openflowplugin.git
9 years agoBUG-1075: ingress back pressure 58/7458/9
Michal Rehak [Tue, 27 May 2014 11:43:42 +0000 (13:43 +0200)]
BUG-1075: ingress back pressure

- added fair competition queue keeper implementation
- added queueProcessor (harvesting and processing messages)
- per-connection-queues register for queueProcessor and are blocked if polling queue is full
- removed netty thread blocking by publish
- added dropping of messages
- added meaningful names to threads
- added harvester, queues zipper, messageSource registration stuff
- changed ticketPool to blocking if poolQueue full
- doubleCheckedLocking for harverster starving
- handshake pool switched to logging implementation
- onHello method is no more synchronized
- for messages {hello, features} timeout = 8s
- optimized synchronized singleton methods
- used ThreadFactoryBuilder
- processed comments

Change-Id: Ic0b4ebc2df4e19793fa1d1d125a0d31dd2f9c6c8
Signed-off-by: Michal Rehak <mirehak@cisco.com>
9 years agoBug 1185 :of-flow: push-mpls-action different in config and operational data store 02/8002/1
Anilkumar Vishnoi [Fri, 13 Jun 2014 20:45:13 +0000 (02:15 +0530)]
Bug 1185 :of-flow: push-mpls-action different in config and operational data store

Change-Id: I2fac5b3eaeced2a1f93f963fc7a180dc164fc589
Signed-off-by: Anilkumar Vishnoi <avishnoi@in.ibm.com>
9 years agoKaraf integration with plugin 20/7620/9
Prasanna Huddar [Tue, 3 Jun 2014 06:13:25 +0000 (11:43 +0530)]
Karaf integration with plugin

Signed-off-by: Prasanna Huddar <prasanna.huddar@ericsson.com>
Change-Id: I1ffa7ff3ae2d88b63c69a3c7bdc6a8c28ce2db4d

9 years agoBug 1181 : of-flow: Flow match criteria (ipv6-source,ipv6-destination) is different... 57/7957/2
Anilkumar Vishnoi [Thu, 12 Jun 2014 19:06:35 +0000 (00:36 +0530)]
Bug 1181 : of-flow: Flow match criteria (ipv6-source,ipv6-destination) is different in config and operational database

- added unit test

Change-Id: I186d9bdd67a62fa8f98606a8f3ad76d99ef56fec
Signed-off-by: Anilkumar Vishnoi <avishnoi@in.ibm.com>
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoMerge "Use openflowjava-supplied MAC address utilities"
michal rehak [Thu, 12 Jun 2014 06:05:49 +0000 (06:05 +0000)]
Merge "Use openflowjava-supplied MAC address utilities"

10 years agoMerge "Bug-1152:Converting PopVlan to StripVlan for OF1.0"
michal rehak [Thu, 12 Jun 2014 06:04:51 +0000 (06:04 +0000)]
Merge "Bug-1152:Converting PopVlan to StripVlan for OF1.0"

10 years agoBug-1152:Converting PopVlan to StripVlan for OF1.0 87/7887/2
Kamal Rameshan [Wed, 11 Jun 2014 02:09:23 +0000 (19:09 -0700)]
Bug-1152:Converting PopVlan to StripVlan for OF1.0

Change-Id: If2d3e8f1ffdc6bbd089d5d073fab55d88fac9900
Signed-off-by: Kamal Rameshan <kramesha@cisco.com>
Bug-1152:Converting StripVlan to PopVlan for OFP-SAL traffic

Change-Id: If2d3e8f1ffdc6bbd089d5d073fab55d88fac9900
Signed-off-by: Kamal Rameshan <kramesha@cisco.com>
10 years agoMerge "Bug 1072 -Add OF13 pre-req for PCP match in TestCommandProvider"
michal rehak [Wed, 11 Jun 2014 16:09:40 +0000 (16:09 +0000)]
Merge "Bug 1072 -Add OF13 pre-req for PCP match in TestCommandProvider"

10 years agoBug 1072 -Add OF13 pre-req for PCP match in TestCommandProvider 73/7373/2
Brent Salisbury [Sun, 25 May 2014 05:02:18 +0000 (01:02 -0400)]
Bug 1072 -Add OF13 pre-req for PCP match in TestCommandProvider

-OXM_OF_VLAN_PCP requires a VID match pre-req. The function
f80() results in a OFPBAC_MATCH_INCONSISTENT since it is
only matching on nw_dst and not a vid.

-Patching this seperate from Bug-1072/Patch-ba79f4f in
case there was a specific reason for this combination
that is outside the spec.

Field: OXM_OF_VLAN_PCP
Dependency: VLAN VID!=NONE
Reference: See table 10, section A.2.3.5 for validation

Change-Id: Ie88b5b9bbfdfa9634252b79242fc333c0b966f77
Signed-off-by: Brent Salisbury <brent.salisbury@gmail.com>
10 years agoMerge "Bug-1137:Handling OF-SAL Match for Vlan range of FFFF"
michal rehak [Tue, 10 Jun 2014 11:35:44 +0000 (11:35 +0000)]
Merge "Bug-1137:Handling OF-SAL Match for Vlan  range of FFFF"

10 years agoBug 1035 - Invalid mask in CommandProvider for uint64/byte conversion 74/7374/3
Brent Salisbury [Sun, 25 May 2014 07:37:14 +0000 (03:37 -0400)]
Bug 1035 - Invalid mask in CommandProvider for uint64/byte conversion

Patchset3: typo on last commit. This resolves both the mask
and goto table issues.

Patchset2: Added Michal's recommendation on the tunnel mask
and that looks good to go now.

After digging I saw the root of tunnelId failures was
actually due to the GotoTable Instructions function
was setting the GotoTable->2 which was also the same
table as the writeTable->2 transaction. Since the OF
spec requires forward progress at the end of each table
this was causing function f48() but also f6() to fail.

