Robert Varga [Mon, 23 Feb 2015 17:30:31 +0000 (18:30 +0100)]
Pass down a collection of cohorts
Rather than using an iterable, use a Collection. This saves a bit of
code and makes the size available wherever the collection goes.
Change-Id: I7ab9fa594676c221c8a4a0b3e227625af462c338
Signed-off-by: Robert Varga <rovarga@cisco.com>
Tom Pantelis [Tue, 24 Feb 2015 00:47:32 +0000 (00:47 +0000)]
Merge "BUG 2676 : Use notification-dispatcher for DataChangeListener actors"
Tom Pantelis [Tue, 24 Feb 2015 00:47:19 +0000 (00:47 +0000)]
Merge "BUG 2676 : Use transaction-dispatcher for ShardTransaction"
Tom Pantelis [Tue, 24 Feb 2015 00:47:03 +0000 (00:47 +0000)]
Merge "BUG 2676 : Use shard-dispatcher for ShardManager and Shard actors"
Tom Pantelis [Tue, 24 Feb 2015 00:46:52 +0000 (00:46 +0000)]
Merge "BUG 2676 : Use custom client-dispatcher when configured"
Tom Pantelis [Mon, 23 Feb 2015 23:13:35 +0000 (23:13 +0000)]
Merge "BUG 2676 : Introduce API for accessing akka dispatchers"
Tom Pantelis [Mon, 23 Feb 2015 23:11:02 +0000 (23:11 +0000)]
Merge "BUG 2718 : Create a diagnostic utility to track append entries replies"
Tom Pantelis [Mon, 23 Feb 2015 19:20:51 +0000 (19:20 +0000)]
Merge "BUG 2741 : Fix issue with real snapshotting when replicatedToAllIndex=-1"
Ed Warnicke [Mon, 23 Feb 2015 18:42:41 +0000 (18:42 +0000)]
Merge "Bug 2673: Binding cursor-based data change API"
Tony Tkacik [Mon, 23 Feb 2015 16:52:26 +0000 (16:52 +0000)]
Merge "Re-enable running of SingleFeaturesTest"
Tony Tkacik [Mon, 23 Feb 2015 16:14:16 +0000 (16:14 +0000)]
Merge "Startup archetype: remove 'Impl' from config subsystem Module name."
Tony Tkacik [Mon, 23 Feb 2015 16:14:06 +0000 (16:14 +0000)]
Merge "Startup archetype: Add basic unit tests for impl."
Tony Tkacik [Mon, 23 Feb 2015 16:12:47 +0000 (16:12 +0000)]
Merge "Startup arch - remove artifactId prefix from dir names."
Tony Tkacik [Mon, 23 Feb 2015 16:08:58 +0000 (16:08 +0000)]
Merge "Cleanup root pom "name"."
Tony Tkacik [Mon, 23 Feb 2015 11:18:23 +0000 (11:18 +0000)]
Merge "Serialize capabilities for notification properly"
Tony Tkacik [Mon, 23 Feb 2015 11:18:02 +0000 (11:18 +0000)]
Merge "BUG-2600 Proper reset caching of notifications upon disconnect"
Tony Tkacik [Mon, 23 Feb 2015 11:17:50 +0000 (11:17 +0000)]
Merge "BUG-2635 Netconf monitoring for md-sal netconf northbound"
Tony Tkacik [Mon, 23 Feb 2015 11:12:25 +0000 (11:12 +0000)]
Merge "DOMDataTree APIs"
Tony Tkacik [Mon, 23 Feb 2015 10:46:08 +0000 (10:46 +0000)]
Merge "Update JMX generator"
Tony Tkacik [Mon, 23 Feb 2015 10:45:09 +0000 (10:45 +0000)]
Merge "xsql should pull junit directly"
Maros Marsalek [Fri, 20 Feb 2015 14:13:03 +0000 (15:13 +0100)]
Serialize capabilities for notification properly
Added and removed capabilities in capabilities change notification were serialized as QName.toString but should be na Uri
Change-Id: Id9783c060ecf45d3a7f506cf605051c779a6d21a
Signed-off-by: Maros Marsalek <mmarsale@cisco.com>
Maros Marsalek [Fri, 20 Feb 2015 09:40:12 +0000 (10:40 +0100)]
BUG-2600 Proper reset caching of notifications upon disconnect
Change-Id: I9269aef2fee21138b30552ffb0762d73ba98d4be
Signed-off-by: Maros Marsalek <mmarsale@cisco.com>
