Alex Heneveld [Fri, 22 Jan 2016 11:46:50 +0000 (11:46 +0000)]
also improve jmx port allocation in tests
to fix another observed test failure,
at https://builds.apache.org/job/incubator-brooklyn-pull-requests/org.apache.brooklyn$brooklyn-software-base/2439/testReport/junit/org.apache.brooklyn.feed.jmx/JmxFeedTest/setUp/
Port already in use: 40125; nested exception is:
java.net.BindException: Address already in use
Alex Heneveld [Fri, 22 Jan 2016 09:41:00 +0000 (09:41 +0000)]
fix race in DSTTest.testCancel
seeing errors on jenkins server though it runs fine locally;
have patched what looks like one race, but not sure it explains the failure
at https://builds.apache.org/job/incubator-brooklyn-pull-requests/org.apache.brooklyn$brooklyn-core/2440/testReport/junit/org.apache.brooklyn.util.core.task/DynamicSequentialTaskTest/testCancelled/
where we are seeing an extra semaphore at the end of the test;
have added comments and logging in case it is seen again.
(locally all is well.)
Alex Heneveld [Thu, 21 Jan 2016 09:52:14 +0000 (09:52 +0000)]
fix location of yaml file read in by docs
docs build failed without this
Alex Heneveld [Thu, 21 Jan 2016 09:48:00 +0000 (09:48 +0000)]
This closes #1148
Alex Heneveld [Thu, 21 Jan 2016 01:22:59 +0000 (01:22 +0000)]
This closes #1164
Alex Heneveld [Thu, 21 Jan 2016 01:19:12 +0000 (01:19 +0000)]
address minor shutdown cleanup code review comments
Alex Heneveld [Wed, 20 Jan 2016 20:49:13 +0000 (20:49 +0000)]
suppress explicit termination in main thread;
the shutdown hook is sufficient and it's a bit of work to terminate properly in-thread
Alex Heneveld [Wed, 20 Jan 2016 18:26:56 +0000 (18:26 +0000)]
background the dev-mode web server shutdown,
and Entities.destroyAll is more graceful on concurrent shutdown
Alex Heneveld [Wed, 20 Jan 2016 18:13:35 +0000 (18:13 +0000)]
prevent deadlock if requests coming in during startup
Alex Heneveld [Wed, 20 Jan 2016 15:56:07 +0000 (15:56 +0000)]
better shutdown
call Main.terminate() in the main thread, rather than relying on shutdown hooks
fixes rest-initiated shutdown when using BrooklynJavascriptGuiLauncher
(looks like that has been broken since #771)
Alex Heneveld [Wed, 20 Jan 2016 21:04:20 +0000 (21:04 +0000)]
This closes #1160
Alex Heneveld [Wed, 20 Jan 2016 21:04:15 +0000 (21:04 +0000)]
This closes #1162
Alex Heneveld [Wed, 20 Jan 2016 21:01:20 +0000 (21:01 +0000)]
code review for task-cancellation mode
Alex Heneveld [Wed, 20 Jan 2016 20:53:04 +0000 (20:53 +0000)]
address code review for #1162
Alex Heneveld [Tue, 19 Jan 2016 12:48:15 +0000 (12:48 +0000)]
clearer semantics for cancel, and cancel dependent submitted tasks by default
prevents leaks where dependent tasks (eg resolveValue) are submitted in the background,
then the caller is cancelled; the interruption was not propagated. now by default it is,
to children tasks and to submitted transients, with options for other (weaker and stronger) cancellations.
see TaskInternal.cancel(TaskCancellationOptions), and new tests for cancelling children in DynamicSequentialTaskTest.
also remove deprecated loose-typing ExecutionUtils.invoke,
more TRACE logging for activities,
and wrap batch config in a task so more resolutions are nested
Aled Sage [Wed, 20 Jan 2016 16:09:17 +0000 (16:09 +0000)]
This closes #1163
Alex Heneveld [Wed, 20 Jan 2016 12:08:31 +0000 (12:08 +0000)]
for chained references, prefer outer catalog item id, and outer tags
as described in changes to release-notes.md
Aled Sage [Tue, 19 Jan 2016 14:57:47 +0000 (14:57 +0000)]
Adds EmptyWindowsProcess
- Includes EmptyWindowsProcess.USE_WINRM_MONITORING, for disabling
checking of WinRM connectivity.
- Adds EmptySoftwareProcess.USE_SSH_MONITORING, for disabling checking
of ssh connectivity.
With EmptyWindowsProcess, one can provision an entity that will not
execute anything on the box. That makes it usable even if there is
no direct WinRM connectivity to the box.
