20 months agoSend commitIndex updates to followers as soon as possible 98/82498/1
Robert Varga [Tue, 11 Jun 2019 13:42:34 +0000 (15:42 +0200)]
Send commitIndex updates to followers as soon as possible

When the commitIndex gets updated when we reach consensus, we should
propagate this fact to the sender peer. Otherwise data tree change
listeners on peers may be unnecessarily delayed, as they end up
waiting for heartbeat interval (or non-batched entries) to be propagated

The results in observably-better behaviour between leader and follower,
as the leader is sending out notifications of consensus, without
delayling for heartbeat -- which is reflected in the changes to

Furthermore, DelayedMessagesElectionScenarioTest is detecting the change,
as two members may achieve asynchronous consensus -- hence the leader
would be sending two messages (request to persist, commitIndex) to push
the consensus out. This exposes a race in the test, where we do not know
the order between TimeoutNow and AppendEntries. If AppendEntries is
processed by member 2 after TimeoutNow, leader will know about the new
term due to AppendEntriesReply we send out. The cluster will still
converge, but the assert for leader will be different if it manages to
process the message. Fix this by isolating member 2.

Change-Id: I695ef25c7a4cf8799c9c5e04c2c33fbf3e2f21df
Signed-off-by: Robert Varga <>
(cherry picked from commit c2b0e92d2c3ba05abf0fefb24413c68fd1c56bea)

20 months agoSimplify recoveryTime computation 58/82458/3
Robert Varga [Fri, 7 Jun 2019 22:44:58 +0000 (00:44 +0200)]
Simplify recoveryTime computation

Stopwatch.stop() returns self, take advantage of that to shorted
the code to compute the formatting string. Also drop toString()
call, as it is implied by string concat.

Change-Id: I4fa645e42fc38b5f9f0775d986fd17aa693725b1
Signed-off-by: Robert Varga <>
20 months agoSwitch default output stream version 13/82313/27
Robert Varga [Wed, 29 May 2019 16:17:13 +0000 (18:17 +0200)]
Switch default output stream version

All versioned callers have been converted to use appropriate
stream versions, this patch bumps the default unversioned stream
to Sodium version.

Since we're changing the serialization format for ExecuteRpc,
make sure we fully embrace writer/reader, so that we get better

Change-Id: I17deeda1a49394600a5b0f93bd46d3b2c812123c
Signed-off-by: Robert Varga <>
20 months agoBump persisted PayloadVersion 14/82314/25
Robert Varga [Wed, 29 May 2019 16:23:04 +0000 (18:23 +0200)]
Bump persisted PayloadVersion

Since the NormalizedNodeStream format has changed, shard persisted
state is affected.

This patch bumps PayloadVersion and applies that bump to snapshots
and CommitTransactionPayload.

On recovery, a snapshot's need to migrate is reflected in its state,
and it is examined just as MigratedSerializables are.
CommitTransactionPayload is not examined, as understanding its
stream version would require deserializing at least its header.

Change-Id: I678527be4487ee1729123ba8b9dcd2269e6cf262
Signed-off-by: Robert Varga <>
20 months agoBump datastore version 11/82311/21
Robert Varga [Wed, 29 May 2019 14:59:01 +0000 (16:59 +0200)]
Bump datastore version

This bumps the ask-based protocol datastore version, so that
SODIUM_VERSION is using Sodium NormalizedNode{Input,Output}.

Change-Id: Ib4ad1c27566179db958ad7776f81417ab99a78cf
Signed-off-by: Robert Varga <>
20 months agoAdd QNameModule coding 26/82426/5
Robert Varga [Thu, 6 Jun 2019 17:36:06 +0000 (19:36 +0200)]
Add QNameModule coding

Coding on QNames is effective when there are a lot of same QNames,
but there are cases where QNames themeselves are not reused --
for example YangInstanceIdentifiers and small read subtrees.

These typically end up using different QNames from a small set
of modules, hence coding modules actually helps them quite a bit.

This patch adds a namespace/revision cache, so that encoding such
QNames is also efficient.

Change-Id: I6cd0885ef605e4a1bf7b768ffe215e9eb3e2fbc5
Signed-off-by: Robert Varga <>
20 months agoAdd buffering to LocalSnapshotStore 39/82439/3
Tomas Cere [Tue, 4 Jun 2019 13:32:20 +0000 (15:32 +0200)]
Add buffering to LocalSnapshotStore

These improve snapshot save/load load times by around 15-20%
on my environment.

Change-Id: I67236f7e97f519993031d462c947588004857c33
Signed-off-by: Tomas Cere <>
20 months agoRemove use of thread-local input 91/82391/9
Robert Varga [Mon, 3 Jun 2019 19:37:43 +0000 (21:37 +0200)]
Remove use of thread-local input

All callers are only called from top-level constructs, thus
not having a relationship with the context in which they are

Change-Id: I856f9109480fb83475a20ce4430d02a1dd055522
Signed-off-by: Robert Varga <>
20 months agoTake advantage of AugmentationIdentifier.create() 99/82399/2
Robert Varga [Tue, 4 Jun 2019 17:51:43 +0000 (19:51 +0200)]
Take advantage of AugmentationIdentifier.create()

When we are reading an augmentation identifier for the first
time we can optimize the number of identifiers by consulting
the global weak cache.

Change-Id: I5cd2d87e0a7c3f01357beb1b097d4a16a6eda3e0
Signed-off-by: Robert Varga <>
20 months agoDo not use NodeIdentifierWithPredicates constructors 97/82397/3
Robert Varga [Tue, 4 Jun 2019 17:05:55 +0000 (19:05 +0200)]
Do not use NodeIdentifierWithPredicates constructors

This class will transition into a value-based object for memory
efficiency reasons. Switch to using its static factory methods.

Change-Id: I7e21a6ccf1ad999d2d0427769b26dffeac166274
Signed-off-by: Robert Varga <>
20 months agoOptimize readQNameSet() 98/82398/2
Robert Varga [Tue, 4 Jun 2019 17:54:22 +0000 (19:54 +0200)]
Optimize readQNameSet()

We are providing a bad hint to HashSet, as we are not taking
into consideration its load factor. Use utility methods from Guava
to get a properly-sized set.

Change-Id: I568be3040b924a8ed77aa0224ab45673104c4664
Signed-off-by: Robert Varga <>
20 months agoRemove sal-binding-broker dependency on javassist 95/82395/2
Robert Varga [Tue, 4 Jun 2019 12:05:27 +0000 (14:05 +0200)]
Remove sal-binding-broker dependency on javassist

We are not using Javassist anymore, remove the dependency on it.