Maros Marsalek [Wed, 18 Feb 2015 16:31:39 +0000 (17:31 +0100)]
BUG-2635 Netconf monitoring for md-sal netconf northbound
Monitoring for mdsal's netconf stores all available monitoring data in the datastore and provides get-schema operation
Change-Id: I573c5d57e5cf25d7688f3355b602327c7af75c65
Signed-off-by: Maros Marsalek <mmarsale@cisco.com>
Tony Tkacik [Mon, 23 Feb 2015 10:44:52 +0000 (10:44 +0000)]
Merge "Migrate JMX plugin's use of BindingGeneratorUtil"
Tony Tkacik [Mon, 23 Feb 2015 10:43:45 +0000 (10:43 +0000)]
Merge "Migrate deprecated guava methods"
Tony Tkacik [Mon, 23 Feb 2015 10:43:27 +0000 (10:43 +0000)]
Merge "Do not override artifact versions"
Tony Tkacik [Mon, 23 Feb 2015 10:43:13 +0000 (10:43 +0000)]
Merge "Use Preconditions"
Tony Tkacik [Mon, 23 Feb 2015 10:43:00 +0000 (10:43 +0000)]
Merge "Do not override jsr305 version"
Robert Varga [Thu, 5 Feb 2015 17:30:28 +0000 (18:30 +0100)]
DOMDataTree APIs
This patch introduces the end-user visible API model for interacting
with the conceptual data tree. Unlike our previous APIs, these revolve
around applications explicitly binding their inputs (former data change
listeners) and outputs (former explicit writes). This allows efficient
inter-datastore interactions to be maintained by understanding what
sort of output applications make and what sort of indexes are required
to feed them the data they require. Beryllium evolution will see the
DOMDataTreeListener being subscribed to an XQuery expression, thus
providing a dense definition of what data the applciation actually
extracts out of the subtree.
Change-Id: I11446ce7f8dbd7cf61540f72e463460b483ff15f
Signed-off-by: Robert Varga <rovarga@cisco.com>
Maros Marsalek [Fri, 13 Feb 2015 12:32:17 +0000 (13:32 +0100)]
BUG-2635 Prepare netconf monitoring service for md-sal monitoring.
So that the monitoring service provides all the information in one place.
Also move get-schema rpc into netconf-monitoring from netconf-impl.
Change-Id: Ia7dc47ea4997df417795f079a7c045709dbaaf47
Signed-off-by: Maros Marsalek <mmarsale@cisco.com>
Maros Marsalek [Thu, 19 Feb 2015 12:53:38 +0000 (13:53 +0100)]
Remove required source element from Get operation for MDSAL
Change-Id: I2c1495dfac8ecad00605f4df98b310b3d99ac4ed
Signed-off-by: Maros Marsalek <mmarsale@cisco.com>
Tony Tkacik [Mon, 23 Feb 2015 08:50:12 +0000 (08:50 +0000)]
Merge "Make reconnect on capability changed truly optional"
Tony Tkacik [Mon, 23 Feb 2015 08:49:48 +0000 (08:49 +0000)]
Merge "Prevent multiple config pushers"
Moiz Raja [Sun, 22 Feb 2015 18:17:44 +0000 (10:17 -0800)]
BUG 2741 : Fix issue with real snapshotting when replicatedToAllIndex=-1
Change-Id: I59b236fd88a329021365f204077170ca260e0b32
Signed-off-by: Moiz Raja <moraja@cisco.com>
tpantelis [Fri, 20 Feb 2015 10:23:09 +0000 (05:23 -0500)]
Fix spurious failures in RaftActorTest due to RecoveryComplete
testSaveSnapshotSuccessCallsDataPersistence fails sporadically due
to RecoveryComplete interfering with the test. In
handleCaptureSnapshotReply it sets replicatedToAllIndex in the
behavior. On RecoveryCompleted, it initializes the behavior which
creates a new Follower with replicatedToAllIndex initialized to -1.
