Ed Warnicke [Wed, 28 Aug 2013 05:07:22 +0000 (22:07 -0700)]
Added an archetype odl-model-project
Change-Id: I4b6c14f34a4657a21626285475628c7ab72738d5
Signed-off-by: Ed Warnicke <eaw@cisco.com>
Giovanni Meo [Tue, 27 Aug 2013 13:43:00 +0000 (13:43 +0000)]
Merge changes I5809ac06,I4362cd4e
* changes:
Added Intellij files to gitignore
Fixed typo and remove unecessary java.lang import
Alissa Bonas [Tue, 27 Aug 2013 12:43:32 +0000 (15:43 +0300)]
Added Intellij files to gitignore
Added IntelliJ IDE specific files extensions
to .gitignore.
Change-Id: I5809ac06ab24215c593313ea48a8cb5a937f7593
Signed-off-by: Alissa Bonas <abonas@redhat.com>
Alissa Bonas [Tue, 27 Aug 2013 10:29:45 +0000 (13:29 +0300)]
Fixed typo and remove unecessary java.lang import
Fixed typo in comment and remove unecessary java.lang import
(as proposed by Sonar report)
Change-Id: I4362cd4e25965c9a5c9322a5a25e189241599c18
Signed-off-by: Alissa Bonas <abonas@redhat.com>
Giovanni Meo [Mon, 26 Aug 2013 10:39:05 +0000 (12:39 +0200)]
Allow controller to be managed via JMX remotely
- In order to inspect the controller remotely JMX comes very handy
especially in clustered environment, this patch allow to start the
controller and to enable JMX remotely, this is still only the unsecure
version, for secure version a subsequent patch will be provided
Change-Id: I47c8651f19e4207b37b064161bb88476f3e31d0b
Signed-off-by: Giovanni Meo <gmeo@cisco.com>
Madhu Venugopal [Tue, 27 Aug 2013 03:08:12 +0000 (20:08 -0700)]
Fixing a Northbound IT test breakage due to a recent NB change
Change-Id: I1699413cf7fa6f02b713f72a669c2fc3115c59cd
Signed-off-by: Madhu Venugopal <vmadhu@cisco.com>
Madhu Venugopal [Tue, 27 Aug 2013 02:26:17 +0000 (02:26 +0000)]
Merge "Northbound cleanup for Flow programmer"
Madhu Venugopal [Tue, 27 Aug 2013 02:19:01 +0000 (02:19 +0000)]
Merge "Northbound cleanup for static routing"
Pramila Singh [Mon, 26 Aug 2013 17:08:22 +0000 (10:08 -0700)]
Northbound cleanup for Flow programmer
Change-Id: I687b3558a4f19a17a0a9ad9247bb25327f4bb676
Signed-off-by: Pramila Singh <pramisin@cisco.com>
Madhu Venugopal [Tue, 27 Aug 2013 02:14:53 +0000 (02:14 +0000)]
Merge "Cleaup Statistics northbound and add exmaples"
Prasanth Pallamreddy [Mon, 26 Aug 2013 23:45:45 +0000 (16:45 -0700)]
Northbound cleanup for static routing
Change-Id: I252c0338780be7eb80d577bbcdfe8455aa171c8c
Signed-off-by: Prasanth Pallamreddy <ppallamr@cisco.com>
Yevgeny Khodorkovsky [Mon, 26 Aug 2013 22:48:15 +0000 (15:48 -0700)]
Cleaup Statistics northbound and add exmaples
- Clean up stats northbound to align with common REST conventions,
add more descriptive javadoc with examples
Change-Id: I9508e72b9f62dc576ab4a569f699b75e7bfb29a9
Signed-off-by: Yevgeny Khodorkovsky <ykhodork@cisco.com>
Jason Ye [Mon, 26 Aug 2013 17:56:22 +0000 (10:56 -0700)]
OpenFLow plugin & SAL API documentation
Change-Id: I0947e48f4ab52cf61e84973dcc808b10045e3382
Signed-off-by: Jason Ye <yisye@cisco.com>
Madhu Venugopal [Mon, 26 Aug 2013 21:14:29 +0000 (21:14 +0000)]
Merge "Fix to remove camel case for Northbound API"
Pramila Singh [Mon, 26 Aug 2013 16:50:34 +0000 (09:50 -0700)]
Fix to remove camel case for Northbound API
Change-Id: I2d13e9d53b0d73504c15d32a57207110d70afb3a
Signed-off-by: Pramila Singh <pramisin@cisco.com>
lakshya [Mon, 26 Aug 2013 18:44:50 +0000 (11:44 -0700)]
Fix to show Node name if present when adding ports to Subnet Gateway on the devices tab.