Change-Id: If975da5e872115495a7c1b38553a5e2bfe4a18ea
Signed-off-by: Brent Salisbury <brent.salisbury@gmail.com>
10 years agoBug-1137:Handling OF-SAL Match for Vlan range of FFFF 44/7844/1
Kamal Rameshan [Tue, 10 Jun 2014 02:08:22 +0000 (19:08 -0700)]
Bug-1137:Handling OF-SAL Match for Vlan  range of FFFF

Change-Id: I14b7284940e45254445fc89077fc231039c412ed
Signed-off-by: Kamal Rameshan <kramesha@cisco.com>
10 years agoBug-1137:Vlan-id defaulted to 65535 for OF1.0. Depends on yangtools patch 7793 94/7794/3
Kamal Rameshan [Sat, 7 Jun 2014 00:11:06 +0000 (17:11 -0700)]
Bug-1137:Vlan-id defaulted to 65535 for OF1.0. Depends on yangtools patch 7793

Change-Id: I942576f2269a23e8d938d80d43feaa294f476cb1
Signed-off-by: Kamal Rameshan <kramesha@cisco.com>
10 years agoUse openflowjava-supplied MAC address utilities 60/7660/4
Robert Varga [Tue, 3 Jun 2014 21:57:54 +0000 (23:57 +0200)]
Use openflowjava-supplied MAC address utilities

The implementations provided here are trivial and do not perform as well
as they could. openflowjava has optimized versions of these, so use
those.

Change-Id: I895b9e2d0268b8497590438195c644c97967abfe
Signed-off-by: Robert Varga <robert.varga@pantheon.sk>
10 years agoUse non-sign-extending shift 79/7679/3
Robert Varga [Wed, 4 Jun 2014 11:59:59 +0000 (13:59 +0200)]
Use non-sign-extending shift

Just a precaution of intent: use shift which introduces 0, not
sign-dependent.

Change-Id: I370805c1d1dd5e9ba3893768767f1cb6047acbbd
Signed-off-by: Robert Varga <robert.varga@pantheon.sk>
10 years agoDrop-test: remove minor janitorial things 74/7674/3
Robert Varga [Wed, 4 Jun 2014 10:29:21 +0000 (12:29 +0200)]
Drop-test: remove minor janitorial things

- remove superfluous constructors
- remove supresswarnings
- optimize sb.append a tiny bit

Change-Id: Id9e4662f4212e0160bf4095f21a66d709f179372
Signed-off-by: Robert Varga <robert.varga@pantheon.sk>
10 years agoDroptest: optimize builder allocation 73/7673/3
Robert Varga [Wed, 4 Jun 2014 08:58:47 +0000 (10:58 +0200)]
Droptest: optimize builder allocation

Builder can be reused between invocations, so let's cache it in a
thread-local variable.

Change-Id: Ib76f1d10aabd95d896d81c5be33d42230ee3f366
Signed-off-by: Robert Varga <robert.varga@pantheon.sk>
10 years agoDroptest: unify packet handling 72/7672/3
Robert Varga [Wed, 4 Jun 2014 07:43:53 +0000 (09:43 +0200)]
Droptest: unify packet handling

Rework the two implementation to make sure they do equivalent
processing, such that the difference between the two really is only the
way they propage changes through the system.

Change-Id: I01cb2daea78747a5659d04da65c810188b3e38c6
Signed-off-by: Robert Varga <robert.varga@pantheon.sk>
10 years agoBUG-1134: Added TCP flag conversions between md-sal and openflowjava 07/7607/3
Brent Salisbury [Mon, 2 Jun 2014 19:03:25 +0000 (15:03 -0400)]
BUG-1134: Added TCP flag conversions between md-sal and openflowjava

-Conversions from high level model (model-flow-base)
to low level models (ofjava).

-Test included in OpenflowpluginTestCommandProvider
as function addMDFlow f83() for testing.

-Test results here:
https://gist.github.com/04f96a080eb7ee32413b

---- Extensions info (see BUG-1133) ----
Support for extensions in OFPlugin is not going to be implemented
this way (for POC it is temporarily ok). Vendor specific messages
should be modeled in vendor bundle (augmenting MD-SAL or OFJAVA-API
models). Current strategy is based on trasnparent forwarding of
vendor messages through OFPlugin.
More info will be shortly at
https://wiki.opendaylight.org/view/OpenDaylight_OpenFlow_Plugin:Backlog:Extensibility

Change-Id: Ifaa835a43ec8e0b2fb23e84abf121aefdbcde339
Signed-off-by: Brent Salisbury <brent.salisbury@gmail.com>
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoOptimize string splitting by using Splitter class 59/7659/1
Robert Varga [Tue, 3 Jun 2014 21:53:29 +0000 (23:53 +0200)]
Optimize string splitting by using Splitter class

Instead of performing costli String.split() operation, instantiate a
shared Splitter and use it to split the string. Requires slight
alterations to logic, but those are rather obvious.

Change-Id: Icec3c4588e062373c090b3479635a93428f278f1
Signed-off-by: Robert Varga <robert.varga@pantheon.sk>
10 years agoImprove InjectionKey performance 43/7643/1
Robert Varga [Tue, 3 Jun 2014 12:12:01 +0000 (14:12 +0200)]
Improve InjectionKey performance

As it turns out we put these into a HashMap, so not dispersing on class
name is actively hurting us. The trade-off here is between hashCode()
performance and its effectiveness.

The design of this class called for Strings to be used as identifiers,
and hashCode() of strings is generally slow. As it turns out, though,
all the strings used were derived from classes -- which is good news.

This patch switches to using Class<?> objects as identifiers, as they
have fast hashCode/equals and the lookups are equivalent.

