Jozef Bacigal [Fri, 15 Apr 2016 11:13:57 +0000 (13:13 +0200)]
Bug-5523 RoleManager changes
- all functionality moved to role manager
- role context hold only information about status
- life cycle conductor defined as listener to role change
- able to register listener to start/stop master services
- timer for initialization device disabled
- added thrown exception when not able to write to transaction
- added junit tests
Change-Id: I741cde5c8161dde8c45f27c6a2e1f99617a613bf
Signed-off-by: Jozef Bacigal <jbacigal@cisco.com>
Abhijit Kumbhare [Thu, 5 May 2016 21:31:00 +0000 (21:31 +0000)]
Merge "Remove nexusproxy property"
Jozef Bacigal [Wed, 23 Mar 2016 14:04:35 +0000 (15:04 +0100)]
WIP: BUG-5464: scaling issue
- pull up table-features to flow-cabable-node
- fix adjanced logic
- remove augmentation of flow-capable-node (redundant table-features)
- fixed unit test
Change-Id: I798f87420330cfd71045eac3de61291a10abdafa
Signed-off-by: Michal Rehak <mirehak@cisco.com>
Signed-off-by: Jozef Bacigal <jbacigal@cisco.com>
Jozef Bacigal [Tue, 26 Apr 2016 07:11:23 +0000 (07:11 +0000)]
Merge "Derive distributuon/karaf pom from karaf-parent"
Abhijit Kumbhare [Mon, 25 Apr 2016 23:06:30 +0000 (23:06 +0000)]
Merge "Log at ERROR level when add/remove flow fail"
Alexis de Talhouët [Mon, 25 Apr 2016 19:38:06 +0000 (15:38 -0400)]
Log at ERROR level when add/remove flow fail
Change-Id: I78c14450e6ba235c99a74efff68c0a38ed4ce739
Signed-off-by: Alexis de Talhouët <adetalhouet@inocybe.com>
Abhijit Kumbhare [Mon, 25 Apr 2016 21:22:10 +0000 (21:22 +0000)]
Merge "Log at ERROR level when request context failed to close"
Alexis de Talhouët [Mon, 25 Apr 2016 20:03:14 +0000 (16:03 -0400)]
Log at ERROR level when request context failed to close
And also log the whole exception as the message can be null
RequestContextUtil | 190 - org.opendaylight.openflowplugin.impl - 0.3.0.SNAPSHOT | Request context wasn't closed. Exception message: null
Change-Id: Ib5d3f263db0baf83bc2d978a2c3e9f2cb1317b0c
Signed-off-by: Alexis de Talhouët <adetalhouet@inocybe.com>
Alexis de Talhouët [Mon, 25 Apr 2016 19:53:00 +0000 (15:53 -0400)]
Log at ERROR level when the queue isn't set...
... as this results in failing the request.
This helps debug, and I believe should be turned
at error level
Change-Id: Ib5152371f449bc81d69b216a2061db7eed624d5a
Signed-off-by: Alexis de Talhouët <adetalhouet@inocybe.com>
Tom Pantelis [Mon, 25 Apr 2016 14:41:16 +0000 (10:41 -0400)]
Derive distributuon/karaf pom from karaf-parent
This picks up the recent changes to the karaf distro from a common
place.