A minor change from Mac to MAC in nodes learnt popout.
Change-Id: I600a8437d757b72bdaf8f0e5d8ffa624a08c4d2b
Signed-off-by: lakshya <lakshya@cisco.com>
Pramila Singh [Mon, 26 Aug 2013 20:36:27 +0000 (13:36 -0700)]
Fix for NorthboundIT test failure
Change-Id: I0d3c2216bf2cdb9ba23b186307ad49b8f2b36184
Signed-off-by: Pramila Singh <pramisin@cisco.com>
Madhu Venugopal [Mon, 26 Aug 2013 19:33:07 +0000 (19:33 +0000)]
Merge "HostTracker immediately ages out hosts on coordinator"
Ed Warnicke [Mon, 26 Aug 2013 19:19:18 +0000 (19:19 +0000)]
Merge "Integration of MDSAL into distribution"
Madhu Venugopal [Mon, 26 Aug 2013 18:52:42 +0000 (18:52 +0000)]
Merge "Northbound cleanup for Switch Manager"
Tony Tkacik [Mon, 26 Aug 2013 17:27:57 +0000 (12:27 -0500)]
Integration of MDSAL into distribution
- Fixed Class Loader bug in runtime code generator
Change-Id: I8282b16247a9d80d74f8646d2ef56f364c2d349f
Signed-off-by: Tony Tkacik <ttkacik@cisco.com>
Giovanni Meo [Mon, 26 Aug 2013 17:53:20 +0000 (19:53 +0200)]
HostTracker immediately ages out hosts on coordinator
- Let say an Host 1.1.4.1 is learnt by CTRL2 and CTRL1 is doing
cleanup of the host tracker database, the CTRL1 was cleaning
immediately the hosts learnt on other controllers because the host
count was not serialized appropriately
Change-Id: I1a0542c5e44cf3fc3f33d6fe95d129e83825cc8c
Signed-off-by: Giovanni Meo <gmeo@cisco.com>
Andrew Kim [Mon, 26 Aug 2013 17:11:15 +0000 (17:11 +0000)]
Merge "Fixing a bug to show node name if present instead of node id while adding SPAN ports on the devices page."
Pramila Singh [Mon, 26 Aug 2013 17:01:46 +0000 (10:01 -0700)]
Northbound cleanup for Switch Manager
Change-Id: If312264096eeefd0f32c2556fd8387d8a3d7a3de
Signed-off-by: Pramila Singh <pramisin@cisco.com>
lakshya [Mon, 26 Aug 2013 06:25:28 +0000 (23:25 -0700)]
Fixing a bug to show node name if present instead of node id while adding
SPAN ports on the devices page.
Change-Id: Id9745acec14bb83dd36f2b35f2811cf1706458e1
Signed-off-by: lakshya <lakshya@cisco.com>
Madhu Venugopal [Mon, 26 Aug 2013 00:34:05 +0000 (17:34 -0700)]
Fixing a minor bug in Cache replace code in Connection Manager.
Change-Id: Ic885dc0890860abb23b067458353ba16d21b275b
Signed-off-by: Madhu Venugopal <vmadhu@cisco.com>
Alessandro Boch [Sat, 24 Aug 2013 14:55:22 +0000 (14:55 +0000)]
Merge "Fix Connection manager to retrieve inventory when is up"
Chi-Vien Ly [Sat, 24 Aug 2013 03:30:22 +0000 (03:30 +0000)]
Merge "Fix Match intersection logic"
Alessandro Boch [Sat, 24 Aug 2013 02:43:32 +0000 (19:43 -0700)]
Fix Match intersection logic
- Change in Match.java: Empty match represents all packets (universal set) while empty set is the null match.