Change-Id: I51e572dc98088733eca15c35ee6d0bac991c5da7
Signed-off-by: Robert Varga <robert.varga@pantheon.sk>
10 years agoOptimize conversionMapping table 39/7639/1
Robert Varga [Tue, 3 Jun 2014 10:50:26 +0000 (12:50 +0200)]
Optimize conversionMapping table

Once the table has been initialized, we can switch it to immutable, as
the keys (Short) provide fast hashCode/equals. Saves memory and
increases lookup speed.

Change-Id: I4698f44ce5a8e83c7c54e544e5ab72487f22a92e
Signed-off-by: Robert Varga <robert.varga@pantheon.sk>
10 years agoDroptest: optimize macToString 38/7638/1
Robert Varga [Tue, 3 Jun 2014 10:01:12 +0000 (12:01 +0200)]
Droptest: optimize macToString

This elides the creation of multiple objects and rather formats the
bytes quickly.

Change-Id: Iabac79bf86c0b2098bffe69f221797f2fd1c326f
Signed-off-by: Robert Varga <robert.varga@pantheon.sk>
10 years agoMerge changes Id2bb5c7d,I341f0ca7
michal rehak [Tue, 3 Jun 2014 09:27:09 +0000 (09:27 +0000)]
Merge changes Id2bb5c7d,I341f0ca7

* changes:
  Use singleton list for simple packetIn
  Prefer more threads than deeper queue

10 years agoMerge "Fixed incorrect help message."
michal rehak [Tue, 3 Jun 2014 09:12:35 +0000 (09:12 +0000)]
Merge "Fixed incorrect help message."

10 years agoUse singleton list for simple packetIn 32/7632/1
Robert Varga [Mon, 2 Jun 2014 17:19:26 +0000 (19:19 +0200)]
Use singleton list for simple packetIn

Tracing has shown we are spending quite some time constructing
CopyOnWriteArrayList -- just to throw it away. Let's use a singleton
list, which is way faster.

Change-Id: Id2bb5c7d548d5d4257170828667a66543099fa38
Signed-off-by: Robert Varga <robert.varga@pantheon.sk>
10 years agoPrefer more threads than deeper queue 31/7631/1
Robert Varga [Tue, 3 Jun 2014 05:56:29 +0000 (07:56 +0200)]
Prefer more threads than deeper queue

ThreadPoolExecutor does not start spawning new threads until the
underlying queue's offer() method return false. This means that we would
completely fill the queue before kicking off more threads.

This patch introduces a ForwardingBlockingQueue, which always returns
false from offer(), thus forcing the threadpool to expand. The policy
then uses put() to schedule (or block on scheduling) the tasks.

Change-Id: I341f0ca7061f6b76ae1f3e049b12704f35140633
Signed-off-by: Robert Varga <robert.varga@pantheon.sk>
10 years agoMerge "Fixed incorrect help message, converted to more standard format."
michal rehak [Tue, 3 Jun 2014 08:49:22 +0000 (08:49 +0000)]
Merge "Fixed incorrect help message, converted to more standard format."

10 years agoFixed incorrect help message. 96/7596/1
Daniel Farrell [Mon, 2 Jun 2014 15:49:14 +0000 (11:49 -0400)]
Fixed incorrect help message.

Change-Id: Id1d62c2ea0a2e7c3f08e3275e12b5a3d9453f4ef
Signed-off-by: Daniel Farrell <dfarrell@redhat.com>
10 years agopython tests fix: 82/7582/3
Vaclav Demcak [Mon, 2 Jun 2014 08:32:57 +0000 (10:32 +0200)]
python tests fix:

* SAL fix build xml update input (flow, meter, grup)
* fix the URL address for a flow in operational DS
* set default coloring disabled
* add logging wait time info

Change-Id: Ibd839ab64313ba9186e47d4f3b91b026c6d57cf8
Signed-off-by: Vaclav Demcak <vdemcak@cisco.com>
10 years agoBUG-529: Remove Reactor getter contention 41/7541/1
Robert Varga [Fri, 30 May 2014 15:55:17 +0000 (17:55 +0200)]
BUG-529: Remove Reactor getter contention

Performance testing has identified contention on getInstance() methods.
It turns out that we can pre-create the instance and then get it in a
lock-free manner.

Change-Id: I49c219d5954fa792218040a0ae66bf344992413a
Signed-off-by: Robert Varga <robert.varga@pantheon.sk>
10 years agoBUG-529: imrpove statistics collection 40/7540/1
Robert Varga [Fri, 30 May 2014 15:54:31 +0000 (17:54 +0200)]
BUG-529: imrpove statistics collection

This removes synchronized blocks in statistics gathering, as these are
multi-threaded codepaths and we saw contention here.

Change-Id: Ie99ea3fd2e16b66af533e773af04d7978c2c0ac1
Signed-off-by: Robert Varga <robert.varga@pantheon.sk>
10 years agoBUG-529: improve droptest performance 39/7539/1
Robert Varga [Fri, 30 May 2014 15:53:45 +0000 (17:53 +0200)]
BUG-529: improve droptest performance

This eliminates the use of synmchronized blocks to perform statistics
maintenance.

Change-Id: I18ba3b0bcf2d04e2ce8ba993e0fa776cd6d67a72
Signed-off-by: Robert Varga <robert.varga@pantheon.sk>
10 years agoFixed incorrect help message, converted to more standard format. 25/7525/1
Daniel Farrell [Thu, 29 May 2014 20:44:52 +0000 (16:44 -0400)]
Fixed incorrect help message, converted to more standard format.