Signed-off-by: Tom Pantelis <tpanteli@brocade.com>
Jozef Bacigal [Mon, 25 Apr 2016 07:41:54 +0000 (07:41 +0000)]
Merge "Bug 5589 - Deprecate PortNumberCache"
Jozef Bacigal [Thu, 21 Apr 2016 12:53:18 +0000 (12:53 +0000)]
Merge changes from topic 'bug-5574'
* changes:
BUG-5574: added flat-bulk service registration
BUG-5574: flat batch proposal
BUG-5574: bulk meters proposal
BUG-5574: introduces sal-groups-batch service impl
BUG-5574: sal-flows-batch proposal
Michal Rehak [Mon, 4 Apr 2016 18:17:05 +0000 (20:17 +0200)]
BUG-5574: added flat-bulk service registration
- added registration/unregistation for flat-batch routed service
Change-Id: I95d005aeb12d1b817c643ffc11a6f5343361d5a4
Signed-off-by: Michal Rehak <mirehak@cisco.com>
(cherry picked from commit
b5ffe8525d85dcf92d38a1eab96e4f3d2836aad0)
Signed-off-by: andrej.leitner <anleitne@cisco.com>
Michal Rehak [Mon, 21 Mar 2016 17:02:39 +0000 (18:02 +0100)]
BUG-5574: flat batch proposal
- yang model
- implementation proposal
- flat batch util
- unit tests
Change-Id: I4406eaa19024348e63e2cc6d07039949ba8aa941
Signed-off-by: Michal Rehak <mirehak@cisco.com>
(cherry picked from commit
60ddc33574168296cf1b59f59a1cbb998cecfc36)
Signed-off-by: andrej.leitner <anleitne@cisco.com>
Michal Rehak [Wed, 16 Mar 2016 12:48:00 +0000 (13:48 +0100)]
BUG-5574: bulk meters proposal
- added bulk API implementation for groups
- with parametrized barrier-after support
- MeterUtil
- unit tests
Change-Id: I9146894ae0b1f630706b8aea6b85a13c8ce3364b
Signed-off-by: Michal Rehak <mirehak@cisco.com>
(cherry picked from commit
26fa4ba6f87c23e159b74526f6fb849f0ceb4783)
Signed-off-by: andrej.leitner <anleitne@cisco.com>
Tomas Slusny [Tue, 19 Apr 2016 12:08:09 +0000 (14:08 +0200)]
Bug 5589 - Deprecate PortNumberCache
- New class - Port to NodeConnectorRef converter
- New test - Conversion between port and NodeConnectorRef
- Methods in PortNumberCache are now deprecated
- Replaced use of above methods in PacketReceivedTranslator with new methods
- Modified tests for PacketReceivedTranslator to work with new methods
Change-Id: Icb04e3f8699d436b549ea670b1e3c42d2e057026
Signed-off-by: Tomas Slusny <tomas.slusny@pantheon.sk>
Michal Rehak [Thu, 7 Apr 2016 18:45:34 +0000 (20:45 +0200)]
Bug 5747 Introduced snapshot gathering marks
Added timestamp of statistics collection start and finish
and successful collection flag.
Change-Id: If00fbb7532f292d69602de7857db73fabc5fd49f
Signed-off-by: Andrej Leitner <anleitne@cisco.com>
Michal Rehak [Tue, 15 Mar 2016 17:54:41 +0000 (18:54 +0100)]
BUG-5574: introduces sal-groups-batch service impl
- added bulk API implementation for groups
- with parametrized barrier-after support
- added tests
Change-Id: If3e3d70d4ceb8eeb4a9902e2ea7f4bad26d9c934
Signed-off-by: Michal Rehak <mirehak@cisco.com>
(cherry picked from commit
134bd7a822a5a3d76ec82f3bd7ad6abf1a933a60)
Signed-off-by: andrej.leitner <anleitne@cisco.com>
Michal Rehak [Mon, 14 Mar 2016 15:44:21 +0000 (16:44 +0100)]
BUG-5574: sal-flows-batch proposal
- added common model for attaching barrier
- added bulk API implementation for flows
- with parametrized barrier-after support
- added tests
Change-Id: I4e941d3654dc142e1d1d64a2258e629016e95a0e
Signed-off-by: Michal Rehak <mirehak@cisco.com>
(cherry picked from commit
79e8e834f0cbbccfbff99e51e4ddd854dd812d19)
Signed-off-by: andrej.leitner <anleitne@cisco.com>
Michal Rehak [Tue, 19 Apr 2016 07:14:53 +0000 (09:14 +0200)]
Revert "Fix "Bug 5589 - Deprecate PortNumberCache""
This reverts commit
0cd77f8ef3c8045fba03dde20044a08ec44e29f5.
Signed-off-by: Michal Rehak <mirehak@cisco.com>
Change-Id: I852bf5dc979ef8a581b3d8326413451cd21cc83a
Luis Gomez [Mon, 18 Apr 2016 23:09:55 +0000 (16:09 -0700)]
Remove nexusproxy property
Change-Id: Ie6cc211e3e4a5b39ef6a09c23cb737eab6c57a96
Signed-off-by: Luis Gomez <ecelgp@gmail.com>
Jozef Bacigal [Mon, 18 Apr 2016 12:28:36 +0000 (12:28 +0000)]
Merge "Bug 5592 - DeviceContext -> DeviceState in *ToNotificationTransformer classes"
Jozef Bacigal [Mon, 18 Apr 2016 12:18:45 +0000 (12:18 +0000)]
Merge "Fix "Bug 5589 - Deprecate PortNumberCache" - New class that can convert DeviceState to NodeConnectorRef and NodeConnectorRef to port number. - 1 test that will verify if conversion between DeviceState and NodeConnectorRef is valid. - Made methods in PortNumberCache deprecated and replaced use of them in PacketReceivedTranslator with new methods. - Modified tests for PacketReceivedTranslator to work with new methods."