Aled Sage [Tue, 19 Jan 2016 14:55:28 +0000 (14:55 +0000)]
Fix EmptySoftwareProcessYamlTest
See https://github.com/brooklyncentral/brooklyn/issues/1377
Aled Sage [Wed, 20 Jan 2016 10:27:02 +0000 (10:27 +0000)]
This closes #1158
Aled Sage [Mon, 18 Jan 2016 21:29:13 +0000 (21:29 +0000)]
winrm: switch to winrm4j v0.2.0, from 0.1.0
- This is a pure-java WinRM client, so removes the jython depenpdency.
- Also changes WinRmTool.executeScript(...) to executeCommand(…),
and WinRmMachineLocation.executeScript to executeCommand.
- Fixes WindowsPerformanceCounterFeedLiveTest
- Renames Winrm4jTool’s package (from pywinrm to winrm4j).
- Fixes imports that used org.python
Alex Heneveld [Tue, 19 Jan 2016 15:50:59 +0000 (15:50 +0000)]
This closes #1161
Alex Heneveld [Tue, 19 Jan 2016 15:36:07 +0000 (15:36 +0000)]
This closes #1123
Alex Heneveld [Tue, 19 Jan 2016 13:52:09 +0000 (13:52 +0000)]
tidy several other places start(Collection<Location>) is assuming locations are always passed in, not inherited
Alex Heneveld [Tue, 19 Jan 2016 14:38:48 +0000 (14:38 +0000)]
This closes #1155
Alex Heneveld [Tue, 19 Jan 2016 14:38:40 +0000 (14:38 +0000)]
This closes #1132
Guglielmo Nigri [Tue, 19 Jan 2016 14:34:12 +0000 (15:34 +0100)]
Add more tests. Correct typo.
Alex Heneveld [Tue, 19 Jan 2016 13:33:05 +0000 (13:33 +0000)]
clean up how start(Location) works, for Dynamic{Cluster,Fabric}
Alex Heneveld [Tue, 19 Jan 2016 13:08:13 +0000 (13:08 +0000)]
This closes #1159
Aled Sage [Tue, 19 Jan 2016 12:20:09 +0000 (12:20 +0000)]
Fix DynamicFabric locations
- When locations are defined at the YAML top-level, then ensure the
DynamicFabric finds those.
Aled Sage [Mon, 18 Jan 2016 22:14:51 +0000 (22:14 +0000)]
This closes #1157
Andrew Kennedy [Mon, 18 Jan 2016 17:15:23 +0000 (17:15 +0000)]
This closes #1154
* github/pr/1154:
Updated JClouds to 1.9.2
Valentin Aitken [Mon, 18 Jan 2016 13:04:18 +0000 (15:04 +0200)]
Set error logging for MySqlNode
Yavor Yanchev [Mon, 18 Jan 2016 15:51:22 +0000 (17:51 +0200)]
Additional security for MySQL
- it changes the default of sql_mode to include "NO_AUTO_CREATE_USER"
- it will prevent the GRANT statement from automatically creating new
user accounts unless authentication information is specified.
Sam Corbett [Mon, 18 Jan 2016 12:15:50 +0000 (12:15 +0000)]
This closes #1131
WinRm documentation - In Powershell use Start-Process ... -Wait
Sam Corbett [Mon, 18 Jan 2016 12:12:51 +0000 (12:12 +0000)]
This closes #1153
WinRM docs for correct preparation of windows templates
Sam Corbett [Mon, 18 Jan 2016 12:09:04 +0000 (12:09 +0000)]
This closes #1141
Fixed issue with exceptions in Rest Client
Graeme-Miller [Mon, 18 Jan 2016 10:59:59 +0000 (10:59 +0000)]
Updated JClouds to 1.9.2
Yavor Yanchev [Mon, 18 Jan 2016 10:41:43 +0000 (12:41 +0200)]
WinRM docs for correct preparation of windows template
- Short troubleshooting section that incorrect windows template can
cause timeouts or fail some deployments
- Settings to Use for an Unattended Installation in known limitations
section
Andrew Kennedy [Mon, 18 Jan 2016 00:37:05 +0000 (00:37 +0000)]
This closes #1142
* github/pr/1142:
deprecate group.members.delegate and prevent delegate creation when adding to quarantine - use of delegate to denote quarantine group membership in the jsgui is no longer required due to #929
Ciprian Ciubotariu [Mon, 18 Jan 2016 00:30:04 +0000 (02:30 +0200)]
This closes #1152
Cristian Tarsoaga [Sat, 16 Jan 2016 01:26:52 +0000 (03:26 +0200)]
[BROOKLYN-183] Add karaf features
Add brooklyn-software-base, brooklyn-software-winrm and brooklyn-policy features.