So, in case of no intersection, return a null Match.
intersect() and getIntersection() optimized in case of universal and empty sets
- Changes in MatchTest.java to junit the above changes
- Change in FlowEntry.java: Match.mergewithFilter() is now returning a new Match instance
Change-Id: Idd3ff80d2fbcc11c3f83757268806596b7b2fc50
Signed-off-by: Alessandro Boch <aboch@cisco.com>
Jason Ye [Fri, 23 Aug 2013 20:49:36 +0000 (13:49 -0700)]
Northbound cleanup for topology manager
Change-Id: I49862df3d3ac616127c82e7622e7c158ca58849e
Signed-off-by: Jason Ye <yisye@cisco.com>
Alessandro Boch [Sat, 24 Aug 2013 00:20:20 +0000 (00:20 +0000)]
Merge "Fixed run.sh to autodetect JAVA_HOME on Mac OS X"
Jason Ye [Fri, 23 Aug 2013 20:43:09 +0000 (20:43 +0000)]
Merge "Add intersection methods to Match"
Andrew Kim [Fri, 23 Aug 2013 17:43:42 +0000 (17:43 +0000)]
Merge "Delete Multiple flows in a single click"
Andrew Kim [Fri, 23 Aug 2013 17:37:08 +0000 (17:37 +0000)]
Merge "Change nodeconnector string shown on the troubleshoot tab when clicked on ports to show node name if it is available and node id otherwise."
Sapan Shah [Fri, 23 Aug 2013 07:43:04 +0000 (00:43 -0700)]
Delete Multiple flows in a single click
Change-Id: I0b61799e66e05aed2b7d89cb885f7b29fa032663
Signed-off-by: Sapan Shah <sapshah@cisco.com>
Hideyuki [Thu, 22 Aug 2013 15:46:05 +0000 (11:46 -0400)]
Enabled classes not in org.opendaylight.controller package to output log.
Change-Id: I85819272500c1f4ed3a5eb3f6580b6aa6aac75d9
Signed-off-by: Hideyuki <h-tai@cd.jp.nec.com>
lakshya [Fri, 23 Aug 2013 00:42:41 +0000 (17:42 -0700)]
Change nodeconnector string shown on the troubleshoot tab when clicked on ports
to show node name if it is available and node id otherwise.
Change-Id: I7ccb73ac56f963feaf31380ddbd87417c905804e
Signed-off-by: lakshya <lakshya@cisco.com>
Ed Warnicke [Thu, 22 Aug 2013 19:24:39 +0000 (14:24 -0500)]
Fixed run.sh to autodetect JAVA_HOME on Mac OS X
Change-Id: I3043d19a90f90258883a870d65acd8c26698e3e1
Signed-off-by: Ed Warnicke <eaw@cisco.com>
Alessandro Boch [Thu, 22 Aug 2013 16:57:50 +0000 (09:57 -0700)]
Add intersection methods to Match
- Add method to test and get intersection of two Match objects and correspondent Junit
- Fixed bug in NetUtils.inetAddressConflict()
- Made Match.toString() less verbose
Change-Id: I46cb93f7fc1a46aaaf3a3c341477166d8c1c8822
Signed-off-by: Alessandro Boch <aboch@cisco.com>
Giovanni Meo [Thu, 22 Aug 2013 20:53:26 +0000 (22:53 +0200)]
Fix Connection manager to retrieve inventory when is up
When ConnectionManager comes up, it's very much possible that may have
missed some inventory updates, so it needs to retrieve them to make
sure holes are covered, this is seen during integration test with the
stub protocol plugin, hence also make sure stub protocol plugin
provides the necessary hooks.
Change-Id: I41d5c71dd70aad909f7a75e0f596ec69a212400c
Signed-off-by: Giovanni Meo <gmeo@cisco.com>
Tony Tkacik [Thu, 22 Aug 2013 15:23:37 +0000 (17:23 +0200)]
Added startImpl to provide BundleContext to providers
Change-Id: Id92b7c469ffa6da22745854fc5d58862f232d6a2
Signed-off-by: Tony Tkacik <ttkacik@cisco.com>
Tony Tkacik [Thu, 22 Aug 2013 09:42:35 +0000 (11:42 +0200)]
Fixed reference to missing file.