Tomas Slusny [Thu, 24 Mar 2016 13:08:02 +0000 (14:08 +0100)]
Fix "Bug 5589 - Deprecate PortNumberCache"
- New class that can convert DeviceState to NodeConnectorRef and NodeConnectorRef to port number.
- 1 test that will verify if conversion between DeviceState and NodeConnectorRef is valid.
- Made methods in PortNumberCache deprecated and replaced use of them in PacketReceivedTranslator with new methods.
- Modified tests for PacketReceivedTranslator to work with new methods.
Change-Id: I4960fd701ce0bf905208806c94e6ba1b7684a0cd
Signed-off-by: Tomas Slusny <tomas.slusny@pantheon.sk>
Thanh Ha [Wed, 16 Mar 2016 22:33:40 +0000 (18:33 -0400)]
Do not deploy the karaf artifact
The karaf artifact is only used for testing. It does not need to be
released to the world.
Change-Id: Ieca324babbf2cf8ac710808fdf40134981ed38e5
Signed-off-by: Thanh Ha <thanh.ha@linuxfoundation.org>
Kamal Rameshan [Wed, 16 Mar 2016 00:24:01 +0000 (17:24 -0700)]
Bug-5534:RoleService does not take care of rpc errors
Change-Id: Id65e597f5952b689dec176c29ae9e78284123ce2
Signed-off-by: Kamal Rameshan <kramesha@cisco.com>
Tomas Slusny [Thu, 31 Mar 2016 07:02:05 +0000 (09:02 +0200)]
Bug 5592
- DeviceContext -> DeviceState in *ToNotificationTransformer classes
Change-Id: I60cb6c908798115addc9ec10e8ed31e556a56f8d
Signed-off-by: Tomas Slusny <tomas.slusny@pantheon.sk>
Jozef Bacigal [Tue, 29 Mar 2016 08:30:12 +0000 (10:30 +0200)]
BUG 5523 RoleCtx changes
*add onDeviceContextLevelDown method
* simplify RoleContext close process
* add timer for clean (fix cluster sync. problems)
*simplify ownershipChange process
*RoleContext with one semaphor only
*add new method for prevent misunderstanding process in RoleContext
*capsulate Semafors locks in RoleContext
*add new state markers (isMainCandidateRegister, isTxCandidateRegistered,
getClusterRole)
Change-Id: I14438aafc8fe62264a62ccca0bd55a4520ffaeda
Signed-off-by: Jozef Bacigal <jbacigal@cisco.com>
Jozef Bacigal [Tue, 29 Mar 2016 08:18:40 +0000 (10:18 +0200)]
BUG 5523 Add onDeviceContextLevelDown for Statistics and Rpc Managers
*change Termination order in OFpluginProvider
*add onDeviceContextLevelDown methods
*add pair NodeId with Context for every manager
*only NodeId is keyProtection with not finished unregistration
for every shutting down order
Change-Id: Iec0defc9efeb276668404a0935114e523ab3fc10
Signed-off-by: Jozef Bacigal <jbacigal@cisco.com>
Jozef Bacigal [Tue, 29 Mar 2016 08:17:17 +0000 (10:17 +0200)]
BUG 5523 DeviceManager handler for connection and disconnection too
*Init and close DeviceCtx suite has contract with ConnectionCtx
so we'd like to have only one way for Clean DeviceCtx suite :
onDeviceDisconnected event for primary connectionContext.
*add timer for shuttinDown TxManager because Cluster
*clean TxManager close txChain handling
Change-Id: Ifba1c538b3369bd125908a114786d5bf92a15c7f
Signed-off-by: Jozef Bacigal <jbacigal@cisco.com>
Jozef Bacigal [Tue, 29 Mar 2016 08:15:12 +0000 (10:15 +0200)]
BUG 5523 - DeviceContext additional functionality
*add shutdonwConnection for use in OFP instead deviceCtx.close
*add DeviceContext states object
*add method for get actualState
*TransactionChainManager
*remove read DS by shuttingDown process
*remove close tx possibility
*Guard submitIsEnabled property (only active TxChainFactory
is real marker for set property to true)
Change-Id: I44789c04b819eed27a297372d6d819727e45f26c
Signed-off-by: Jozef Bacigal <jbacigal@cisco.com>
Jozef Bacigal [Tue, 29 Mar 2016 08:12:06 +0000 (10:12 +0200)]
BUG 5523 - API changes
Fixing closing process for DeviceContext
* rename existing Interfaces
* add method for registration TerminationHandler
* implement TerminationOrder
* fix tests
Change-Id: I69e5305ea45ab7bb34b5133980d2ca29b282cf4b
Signed-off-by: Jozef Bacigal <jbacigal@cisco.com>
(cherry picked from commit
e27f3af11009d085518be491e96b29bf7cf9f7de)
Kamal Rameshan [Thu, 10 Mar 2016 18:36:33 +0000 (10:36 -0800)]
Bug-5486:Device disconnected handers on disconnection get closed in no particular order
HashSets dont gaurantee any retrieval order. Further we dont need hashing as we just iterate through it.