Ignore for now python imports in brooklyn-software-base.
Aled Sage [Fri, 15 Jan 2016 22:37:24 +0000 (22:37 +0000)]
This closes #1130
Alex Heneveld [Fri, 15 Jan 2016 15:11:05 +0000 (15:11 +0000)]
This closes #1151
Alex Heneveld [Fri, 15 Jan 2016 15:07:50 +0000 (15:07 +0000)]
update licenses with latest metadata and dependencies
Alex Heneveld [Fri, 15 Jan 2016 15:04:11 +0000 (15:04 +0000)]
correct references used in license-readme-generation for new project structure
Aled Sage [Fri, 15 Jan 2016 14:54:57 +0000 (14:54 +0000)]
This closes #1136
Aled Sage [Fri, 15 Jan 2016 14:11:06 +0000 (14:11 +0000)]
This closes #1150
Aled Sage [Fri, 15 Jan 2016 13:18:13 +0000 (13:18 +0000)]
BROOKLYN-215: fix NPE when SshMachineLocation.port==null
Alex Heneveld [Thu, 14 Jan 2016 23:53:09 +0000 (23:53 +0000)]
This closes #1145
Alex Heneveld [Thu, 14 Jan 2016 23:52:36 +0000 (23:52 +0000)]
This closes #1146
Alex Heneveld [Thu, 14 Jan 2016 23:48:07 +0000 (23:48 +0000)]
This closes #1149
Alex Heneveld [Thu, 14 Jan 2016 23:45:42 +0000 (23:45 +0000)]
This closes #1143
John McCabe [Thu, 14 Jan 2016 22:58:57 +0000 (22:58 +0000)]
turn off sticky sessions to allow easy balancing validation
Alex Heneveld [Thu, 14 Jan 2016 22:00:09 +0000 (22:00 +0000)]
fix transformation of values which are Sets
fixing test failure introduced, and expand test coverage;
the problem was not introduced here, but i think it was hidden
because we weren't transforming values in some cases.
presumably now we are transforming better+more,
and with this we're also transforming Set instances correctly.
Alex Heneveld [Thu, 14 Jan 2016 22:19:15 +0000 (22:19 +0000)]
unwrap type being extended as soon as it is loaded
Guglielmo Nigri [Thu, 14 Jan 2016 10:42:51 +0000 (11:42 +0100)]
Better control over opening inbound ports
Aled Sage [Thu, 14 Jan 2016 16:46:42 +0000 (16:46 +0000)]
This closes #1125
Aled Sage [Thu, 14 Jan 2016 16:44:57 +0000 (16:44 +0000)]
This closes #1147
Alex Heneveld [Thu, 14 Jan 2016 16:13:13 +0000 (16:13 +0000)]
add release notes mention of location changes
Alex Heneveld [Thu, 14 Jan 2016 14:32:55 +0000 (14:32 +0000)]
parameter-inheritance now marked with TODO, and prev behaviour largely restored
Alex Heneveld [Wed, 13 Jan 2016 23:08:29 +0000 (23:08 +0000)]
WIP - change semantics of parameter addition
api EntitySpec.parameters(...) should *add* parameters, like the other methods;
since parameters have just been introduced it is fine to do this
also comments and tidy of other merges
Alex Heneveld [Thu, 14 Jan 2016 15:49:52 +0000 (15:49 +0000)]
revise tests for new semantics of inheriting locations
also allow nulls in the camp objects since they can come in via `{ location: null }`
Svetoslav Neykov [Thu, 14 Jan 2016 15:17:25 +0000 (17:17 +0200)]
Test catalog item tags serialization
Svetoslav Neykov [Thu, 14 Jan 2016 15:01:38 +0000 (17:01 +0200)]
Fix catalog XML deserializing
Fixes the serialized XML format so it can be deserialized later.
Alex Heneveld [Thu, 14 Jan 2016 13:36:05 +0000 (13:36 +0000)]
in AbstractApplication, don't infer locations to pass to start(...)
with this, locations defined on an app are no longer inserted automatically by BasicApplication.start(). this is because locations are now normally defined on yaml. entities typically prefer a location if supplied to start(...), but if none are supplied there they look in local configuration and then ancestor configuration. if BasicApplication passes their configured location to the start method, that is trumping any location defined on they descendant entity, which is probably not desired. if an entity does not define a location, there is no change to behaviour as it looks up ancestor location regardless. but this fixes a bug where if a location A is defined on an application, and another location B defined on a particular entity in that application, the entity now uses B whereas previously it was using A. this makes explicit entity startup behaviour consistent with how an entity spec used in a cluster works (which has and still prefers the location defined in the entity spec).