Change-Id: I46e8f30e002d13d324779f461ac0365b67a6f70a
Signed-off-by: Robert Varga <>
20 months agoRemove use of thread-local output 90/82390/6
Robert Varga [Mon, 3 Jun 2019 19:12:44 +0000 (21:12 +0200)]
Remove use of thread-local output

All callers can be safely migrated to using a non-shared writer,
as they all are expected to be apex implementations.

Change-Id: I87cfa1d3ec415b79c40c29a91ccbd8da2869b27a
Signed-off-by: Robert Varga <>
20 months agoBump mdsal to 4.0.2 66/82266/6
Robert Varga [Tue, 28 May 2019 07:18:32 +0000 (09:18 +0200)]
Bump mdsal to 4.0.2

This picks up the latest fixes from upstream.

Change-Id: I34ef8d8f5d61c6ca9f795369ab99b06f4488d5f3
Signed-off-by: Robert Varga <>
20 months agoBump yangtools to 3.0.1 65/82265/5
Robert Varga [Tue, 28 May 2019 07:18:08 +0000 (09:18 +0200)]
Bump yangtools to 3.0.1

This bumps yangtools to 3.0.1, picking up latest fixes and

Change-Id: Id7ca8d3e23865070896b9da31e7d09a509b21578
Signed-off-by: Robert Varga <>
20 months agoSwitch CompositeModification to bypass thread-local streams 88/82388/5
Robert Varga [Mon, 3 Jun 2019 17:43:22 +0000 (19:43 +0200)]
Switch CompositeModification to bypass thread-local streams

Thread-local streams have little control over the version being
used, as the stream is allocated all over the place.

This model cannot function without explicit setup, which is used
by CompositeModification. This patch migrates CompositeModification
to use a new NormalizedNodeDataInput-based interface, so
thread-locals become obsolete.

Change-Id: I1111be5d6fc8c0129fb72232c3e713004068f152
Signed-off-by: Robert Varga <>
20 months agoMove byte-based serialization method 89/82389/2
Robert Varga [Mon, 3 Jun 2019 18:52:36 +0000 (20:52 +0200)]
Move byte-based serialization method

This is only used in testing, move the method out to testing class.

Change-Id: Ib18be55467561a2471a74285b062b40d4e9e08c8
Signed-off-by: Robert Varga <>
20 months agoMake Modification/Externalizable interface visible 87/82387/1
Robert Varga [Mon, 3 Jun 2019 16:56:44 +0000 (18:56 +0200)]
Make Modification/Externalizable interface visible

Finding direct callers of these methods is not easy, override
them to get exact call sites. This will help us with versioning
of the output stream.

Change-Id: Idc9bb036fac786b4480c098bd592e272d5a4c0ad
Signed-off-by: Robert Varga <>
20 months agoRemove ask-based close transaction chain persistence 37/79037/9
Robert Varga [Thu, 23 May 2019 16:57:00 +0000 (18:57 +0200)]
Remove ask-based close transaction chain persistence

ask-based protocol has tracking disabled when encountered, hence
it does not make sense to persist anything.

Change-Id: I1ec87a9905aa76f3e2a57ae2440c8c693ff60848
Signed-off-by: Robert Varga <>
20 months agoRemove sal-connector-api 79/82379/2
Robert Varga [Mon, 3 Jun 2019 09:18:49 +0000 (11:18 +0200)]
Remove sal-connector-api

This artifact is not used anywhere, remove it.

Change-Id: I43a9df5d77abf28d652ea9289025b49638dad702
Signed-off-by: Robert Varga <>
21 months agoBump cds-access-api ABIVersion 12/82312/3
Robert Varga [Wed, 29 May 2019 15:33:08 +0000 (17:33 +0200)]
Bump cds-access-api ABIVersion

In order to take advantage of improvements to NormalizedNode stream,
we need to declare a new ABI version, which will actually use
the new stream version.

Change-Id: I458a24586888e0b2fcc1c636e16d0b8e9e6e681f
Signed-off-by: Robert Varga <>
21 months agoOptimize NodeIdentifier reading 79/82279/13
Robert Varga [Wed, 29 May 2019 10:10:52 +0000 (12:10 +0200)]
Optimize NodeIdentifier reading

NodeIdentifiers are just an alias for QName, which is heavily
reused. This patch adds a secondary cache for interpreting QNames
as NodeIdentifiers, with instance reuse.

This allows us to reduce the number of NodeIdentifier instances,
while not relying on NodeIdentifier.create() for the common lookup

Change-Id: Ifa0c5d572f7d39da49707529c7ddd9c9f36f2dab
Signed-off-by: Robert Varga <>
21 months agoAdd support for coded QNames/AugmentationIdentifiers 04/82304/7
Robert Varga [Wed, 29 May 2019 09:27:21 +0000 (11:27 +0200)]
Add support for coded QNames/AugmentationIdentifiers

Profiling is showing we are spending about 46% in readQName(),
which further breaks down to 30% in readCodedString() and 61%
in QNameFactory.create().

Out of 1.7M invocations of readQName(), we end up loading only
508 QNames -- which is understandable, as the number of QNames is
bound by the SchemaContext contents. This also applies to

This patch defines a new streaming format version, which uses
the same linear coding approach we take for Strings, taking
hash-based lookups out of the hot path -- bringing reconstruction
time down from 36 to 19-24 seconds.

It has also the effect of reducing the snapshot size, in our testing
bringing a 388.8MiB snapshot down to 210.6MiB.

Change-Id: Ibbc96c286d52b3261b04dd7de02e13bc6b44e803
Signed-off-by: Robert Varga <>
21 months agoSeparate lazy-versioned NormalizedNodeDataInput 78/82278/6
Robert Varga [Tue, 28 May 2019 13:28:55 +0000 (15:28 +0200)]
Separate lazy-versioned NormalizedNodeDataInput

NormalizedNodeInputStreamReader is really only one implementation
of the streaming format, which happens to be bound to LITHIUM_VERSION.

This effectively means NormalizedNodeInputOutput can give out either
a version-bound reader or an unbound reader -- which we facilitate by
creating VersionedNormalizedNodeDataInput which is a simple forwarder
with lazily-initialized delegate.

Since this increases the number of (NormalizedNode)DataInput
implementations, we share as much code as possible through the use
of forwarding implementations.

Change-Id: I28713346730ed5bf7f912bf8afa5d8722b202035
Signed-off-by: Robert Varga <>
21 months agoDo not format QNames to string on input 76/82276/4
Robert Varga [Tue, 28 May 2019 11:14:34 +0000 (13:14 +0200)]
Do not format QNames to string on input

Profiling has revelead the creation of String for QNameFactory
lookup is taking around 19% of overall cost of

Rather than using concatenation, use a dedicated Key object, which
holds the separated-out localname/namespace/revision strings,
thus allocating a small object instead of a full-blown String.