Recovery is done asyn so if RecoveryCompleted occurs after
CaptureSnapshotReply, the test fails because replicatedToAllIndex
isn't the expected value.
We need to call waitForInitializeBehaviorComplete after creating the
MockRaftActor so recovery and behavior initialization is complete before
starting the tests.
Change-Id: I3176afbf7ac937550c33bdb57d14d6968dbc2514
Signed-off-by: tpantelis <tpanteli@brocade.com>
Moiz Raja [Sat, 21 Feb 2015 21:45:00 +0000 (21:45 +0000)]
Merge "Fix RaftActorTest.testRaftRoleChangeNotifier failure"
Moiz Raja [Wed, 18 Feb 2015 20:33:53 +0000 (12:33 -0800)]
BUG 2733 : Leader should always schedule heartbeats
This ensures that lagging or new followers are not starved as this
can potentially destabilize the cluster
Change-Id: Ic83d9d595a1222550286f3eb29d4cdef7bf1bccd
Signed-off-by: Moiz Raja <moraja@cisco.com>
Moiz Raja [Sat, 21 Feb 2015 17:59:06 +0000 (09:59 -0800)]
BUG 2676 : Use notification-dispatcher for DataChangeListener actors
Change-Id: I154e1598a3058a978902fd6db5512d51dff36a62
Signed-off-by: Moiz Raja <moraja@cisco.com>
Moiz Raja [Sat, 21 Feb 2015 17:19:14 +0000 (09:19 -0800)]
BUG 2676 : Use transaction-dispatcher for ShardTransaction
Change-Id: If70e141ff4e0eae0bd41ba70d4b0c126eb752fa1
Signed-off-by: Moiz Raja <moraja@cisco.com>
Moiz Raja [Sat, 21 Feb 2015 17:00:42 +0000 (09:00 -0800)]
BUG 2676 : Use shard-dispatcher for ShardManager and Shard actors
Change-Id: Ib6aed392eb861bb1da16d4780f08b27fad6d17db
Signed-off-by: Moiz Raja <moraja@cisco.com>
Moiz Raja [Sat, 21 Feb 2015 16:43:33 +0000 (08:43 -0800)]
BUG 2676 : Use custom client-dispatcher when configured
Use a client dispatcher for all code which originates at the client
end and requires a dispatcher for completion of a Future that was
created by an ask
Change-Id: If2c1fa064e368152539ad21622985cc044f305d3
Signed-off-by: Moiz Raja <moraja@cisco.com>
Ed Warnicke [Sat, 21 Feb 2015 16:41:33 +0000 (09:41 -0700)]
Re-enable running of SingleFeaturesTest
For reasons I have not yet figured out, depending on the
yangtools version of features-test causes features tests
to no longer run. This patch changes controller
to depending on the features-tests from odlparent
Change-Id: I45397f042cc367addde2ada74a27aba98fc98126
Signed-off-by: Ed Warnicke <eaw@cisco.com>
Moiz Raja [Sat, 21 Feb 2015 14:58:33 +0000 (06:58 -0800)]
BUG 2676 : Introduce API for accessing akka dispatchers
Change-Id: I2a9b0f0848da1a6181a3e6c8d61d0d96499a61d0
Signed-off-by: Moiz Raja <moraja@cisco.com>
Moiz Raja [Sat, 21 Feb 2015 01:59:40 +0000 (17:59 -0800)]
BUG 2718 : Create a diagnostic utility to track append entries replies
This patch introduces a utility to keep track of all messages that were
processed between the time a certain message is received. It is used for
tracking why AppendEntriesReply do not come in a timely manner. This is
important to know because if a leader will switch to IsolatedLeader if
these messages do not arrive within the ElectionTimeout interval causing
disruption in service.
Change-Id: I13bab798aad8a225d22d2ebd3775cbf0bf1d1592
Signed-off-by: Moiz Raja <moraja@cisco.com>
Tom Pantelis [Sat, 21 Feb 2015 13:36:49 +0000 (13:36 +0000)]
Merge "Fix for Bug 2727 - Upgrade Akka from 2.3.4 to 2.3.9"
Robert Varga [Fri, 20 Feb 2015 11:09:37 +0000 (12:09 +0100)]
Update JMX generator
This takes advantage of the fact maven 3.1.1 supports slf4j natively.