Some log changes.
Change-Id: Ia31ce404e44e78bd0a3483e969da06119afe75f0
Signed-off-by: Kamal Rameshan <kramesha@cisco.com>
michal rehak [Thu, 24 Mar 2016 09:38:37 +0000 (09:38 +0000)]
Merge "Bug-5551:Audit of Connection Events"
Sai MarapaReddy [Sat, 19 Mar 2016 06:36:20 +0000 (23:36 -0700)]
Bug-5562 - Ipv4 L3-ArbitraryMask patch
Current code expects the Arbitrary Mask to be normalized from switch.
Example:- 3.3.3.3/255.0.255.0 to 3.0.3.0/255.0.255.0
Not all switches normalizes arbitrary bit mask in the flows when they are injected.
Example:- 3.3.3.3/255.0.255.0 to 3.3.3.3/255.0.255.0
This patch ensures that the operational data store is in congruent with the flows
that are shown on the switch. Patch is built on checking the mask & comparing the
normalized stores & stats flows which ensures Ipv4 L3-ArbitraryMask works independant
of how the switches normalizes the ip addresses when flows are pushed.
Conflicts:
openflowplugin/src/main/java/org/opendaylight/openflowplugin/openflow/md/core/sal/convertor/IpConversionUtil.java
openflowplugin/src/main/java/org/opendaylight/openflowplugin/openflow/md/core/sal/convertor/match/MatchConvertorImpl.java
Change-Id: Icfc6d4805c58f6483fc3ca0458b827b14c0a0cb1
Signed-off-by: Sai MarapaReddy <sai.marapareddy@gmail.com>
Kamal Rameshan [Thu, 17 Mar 2016 20:18:51 +0000 (13:18 -0700)]
Bug-5551:Audit of Connection Events
Change-Id: Ida3e632e80add4491e29e6c014a94a19c7ed9909
Signed-off-by: Kamal Rameshan <kramesha@cisco.com>
Anil Vishnoi [Mon, 21 Mar 2016 22:13:19 +0000 (22:13 +0000)]
Merge "BUG-5489 Group Delete not working with ovs2.5"
michal rehak [Tue, 15 Mar 2016 09:23:12 +0000 (09:23 +0000)]
Merge "Fix Mockito Generics"
Jozef Bacigal [Fri, 11 Mar 2016 14:20:43 +0000 (15:20 +0100)]
Bug 5521 - flows reconciliation
Fix missing flows by initialization/reconciliation
Change-Id: I2b3d892cb53abc1603db45c7937ae98ac4a36a79
Signed-off-by: Jozef Bacigal <jbacigal@cisco.com>
andrej.leitner [Mon, 14 Mar 2016 14:54:06 +0000 (15:54 +0100)]
Fix Mockito Generics
*fix Generic for mockito matcher
*fix testGroupModConvertSortedBuckets
Change-Id: I8148c94cf572440a1cb9be7a0d962c8501cb3f42
Signed-off-by: andrej.leitner <anleitne@cisco.com>
michal rehak [Fri, 11 Mar 2016 14:42:51 +0000 (14:42 +0000)]
Merge "Correction classes names in loggers."
michal rehak [Fri, 11 Mar 2016 14:40:25 +0000 (14:40 +0000)]
Merge "Post "Clustering optimization" updates"
Shuva Jyoti Kar [Thu, 10 Mar 2016 09:43:19 +0000 (15:13 +0530)]
BUG-5489 Group Delete not working with ovs2.5
Change-Id: Ie63c46e5d9def2c89e5e22e6d5ca56678b43424d
Signed-off-by: Shuva Jyoti Kar <shuva.jyoti.kar@ericsson.com>
Hideyuki Tai [Fri, 26 Feb 2016 14:27:39 +0000 (06:27 -0800)]
Suppressed INFO log messages.
According to the Logging Best Practices in the OpenDaylight wiki,
messages at INFO level should be for operators and administrators.
However, some log messages for developers inappropriately used the INFO
level. This patch changed the log level of the messages from INFO to
TRACE.