This improves snapshot deserialization perfomance on a 350MiB
snapshot by about 16%, also resulting in lower memory pressure.

Change-Id: If3e9e4b332b0a5acbe33aa10fca189bbc7da8c81
Signed-off-by: Robert Varga <>
21 months agoUpdate QNameFactory caching 75/82275/3
Robert Varga [Tue, 28 May 2019 10:42:02 +0000 (12:42 +0200)]
Update QNameFactory caching

This eliminates the use of soft references in favor of weak ones,
so that we do not rely on GC heuristics for retention.

We also make the maximum cache size runtime-configurable via

Finally we use QName.intern() to defer to the global QName cache,
resulting in better QName instance sharing.

Change-Id: Ice3596d23b96c4ecfc5775cb21968233b042a984
Signed-off-by: Robert Varga <>
21 months agoFix write transaction shard stats 89/82189/3
Tomas Cere [Tue, 21 May 2019 11:42:47 +0000 (13:42 +0200)]
Fix write transaction shard stats

With the changes in ask protocol due to to tell based
integration, shards stats had incorrectly reported
write transaction counts. Since backend doesnt care
about the difference between write-only and read-write
transactions and all are read-write from backend point of view
we can safely remove writeOnly counter.

Change-Id: I18046ad3b229755bbd0daa788a3fe62f59e4140d
Signed-off-by: Tomas Cere <>
21 months agoUse Files.delete() in LocalSnapshotStore 13/82213/2
Tomas Cere [Thu, 23 May 2019 08:04:16 +0000 (10:04 +0200)]
Use Files.delete() in LocalSnapshotStore

If delete fails, it needs to be logged so change this up.

Change-Id: If4051cb3e5b83dee51f919fc1713cc17f032597e
Signed-off-by: Tomas Cere <>
21 months agoDo not enque purges for ask based protocol 37/82137/16
Tomas Cere [Thu, 16 May 2019 12:12:01 +0000 (14:12 +0200)]
Do not enque purges for ask based protocol

With metadata disabled for ask based protocol, there is no need
to purge transaction on the backend and pollute the journal.

Tell based protocol has purges handled by the fronted so it
should be safe for us to remove them on the backend for ask

Change-Id: Ia5c9470782bb4d59b78bbaa605ee5d36c7fb1644
Signed-off-by: Tomas Cere <>
21 months agoRemove PersistAbortTransactionPayload 27/82027/14
Tomas Cere [Tue, 14 May 2019 10:56:04 +0000 (12:56 +0200)]
Remove PersistAbortTransactionPayload

With the metadata tracking disabled for ask based protocol
there is no need to track aborts for read only transactions
on the backend.

Change-Id: I189ae3231bb2f3c0eaa0bbe21a14342446708c5f
Signed-off-by: Tomas Cere <>
21 months agoDo not leak DataTree from backend actor 82/82182/1
Robert Varga [Tue, 21 May 2019 08:47:34 +0000 (10:47 +0200)]
Do not leak DataTree from backend actor

YANG tools is now exposing the read-only part of DataTree as
ReadOnlyDataTree. Use this interface to restrict the set of
methods accessible to frontend.

Change-Id: I9426d4cee1e68e3cc21f7a39322a0df09739e770
Signed-off-by: Robert Varga <>
21 months agoDisable transaction tracking for ask-based protocol 40/82140/8
Robert Varga [Thu, 16 May 2019 13:54:29 +0000 (15:54 +0200)]
Disable transaction tracking for ask-based protocol

When we encounter an ask-based protocol message on a leader,
we persist a payload to make sure no state tracking occurs
for that client.

Change-Id: I3d12a06ce9e5b65ada5b59bde9f9b5486e5e4ef7
Signed-off-by: Robert Varga <>
Signed-off-by: Tomas Cere <>
21 months agoAllow transaction tracking to be disabled 83/81983/13
Robert Varga [Fri, 10 May 2019 10:25:31 +0000 (12:25 +0200)]
Allow transaction tracking to be disabled

Ask-based protocol does not need tracking of transactions and
histories, as it is not retransmitting requests. It also does
not inform backend about purely-local aborted transactions
(read-write and read-only), which leads to transaction tracking
rangesets having holes where those IDs are used.

This adds the prerequisite handling of disabling from the leader
without adding the actual mechanics.

Change-Id: I133e7688b492336937f394f0f6c3f080a05a820f
Signed-off-by: Robert Varga <>
Signed-off-by: Tomas Cere <>
21 months agoRemove QueuedNotificationManagerMXBean 71/82171/1
Robert Varga [Mon, 20 May 2019 10:43:49 +0000 (12:43 +0200)]
Remove QueuedNotificationManagerMXBean

This interface is available from yangtools directly, remove
it from controller.

Change-Id: Ic894c0e9fc5d114ebff07bb11f67f1ec67611031
Signed-off-by: Robert Varga <>
21 months agoDerive config-artifacts from odlparent-lite 71/82071/1
Robert Varga [Wed, 15 May 2019 07:58:52 +0000 (09:58 +0200)]
Derive config-artifacts from odlparent-lite

This is to fix failures seen on promoting to Central.

Change-Id: I6204d68c5144437e5af50bb5f2cb29431841d394
Signed-off-by: Robert Varga <>
21 months agoBump mdsal to 4.0.1 23/81823/8
Robert Varga [Tue, 30 Apr 2019 10:55:09 +0000 (12:55 +0200)]
Bump mdsal to 4.0.1

Upgrade mdsal to 4.0.1 and remove most references to Javassist
and its associated streamer implementation.

Change-Id: Ia9dc2c886217f1c353db6523b654e993ed9a397a
Signed-off-by: Robert Varga <>
21 months agoRemove sal-dom-broker dependency on in-memory datastore 34/81934/5
Robert Varga [Tue, 7 May 2019 10:02:55 +0000 (12:02 +0200)]
Remove sal-dom-broker dependency on in-memory datastore

This dependency should be purely scope=test, so that we do not
depend on a particular datastore.

Change-Id: I24962d09b80d9f1a68ba777980d129387bc41380
Signed-off-by: Robert Varga <>
21 months agoSplit odl-mdsal-broker-local and odl-mdsal-broker features 33/81933/4
Robert Varga [Tue, 7 May 2019 09:50:00 +0000 (11:50 +0200)]
Split odl-mdsal-broker-local and odl-mdsal-broker features

These two features are supposed to be exclusive -- either we want
local-only services (-local) or we want clustered services.