Also fixup some of the coding style issues still present.
Change-Id: I37724cc0a5c194fc333c577f39deaf550c341d34
Signed-off-by: Robert Varga <rovarga@cisco.com>
Robert Varga [Sat, 21 Feb 2015 10:20:45 +0000 (11:20 +0100)]
xsql should pull junit directly
Currently we pull junit transitively, which means it will break once the
transitive dependencies change.
Change-Id: I50ad5143811fb21d4719bd47085f366899ae7409
Signed-off-by: Robert Varga <rovarga@cisco.com>
Tom Pantelis [Sat, 21 Feb 2015 05:18:44 +0000 (05:18 +0000)]
Merge "BUG 2702 : Switch to a low impact scheduling mechanism for IsolatedLeaderCheck"
Tom Pantelis [Sat, 21 Feb 2015 05:17:41 +0000 (05:17 +0000)]
Merge "BUG 2676 : Add a diagnostic log message to detect delay in ApplyState processing"
Ed Warnicke [Sat, 21 Feb 2015 05:09:23 +0000 (05:09 +0000)]
Merge "Revert "BUG-2637: migrate features""
Ed Warnicke [Sat, 21 Feb 2015 04:34:14 +0000 (04:34 +0000)]
Revert "BUG-2637: migrate features"
This reverts commit
300d247c3c68387bf0bed3fc9dd90558ebe01f3f.
300d247c3c68387bf0bed3fc9dd90558ebe01f3f broke because the flow-features were not migrated properly to openflowplugin.
Change-Id: I6502a93b10b4c5ee41d037da216092cdf003d17f
Signed-off-by: Ed Warnicke <eaw@cisco.com>
Anil Vishnoi [Fri, 20 Feb 2015 20:01:18 +0000 (20:01 +0000)]
Merge "BUG-2637: migrate features"
Moiz Raja [Fri, 20 Feb 2015 12:37:46 +0000 (04:37 -0800)]
BUG 2702 : Switch to a low impact scheduling mechanism for IsolatedLeaderCheck
Using the akka scheduler's schedule mechanism to schedule tasks can cause
disruption in processing messages as scheduled messages can get processed before
other messages. To avoid this we now generate an isolated leader check from
the heartbeat processor and avoid an extra scheduled task.
Change-Id: Ied6b518796da92742e2c61eebcd1e0a3415df66a
Signed-off-by: Moiz Raja <moraja@cisco.com>
Moiz Raja [Fri, 20 Feb 2015 12:15:27 +0000 (04:15 -0800)]
BUG 2676 : Add a diagnostic log message to detect delay in ApplyState processing
Log a message when ApplyState is processed 50 milliseconds after being generated
Change-Id: I7e101dab292b4a72d23904798dc2362a740e4c04
Signed-off-by: Moiz Raja <moraja@cisco.com>
Tom Pantelis [Fri, 20 Feb 2015 19:18:47 +0000 (19:18 +0000)]
Merge "BUG 2676 : Add more context to the Initiate snapshot capture log message"
Tom Pantelis [Fri, 20 Feb 2015 19:17:17 +0000 (19:17 +0000)]
Merge "BUG 2676 : Reduce default logging in Dummy Data Store"
Tom Pantelis [Fri, 20 Feb 2015 19:16:09 +0000 (19:16 +0000)]
Merge "BUG 2676 : Add diagnostic message when AppendEntriesReply comes in very late"
Tom Pantelis [Fri, 20 Feb 2015 19:14:47 +0000 (19:14 +0000)]
Merge "BUG 2676 : Add some tuning parameters to improve performance of CDS"
Michal Rehak [Fri, 23 Jan 2015 21:12:26 +0000 (22:12 +0100)]
BUG-2637: migrate features
- remove migrated features
- integration test fix:
- commented out dependencies to models affected by migration
- adapted sal-binding-it in order not to push those models
into pax-exam context
- adapted sal-binding-broker-impl to use those models only in
test scope
- cleaned commented code/config
Change-Id: I8e09a7cd9ff9ba0f041abe1bf739be755aed3e8d
Signed-off-by: Michal Rehak <mirehak@cisco.com>
Moiz Raja [Fri, 20 Feb 2015 11:57:36 +0000 (03:57 -0800)]
BUG 2676 : Add more context to the Initiate snapshot capture log message
Add more context in the Initiate snapshot capture log message so that when
a snapshot capture occurs due to lack of memory we can get a little more
detail from the logs.