Alex Heneveld [Thu, 14 Jan 2016 12:16:23 +0000 (12:16 +0000)]
allow locations to be overridden when referring to a type
not sure of the cleanest way to achieve this, but where a type includes a location,
a caller may wish to override it, such as in the default catalog references to template 2.
previously i think locations in referenced types were just dropped (?)
but then it was changed to be additive, which makes sense and is consistent, but
breaks the examples and means there was no way to override a location in a referenced type.
this changes it so that *if* a location block is included on the *referencer*, even if empty,
it overrides the locations on the referenced type. examples are updated accordingly.
an alternative would be to introduce a new explicit field to clear locations;
or a very different alternative would be to put locations in comments in the examples,
and clarify in the code templates/applications (as presently the presence of a location
is what triggers something being editable!); those are worth considering, especially the latter,
but the present change actually seems sensible as it is rare a caller would want to *add*
a location to a referenced type!
Alex Heneveld [Thu, 14 Jan 2016 11:45:20 +0000 (11:45 +0000)]
better logging and rendering of errors
Graeme-Miller [Thu, 14 Jan 2016 13:44:51 +0000 (13:44 +0000)]
moved new QA classes to correct location
Martin Harris [Thu, 14 Jan 2016 11:56:49 +0000 (11:56 +0000)]
Adds all interfaces to proxy if deploying concrete application
Aled Sage [Thu, 14 Jan 2016 10:43:50 +0000 (10:43 +0000)]
BROOKLYN-212: more improvements
- Resizable.resize: throw InsufficientCapacityException if didn’t
manage to reach target size, rather than only if didn’t manage to
increase in size at all.
- DynamicCluster: delete nodes that throw NoMachinesAvailableException,
rather than putting them in quarantine.
- Fix AutoScalerPolicy’s max-capacity high-water mark, when
resizeUpStabilizationDelay is used.
Aled Sage [Mon, 11 Jan 2016 11:39:51 +0000 (11:39 +0000)]
AutoScaler test: remove unnecessary timeout config
succeedsEventually calls can just use the default timeout (unless it’s
a performance test where it must complete in a faster time!)
Aled Sage [Fri, 8 Jan 2016 21:33:25 +0000 (21:33 +0000)]
BROOKLYN-212: AutoScaling doesn’t retry if InsufficientCapacity
- Adds Resizable.InsufficientCapacityException, thrown by
Resizable.resize() if could not grow at all.
- DynamicCluster catches NoMachinesAvailableException, and rethrows
as InsufficientCapacityException in resize().
- AutoScalerPolicy catches InsufficientCapacityException, and sets
insufficientCapacityHighWaterMark to record the max size it can get
to. Does not try again to resize above that, unless the highWaterMark
is explicitly cleared by reconfiguring that config value.
- Tests:
- Changes TestCluster to include history of sizes and desiredSizes
- Changes TestCluster, so can throw InsufficientCapacityException
when gets to a particular size.
- Test for DynamicCluster throwing InsufficientCapacityException
- Test for AutoScalerPolicyMetricTest, to not resize above the failure
level again.
- Test for AutoScalerPolicyNoMoreMachinesTest, for when BYON location
has run out of machines in a DynamicCluster.
