7 nodes.
BZ 880795 ] upgrade augeas library to 0.9.0-4.
BZ 871535 ] update product types to support EPP->JPP rename.
Update the zip and chmod processing filters for the agent to include all of the shared library file extensions.
adding new public version strings for comparison testing.
Add linux script to build remote branches from current state on source branch.
BZ 905632 ] fix MetadataManager query to use unlimited criteria queries.
BZ 905632 ] i)updating fix to instead chunk through type removal until done ii)move null check earlier in uninventory asynch logic to avoid unnecessary ERROR logging/removals that occur more now with chunking.
BZ 905632 ] This is a unit test only fix to defend against regressions here. Failure was undetected before.
BZ 906844 ] updating product name so discovery detail lookup over rest works to eliminate IllegalArgumentExceptions from agent log.
Stefan Negrea <snegrea@redhat.com> (52):
BZ 803737 ] - Adding the ignore rules for intentional API changes in AS7 plugin.
BZ 803737 ] - Update the ignore rules for AS7 plugin to use path expressions for class names.
BZ 803737 ] Update difference type.
Updated definitions for standalone and domanin to GA.
BZ 861127 ] Update DBInstallationTest with RHQ 4.4.0 schema and data files.
Revert "Trivial: Comment on skipped db-upgrade version 2.124"
BZ 861127 ] Revert back to sequential db upgrade ids. (cherry picked from commit 84e7e552a2ee7d05aff8eb489c66cdfe6adf2bd0)
BZ 869730 ] Move static block of code that was checking for Sigar availability to a static method to be able to use correctly at startup both system properties and stored agent preferences. This change will avoid attempts to load the Sigar library even if --nonative flag or rhq.agent.disable-native-system property are set to true.
Add support for terminal branches. This will allow recutting the tag name or cutting a new branch from branches that are not expected to get new development.
Do not attempt to commit changes if there are no changes available. This can happen if and only if there is no effective change in the release version and the current build it is attempting to re-cut the release tag.
Avoid uncessary attempts to merge remote branches when local branches are created for test builds with tagging.
Add option to override the tag name generated by the script. This feature should be used for continuous integration purposes only.
Add tag name variable to the debug output.
Test for the release_tag variable and not the actual string.
Force versions 1.3.1 until upstream JIRA for the plugin is resolved.
Update Eclipse classpath entries for the Cassandra plugin.
Fix all code warnings. Also update the call to the repair method of StorageServiceComponent.
Update plugin version to be in sync with RHQ master version.
Add support for restart and compact operations. Also update repair operations to use dependent values for keyspace.
Moved keyspace and column family specific operations away from storage service to respective resources. Added compaction manager resource. Added metrices to commit log. Plus various other minor updates.
Enable snapshot taking for the entire node, keyspace, and column family. Add all meaningful metrics, operations and properties to column family resources.
Enable client request metrics. Also included is a compaction of the XML plugin descriptor.
Added support for cache service. Also updated storage proxy, storage manager, and compaction manager.
Add server and internal metrics to the list of metrics.
Add support for resource self-referencing option sources.
Fix compilation error in unused plugin.
Updating the JMX plugin to check if the connection is still open or valid before returning the connection object. In case the connection is no longer valid, the code will attempt to reconnect to the server.
Correct parameter name to match execution code.
Mark connection and connectionProvider as volatile since they can be accesses from multiple threads at the same time. This will avoid issues where a thread will have a stale cached value of either attribute.
Update self-referencing option sources to use explicit keyword 'self'. This reduces the confusion that an implicit self reference would pose for plugin writers.
Add the last set of exposed management interfaces to the resource descriptor.
One more updated for the code that attempts to reconnect to the JMX server. Before attempting to reconnect, attempt to close the existing connection to avoid thread leaks from underlying resources.
Add a generic map/set property parser covers cases specific to Cassandra MBeans.
Enable full support for StorageService properties and metrics.
Get the list of keyspaces directly from the StorageService bean instead of using hector.
A few plugin descriptor and code updates based on integration test results.
Add cassandra integration tests.
Avoid logging warning due to missing dependencies.
Make thrift port configurable via server resource connection settings. Also remove CassandraUtil to ensure that port updates from the configuration are avaiable to child components right away.
Disable tests due to missing Cassandra common infrastructure.
Update generic map reader to match latest API changes for PropertyDefinitionMap.
Add back Cassandra plugin specific .classpath updates.
Update Cassandra plugin pom to include the dev profile for dev container deployments.
Add korean locale to the list of default locales build by core gui.
Update the test to take into account the possibility that the TagManager is disabled.
Arquillian container version should be independent from the container version due to artifact dependencies.
Move arquillian container dependency to a profile that is activated if tests are not skipped.
Move test-jar dependencies to the special test profile.
Move test jar dependency to test specific profile.
Move more dependencies to special test profile to avoid resolution if tests are disabled.
Move few more test-jar dependencies to the test specific profile.
BZ 912478 ] Add managed server deployment as parent for the web subsystem resource type.
SungUk Jeon <allnewangel@gmail.com> (1):
Korean translations
Thomas Segismont <tsegismo@redhat.com> (20):
Fix Eclipse classpath file GWT -> 2.5.0 joda-time was missing
BZ 879583 ] Monitoring : process availability gives wrong status (SIGAR)
BZ 879583 ] Monitoring : process availability gives wrong status (SIGAR)
BZ 871873 ] Availability check change is not automatically picked up
BZ 858138 rhq-agent-env.sh gives example of bad su command
BZ 887320 - Agent raise OutOfMemoryError when adding new Deployment child resource on AS7
Revert "Bug 887320 - Agent raise OutOfMemoryError when adding new Deployment child resource on AS7"
Add a regression test for ProcessInfo
Fix concurrent access to static counter in AgentUpdateServlet
AS7 plugin upload connection now with httpclient
BZ 885664 OpenSSHD and MySQL availability check may report stale data
BZ 885664 OpenSSHD and MySQL availability check may report stale data
Fix Eclipse classpath file: CDI and javax.inject
BZ 905000 - RFE: RHQ fails to set JBOSS_PIDFILE and LAUNCH_JBOSS_IN_BACKGROUND when importing JBoss AS7
BZ 885664 OpenSSHD and MySQL availability check may report stale data part 2
Disable JMXPluginTest (Sigar getProcCredName issue)
Fix JMX plugin test failure
Fix AS7 plugin itest failures due to misuse of context.getNativeProcess
BZ 911172 - Creating mail server fails with EAP 6.1.0.DR4 if outbound-socket-binding-ref is not filled
netservices plugin strengthening