Change-Id: Ib87d4a207c8604d7fefc9b7587434121768b6f1c
Signed-off-by: Tony Tkacik <ttkacik@cisco.com>
Chi-Vien Ly [Thu, 22 Aug 2013 04:01:54 +0000 (04:01 +0000)]
Merge "Added distributionManagement section to md-sal parent pom."
Chi-Vien Ly [Thu, 22 Aug 2013 03:53:25 +0000 (03:53 +0000)]
Merge "1. Controller switchEvents queue should be priority based. The queue holds switch up, down, error and message events. Assign higher priority to switch up/down events. 2. When switch is going to disconnect, SwitchHandler can clean up some internal states to speed up the process in order to avoid unnecessary messages backing up in the queue. 3. In SwitchManager, don't send additional port down notifications after node down event is received. 4. Fixed an issue where links are not recovered when the switch is reconnected to the controller. 5. Fixed TLS exception@java.security.ProviderException."
Ed Warnicke [Thu, 22 Aug 2013 03:37:35 +0000 (22:37 -0500)]
Added distributionManagement section to md-sal parent pom.
Change-Id: I553acfa7476e88a389002b3a3a864ded9e3ba3ac
Signed-off-by: Ed Warnicke <eaw@cisco.com>
Andrew Kim [Wed, 21 Aug 2013 23:52:12 +0000 (23:52 +0000)]
Merge "Minor correction to UI element"
Andrew Kim [Wed, 21 Aug 2013 23:45:57 +0000 (23:45 +0000)]
Merge "Enhancements to common.northbound"
Andrew Kim [Wed, 21 Aug 2013 03:42:10 +0000 (22:42 -0500)]
Minor correction to UI element
Change-Id: I4fab95bdf8d9fa73c53c9ee6d7dd9a52961cae6e
Signed-off-by: Andrew Kim <andrekim@cisco.com>
Alessandro Boch [Wed, 21 Aug 2013 23:09:20 +0000 (23:09 +0000)]
Merge "Frm to install/uninstall flow when port is added/removed from slice"
lakshya [Wed, 21 Aug 2013 22:44:27 +0000 (15:44 -0700)]
Fix for JS error on Flows tab.
This fixes the JS error "TypeError: $span.data(...) is null" on the flows tab by introducing a null check.
Change-Id: Ibbf2141bdca83b445e0d5d162cfe8a3e9a1fe9a0
Signed-off-by: lakshya <lakshya@cisco.com>
Jason Ye [Wed, 21 Aug 2013 20:44:44 +0000 (13:44 -0700)]
1. Controller switchEvents queue should be priority based. The queue holds switch up, down, error and message events. Assign higher priority to switch up/down events.
2. When switch is going to disconnect, SwitchHandler can clean up some internal states to speed up the process in order to avoid unnecessary messages backing up in the queue.
3. In SwitchManager, don't send additional port down notifications after node down event is received.
4. Fixed an issue where links are not recovered when the switch is reconnected to the controller.
5. Fixed TLS exception@java.security.ProviderException.
Change-Id: I4ba190a59f8d046df999401572d3c6822b15429b
Signed-off-by: Jason Ye <yisye@cisco.com>
lakshya [Wed, 21 Aug 2013 20:20:43 +0000 (13:20 -0700)]
Increase the number of rows per page to 10 for a popout.
Change-Id: Ia516bc3f76b7e03ecd302c006cda2bc8662fe8f5
Signed-off-by: lakshya <lakshya@cisco.com>
Diti Bhatia [Wed, 21 Aug 2013 00:13:50 +0000 (17:13 -0700)]
Frm to install/uninstall flow when port is added/removed from slice
Change-Id: I462545382e55f480f7bdc1c5084e0c09907a8a8f
Signed-off-by: Diti Bhatia <dibhatia@cisco.com>
Alessandro Boch [Tue, 20 Aug 2013 23:36:45 +0000 (16:36 -0700)]
Enhancements to common.northbound
- Added BadRequest and NotImplemented exceptions
Change-Id: I9a021ddd103c16122c39626e9cca044bd8a33ea6
Signed-off-by: Alessandro Boch <aboch@cisco.com>
Ed Warnicke [Tue, 20 Aug 2013 15:52:55 +0000 (15:52 +0000)]
Merge "Added YANG models for flow configuration, statistics and inventory."