Change-Id: I7e666149a7ea3dbdcddf7f157a5010d4fb5aee63
Signed-off-by: Hideyuki Tai <Hideyuki.Tai@necam.com>
(cherry picked from commit
a7b507c778ead7c2e316b1acadd55653dcb2dcaa)
Kamal Rameshan [Wed, 9 Mar 2016 17:33:00 +0000 (17:33 +0000)]
Merge "Bug:3026 - Echo response timeout needs be exported to configuration"
Jozef Bacigal [Wed, 9 Mar 2016 11:24:40 +0000 (12:24 +0100)]
Correction classes names in loggers.
- this are probably copy paste errors
Change-Id: I7ac309b218fc8250edd4ba2e65e2174fab2babcb
Signed-off-by: Jozef Bacigal <jbacigal@cisco.com>
Jozef Bacigal [Wed, 2 Mar 2016 11:25:23 +0000 (12:25 +0100)]
Post "Clustering optimization" updates
3/2/16
- small updates after clustering optimizations
- grammar policy for method names, class name and logs
- clean ups
3/3/16
- deviceState class, set SLAVE after init
- deviceStateTest updates
Change-Id: Ia38b125cb974fc15a49ce288f9880278d4db6288
Signed-off-by: Jozef Bacigal <jbacigal@cisco.com>
michal rehak [Wed, 9 Mar 2016 08:55:15 +0000 (08:55 +0000)]
Merge "Quick fix RPCs and DevicCtx.close"
Kamal Rameshan [Fri, 26 Feb 2016 23:39:50 +0000 (15:39 -0800)]
Bug:3026 - Echo response timeout needs be exported to configuration
Takes care of Bug-4677 : Li: Knob to configure Barrier duration
Also incorporated Jozef bacigal's hot fix in stable
Change-Id: I42697dbbb93cc40c1547b92e9665851e7f5f0b98
Signed-off-by: Kamal Rameshan <kramesha@cisco.com>
(cherry picked from commit
e22904723888cfe73fff63e280e4c4c5af51b910)
Sai MarapaReddy [Wed, 10 Feb 2016 23:26:04 +0000 (15:26 -0800)]
Bug-5119 - Adding support for L3 address arbitraty bit mask with Unittests coverage & indentation & addressed latest comments & rebased
Author: Sai MarapaReddy <sai.marapareddy@gmail.com>
Commit: Sai MarapaReddy <sai.marapareddy@gmail.com>
Change-Id: I26ef6aef4ffdbc5047c5a2282fbdc42d7bbfcd27
Signed-off-by: Sai MarapaReddy <sai.marapareddy@gmail.com>
Jozef Bacigal [Fri, 4 Mar 2016 15:11:06 +0000 (16:11 +0100)]
Quick fix RPCs and DevicCtx.close
Change-Id: I1cd92d91e1025ae3aee00418b86912f2e86580b4
Signed-off-by: Vaclav Demcak <vdemcak@cisco.com>
Signed-off-by: Jozef Bacigal <jbacigal@cisco.com>
Anil Vishnoi [Tue, 8 Mar 2016 03:31:07 +0000 (19:31 -0800)]
Bug 5404 - He plugin: Meter stats not reported the first time switch connects to controller
Change-Id: I970ed9bd02fd20054d552e3a6127f9766cd8ab92
Signed-off-by: Anil Vishnoi <vishnoianil@gmail.com>
(cherry picked from commit
60fd94868c75a9d49eb968f73a1ed337a38ed727)
Robert Varga [Fri, 4 Mar 2016 14:36:41 +0000 (15:36 +0100)]
Do not retain DeviceState
TransactionChainManager only needs the instance identifier (or Node Id,
available from there), do not retain the entire state.
Change-Id: I93ff1fc08ebf469a907493452e9e771d59dd849f
Signed-off-by: Robert Varga <rovarga@cisco.com>
Robert Varga [Fri, 4 Mar 2016 14:16:52 +0000 (15:16 +0100)]
Do not leak internal state
The state transition enum should not be visible from outside, hide it.
Change-Id: I2b3fa8af7829509b5a968eb952a53c829e189aaa
Signed-off-by: Robert Varga <rovarga@cisco.com>
Robert Varga [Fri, 4 Mar 2016 13:41:27 +0000 (14:41 +0100)]
Do not lose thread on close
We need to make sure we shutdown the spypool on close.
Change-Id: I07d7144b9ca29e35a6dc24231b525c64c8e3891a
Signed-off-by: Robert Varga <rovarga@cisco.com>
Robert Varga [Fri, 19 Feb 2016 14:37:42 +0000 (15:37 +0100)]
Use IetInetUtil conversion
Instead of home-grown covnersion, use the utilities provided.