Change-Id: If346d70baba1ec11c310cbafcb2040e64a874fe9
Signed-off-by: Daniel Farrell <dfarrell07@gmail.com>
10 years agoMerge "BUG-1074: add javadoc for QueueKeeper"
Abhijit Kumbhare [Tue, 27 May 2014 16:51:04 +0000 (16:51 +0000)]
Merge "BUG-1074: add javadoc for QueueKeeper"

10 years agoRemoved unnecessary libxml2 import 21/7421/1
Martin Bobak [Tue, 27 May 2014 11:24:46 +0000 (13:24 +0200)]
Removed unnecessary libxml2 import

Signed-off-by: Martin Bobak <mbobak@cisco.com>
10 years agoCRUD ODL python RESTconf tests 90/7390/3
Vaclav Demcak [Fri, 24 Jan 2014 17:30:44 +0000 (18:30 +0100)]
CRUD ODL python RESTconf tests

* move to python2.6 and python script refactoring
* fix xml
* add CRUD tests (flow, meter, group)

https://wiki.opendaylight.org/view/OpenDaylight_OpenFlow_Plugin::Python_test_scripts

Signed-off-by: Vaclav Demcak <vdemcak@cisco.com>
Change-Id: I3c8723dae0cbd1500f9fe318ec90115ff739f3db

10 years agoBUG-1074: add javadoc for QueueKeeper 01/7401/1
Michal Rehak [Mon, 26 May 2014 15:22:22 +0000 (17:22 +0200)]
BUG-1074: add javadoc for QueueKeeper

- added javadoc for QueueKeeper and QueueKeeperImpl

Change-Id: I5cf4d5c2798766435985bb9a93c45b07e18d4d8b
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoBUG-770: NumberFormatException for datapathId 99/7399/2
Michal Rehak [Mon, 26 May 2014 12:50:31 +0000 (14:50 +0200)]
BUG-770: NumberFormatException for datapathId

- changed long to bigInteger in order to avoid negative Long
- added unit tests for touched util-methods

Change-Id: Ibdebd3965baf552bd92702af0399961b70defe56
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoBug 1071: Invalid value used for frame priority code point in OFCommandProvider 71/7371/1
Brent Salisbury [Sun, 25 May 2014 03:29:16 +0000 (23:29 -0400)]
Bug 1071: Invalid value used for frame priority code point in OFCommandProvider

-Adjusted to a magic number in bounds for PCP (0-7) to enable the test to stop failing for the function f80 in OpenFlowPluginTestCommandProvider.

Change-Id: Ib146f228e174b6785dfe9fecf93fa7cabba79f4f
Signed-off-by: Brent Salisbury <brent.salisbury@gmail.com>
10 years agoBUG-1065: translation of Ipv6ExtHeader flags to int 38/7338/1
Michal Rehak [Thu, 22 May 2014 15:51:19 +0000 (17:51 +0200)]
BUG-1065: translation of Ipv6ExtHeader flags to int

- moved translation to separate method + test
- fixed the false part of ternary operator

Change-Id: I498dbf6c37f2d7348743d33bd49fe2ff46870621
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoopenflowPlugin still uses for IT unsupported version of nagasena 02/7302/2
Giovanni Meo [Tue, 20 May 2014 14:45:00 +0000 (16:45 +0200)]
openflowPlugin still uses for IT unsupported version of nagasena

Updated the openflowplugin IT to use the newer version of the nagasena
artifacts that not longer comes from controller. Without this patch
this project cannot release artifacts because will depends on a
SNAPSHOT that cannot be made RELEASED

Signed-off-by: Giovanni Meo <gmeo@cisco.com>
10 years agoBug:275 - Modifying a flow is not applying the changes to the switch 99/7099/2
Kamal Rameshan [Fri, 16 May 2014 01:21:21 +0000 (18:21 -0700)]
Bug:275 - Modifying a flow is not applying the changes to the switch

Change-Id: I75847446791a4dab3aa2e3b076c8b369c7337b64
Signed-off-by: Kamal Rameshan <kramesha@cisco.com>
10 years agoBug:995-Controller does not learn node connectors from mininet OF10 27/7127/3
Michal Rehak [Mon, 19 May 2014 13:35:52 +0000 (15:35 +0200)]
Bug:995-Controller does not learn node connectors from mininet OF10

- replaced shutdownNow with shutdown+purge

Change-Id: Ia9f299fd5e46d13a9bbd9d28925ca4e758ec46ec
Signed-off-by: Kamal Rameshan <kramesha@cisco.com>
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoOptimize list manipulation 42/7142/3
Robert Varga [Sat, 17 May 2014 20:29:15 +0000 (22:29 +0200)]
Optimize list manipulation

This patch optimizes the various allocations of ArrayList where we know
the target size.
- by versionOrder the size equals to the highest version value + 1

Change-Id: I2a54a4cacc21fe8d382080954de5fd8d2c16f472
Signed-off-by: Robert Varga <robert.varga@pantheon.sk>
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoImprove droptest logging 52/7152/1
Robert Varga [Sun, 18 May 2014 09:49:15 +0000 (11:49 +0200)]
Improve droptest logging

Print out the full exceptions, such that we can reason about it.

Change-Id: Ie0ec723253a8e65e255bdc06f6322b1d72033e22
Signed-off-by: Robert Varga <robert.varga@pantheon.sk>
10 years agoOptimize drop-test memory usage 40/7140/2
Robert Varga [Sat, 17 May 2014 20:12:17 +0000 (22:12 +0200)]
Optimize drop-test memory usage

This patch makes sure drop-test uses the Collections.singletonList(),
so as to not waste memory on ArrayList, since we are only forwarding a
single entry.