Change-Id: I11c261b6c97471356645826c00ea052ffbe7b82b
Signed-off-by: Robert Varga <>
21 months agoCentralize read{Augmentation,Node}Identifier() methods 08/82008/3
Robert Varga [Mon, 13 May 2019 05:53:27 +0000 (07:53 +0200)]
Centralize read{Augmentation,Node}Identifier() methods

We are using the same construct in multiple places, make sure we
centralize it and mark a FIXME for some efficiency increase.

Change-Id: Ie65089a1a7bb6a2752887498d354a61cb32f0e40
Signed-off-by: Robert Varga <>
21 months agoMigrate deprecated yangtools method 10/82010/2
Robert Varga [Mon, 13 May 2019 07:26:34 +0000 (09:26 +0200)]
Migrate deprecated yangtools method

DataTreeCandidateNodes.fromNormalizedNode() has been deprecated
in favor of written() -- migrate this user.

Change-Id: I355058568c2c1512fbca083142b17a3a515e9329
Signed-off-by: Robert Varga <>
21 months agoPoison entries outside of main lock 49/81949/4
Robert Varga [Tue, 7 May 2019 17:42:26 +0000 (19:42 +0200)]
Poison entries outside of main lock

Poisoning entries may involve reaction from their callbacks, which
can attempt to circle back through connections.

Make sure we poison them outside of lock context, so that any
callbacks end up seeing a poisoned connection, but without the lock
being held -- hence the locks can be acquired in-order.

Change-Id: I26551d052307812e76f3e45024a77dbb83312b17
Signed-off-by: Robert Varga <>
21 months agoAllow SnapshotBackedReadTransaction customization 55/81955/1
Robert Varga [Wed, 8 May 2019 10:28:24 +0000 (12:28 +0200)]
Allow SnapshotBackedReadTransaction customization

In some specific cases we need to customize abort-like handling,
which is already implemented for write-like transactions, but is
not present for read-only transaction.

This patch adds the capability to attach a close() handler and
makes sure AbstractSnapshotBackedTransactionChain takes advantage
of it.

Change-Id: Ic7027956556b5dd25120ee81613a6151e5dbc501
Signed-off-by: Robert Varga <>
21 months agoEnqueue purge snapshot directly 45/81945/1
Robert Varga [Tue, 7 May 2019 14:23:19 +0000 (16:23 +0200)]
Enqueue purge snapshot directly

When we are sending abort+purge, make sure we propagate it out
as soon as possible.

Change-Id: I2b63aa20867a8b99b779115f0bb448c0e8ac9a22
Signed-off-by: Robert Varga <>
21 months agoMake sure we purge transaction on abort 39/81939/2
Robert Varga [Tue, 7 May 2019 11:22:10 +0000 (13:22 +0200)]
Make sure we purge transaction on abort

In case of read-only transactions it is not sufficient to just
persist the abort payload, we need to also purge the transaction.

Change-Id: I8761040d074743b5a1c5eebbecc6449c30abe47e
Signed-off-by: Robert Varga <>
21 months agoRefactor odl-mdsal-remoterpc-connector 17/81917/4
Robert Varga [Mon, 6 May 2019 17:48:51 +0000 (19:48 +0200)]
Refactor odl-mdsal-remoterpc-connector

RPC connector has a logical dependency on components present
in the distributed datastore. This refactors it to take that
into account.

Change-Id: Icd0145b35310133822ee02cdf4a3d4cffbc6e257
Signed-off-by: Robert Varga <>
21 months agoRefactor odl-mdsal-distributed-datastore 16/81916/3
Robert Varga [Mon, 6 May 2019 18:04:10 +0000 (20:04 +0200)]
Refactor odl-mdsal-distributed-datastore

odl-mdsal-distributed-datastore forms the baseline for supporting
akka-based cluster. This refactors it to pull its dependencies in
so it can be properly tested.

This also moves CommitStatsMXBean, so we do not depend on controller's
broker at all.

Finally we co-locate all configuration files into this feature, so
it can boot up.

Change-Id: I9319de78d61d63e3c49825e940af5c8c70657185
Signed-off-by: Robert Varga <>
21 months agoSplit out odl-controller-blueprint 14/81914/2
Robert Varga [Mon, 6 May 2019 17:39:00 +0000 (19:39 +0200)]
Split out odl-controller-blueprint

Blueprint only needs controller APIs, it does not need implementation
and it should be packaged in odl-config-netty.

This patch splits it out in a separately-tested feature

Change-Id: I4b12a3880b32fc21515dbb1e69f7e21b7c4d5c57
Signed-off-by: Robert Varga <>
21 months agoRefactor odl-mdsal-clustering-commons 13/81913/2
Robert Varga [Mon, 6 May 2019 16:41:46 +0000 (18:41 +0200)]
Refactor odl-mdsal-clustering-commons

odl-mdsal-clustering-commons does not have a dependency on
odl-mdsal-broker-local, hence we can factor it out to pull in
its upstream dependencies, thus fixing its tests.

Change-Id: Ie1df487329df8ea1295e0e2dc3b0e0f07e2e4037
Signed-off-by: Robert Varga <>
21 months agoSplit out odl-controller-binding-api 12/81912/2
Robert Varga [Mon, 6 May 2019 16:32:43 +0000 (18:32 +0200)]
Split out odl-controller-binding-api

This refactors packaging so that we package Binding-level APIs
in a corresponding feature -- mirroring MD-SAL's setup.

Change-Id: I92adc8f14462633a253136d6c37e708a30cc4a37
Signed-off-by: Robert Varga <>
21 months agoReset snapshot progress after timeout has been hit 16/81816/7
Tomas Cere [Tue, 30 Apr 2019 08:49:02 +0000 (10:49 +0200)]
Reset snapshot progress after timeout has been hit

Add a stopwatch to LeaderInstallSnapshotState that allows
us to keep track whether a follower took too much time to respond
to InstallSnapshot for whatever reason allowing us to retry
chunks that time out.

Change-Id: Id443a4cc7a069ad4d9982d537d9a8d82b845ac35
Signed-off-by: Tomas Cere <>
21 months agoSplit out odl-controller-dom-api 11/81911/1
Robert Varga [Mon, 6 May 2019 16:12:41 +0000 (18:12 +0200)]
Split out odl-controller-dom-api

This refactors packaging so that we package DOM-level APIs
in a corresponding feature -- mirroring MD-SAL's setup. We also fix
dependencies on MD-SAL features to use ranges instead of fixed

Change-Id: Ib5e0649552b34662a91b4fb32da9af69f589bb6b
Signed-off-by: Robert Varga <>
21 months agoFix memory leak in BA mount service 91/81891/5
Maros Marsalek [Fri, 3 May 2019 12:04:56 +0000 (14:04 +0200)]
Fix memory leak in BA mount service

The cache used in BA mount service uses weakKeys() so the keys (DOM mountpoint)
could be GCed when no ono, except cache, referenced them).
However the values in the cache (BA mountpoint) also kept a reference to
the key, so the key was never weakly reachable and thus never GCed
resulting in a memory leak.