Change-Id: I85647adcf74d7eb36f767ba5dd6824e64a802811
Signed-off-by: Moiz Raja <moraja@cisco.com>
Moiz Raja [Fri, 20 Feb 2015 11:49:03 +0000 (03:49 -0800)]
BUG 2676 : Reduce default logging in Dummy Data Store
Reduce the default log level to error and added a diagnostic
log message to detect situations where an append entries does
not get sent in a reasonable amount of time (5 seconds)
Change-Id: I46b7b20688207318dcfec2b209519761ecf1d93d
Signed-off-by: Moiz Raja <moraja@cisco.com>
Moiz Raja [Fri, 20 Feb 2015 11:38:18 +0000 (03:38 -0800)]
BUG 2676 : Add diagnostic message when AppendEntriesReply comes in very late
Log an error message when an AppendEntriesReply arrives beyond the ElectionTimeout
period. This may help detect cases where messages are not being processed
fast enough by the actor system. This particular condition can cause a Leader to
become an IsolatedLeader even if the remote Follower is receiving heartbeats.
Change-Id: Id6a797820b63cb4f18c3d1bbd2781e72674aa9d8
Signed-off-by: Moiz Raja <moraja@cisco.com>
Robert Varga [Fri, 20 Feb 2015 11:45:15 +0000 (12:45 +0100)]
Migrate JMX plugin's use of BindingGeneratorUtil
The functionality was moved to BindingMapping, so migrate this user.
Change-Id: I23581636182b1e63a82d8b50595c5aba5e5ffc5a
Signed-off-by: Robert Varga <rovarga@cisco.com>
Moiz Raja [Fri, 20 Feb 2015 11:20:38 +0000 (03:20 -0800)]
BUG 2676 : Add some tuning parameters to improve performance of CDS
Added two parameters,
- throughput = 1 this is to make the dispatcher fair, which helps in preventing
actor starvation
- SingleConsumerOnlyUnboundedMailbox which is a more efficient and recommended
Mailbox to use in our case
I've tested both these settings and they seem to work pretty well and do not
contribute to destabilization
Change-Id: I5072f7dad7ad862eaafaada3e88d78600fb3665d
Signed-off-by: Moiz Raja <moraja@cisco.com>
Moiz Raja [Fri, 20 Feb 2015 10:53:20 +0000 (02:53 -0800)]
BUG 2734 : Use the 95th percentile number to determine backpressure
When running tests with OpenFlow I see the following transaction completion
times consistently for 127 and 255 switches with statistics on.
50% - 5-6 ms
75% - 7-8 ms
95% - 20-60 ms
98% - 100-160 ms
By using the 95th percentile number to determine transaction creation backpressure
we can safely increase our throughput without causing timeouts due to outliers.
Change-Id: I0d9739b3045afb389583f2dd18ac431215a28f84
Signed-off-by: Moiz Raja <moraja@cisco.com>
Maros Marsalek [Thu, 19 Feb 2015 12:53:08 +0000 (13:53 +0100)]
Make reconnect on capability changed truly optional
There was true by default hardcoded in the code ignoring configuration
Change-Id: If6d8acdfba350eaed65e660c7b9e994751e9ddf7
Signed-off-by: Maros Marsalek <mmarsale@cisco.com>
Maros Marsalek [Thu, 19 Feb 2015 10:46:17 +0000 (11:46 +0100)]
Prevent multiple config pushers
Up to this point, it was possible to spawn multiple config pushers.
This was probably already happening with no observable effect other than relatively frequent Optimistic lock failed exceptions from config subsystem.