Change-Id: Iaf4e7e408ed147177a40355b3c0ab2d4e18d133c
Signed-off-by: Robert Varga <robert.varga@pantheon.sk>
10 years agoBUG-542 - adding overall statictics 24/7124/1
Michal Rehak [Fri, 16 May 2014 19:11:08 +0000 (21:11 +0200)]
BUG-542 - adding overall statictics

- temporary extended messageSpy code
- added logging to rpc threadPool - afterExecution
- cleaned imports
- deprecated hard coded configurations for OFJava SwitchConnectionProvider
- simplified signature of MessageSpy
- added messageSpy and rpcPool into SessionManager

Change-Id: If41e33c885dab0907fdf642ad1f483d93f74b19c
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoBUG-956 deadlock by rpc invocation - phase2 17/7117/1
Michal Rehak [Thu, 8 May 2014 20:09:43 +0000 (22:09 +0200)]
BUG-956 deadlock by rpc invocation - phase2

- tuning Future chaining
- introduced ListeningExecutorService
- dereference used
- clean up after bulkTransactionCache removal
- add/update meter and group refactored in order to follow future responses and run in pool

Change-Id: I5c9367512c3ab6f1799f30dad1344667fde63991
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoBUG-1006 - removal of bulkTransactionCache 24/7024/2
Michal Rehak [Thu, 15 May 2014 15:56:44 +0000 (17:56 +0200)]
BUG-1006 - removal of bulkTransactionCache

- removed bulkTransactionCache - this contract is considered a bad tradeoff:
  Application which invoked rpc bound with device will receive transaction id
  in response. If the app needs to analyze error it is up to app
  - to listen to appropriate errorNotification and
  - to keep track of last input value and pair it by transactionId with
    error notification

Change-Id: Icf38f82986451bc9cd46ce84996fe9631295bc16
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoBUG-1006 - limiting bulkTransactionCache 94/6994/1
Michal Rehak [Wed, 14 May 2014 17:24:58 +0000 (19:24 +0200)]
BUG-1006 - limiting bulkTransactionCache

- size
- soft values
- shorter lifetime
- removing values after use

Change-Id: Ia02c07c02726291834c1f73bb19b03610124e9b8
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoBug 998 - Fix for update flow process 36/6936/1
Vaclav Demcak [Tue, 13 May 2014 16:28:15 +0000 (18:28 +0200)]
Bug 998 - Fix for update flow process

- change FlowAddBuilder to FlowUpdateBuilder for update process

Change-Id: I727ae03332bee6079b7927e98660b45e9dc537ef
Signed-off-by: Vaclav Demcak <vdemcak@cisco.com>
10 years agoBug 998 - fix for update flow 35/6935/1
Vaclav Demcak [Tue, 13 May 2014 16:07:52 +0000 (18:07 +0200)]
Bug 998 - fix for update flow

Change-Id: Iad157785a9f42e89d6b99d4211fdce591efcea41
Signed-off-by: Vaclav Demcak <vdemcak@cisco.com>
10 years agofix for BUG-947 - ErrorHandler classLoader issue 25/6725/4
Michal Rehak [Mon, 5 May 2014 15:05:20 +0000 (17:05 +0200)]
fix for BUG-947 - ErrorHandler classLoader issue

- exception must be logged in the same thread as there is thread classLoader context, which might differ from the previously dedicated thread for exception logging

Change-Id: Ia2f911fdd5845280575fcb6a57c16aafdc0c2b53
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agofix for BUG-709 - connectionCookie generate, use 61/6661/5
Michal Rehak [Fri, 2 May 2014 07:47:24 +0000 (09:47 +0200)]
fix for BUG-709 - connectionCookie generate, use

- session cookie is based on datapathId only (not exposed, based in BigInteger)
- connection cookie is based on auxiliaryId, unique hashed key is exposed

Change-Id: Ifc293f876b3e57be38622141e92f8cdb0e47d056
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoMerge "BUG-985 - finisher queue is unbound"
michal rehak [Mon, 12 May 2014 17:57:37 +0000 (17:57 +0000)]
Merge "BUG-985 - finisher queue is unbound"

10 years agoBUG-985 - finisher queue is unbound 85/6885/4
Jan Medved [Mon, 12 May 2014 05:39:50 +0000 (22:39 -0700)]
BUG-985 - finisher queue is unbound

and causes out of memory error under stress

- replaced queue.offer with queue.put (blocking behavior)

Changed the initialization of the TicketFinisher queue to limit the size of the queue to 1,000 elements. Backpressure will be exerted when the max queue size has been reached. This is to fix an out-of-memory condition that occurs under heavy load when the TicketFinisher queue grows w/o bounds.

Change-Id: I619d1c27db98fec95380bef2504add3025195abc
Signed-off-by: Jan Medved <jmedved@cisco.com>
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoMerge "Added statistics for received packets in and for sent addFlows to the RPC...
michal rehak [Mon, 12 May 2014 17:06:59 +0000 (17:06 +0000)]
Merge "Added statistics for received packets in and for sent addFlows to the RPC and Data Store drop tests. BUG-991 (https://bugs.opendaylight.org/show_bug.cgi?id=991)"

10 years agoAdded statistics for received packets in and for sent addFlows to the RPC and Data... 55/6755/6
Jan Medved [Tue, 6 May 2014 17:57:57 +0000 (10:57 -0700)]
Added statistics for received packets in and for sent addFlows to the RPC and Data Store drop tests. BUG-991 (https://bugs.opendaylight.org/show_bug.cgi?id=991)

Change-Id: I3a99bb81c8eadf87d731715b1f40f36af9734ba9
Signed-off-by: Jan Medved <jmedved@cisco.com>
10 years agoBUG-986 - remove catch Throwable (sonar) 92/6892/2
Michal Rehak [Mon, 12 May 2014 12:05:28 +0000 (14:05 +0200)]
BUG-986 - remove catch Throwable (sonar)

- avoided catch of Throwable
- decreased some debug levels to trace

Change-Id: I26b6a8f260fc6db5596df07f9408010118a72ea1
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoFix for Bug 957. 35/6835/1
Ravikumar Chiguruvada [Fri, 9 May 2014 13:07:36 +0000 (18:37 +0530)]
Fix for Bug 957.