Which is especially visible in case of frequent reconnects of netconf
mountpoints where after every reconnect a new DOM mountpoint is created,
requiring a new BA mountpoint creating new entry in the cache of BA mount service...

Added also a simple test verifying proper cache cleanup

Change-Id: I9c109dd6d499d6185842d6a1ad5a59d75565af5c
Signed-off-by: Maros Marsalek <>
22 months agoImprove segmented journal JMX beans 90/81890/1
Robert Varga [Fri, 3 May 2019 11:01:14 +0000 (13:01 +0200)]
Improve segmented journal JMX beans

This patch updates metric naming to properly tie them to individual
journals -- so that the statistics are actually useful.

Change-Id: I6f4f398894860566e234562d731462535b62801e
Signed-off-by: Robert Varga <>
22 months agoChange segmented journal naming 89/81889/1
Robert Varga [Fri, 3 May 2019 10:48:34 +0000 (12:48 +0200)]
Change segmented journal naming

Using Base64 encoding is not quite nice for debugging purposes,
use URLEncoder, so that we are consistent with snapshot store

Change-Id: I55f7bef79b7d5c14ab10180444770fedcc23eab9
Signed-off-by: Robert Varga <>
22 months agoHarden ShardCommitCoordinator to survive EOS write retries 90/81790/1
Tomas Cere [Fri, 26 Apr 2019 10:37:38 +0000 (12:37 +0200)]
Harden ShardCommitCoordinator to survive EOS write retries

EntityOwnershipShard can break the usual shard rule of only one 3PC
running at one time, causing the retrying write attempting to modify
an already sealed modification.

Change-Id: Iaa56b68a30f9191f7aa14428f39d4d7524aeb09d
Signed-off-by: Tomas Cere <>
22 months agoDo not allow persistence callbacks to throw Exception 40/81640/6
Robert Varga [Sat, 13 Apr 2019 12:55:13 +0000 (14:55 +0200)]
Do not allow persistence callbacks to throw Exception

Since we are indirecting through an executor, we are forced
to wrap any exception -- just do not bother, as the callbacks
are executed in the context of an actor anyway (dealing with
RuntimeExceptions) and users are not throwing checked exceptions.

Change-Id: I6cea19ab7192fa42ad3c346d554411cd0d558a64
Signed-off-by: Robert Varga <>
22 months agoBump mdsal to 4.0.0 30/81130/38
Robert Varga [Fri, 22 Mar 2019 18:33:21 +0000 (19:33 +0100)]
Bump mdsal to 4.0.0

This bumps mdsal to 4.0.0, adjusting for changes it brings, notably:
- FluentFuture removed from DOM RPC invocation services
- ping-pong functionality directly part of DOMDataBroker API and not
  via a dedicated DOMDataBroker instance
- DataContainer.getImplementedInterface() renamed to
- DOMBrokerReadWriteTransaction.close() method being removed
- BindingCodecTreeNode being refactored

Change-Id: I5af13aa82036373826291a54661f7fd20ff387c3
Signed-off-by: Robert Varga <>
22 months agoBump yangtools to 3.0.0 29/81129/31
Robert Varga [Fri, 22 Mar 2019 18:32:29 +0000 (19:32 +0100)]
Bump yangtools to 3.0.0

This is bumps yangtools to 3.0.0, adjusting for changed APIs. The
changes specifically include:

- use ListenerRegistry.getRegistrations() instead of it implementing
- adjust for NormalizedNodeStreamWriter interface change
- remove use of *AttrBuilder, as NormalizedNodes no longer hold
- DataTreeCandidateNodes.empty() instead of home-grown implementation
- adjust for DataTreeCandidateNode.getModifiedChild() returning Optional
- use QueuedNotificationManager.BatchedInvoker

Change-Id: Ia30d73ccf6380e3759414b3026b822df0950a481
Signed-off-by: Robert Varga <>
22 months agoBump odlparent to 5.0.0 28/81128/25
Robert Varga [Fri, 22 Mar 2019 18:31:50 +0000 (19:31 +0100)]
Bump odlparent to 5.0.0

This bumps odlparent references to 5.0.0, removing use of the JSR305
remnants. Also a useless javadoc plugin override is removed.

Change-Id: Iae5355e49115afdd73b60bd26ac85f7f542bbff8
Signed-off-by: Robert Varga <>
22 months agoDo not break actor containment 82/81482/7
Tomas Cere [Tue, 9 Apr 2019 08:46:18 +0000 (10:46 +0200)]
Do not break actor containment

During async persistence the actor containment is broken which
results in applyState being called directly from the persistence actor.
This means that the ClientRequestTracker might be missing an entry
for the transaction and the transaction being applied as a foreign candidate
in ShardDataTree.
If this happens there will be a transaction stuck in COMMIT_PENDING state,
blocking any further progress in the shard until its restarted.

Change-Id: I944b233d13103df08b68baeaf3907c064d1d526e
Signed-off-by: Tomas Cere <>
Signed-off-by: Robert Varga <>
23 months agoIntegrate SegmentedJournalActor with metrics/JMX 53/81353/1
Robert Varga [Tue, 2 Apr 2019 21:40:20 +0000 (23:40 +0200)]
Integrate SegmentedJournalActor with metrics/JMX

This adds the wiring needed to expose segmented journal metrics
through JMX.

Change-Id: I42f0b29985410a34ec91639ce8be2903bbe7dd5b
Signed-off-by: Robert Varga <>
23 months agoFixup HTML5 javadoc compatibility 39/81339/1
Robert Varga [Mon, 1 Apr 2019 18:05:15 +0000 (20:05 +0200)]
Fixup HTML5 javadoc compatibility

This fixes use of 'summary' attribute.

Change-Id: I872afdef8b678a68f4e1781330c3f11f0e010925
Signed-off-by: Robert Varga <>
23 months agoMigrate nullness annotations 18/81318/2
Robert Varga [Fri, 29 Mar 2019 01:53:09 +0000 (02:53 +0100)]
Migrate nullness annotations

This mass-migrates sal-distributed-datastore to use JDT nullness
annotations instead of JSR305.