Change-Id: I9a44b8eb87f297c8d456182b7080fd9f447802c3
Signed-off-by: Robert Varga <rovarga@cisco.com>
Robert Varga [Fri, 4 Mar 2016 13:31:08 +0000 (14:31 +0100)]
Fix a guaranteed NPE
If we fail to decode a packet, we will end up causing a NPE in
statistics. Use the packetIn class for that instead.
Change-Id: Ib53ca939ee4a08319c2595d560bfdd99c013eba2
Signed-off-by: Robert Varga <rovarga@cisco.com>
Robert Varga [Fri, 4 Mar 2016 13:34:01 +0000 (14:34 +0100)]
Do not override java source/target
odlparent already provides correct definitions (which are Java 8), so
use those.
Change-Id: Ib6d19bbe1aff0930fd9f0151e44a643fa7854e38
Signed-off-by: Robert Varga <rovarga@cisco.com>
michal rehak [Wed, 2 Mar 2016 16:46:59 +0000 (16:46 +0000)]
Merge "Bug 4957 Cleaning code and fixing issues from manual tests"
michal rehak [Wed, 2 Mar 2016 16:46:56 +0000 (16:46 +0000)]
Merge "Bug 4957: OF Role processing changes"
michal rehak [Wed, 2 Mar 2016 16:46:53 +0000 (16:46 +0000)]
Merge "Bug-4957 Cluster Role change fix"
michal rehak [Wed, 2 Mar 2016 16:27:38 +0000 (16:27 +0000)]
Merge "Bug-4957: Make async operational DS Read"
michal rehak [Wed, 2 Mar 2016 16:27:35 +0000 (16:27 +0000)]
Merge "Bug-4957 Fix SalRoleService blocking call"
michal rehak [Wed, 2 Mar 2016 16:27:32 +0000 (16:27 +0000)]
Merge "Bug-4957: Double candidate, close processing"
michal rehak [Wed, 2 Mar 2016 16:27:28 +0000 (16:27 +0000)]
Merge "Bug-4957: Double candidate, introduced roleCtx states"
michal rehak [Wed, 2 Mar 2016 16:27:25 +0000 (16:27 +0000)]
Merge "Bug-4957: Double candidate, clean up"
michal rehak [Wed, 2 Mar 2016 16:27:22 +0000 (16:27 +0000)]
Merge "Bug-4957 Double candidate Onwership initial commit"
michal rehak [Wed, 2 Mar 2016 16:27:19 +0000 (16:27 +0000)]
Merge "Bug 4957 Fix Cluster Role change for Statistics Polling"
michal rehak [Wed, 2 Mar 2016 16:27:15 +0000 (16:27 +0000)]
Merge "Bug 4957 Fix Cluster Role change for RPC registration"
Vaclav Demcak [Thu, 25 Feb 2016 13:15:04 +0000 (14:15 +0100)]
Bug 4957 Cleaning code and fixing issues from manual tests
*Initialization Device info collecting
-Device Info collecting is async function and it could finish
with unactive connection (bug stat collecting always return true)
so we need to check actual state before we finish MASTER
roleChange.
-We don't wish to commit any data and we are able to close
NEW Transaction. So we try to do it.
*SalRoleServiceImpl changes
-add result FutureCallback for release semafor guard
-add check for cancel for every loop
-fix return failed future for InterruptedException
*Change close method for all Managers
-Fix close method in RoleManager we not suspend TxEntity
-Use consumingIterator for cleaning manager's contexts
-Use concumingIterator for close in RpcContext
*Additional clean
-clean RoleContext for TxLockOwned
-fix not propagated Excepton from DeviceContext#onDeviceContextLevelUp
-clean RoleManager
-add check for Connection state for statistics
-fix tests
Change-Id: I04d19e4c0a0e87e4158cd2ca25da32fe15cfef22
Signed-off-by: Vaclav Demcak <vdemcak@cisco.com>
Signed-off-by: Jozef Bacigal <jbacigal@cisco.com>
michal rehak [Tue, 1 Mar 2016 14:00:05 +0000 (14:00 +0000)]
Merge "Pre-compute ipv4 masks"
Vaclav Demcak [Mon, 22 Feb 2016 14:52:38 +0000 (15:52 +0100)]
Bug 4957: OF Role processing changes
* change init order to CM -> DM -> SM -> RPC -> Role -> DM
* set default Role setup to Slave by init phase
* Fix order of verify for put to contextsList before call
addDeviceCotextCloseHander for every CtxManagers
- prevention for unexpected not finished Contexts
* add more precise logging
* fix tests
Change-Id: I9492a96d0725823d7901e98af81e0e3961e9f414
Signed-off-by: Vaclav Demcak <vdemcak@cisco.com>
Signed-off-by: Jozef Bacigal <jbacigal@cisco.com>