Signed-off-by: Ravikumar Chiguruvada <ravikumar.chiguruvada@ericsson.com>
Change-Id: Ifa471d462f4800f553bb68d3ed7fe1cc71a17fcc

10 years agoMerge "BUG-976: jenkins slowed down by dowload from nexus"
Michal Rehak [Thu, 8 May 2014 10:08:21 +0000 (10:08 +0000)]
Merge "BUG-976: jenkins slowed down by dowload from nexus"

10 years agoBUG-976: jenkins slowed down by dowload from nexus 12/6812/2
Michal Rehak [Thu, 8 May 2014 08:47:59 +0000 (10:47 +0200)]
BUG-976: jenkins slowed down by dowload from nexus

- removed opendaylight.release repository (contained in groups/public)

Change-Id: I190e3ccd9482ab58dd5502567b4d2a7e14b5e68f
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoFix for Bug 957 02/6802/1
Ravikumar Chiguruvada [Thu, 8 May 2014 06:17:00 +0000 (11:47 +0530)]
Fix for Bug 957

Change-Id: Ifea346315b62b36d3c9c8bf2a180c7db157f4b79

Signed-off-by: Ravikumar Chiguruvada <ravikumar.chiguruvada@ericsson.com>
10 years agoMerge "additional fix fro BUG-782 unregistering switch providers"
Michal Rehak [Wed, 7 May 2014 22:43:47 +0000 (22:43 +0000)]
Merge "additional fix fro BUG-782 unregistering switch providers"

10 years agofix for BUG-956 - deadlock by rpc invocation 38/6738/3
Tony Tkacik [Tue, 6 May 2014 10:20:37 +0000 (12:20 +0200)]
fix for BUG-956 - deadlock by rpc invocation

- fixed future chaining
- added rpcPool for asynchronous processing

Change-Id: I19c9b94e383f3e3f5bfff179a966b1950a942821
Signed-off-by: Tony Tkacik <ttkacik@cisco.com>
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoadditional fix fro BUG-782 unregistering switch providers 60/6660/3
Michal Rehak [Wed, 30 Apr 2014 07:26:29 +0000 (09:26 +0200)]
additional fix fro BUG-782 unregistering switch providers

- CompositeObjectRegistration is stored into sessionContext upon addSession
- upon removeSession the CompositeObjectRegistration is closed

Change-Id: Ieee90f50645db12c5df198fd4501daa804a135d3
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoIT fix BUG-837 - missing dependencies 95/6795/1
Michal Rehak [Wed, 7 May 2014 15:13:06 +0000 (17:13 +0200)]
IT fix BUG-837 - missing dependencies

- thirdparty stuff
- netconf stuff
- config stuff
- logback config

Change-Id: I5f19bf187af6b29ecf65c3eb287da940d4f75fa0
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agofix for BUG-948 - changing log level for portStatusMessage 08/6708/1
Michal Rehak [Mon, 5 May 2014 12:05:28 +0000 (14:05 +0200)]
fix for BUG-948 - changing log level for portStatusMessage

- in translator the port status message content was logged inappropriately using error level

Change-Id: I3b395b09c39b0c88c6a0d3f452ade412d9c8f68a
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agofix2 for BUG-837 - bindingAwareBroker waiting 12/6612/1
Michal Rehak [Wed, 30 Apr 2014 07:45:25 +0000 (09:45 +0200)]
fix2 for BUG-837 - bindingAwareBroker waiting

- updated controller.xml - configSubsystem IT changed in sal-binding-it

Change-Id: Ib28900451951b1dd1ab1207f9c9e7bc7c58c577c
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agofix for BUG 782 - NPE in ErrorTranslator during handshake 13/6513/1 topic/threading
Michal Rehak [Mon, 28 Apr 2014 09:38:39 +0000 (11:38 +0200)]
fix for BUG 782 - NPE in ErrorTranslator during handshake

- there is no session available before registration

Signed-off-by: Michal Rehak <mirehak@cisco.com>
Change-Id: Ib0e420a879c6512e5f86003ad9d024aee789c04d

10 years agoflowCookieMask conversion fix (BUG 840) 57/6357/1
Michal Rehak [Thu, 24 Apr 2014 12:13:23 +0000 (14:13 +0200)]
flowCookieMask conversion fix (BUG 840)

Change-Id: I8c601bc1cd5865f2112dbe87176d2bef5f80b608
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoIT unstability fix - BUG 837 43/6343/2
Michal Rehak [Wed, 23 Apr 2014 16:42:25 +0000 (18:42 +0200)]
IT unstability fix - BUG 837

- registration of node by md-sal keeps failing - if closing of connection will wait for 2 secs after handshake, md-sal should succeed in registering

Change-Id: I510defd8bbb7334eaaebc4b066d019b6de2af040
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoBug:701 - Have Typed ErrorNotification for different node errrors 50/6250/2
Kamal Rameshan [Thu, 17 Apr 2014 03:02:41 +0000 (20:02 -0700)]
Bug:701 - Have Typed ErrorNotification for different node errrors

Change-Id: I77cc6c8b443e6a6598f499043b6e5c242be23a1d
Signed-off-by: Kamal Rameshan <kramesha@cisco.com>
10 years agofix for BUG-709 - mixed up flow cookie and connection cookie 60/6260/2
Michal Rehak [Thu, 17 Apr 2014 15:15:59 +0000 (17:15 +0200)]
fix for BUG-709 - mixed up flow cookie and connection cookie