Change-Id: I2cd739e3db8c074a55e335e17297a23065775d6a
Signed-off-by: Robert Varga <>
23 months agoMigrate to AbstractActor 96/80396/8
Robert Varga [Tue, 19 Feb 2019 14:58:28 +0000 (15:58 +0100)]
Migrate to AbstractActor

UntypedActor has been deprecated, this patch migrates to using
AbstractActor so we eliminate a slew of deprecatation warnings.

Change-Id: I4e3274ce92d5f68df1937f91f823eed905da4d6c
Signed-off-by: Robert Varga <>
Signed-off-by: Ajay Lele <>
23 months agoFix checkstyle in toaster-it 69/81269/1
Robert Varga [Thu, 28 Mar 2019 18:54:43 +0000 (19:54 +0100)]
Fix checkstyle in toaster-it

Simple whitespace cleanup to keep checkstyle happy.

Change-Id: Id51f7b87b632cce92c29cbae8794662fe6feb250
Signed-off-by: Robert Varga <>
23 months agoFix transaction mocking 62/81262/1
Robert Varga [Thu, 28 Mar 2019 16:55:17 +0000 (17:55 +0100)]
Fix transaction mocking

We should not confuse FluentFutures and CheckedFutures, as the latter
is deprecated and with Guava 27 they are not compatible anyway.

Change-Id: I7489b4470df6fa435aab190b6e38eaf154522fe7
Signed-off-by: Robert Varga <>
23 months agoMigrate to use openCursor() 61/81261/1
Robert Varga [Thu, 28 Mar 2019 16:26:24 +0000 (17:26 +0100)]
Migrate to use openCursor()

createCursor() has been deprecated long time ago, use its replacement

Change-Id: Id8862470ca4ad8c9fdec5c05cbac044c6715e153
Signed-off-by: Robert Varga <>
23 months agoFix logging messages 60/81260/5
Robert Varga [Thu, 28 Mar 2019 13:32:44 +0000 (14:32 +0100)]
Fix logging messages

Upgraded spotbugs is finding more violations, fix them up.

Change-Id: I8956c5b7a6eef38286f4c2b6fb07ca6a775e6965
Signed-off-by: Robert Varga <>
23 months agoUse ReusableNormalizedNodePruner in PruningDataTreeModification 58/81258/5
Robert Varga [Thu, 28 Mar 2019 12:01:48 +0000 (13:01 +0100)]
Use ReusableNormalizedNodePruner in PruningDataTreeModification

Rather than instantiating a pruner for each invocation, keep an
instance around for reuse. In order to facilitate free-standing
modifications, we also keep a reference to the DataSchemaContextTree
bound our current SchemaContext.

Change-Id: I5d310b12feb1ff46dc2af0e67294a925d48a4957
Signed-off-by: Robert Varga <>
23 months agoAdd ReusableNormalizedNodePruner 57/81257/4
Robert Varga [Thu, 28 Mar 2019 11:46:34 +0000 (12:46 +0100)]
Add ReusableNormalizedNodePruner

This version of NormalizedNodePruner is reusable across invocations,
and improving CPU and memory efficiency in bulk operations due to
not needing to allocate stack nor lookup DataSchemaContextTree.

Change-Id: I1798f5f6255cd23702bf3792c1f4f5149f92d208
Signed-off-by: Robert Varga <>
23 months agoMove checkNotSealed() to addBuilder() 51/81251/2
Robert Varga [Thu, 28 Mar 2019 10:22:55 +0000 (11:22 +0100)]
Move checkNotSealed() to addBuilder()

addBuilder() is called after previous check for seal from all callers,
who do not have any side-effects aside allocating arguments to

Move the call to checkNotSealed() to addBuilder(), making the code
more straightforward and possibly preventing bugs from creeping in.

Change-Id: I127385378957deb7f38f1203870a0e52cd630644
Signed-off-by: Robert Varga <>
23 months agoRemove NormalizedNodePruner.findSchemaNodeForNodePath() 50/81250/2
Robert Varga [Thu, 28 Mar 2019 10:16:27 +0000 (11:16 +0100)]
Remove NormalizedNodePruner.findSchemaNodeForNodePath()

DataSchemaContextTree.findChild() provides equivalent functionality,
so use that instead of brewing our own.

Change-Id: Ia55ac99ce8fa37daf06ce0e7e1615f1bd0a07123
Signed-off-by: Robert Varga <>
23 months agoDo not use Optional in NormalizedNodeWrapper 49/81249/2
Robert Varga [Thu, 28 Mar 2019 09:58:49 +0000 (10:58 +0100)]
Do not use Optional in NormalizedNodeWrapper

Use nullable schema instead, as the caller is package-private and
ends up unpacking the return anyway. This simplifies
NormalizedNodePruner a bit.

Change-Id: I395dd8266bd7198f66890c4ae3633ecda9281726
Signed-off-by: Robert Varga <>
23 months agoHide NormalizedNodeBuilderWrapper 48/81248/2
Robert Varga [Thu, 28 Mar 2019 07:48:02 +0000 (08:48 +0100)]
Hide NormalizedNodeBuilderWrapper

This class is not instantiable outside of the package is it does
not leak from NormalizedNodePruner -- therefore it cannot be used
by anyone else. Make sure it's package-private and clean it up
a bit.

Change-Id: I17911508273d68c2795d38e2156c9407522f55fe
Signed-off-by: Robert Varga <>
23 months agoRemove NormalizedNodePruner.SimpleStack 34/81234/2
Robert Varga [Wed, 27 Mar 2019 22:51:57 +0000 (23:51 +0100)]
Remove NormalizedNodePruner.SimpleStack

This custom implementation is highly inefficient, because:
1) peek/pop operations perform get(int) and remove(int) operations
   on ArrayList, which are highly inefficient O(N)
2) it allocates an object for each entry

Replace this class with a stock Deque, which takes care of the first
problem, as push/pop/peek are amortized O(1).

For implementation use an ArrayDeque, which takes care of the second
problem, as stack entries are reused.

Change-Id: Ie59e2fb47f841e4e5fc08c3ab7e164ad0cb368ad
Signed-off-by: Robert Varga <>
23 months agoClean up NormalizedNodePruner 33/81233/1
Robert Varga [Wed, 27 Mar 2019 22:45:15 +0000 (23:45 +0100)]
Clean up NormalizedNodePruner

- shorten YangInstanceIdentifier.PathArgument references
- do not declare runtime exceptions as thrown
- use a space after comma

Change-Id: Id0f09d2cd72bc619331c2d16e8f184db784c9976
Signed-off-by: Robert Varga <>
23 months agoDeleting a non-existent node fails 84/28684/8
Valentin Mayamsin [Fri, 16 Oct 2015 22:08:34 +0000 (15:08 -0700)]
Deleting a non-existent node fails

This adds a unit test for deletion of a non-existent node.