Aled Sage [Fri, 8 Jan 2016 21:27:23 +0000 (21:27 +0000)]
Asserts: expectedFailureOfType checks caused-by as well
Alex Heneveld [Wed, 13 Jan 2016 22:58:16 +0000 (22:58 +0000)]
enable and fix test for yaml source being set correctly
also transfer catalog item id and test for that
Alex Heneveld [Wed, 13 Jan 2016 22:34:46 +0000 (22:34 +0000)]
Merge branch 'master' into blueprint-for-app-spec
Alex Heneveld [Wed, 13 Jan 2016 18:05:43 +0000 (18:05 +0000)]
This closes #1115
Alex Heneveld [Wed, 13 Jan 2016 18:02:43 +0000 (18:02 +0000)]
fix tests related to app/entity wrapping/unwrapping
Alex Heneveld [Wed, 13 Jan 2016 16:40:24 +0000 (16:40 +0000)]
tidy up of entity/app unwrapping
Alex Heneveld [Wed, 13 Jan 2016 16:40:06 +0000 (16:40 +0000)]
Merge branch 'master' into blueprint-for-app-spec
Conflicts:
brooklyn-server/camp/camp-brooklyn/src/test/java/org/apache/brooklyn/camp/brooklyn/catalog/CatalogYamlTemplateTest.java -- tests added
John McCabe [Wed, 13 Jan 2016 15:54:01 +0000 (15:54 +0000)]
deprecate group.members.delegate and prevent delegate creation when adding to quarantine
- use of delegate to denote quarantine group membership in the jsgui is no longer required due to #929
Graeme-Miller [Wed, 13 Jan 2016 15:35:17 +0000 (15:35 +0000)]
Fixed an issue, where calling getEntity on BuiltResponsePreservingError would throw a previously caught error. We now throw an IllegalStateException to make it clear where the exception is coming from
Alex Heneveld [Wed, 13 Jan 2016 11:29:26 +0000 (11:29 +0000)]
This closes #1138
Alex Heneveld [Wed, 13 Jan 2016 11:28:03 +0000 (11:28 +0000)]
test (failing) for setting the plan yaml on a spec when it comes from a template registry item
Sam Corbett [Wed, 13 Jan 2016 10:40:59 +0000 (10:40 +0000)]
This closes #1139
Remove karaf-features auto-generated dependencies.xml
Sam Corbett [Wed, 13 Jan 2016 10:31:12 +0000 (10:31 +0000)]
This closes #1137
BROOKLYN-214: fix cancelling of AttributeWhenReady task
Sam Corbett [Wed, 13 Jan 2016 10:19:22 +0000 (10:19 +0000)]
This closes #1133
BROOKLYN-213 Fix localhost external IP loader
Ciprian Ciubotariu [Tue, 12 Jan 2016 17:46:41 +0000 (19:46 +0200)]
[BROOKLYN-183] Remove dangling file
The real stuff is karaf/src/feature/feature.xml
Ciprian Ciubotariu [Tue, 12 Jan 2016 17:35:51 +0000 (19:35 +0200)]
[BROOKLYN-183] Remove karaf dependencies.xml from git
The dependencies.xml file can be used to track dependency changes for
karaf features. For this reason it has to be added to the git repository
so such changes can be detected between commits.
However, developers complained about having this file on git, since it
confusingly keeps showing up in commits. Also, at this stage this
tracking is unnecessary.
Aled Sage [Tue, 12 Jan 2016 16:55:01 +0000 (16:55 +0000)]
YAML: don’t wrap sub-entities in BasicApplication
Previously, any catalog item of type “template” would always be wrapped
in a BasicApplication. This was being done in three places. The first
is ok; the second two are bad:
1. When deploying a top-level “template” from the catalog.
2. When referring to a “template” catalog item in a member-spec (e.g.
for a DynamicCluster)
3. When referring to a “template” catalog item in brooklyn.children.
Aled Sage [Tue, 12 Jan 2016 13:24:43 +0000 (13:24 +0000)]
BROOKLYN-214: fix cancelling of AttributeWhenReady task
iyovcheva [Tue, 12 Jan 2016 08:21:39 +0000 (10:21 +0200)]
Merge pull request #1 from aledsage/iyovcheva/externalised-config-in-catalog-definition-aled
Externalised config in catalog definition
Aled Sage [Mon, 11 Jan 2016 15:17:15 +0000 (15:17 +0000)]
external-config in catalog: avoid parsing items
Don’t parse the “item” or “items” section when populating the catalog.
The camp-yaml DSL section in that is only parsed when the blueprint is
actually being used, rather than when populating the catalog.
Sam Corbett [Mon, 11 Jan 2016 13:59:31 +0000 (13:59 +0000)]
Update commons-lang3 to 3.3.2
Matches Alien4Cloud version, used by brooklyn-tosca
Aled Sage [Mon, 11 Jan 2016 13:27:59 +0000 (13:27 +0000)]
ExternalConfig in catalog: improve tests
Ivana Yovcheva [Tue, 29 Dec 2015 19:48:09 +0000 (21:48 +0200)]
Support "externalised config" in catalog definition
Aled Sage [Fri, 8 Jan 2016 13:59:08 +0000 (13:59 +0000)]
This closes #1134
Sam Corbett [Fri, 8 Jan 2016 12:01:11 +0000 (12:01 +0000)]
Expose BrooklynFeatureEnablement.setDefault
Graeme-Miller [Fri, 8 Jan 2016 11:56:27 +0000 (11:56 +0000)]
add LoopOverGroupMembersTestCase & TargetableTestComponent. Should have been in previous commit but were excluded due to re-org
Matt Champion [Fri, 8 Jan 2016 10:58:01 +0000 (10:58 +0000)]
Remove discontinued IP lookup service - BROOKLYN-213
The IP lookup service http://www.telize.com/ip was discontinued on the 15th of November 2015.