Change-Id: Idf994c1a184a070a32440c57c5ca55a0740c21d6
Signed-off-by: Maros Marsalek <mmarsale@cisco.com>
tpantelis [Fri, 20 Feb 2015 06:47:58 +0000 (01:47 -0500)]
Fix RaftActorTest.testRaftRoleChangeNotifier failure
I saw this test fail once:
java.lang.AssertionError: expected:<3> but was:<2>
at org.junit.Assert.fail(Assert.java:88)
at org.junit.Assert.failNotEquals(Assert.java:743)
at org.junit.Assert.assertEquals(Assert.java:118)
at org.junit.Assert.assertEquals(Assert.java:555)
at org.junit.Assert.assertEquals(Assert.java:542)
at
org.opendaylight.controller.cluster.raft.RaftActorTest$13.<init>(RaftActorTest.java:894)
at
org.opendaylight.controller.cluster.raft.RaftActorTest.testRaftRoleChangeNotifier(RaftActorTest.java:881)
The test has an arbitrary 2 sec wait before retrieving the collected
RoleChanged messages. With the election timeout and arbitrary GC and threading
delays, 2 sec may not be enough.
Arbitrary sleeps in tests are usually trouble - I made it deterministic
by waiting 5 sec for the expected messages.
Change-Id: I51c24a5844321bbd000a0460777d08b51a5f82c8
Signed-off-by: tpantelis <tpanteli@brocade.com>
Robert Varga [Thu, 19 Feb 2015 10:16:54 +0000 (11:16 +0100)]
Migrate deprecated guava methods
sameThreadExecutor() has been deprecated, use its replacement instead.
Change-Id: Icf1fed34977bd64393f48fe9e9829bee7f7db846
Signed-off-by: Robert Varga <rovarga@cisco.com>
Robert Varga [Thu, 19 Feb 2015 08:32:05 +0000 (09:32 +0100)]
Use Preconditions
Instead of an explicit if/throw, use a checkState.
Change-Id: Ia871c93649bfa6bd2f6e2283a3baa7eca6a8ca91
Signed-off-by: Robert Varga <rovarga@cisco.com>
Robert Varga [Thu, 19 Feb 2015 08:26:59 +0000 (09:26 +0100)]
Do not override artifact versions
We already have these defined in appropriate management sections, no
need to override them here.
Change-Id: I529ade39566be065c6992ef1b72a3d8f0cdef076
Signed-off-by: Robert Varga <rovarga@cisco.com>
tpantelis [Wed, 18 Feb 2015 02:26:21 +0000 (21:26 -0500)]
Refactor AbstractRaftActorBehaviorTest and IsolatedLeaderTest
Removed the Within blocks in tests.
Used TestActorFactory to create and cleanup actors.
Added some more assertions in some tests.
Change-Id: Idd790b97dd396c75676843779ceab192b848aa17
Signed-off-by: tpantelis <tpanteli@brocade.com>
Moiz Raja [Fri, 20 Feb 2015 00:13:32 +0000 (00:13 +0000)]
Merge "InstallSnapshot does not pass the term to the protobuff instance"
Tony Tkacik [Thu, 19 Feb 2015 08:15:32 +0000 (08:15 +0000)]
Merge "Initial message bus implementation"
Tony Tkacik [Thu, 19 Feb 2015 08:09:00 +0000 (08:09 +0000)]
Merge "BUG-2634 Config binding for netconf server"
Tony Tkacik [Thu, 19 Feb 2015 08:07:49 +0000 (08:07 +0000)]
Merge "BUG-2633 - Netconf northbound mapping."
Tony Tkacik [Thu, 19 Feb 2015 08:07:18 +0000 (08:07 +0000)]
Merge "BUG-2633 - Added Operation Service Factory Filtering for Netconf."
tpantelis [Wed, 11 Feb 2015 07:43:28 +0000 (02:43 -0500)]
InstallSnapshot does not pass the term to the protobuff instance
Change-Id: If84d7ddc81fd1ea154477fe9795bba9e748af24c
Signed-off-by: tpantelis <tpanteli@brocade.com>
tpantelis [Wed, 11 Feb 2015 02:41:12 +0000 (21:41 -0500)]
Refactor CandidateTest
Removed the Within blocks in tests.
Used TestActorFactory to create and cleanup actors.
Added some more assertions in some tests.