Signed-off-by: Andrej Leitner <anleitne@cisco.com>
Vaclav Demcak [Mon, 22 Feb 2016 14:52:38 +0000 (15:52 +0100)]
Bug-4957 Cluster Role change fix
* TxManager shutdown fail has to clean DS (put empty Node)
* DeviceCtx onRoleChange needs call Device Initialization
functionality but we need to know when we are able to
put fresh data in to Oper DS and when we could jump
write functionality
* DeviceCtx contract change for method OnClusterRoleChange
Note: Cluster Master to Slave change could finish during
initialization phase or TxManager could fail in submitting
last transaction state. So we needs a marker for "something
unexpected is happend" and we have to put fresh data to
Operational DataStore
Change-Id: I18e8c0d75e2e5cf195d9a98c1e096fb5517fab72
Signed-off-by: Jozef Bacigal <jbacigal@cisco.com>
Signed-off-by: Vaclav Demcak <vdemcak@cisco.com>
Anil Vishnoi [Sun, 28 Feb 2016 03:43:22 +0000 (19:43 -0800)]
Bug 2951 - ipv6 addresses containing zeros in input flow should be handled in operational flows
Change-Id: I60955de3c9c28825de46520f8ccdd272ca62838d
Signed-off-by: Anil Vishnoi <vishnoianil@gmail.com>
Vaclav Demcak [Sat, 20 Feb 2016 13:07:14 +0000 (14:07 +0100)]
Bug-4957: Make async operational DS Read
* don't block netty thread for reading oper. DS for clean
Tables in collecting FlowStat
* fix tests
Change-Id: I32df820c69f07e7bc00b71a87c4c0c07d43c9312
Signed-off-by: Vaclav Demcak <vdemcak@cisco.com>
Vaclav Demcak [Sat, 20 Feb 2016 12:13:15 +0000 (13:13 +0100)]
Bug-4957 Fix SalRoleService blocking call
- SalRoleService invoke blocking calls for GenerationId
and for change Role. Both with 10 sec wait. Block any
threads for this time is danger in any way. But archit.
change replace Akka / DS threads with netty and we
get deadlock for netty thread.
Change-Id: If801baf8698977251ebef88d145b6e03d6a80e83
Signed-off-by: Jozef Bacigal <jbacigal@cisco.com>
Signed-off-by: Vaclav Demcak <vdemcak@cisco.com>
Vaclav Demcak [Tue, 2 Feb 2016 10:24:34 +0000 (11:24 +0100)]
Bug-4957: Double candidate, close processing
- close has one extra state: Master unregistred NodeEntity and
some another ClusterNode win election (but it is in close process
too) so we are not able to propagate role change correctly and
we could delegate possible Delete Node from correct ClusterNode.
Change-Id: I5e8235686632699b7d827735c75eb0348fadc36c
Signed-off-by: Jozef Bacigal <jbacigal@cisco.com>
Vaclav Demcak [Tue, 2 Feb 2016 14:09:45 +0000 (15:09 +0100)]
Bug-4957: Double candidate, introduced roleCtx states
- introduced roleContext states
- organized methods used by roleContext lifecycle state
transitions to use uniformed pattern with
exceptions if behavior differs for given combination
of state and role
Change-Id: I53531fd6d168c38b3fdae457d432dacfdf5e5a9c
Signed-off-by: Jozef Bacigal <jbacigal@cisco.com>
Vaclav Demcak [Tue, 2 Feb 2016 12:52:21 +0000 (13:52 +0100)]
Bug-4957: Double candidate, clean up
- getting rid of twisted futures
- simplifying code paths
- decreasing amount of failure handling places
upon the same code path
Change-Id: I4e54f4644a919538aa4297f600a835004db7e3b6
Signed-off-by: Jozef Bacigal <jbacigal@cisco.com>
Jozef Bacigal [Wed, 27 Jan 2016 22:17:12 +0000 (23:17 +0100)]
Bug-4957 Double candidate Onwership initial commit
Note: Doulbe candidate Ownership is fixing missing
TxChain management from Distributed DataStore.
-Add init suport for TxEntity processing
-fix ownershipChange find correct RoleCtx by entity
Change-Id: I0ef898483c33e80a59e19e0e3c1b83b628cf88af
Signed-off-by: Jozef Bacigal <jbacigal@cisco.com>
Jozef Bacigal [Wed, 27 Jan 2016 22:17:12 +0000 (23:17 +0100)]
Bug 4957 Fix Cluster Role change for Statistics Polling
-polling statistics can be done for every Role, what we
realy want to fix, is check an actual Role from DeviceState
before polling.