- flow cookie [64b] is present at receivedPacket (to distinguish responsible flow)
- connection cookie [32b] is present at receivedPacket and is expected
  to be copied into corresponding transmitPacketInput (to distinguish auxiliary connection)
- offtopic: felix generated META-INF moved to target (drop-test, test-provider)
- hotfix - InstanceIdentifier API changes adaptation

requires controller change: https://git.opendaylight.org/gerrit/#/c/6259/

Change-Id: I76d30ba6ba16bf6ac57977288ee30ab059c12659
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agofix for BUG 766 - failing IT because of new dependency 51/6251/1
Michal Rehak [Thu, 17 Apr 2014 09:23:24 +0000 (11:23 +0200)]
fix for BUG 766 - failing IT because of new dependency

- added org.opendaylight.controller.thirdparty:nagasena dependency for pax
- cleaned deps already included in TestHelper

Change-Id: Ie36bc3b0055b9db7b2b5617fad51801917d28aaf
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoadditional config disribution (BUG 541) 32/6232/1
Michal Rehak [Wed, 16 Apr 2014 07:28:30 +0000 (09:28 +0200)]
additional config disribution (BUG 541)

- added subproject in order to distribute configSubsystem xml

Change-Id: Ida89544a9ed1b7d53de797357730ead0a11130dc
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoconfig subsystem integration - BUG 541 84/5884/2
Basheeruddin Ahmed [Thu, 3 Apr 2014 00:55:01 +0000 (17:55 -0700)]
config subsystem integration - BUG 541

added configuration xml for configSubsystem
added sal code generator
changed reference to swichConnectionProvider (now there is collection)
changed startup in order to use configSubsystem wiring
preserved the osgi BunndleContext access for cli stuff
fixed IT (configSubsystem related stuff)
enabled documentation generator
upated Future obtained from OFJava to ListenableFuture
changed wiring to OFJava (1switchConnectionProvider = 1 switchConnectionHandler)

Change-Id: I9b7ca8a173e4a36381411cec08772d8440180a03
Signed-off-by: Basheeruddin Ahmed <syedbahm@cisco.com>
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoBug 651 - cleanup eclipse errors in openflowplugin 08/6008/4
jameshall03885 [Wed, 9 Apr 2014 17:16:05 +0000 (13:16 -0400)]
Bug 651 - cleanup eclipse errors in openflowplugin

- To avoid devs from having to run the IDE profile so eclipse can see the yang
generated source files, the yang source files are now generated under src/main/yang-gen-sals
or src/main/yang-gen-config instead of target/target-ide.

- The maven-clean-plugin was moved to the ofp parent pom and removed from
all child poms where it was specified. Thus, it will be automatically run for all
projects. The plugin's execution will clean all xtend and yang generated source dirs.

- The build-helper-maven-plugin was also moved to the parent pom and removed from
all child poms. It will automatically run for all projects and no longer needs to be
specified in child poms. The plugin's execution will created eclipse source folders
for all xtend and yang generated source.

I also removed two occurances of "&gt;" which seemed extraneous ... but perhaps I'm unaware of some pom.xml magic?

Change-Id: Ie99e141e55e03a3a3ebce3743624e7bea6e42e7b
Signed-off-by: jameshall03885 <ghall@brocade.com>
10 years agoBug:679 - eliminate the use of xtend in openflowplugin 33/6033/2
Kamal Rameshan [Thu, 10 Apr 2014 18:43:42 +0000 (11:43 -0700)]
Bug:679 - eliminate the use of xtend in openflowplugin

- replaced xtend classes with pure java classes
- removed xtend dependency and plugin from pom

Change-Id: Ia0fdf0b7162b04143a5e654971536481f28c3b2a
Signed-off-by: Kamal Rameshan <kramesha@cisco.com>
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agofix for BUG 661 01/5901/1
Michal Rehak [Fri, 4 Apr 2014 07:32:34 +0000 (09:32 +0200)]
fix for BUG 661

- lowered version of commons-codec (to 1.7 - specified in dependencyManagement of parent)

Change-Id: I7f262318deeaac2aaf7abfc7f1221874d725ed0c
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoMerge "Added missing keys. This allows to reuse DTOs for operational/configuration...
michal rehak [Thu, 3 Apr 2014 11:48:16 +0000 (11:48 +0000)]
Merge "Added missing keys. This allows to reuse DTOs for operational/configuration data."

10 years agoAdded missing keys. This allows to reuse DTOs for operational/configuration data. 35/5835/1
Tony Tkacik [Tue, 1 Apr 2014 14:56:01 +0000 (16:56 +0200)]
Added missing keys. This allows to reuse DTOs for operational/configuration data.

Signed-off-by: Tony Tkacik <ttkacik@cisco.com>
10 years agofix BUG 596 (part2) 21/5821/1
Michal Rehak [Mon, 31 Mar 2014 13:14:45 +0000 (15:14 +0200)]
fix BUG 596 (part2)

- reenabled fixed SalPortService interface (providing portUpdate)

Change-Id: I5adbf8e1868abf55f319b1d03d5d44deca88b378
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agovlan ID fix 09/5809/2
Prasanna Huddar [Sat, 29 Mar 2014 11:47:43 +0000 (17:17 +0530)]
vlan ID fix
Signed-off-by: Prasanna Huddar <prasanna.huddar@ericsson.com>
10 years agofix for BUG 609 07/5807/1
Michal Rehak [Fri, 28 Mar 2014 23:06:35 +0000 (16:06 -0700)]
fix for BUG 609

- extra hello message is being disregarded

Change-Id: Idfe6009c5791d2441c1e5fe5311b75b6927d5f5f
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoReenabled integration tests - part2 06/5806/1
Michal Rehak [Fri, 28 Mar 2014 21:38:10 +0000 (14:38 -0700)]
Reenabled integration tests - part2

- adding dirty sleep to have the controller part up and running

Change-Id: I1f6e9b617d10df162631af34598206b7b5d81bf5
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoReenabled integration tests 75/5475/3
Michal Rehak [Tue, 11 Feb 2014 10:30:05 +0000 (11:30 +0100)]
Reenabled integration tests