Change-Id: I558876cd058a0ad7818317ec98fc93c4bde95630
Signed-off-by: tpantelis <tpanteli@brocade.com>
tpantelis [Tue, 10 Feb 2015 21:08:36 +0000 (16:08 -0500)]
Refactor FollowerTest
Removed the Within blocks in tests.
Changed all the tests to use the leaderActor and followerActor class
members for consistency. The actors are killed on tearDown by the TestActorFactory.
The creation of a JavaTestKit block is no longer necessary and was removed from each test.
Added some more assertions in some tests.
Change-Id: Iee793c2b1129b5826dbd3c267c404aae30418fdb
Signed-off-by: tpantelis <tpanteli@brocade.com>
Moiz Raja [Thu, 19 Feb 2015 21:34:47 +0000 (21:34 +0000)]
Merge "Refactor LeaderTest"
Tom Pantelis [Thu, 19 Feb 2015 20:57:45 +0000 (20:57 +0000)]
Merge "Bug 2538: Remove redundant Augmentation checks and tests"
Michal Rehak [Thu, 5 Feb 2015 11:32:33 +0000 (12:32 +0100)]
BUG-2682: switch sal-binding-dom-it to sal test models
- DilcardedDataChangeListenerTest
- ConcurrentImplicitCreateTest
- BrokerIntegrationTest
- ChangeOriginatedInDomBrokerTest
Change-Id: I20f2c6d5d93a535f0ccfb0efc484ec099a75f53b
Signed-off-by: Michal Rehak <mirehak@cisco.com>
Tony Tkacik [Thu, 19 Feb 2015 17:09:32 +0000 (17:09 +0000)]
Merge "Bug 2682 - Switch sal-binding-dom-it to sal-test-model"
Michal Rehak [Thu, 19 Feb 2015 12:39:06 +0000 (13:39 +0100)]
Bug 2682 - Switch sal-binding-dom-it to sal-test-model
* Migration to sal-test-model
Change-Id: If46014ec4cc8b3311a7766bcb2cb9a61b9e71592
Signed-off-by: Jan Hajnar <jhajnar@cisco.com>
Signed-off-by: Martin Bobak <mbobak@cisco.com>
Signed-off-by: Michal Rehak <mirehak@cisco.com>
Robert Varga [Sat, 7 Feb 2015 21:55:46 +0000 (22:55 +0100)]
Initial message bus implementation
This patch follows up on the API definition and builds a simple
implementation.
Change-Id: Ic128538a02d71a40ea44efe53f3c4b4503c068ac
Signed-off-by: Robert Gallas <rgallas@cisco.com>
Signed-off-by: Robert Varga <rovarga@cisco.com>
Robert Varga [Sat, 7 Feb 2015 20:12:07 +0000 (21:12 +0100)]
Introduce message bus models
This patch introduces the basic mechanics for integrating various event
sources across MD-SAL, effectively forming a message bus.
Change-Id: I5cbaf9c0975f38c3e3c05ab9862aacf20d2b57c6
Signed-off-by: Robert Gallas <rgallas@cisco.com>
Signed-off-by: Robert Varga <rovarga@cisco.com>
tpantelis [Tue, 10 Feb 2015 06:07:28 +0000 (01:07 -0500)]
Raft behavior logging
Added more logging in the raft behavior classes where appropriate.
Reduced logging in some places to trace or removed redundant logging to
avoid too much output when debug level is enabled.
Removed isDebugEnabled() checks in various places where not needed to
improve test line coverage.
Also I found it useful to see the behavior's state in the log messages.
So I added a logName field and method to AbstractRaftBehavior that
is printed in each log message instead of just context.getId().
Also added/modified toString() in a few RaftRPC message classes.