Change-Id: I6cedfab0a47022632178af71c54c7c59516b5f83
Signed-off-by: Jozef Bacigal <jbacigal@cisco.com>
Jozef Bacigal [Wed, 27 Jan 2016 22:17:12 +0000 (23:17 +0100)]
Bug 4957 Fix Cluster Role change for RPC registration
-we have to registred RPC for LEADER and unregistred
RPC for FOLLOWER for every role change.
Change-Id: I53e862dd46cb0feb4ba1221345ebe0d542df1097
Signed-off-by: Jozef Bacigal <jbacigal@cisco.com>
Robert Varga [Tue, 23 Feb 2016 21:57:20 +0000 (22:57 +0100)]
Pre-compute ipv4 masks
Instead of computings masks over and over again, store them as
pre-computed arrays and clone them on access. Also make methods static.
Change-Id: I10c379fa0d5aaf8dd15f8d51d7de3ece1ebd48d3
Signed-off-by: Robert Varga <rovarga@cisco.com>
Abhijit Kumbhare [Thu, 11 Feb 2016 23:08:09 +0000 (23:08 +0000)]
Merge "BUG-5020 Handling exception while submission"
Shuva Jyoti Kar [Thu, 11 Feb 2016 20:03:12 +0000 (01:33 +0530)]
BUG-5020 Handling exception while submission
Change-Id: I755c0f85d25fd7f30b77c9ea6765c5e0c185153b
Signed-off-by: Shuva Jyoti Kar <shuva.jyoti.kar@ericsson.com>
Signed-off-by: Anil Vishnoi <vishnoianil@gmail.com>
Anil Vishnoi [Thu, 11 Feb 2016 19:32:57 +0000 (11:32 -0800)]
Bug 5327 - LLDP spoof warning in 3-node cluster
Change-Id: I0bb970d9de99a162dd5f09ae8376151755da617d
Signed-off-by: Anil Vishnoi <vishnoianil@gmail.com>
Anil Vishnoi [Thu, 11 Feb 2016 18:53:07 +0000 (18:53 +0000)]
Merge "BUG-4099: Groups pointing to ports"
Anil Vishnoi [Thu, 11 Feb 2016 06:31:30 +0000 (06:31 +0000)]
Merge "Bug 5050: Correctly deregister routed RPC."
michal rehak [Tue, 9 Feb 2016 18:16:27 +0000 (18:16 +0000)]
Merge "Bug 4957 Add async shuttingDown method for TransactionChainManager"
michal rehak [Tue, 9 Feb 2016 18:16:24 +0000 (18:16 +0000)]
Merge "Bug 4957 Fix Last No Master RoleCtx cleaning"
michal rehak [Tue, 9 Feb 2016 18:16:21 +0000 (18:16 +0000)]
Merge "Bug 4957 Fix blocking call to Init RoleGet"
michal rehak [Tue, 9 Feb 2016 18:16:17 +0000 (18:16 +0000)]
Merge "Bug 4957 OpenflowPluginProviderImpl.close() fixing"
michal rehak [Tue, 9 Feb 2016 18:16:13 +0000 (18:16 +0000)]
Merge "Bug 4957 Fix methods for change TxChainManager"
michal rehak [Tue, 9 Feb 2016 18:16:09 +0000 (18:16 +0000)]
Merge "Bug 4957 Clean unnecessary code"
michal rehak [Tue, 9 Feb 2016 18:15:58 +0000 (18:15 +0000)]
Merge "Bug 4957 Wiring new TxChainManager lifecycle"
michal rehak [Tue, 9 Feb 2016 09:20:42 +0000 (09:20 +0000)]
Merge "Bug 4957 Role lifecycle support for TxChainManager in DeviceContext"
michal rehak [Tue, 9 Feb 2016 09:20:36 +0000 (09:20 +0000)]
Merge "Bug 4957 Fix RoleChangeLifecyle (init, close, life)"
michal rehak [Tue, 9 Feb 2016 09:20:32 +0000 (09:20 +0000)]
Merge "Bug 4957 Fixing DeviceCtx lifeCycle"
michal rehak [Tue, 9 Feb 2016 09:20:27 +0000 (09:20 +0000)]
Merge "Bug 4957 TxChainManager lifecycle startup cleaning"
michal rehak [Tue, 9 Feb 2016 09:20:22 +0000 (09:20 +0000)]
Merge "Bug 4957 RoleContext updated with initialization"