Change-Id: I32d537c7e58a6161147e7e51b80e71cb8aaa0d79
Signed-off-by: Valentin Mayamsin <>
Signed-off-by: Robert Varga <>
23 months agoFix test mocking 27/81127/1
Robert Varga [Fri, 22 Mar 2019 19:40:03 +0000 (20:40 +0100)]
Fix test mocking

MD-SAL store APIs require FluentFuture, not CheckedFuture, fix that

Change-Id: I2222c6a56f8b7a1bcc6569ba84e01ce3b4904a51
Signed-off-by: Robert Varga <>
23 months agoUpdate javadoc links 26/81126/1
Robert Varga [Fri, 22 Mar 2019 19:32:32 +0000 (20:32 +0100)]
Update javadoc links

Our javadoc links are pointing to deprecated Guava methods, update
them to point to their equivalents, as they are going away in our
next upgrade.

Change-Id: Ia82471897ba860f38be45219676b44f3d55d977c
Signed-off-by: Robert Varga <>
23 months agoFix logging arguments 25/81125/1
Robert Varga [Fri, 22 Mar 2019 19:32:08 +0000 (20:32 +0100)]
Fix logging arguments

Upgraded spotbugs is finding these violations, fix them up.

Change-Id: I6f7490cd1d48c6e22d048842265bf71d0ff24afc
Signed-off-by: Robert Varga <>
23 months agoAdd missing license headers 64/68764/3
Stephen Kitt [Mon, 26 Feb 2018 12:02:44 +0000 (13:02 +0100)]
Add missing license headers

Change-Id: Ie8405e6f1de698c958d347cfb53380a111d6a9ef
Signed-off-by: Stephen Kitt <>
23 months agoUse a simple list for string code table 89/80889/1
Robert Varga [Fri, 15 Mar 2019 11:33:05 +0000 (12:33 +0100)]
Use a simple list for string code table

Since the codes are assigned in a linear fashion, we can use
a simple ArrayList to track assignes codes, preventing the need
for boxing integer during population and lookups. Lookups also
end up requiring an array offset lookup rather than needing
to go through multiple indirections.

While this can in theory lead to a large array being allocated,
the set of strings we keep in the table is limited by the
SchemaContext and thus will be typically capped ad couple of

The cost of growing the lookup should not be significantly
higher, even if ArrayList grows slower:
- 10, 16, 25, 38,  58,  88
versus HashMap (with .75 load factor)
- 12, 24, 48, 96, 192, 384

each resize operation is significantly faster.

Change-Id: I00c7a2e0985dbf901b6857a16424bf157e5af39d
Signed-off-by: Robert Varga <>
23 months agoImprove NormalizedNodeInputStreamReader defensiveness 88/80888/2
Robert Varga [Fri, 15 Mar 2019 11:20:24 +0000 (12:20 +0100)]
Improve NormalizedNodeInputStreamReader defensiveness

If face of corrupted streams, for example due to CONTROLLER-1752,
we can end up interpreting unknown bytes as null strings, potentially
corrupting data.

This patch uses explicit type dispatch and detects wrong string
value types, as well as rejecting unknown string value codes.

Change-Id: Ic2e7887bc1d32a1dc7a364d1bff195752edd762d
Signed-off-by: Robert Varga <>
23 months agoAdd integration test for segmented journal 23/80823/3
Tomas Cere [Wed, 13 Mar 2019 11:50:55 +0000 (12:50 +0100)]
Add integration test for segmented journal

Extracts test cases for DistributedDataStore that do not need latching
in InMemoryJournal/Snapshot so we can have these run with segmented journal.

Change-Id: I0ea4e58bff0f0e8e1a40bd4f26b827623139d6c9
Signed-off-by: Tomas Cere <>
23 months agoFix segmented journal replay 22/80822/1
Tomas Cere [Tue, 12 Mar 2019 09:54:47 +0000 (10:54 +0100)]
Fix segmented journal replay

lastDelete was getting set with the incorrect value from the delete journal.
The actual written value from the delete journal needs to be used
instead of the index.
dataJournal compact also needs to be compacted upto lastDelete + 1 since
deleteUpTo is inclusive and compact keeps this value around.

Change-Id: I4a678be67fd1ad09c57273ef0fc3b7398a7d714f
Signed-off-by: Tomas Cere <>
23 months agoIntroduce candidate election timeout divisor 63/80663/2
Robert Varga [Fri, 1 Mar 2019 20:45:48 +0000 (21:45 +0100)]
Introduce candidate election timeout divisor

When a candidate is waiting for an election to occur, it is sometimes
preferable to shorten the timeout interval down, so that we do not
wait the full election timeout.

This patch introduces the knob which serves to divide the normal
timeout -- i.e. it can be halved.

Change-Id: Idbe81c39bf7aecf56b9f13a242bea89a1a4ac4f1
Signed-off-by: Robert Varga <>
23 months agoPrevent a CME during actor instantiation 65/80665/3
Robert Varga [Mon, 4 Mar 2019 12:02:36 +0000 (13:02 +0100)]
Prevent a CME during actor instantiation

This issue was missed in review: createHandler() should not be
touching handlers, as it is already called from computeIfAbsent()
and hence would result in a CME with JDK11:

Change-Id: I8d4940b8bec2cf727d97d68796f1c9cdaaaf89d3
Signed-off-by: Robert Varga <>
2 years agoAdd SegmentedFileJournal 68/80368/10
Robert Varga [Mon, 4 Feb 2019 09:09:57 +0000 (10:09 +0100)]
Add SegmentedFileJournal

This adds an alternative Akka persistence journal implementation
based on Atomix's SegmentedJournal.

SegmentedJournal is a linear append-only log of entries, each of
which has a incrementing 63bit index. This logical structure is
stored in rolling segments -- i.e. entries are appended to current
segment until it is full, at which point a new segment is allocated
and writeout continues there. Old segments can be freed as long as
they are not needed.

This layout makes it good match to how Akka's persistence works,
with the one exception that the SegmentedJournal cannot explicitly
delete entries from the head of the journal.

SegmentedFileJournal allocates one SegmentedJournal for each
persistenceId it encounters and uses a dedicated actor for it. This
provides a major simplification in the implementation as well as
allows for concurrent persistence of multiple PersistentActors.

Change-Id: I78140b154bab44a3e17d5ffb76b040c62add3204
Signed-off-by: Robert Varga <>
2 years agoRename ActorContext to ActorUtils 94/80394/4
Robert Varga [Tue, 19 Feb 2019 15:43:22 +0000 (16:43 +0100)]
Rename ActorContext to ActorUtils

ActorContext is overloaded name even within Akka, without us adding
to it. Furthermore Akka's AbstractActor defines ActorContext as its
nested class, which thoroughly breaks resolution priorities.