Change-Id: Ic4d3c93e25d5caf16319b8433d54cab69b26f5c0
Signed-off-by: tpantelis <tpanteli@brocade.com>
Moiz Raja [Wed, 18 Feb 2015 22:38:50 +0000 (22:38 +0000)]
Merge "Bug-2692:Real snapshots should use the replicatedToAllIndex for clearing in-mem log"
Tom Pantelis [Wed, 18 Feb 2015 21:51:47 +0000 (21:51 +0000)]
Merge "Make ReadData exception logging less noisy"
Tom Pantelis [Wed, 18 Feb 2015 21:50:49 +0000 (21:50 +0000)]
Merge "Minor enhancements to dummy datastore"
Moiz Raja [Tue, 17 Feb 2015 17:46:03 +0000 (09:46 -0800)]
Minor enhancements to dummy datastore
- Log Duplicate AppendEntries
- Enhance AppendEntries logging
- Make logging configurable
- Handle append entries with multiple entries
- Do not delay append entry reply for heartbeats, do so only for messages with entries
Change-Id: Ife856bb469e81c42d22ff4a6f13637ee2cf00cce
Signed-off-by: Moiz Raja <moraja@cisco.com>
Moiz Raja [Tue, 17 Feb 2015 17:52:33 +0000 (09:52 -0800)]
Make ReadData exception logging less noisy
There is some component (openflow) which is reading data on a ClosedTransaction.
This is causing logging on the datastore side. The component itself seems to handle
the error reported by the datastore quietly, so switching the logging the error to
debug and logging full stack trace at trace level.
Change-Id: I3c118765dca7f4a50d26c8dd50eee96c65de1871
Signed-off-by: Moiz Raja <moraja@cisco.com>
Tony Tkacik [Wed, 18 Feb 2015 16:12:06 +0000 (16:12 +0000)]
Merge "Introduce message bus models"
Robert Varga [Wed, 18 Feb 2015 10:41:43 +0000 (11:41 +0100)]
Do not override jsr305 version
The override puts it at 2.0.1, odlparent has it 2.0.3, which causes
compile-time maven warnings. Remove the override.
Change-Id: I99b874977b670d8af811642ecb28c2240d965dfd
Signed-off-by: Robert Varga <rovarga@cisco.com>
Tony Tkacik [Wed, 18 Feb 2015 09:31:01 +0000 (09:31 +0000)]
Merge "BUG-2573: mark CompositeNode-based RPC APIs obsolete"
Tony Tkacik [Wed, 18 Feb 2015 09:27:26 +0000 (09:27 +0000)]
Merge "Marked as deprecated all classes within protocol-framework."
Robert Varga [Mon, 16 Feb 2015 08:46:53 +0000 (09:46 +0100)]
Migrate to MoreObjects.ToStringHelper
Objects.ToStringHelper has been deprecated and moved to MoreObjects.
Adjust to this change.
Change-Id: Ibcaaebfe8e2cbbd77e624a53fc153d857b373a81
Signed-off-by: Robert Varga <rovarga@cisco.com>
tpantelis [Tue, 10 Feb 2015 13:15:07 +0000 (08:15 -0500)]
Refactor LeaderTest
Removed the remaining Within blocks in tests.
Changed each test to use the leaderActor and followerActor class members for
consistency. The actors are killed on tearDown. The creation of a
JavaTestKit block is no longer necessary and was removed from each test.
Added a shorcut method createActorContextWithFollower that creates the
leader context and adds a peer follower. This eliminates some duplicate code
in the tests.
Added an expectFirstMatching method to MessageCollector that waits for
the message in contrast to getFirstMatching.
Added some more assertions in some tests.
Change-Id: Id4ee0291352999b31e40abd9895e3d0237acf432
Signed-off-by: tpantelis <tpanteli@brocade.com>
Kamal Rameshan [Wed, 11 Feb 2015 07:24:55 +0000 (23:24 -0800)]
Bug-2692:Real snapshots should use the replicatedToAllIndex for clearing in-mem log
Similar to fake snapshots, real snapshots should also be using the replicatedToAllIndex.
replicatedToAllIndex is part of the RaftActorBehavior.
A performSnapshotWithoutCapture helper method is part of the AbstractRaftActorBehavior.
On CaptureSnapshotReply, we clear the log based on replicatedToAllIndex and the corresponding term.
Many of the existing tests seem to test out this changes and some were added and enhanced.
Bug-2715: Fix in-mem log cleanup for an Inactive follower
In case if one of the follower is down, rely on the memory usage to clear the log.
This is done in CaptureSnapshotReply
Change-Id: Ifb3ca19691f20735e3f84b968a7adf01398c20e0
Signed-off-by: Kamal Rameshan <kramesha@cisco.com>