- added osgi console and debug support for IT
- pulled dependencies to parent pom (including IT dependencies)
- fixed dependencies (pax, config-manager, commons-codec, junit)
- fixed pax bundles
- tidied up plugin activator + OpenflowPluginProvider
- changing reactor strategy to perClass (speedup testing)
- added AutoCloseable interface to ErrorHandler
- moved errorHandler setup up to MDController
- enabling of SalIntegrationTest
- added check whether the SimpleClient succeeded to connect to controller

- updated versions of sal-api, netconf-connector, config-manager
- added 5s wait on startup - dirty temprary solution untill configSubsysten

- rebased

Change-Id: I21ff052ff6d90ff95f5b8b2a51fb8d736e697034
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agofix for BUG 596 81/5781/2
Michal Rehak [Wed, 26 Mar 2014 19:21:11 +0000 (12:21 -0700)]
fix for BUG 596

- SalPortService wont be supported for now

Change-Id: Ia9f140d969e324630fbe7d64bd03ff8d198bc0c5
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoBUG-592: prepare for API change 88/5788/2
Robert Varga [Thu, 27 Mar 2014 00:35:03 +0000 (01:35 +0100)]
BUG-592: prepare for API change

getPathArguments() was added by mistake with its current return type. It
will be fixed in a follow-up patch, so fall back to getPath(), which does
the same thing, such that it won't break when things change.

Change-Id: Iae3661aa8942e050c0ac638cce0bec6dae9fa290
Signed-off-by: Robert Varga <robert.varga@pantheon.sk>
10 years agoquickfix - added missing org.osgi.core provided dependency 69/5769/1
Michal Rehak [Wed, 26 Mar 2014 02:59:53 +0000 (03:59 +0100)]
quickfix - added missing org.osgi.core provided dependency

- sample-consumer
- learning-switch

Change-Id: Ibb4f5fa3abcfa88103097e95a147712ddf06d2ff
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoMerge "Added required keys to items in DropTestCommiter."
michal rehak [Tue, 25 Mar 2014 02:23:25 +0000 (02:23 +0000)]
Merge "Added required keys to items in DropTestCommiter."

10 years agoAdded required keys to items in DropTestCommiter. 42/5742/1
Tony Tkacik [Tue, 25 Mar 2014 02:02:32 +0000 (03:02 +0100)]
Added required keys to items in DropTestCommiter.

Signed-off-by: Tony Tkacik <ttkacik@cisco.com>
10 years agoAdjust permissions: datafiles should not be executable 40/5740/1
Robert Varga [Tue, 28 Jan 2014 00:19:22 +0000 (01:19 +0100)]
Adjust permissions: datafiles should not be executable

Signed-off-by: Robert Varga <robert.varga@pantheon.sk>
10 years agoFix a typo 39/5739/1
Robert Varga [Sat, 15 Feb 2014 06:28:00 +0000 (07:28 +0100)]
Fix a typo

Signed-off-by: Robert Varga <robert.varga@pantheon.sk>
10 years agoBug 572 - Removing the duplicate translate method call. Tested on the mininet with... 34/5734/1
Lakshman Mukkamalla [Mon, 24 Mar 2014 19:54:42 +0000 (12:54 -0700)]
Bug 572 - Removing the duplicate translate method call. Tested on the mininet with linear,2 switch.

Change-Id: I8386b31e68742c844495da96463330fee61cd5b1
Signed-off-by: lmukkama@cisco.com
10 years agoRefactor and fix another reference bug in PortConvertor 12/5712/1
Moiz Raja [Fri, 21 Mar 2014 23:38:58 +0000 (16:38 -0700)]
Refactor and fix another reference bug in PortConvertor

Another place where an out-parameter was used has been fixed. Also refactored some code to make it a little cleaner.

Change-Id: I3ef970cc09007e84d43ff8f5bbce5cde241ab525
Signed-off-by: Moiz Raja <moraja@cisco.com>
10 years agoPass the appropriate PortModInput to portMod 06/5706/3
Moiz Raja [Fri, 21 Mar 2014 07:15:06 +0000 (00:15 -0700)]
Pass the appropriate PortModInput to portMod

In ModelDrivenSwitchImpl we were building an input but the built input was not passed to portMod. This had issues like missing Xid on the PortModInput
which caused NPE's downstream in the open flow java code

Also we were not correctly population ConfigV10, MaskV10 and FeaturesV10 on the PortModInput which caused NPE in the open flow java code when using
open flow 10

This commit fixes both of those problems. After this updatePort actually works.

Change-Id: Ie2548a36b2ba06f1d1ce0ccf833627bc7cbaf785
Signed-off-by: Moiz Raja <moraja@cisco.com>
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoFix bug in PortConvertor#maskPortConfigFields 05/5705/2
Moiz Raja [Fri, 21 Mar 2014 00:22:53 +0000 (17:22 -0700)]
Fix bug in PortConvertor#maskPortConfigFields

This method was trying to set a passed reference to a different value and expecting to see it set in the caller
This won't work in java because java passes object references also as a value

See http://www.javaworld.com/article/2077424/learn-java/does-java-pass-by-reference-or-pass-by-value.html

Change-Id: I809f0887fb435b5ae2dd2b91304b8c20fa2adb17
Signed-off-by: Moiz Raja <moraja@cisco.com>
Signed-off-by: Michal Rehak <mirehak@cisco.com>
10 years agoRegister for SalPortService 04/5704/2
Moiz Raja [Fri, 21 Mar 2014 00:21:53 +0000 (17:21 -0700)]
Register for SalPortService

This is required so that someone can use the SalPortService as a consumer
- rebased onto merged change

Change-Id: I14069f04420285a783c16926543cf30ec85308b8
Signed-off-by: Moiz Raja <moraja@cisco.com>
Signed-off-by: Michal Rehak <mirehak@cisco.com>