Rename ActorContext to ActorUtils, reducing confusion around class

Change-Id: I140239a8f74ee7deecf9ee848df0cfbbb72f3c4d
Signed-off-by: Robert Varga <>
2 years agoAdd controller aggregate javadocs 68/80568/2
Robert Varga [Sun, 24 Feb 2019 23:20:54 +0000 (00:20 +0100)]
Add controller aggregate javadocs

This adds aggregated javadocs for the controller projects.

Change-Id: If5efaa7d5dda94a7a6416230e6c2e61dd963bd90
Signed-off-by: Robert Varga <>
2 years agoDo not assert seal transition on forward path 11/80411/2
Robert Varga [Wed, 20 Feb 2019 12:56:30 +0000 (13:56 +0100)]
Do not assert seal transition on forward path

Unlike the replay path, where we expect no successor, the forward
path may very well encounter a successor being present.

Move the assertion to replay path only, as that is where it catches
issues. The forward path automatically propagates the seal via its
use of sendRequest(), which is forwarded as needed and therefore
we only note the race via LOG.debug() and not assert it.

Change-Id: I6f95acc49800d456c049d19982c801015fd69420
Signed-off-by: Robert Varga <>
2 years agoImprove aborted transaction logging 15/80415/2
Robert Varga [Wed, 20 Feb 2019 13:47:12 +0000 (14:47 +0100)]
Improve aborted transaction logging

The fact that a snapshot is recorded does not mean it has not been
closed -- in fact write-aspect transactions do not actively remove

When we prune recorded snapshots on transaction chain close, we need
to pay attention to the return value of abort(), as that is
indicating whether we should in fact warn about an unclosed

This means that the actual set of transactions we warn about cannot
be determined until we have traversed all recorded snapshots, hence
modify the logic to record the identifiers that were in fact aborted
and report all of them in a single message along with a stack trace
so the offender may be identifier.

Change-Id: I8b176f1990c9aa9c056cd03203ea50fb9b9549a0
Signed-off-by: Robert Varga <>
2 years agoCleanup java.util.Optional references 10/80410/2
Robert Varga [Wed, 20 Feb 2019 10:46:03 +0000 (11:46 +0100)]
Cleanup java.util.Optional references

As a byproduct of gradual migration, we have a few places where
we use FQCN to reference java.util.Optional eventhough it is among
imports. Fix that.

Change-Id: Ibede76ffc614c02dbc5e29e78e6c421abcd81ec9
Signed-off-by: Robert Varga <>
2 years agoMigrate to ListenerRegistry.create() 93/80393/2
Robert Varga [Tue, 19 Feb 2019 14:24:29 +0000 (15:24 +0100)]
Migrate to ListenerRegistry.create()

As per deprecation, the default constructor will be going away.

Change-Id: Id3216acc23fd56554b30b4225275f5bd2cbf4893
Signed-off-by: Robert Varga <>
2 years agoFix free-standing transaction lookup with module-based shards 92/80392/3
Robert Varga [Tue, 19 Feb 2019 12:50:04 +0000 (13:50 +0100)]
Fix free-standing transaction lookup with module-based shards

When we have a combination of tell-based protocol with module-based
shards, the frontend will use the cookie within LocalHistoryIdentifier,
making it potentially non-zero.

The backend tracks free-standing transactions under a local history,
which has cookie set to zero, hence it will not match when we attempt
to look it up for the purposes commit/abort/purge -- leading to
mismatched leader state for these transactions.

Change-Id: Ic4a493d0728b8de288a779d8c97a157584dc76bb
Signed-off-by: Robert Varga <>
2 years agoRemove use of CheckedFuture 71/80371/1
Robert Varga [Mon, 18 Feb 2019 17:59:03 +0000 (18:59 +0100)]
Remove use of CheckedFuture

YangTextSchemaSource uses ListenableFuture only, hence we can
remove use of deprecated Futures.immediate(Failed)CheckedFuture().

Change-Id: I40e571441894c49bbea7a8c11a3854505147af3d
Signed-off-by: Robert Varga <>
2 years agoReduce JSR305 proliferation 11/80211/6
Robert Varga [Thu, 7 Feb 2019 16:53:09 +0000 (17:53 +0100)]
Reduce JSR305 proliferation

retention=runtime annotations are mixing really badly with Java EE
injection and Java 11.

Make sure we do not use javax.annotation package in APIs and reduce
overall proliferation inside implementations.

Change-Id: I569815f0336efdc0de662c3b80f0fa6e5dd47d8a
Signed-off-by: Robert Varga <>
2 years agoAccount for reported UNMODIFIED nodes 67/80267/4
Robert Varga [Mon, 11 Feb 2019 16:00:14 +0000 (17:00 +0100)]
Account for reported UNMODIFIED nodes

DataObjectModification has no way of expressing UNMODIFIED nodes,
but they can (and are) reported from DataTreeCandidateNode, hence
we end up reporting IllegalStateException when the user tries to
look at those nodes.

Make sure we take the node under consideration and do not report
unmodified children, hence preventing the ISE from happening.

Change-Id: I34d36ae083c5ce3ad793eb584236f175f7a3a906
Signed-off-by: Robert Varga <>
2 years agoBump versions by x.y.(z+1) 84/80084/1
jenkins-releng [Fri, 1 Feb 2019 13:06:29 +0000 (13:06 +0000)]
Bump versions by x.y.(z+1)

Change-Id: I6a04c82981ff3244cf20416216d230d77f58f5dc
Signed-off-by: jenkins-releng <>
2 years agoMove BP xml files to standard OSGI-INF/blueprint 49/75549/7
Tom Pantelis [Tue, 22 Jan 2019 15:46:56 +0000 (10:46 -0500)]
Move BP xml files to standard OSGI-INF/blueprint

We originally used org/opendaylight/blueprint in case we needed
to customize how BP bundles are processed but this wasn't needed.
So let's move to the standard BP dir. We'll keep the
BlueprintBundleTracker as is for backwards compatibility - the only
change was to modify it to look in both dirs for BP xml files
when restarting the container for a bundle.

Change-Id: I777e831ef4d8293b49d7aa6c8cdecec96b97937f
Signed-off-by: Tom Pantelis <>

©2013 OpenDaylight, A Linux Foundation Collaborative Project. All Rights Reserved.
OpenDaylight is a registered trademark of The OpenDaylight Project, Inc.
Linux Foundation and OpenDaylight are registered trademarks of the Linux Foundation.
Linux is a registered trademark of Linus Torvalds.