Giovanni Meo [Tue, 20 Aug 2013 06:53:55 +0000 (06:53 +0000)]
Merge "Fix for hosttracker. Ignore ARP messages from internal nodes."
Tony Tkacik [Mon, 19 Aug 2013 18:37:51 +0000 (20:37 +0200)]
Added YANG models for flow configuration, statistics and inventory.
Added AbstractBindingAwareConsumer and Provider, updated samples.
Change-Id: I72a8707b7e0b4e97d11d9f6f4288f0656d0840e0
Signed-off-by: Tony Tkacik <ttkacik@cisco.com>
Alessandro Boch [Mon, 19 Aug 2013 16:12:53 +0000 (09:12 -0700)]
FRM to expose list of installed flow entries per group
ISSUE:
Given a group/policy name, ForwardingRulesManager only provides a function to return the list of Flow Entry it was requested to install.
In presence of container flows, this list does not represent the actual entries which were installed on the network nodes for that group.
CHANGE:
Add getInstalledFlowEntriesForGroup(String policyName) to IForwardingRulesManager.
Change-Id: Ie26d64c05d5eb49b3073d34d701ea1fa5b05adfa
Signed-off-by: Alessandro Boch <aboch@cisco.com>
Alessandro Boch [Thu, 15 Aug 2013 16:36:38 +0000 (09:36 -0700)]
Fix bug in Static Route configuration path
ISSUE: Multiple equivalent static route configurations with different names not detected
CHANGE: In addStaticRoute() perform proper equality check on existing StaticRoute objects
(using available StaticRoute compareTo() method) instead of checking on StaticRouteConfig
objects which can never be positive as same name config is detected at the beginning of method
Change-Id: I51d2e1a641f612f5e79eb11063a868739a275209
Signed-off-by: Alessandro Boch <aboch@cisco.com>
Pramila Singh [Thu, 15 Aug 2013 22:06:15 +0000 (15:06 -0700)]
Fix: Switch manager should clean the nodeconnector info when a switch disconnects
Change-Id: I2ccdf1a6d2ed38e1e78242ae99d9bb1711f20040
Signed-off-by: Pramila Singh <pramisin@cisco.com>
Alessandro Boch [Thu, 15 Aug 2013 23:47:45 +0000 (16:47 -0700)]
Default network admin password change is not persistent across controller restart
ISSUE: Change default network admin (NA) user password, save config, restart controller.
To be able to login as default NA user you still have to type the default password.
This happens because when parsing startup configuration user config addition are replayed
through public API which has a check to block addiiton/removal of default NA user.
CHANGES:
- Perform checks in a middle layer function and have a internal level function which just update the db
- Call the internal function during startup config replay
- Added missing checks on presence/conflict of user configs when removing/adding a user config object
Change-Id: I38c98bb946b42215cec46d86d3b1a9885c12795b
Signed-off-by: Alessandro Boch <aboch@cisco.com>
Katrina LaCurts [Fri, 16 Aug 2013 20:30:41 +0000 (16:30 -0400)]
Fix for hosttracker. Ignore ARP messages from internal nodes.
Signed-off-by: Katrina LaCurts <katrina.lacurts@plexxi.com>
Change-Id: I2030822045b7dc8bd1d013cf5dfcd4898f5ddc32
Giovanni Meo [Fri, 16 Aug 2013 16:04:08 +0000 (16:04 +0000)]
Merge "Migrating caches to TRANSACTIONAL Caches and enabled use1PcForAutoCommitTransactions. In addition, - Removed all the unused Caches (especially the save event cache). - Changed ARP Handler's ICacheUpdateAware processing to not process the cache event in its own thread (instead of the infinispan's thread)."
Andrew Kim [Fri, 16 Aug 2013 15:33:46 +0000 (15:33 +0000)]
Merge "Devices dashlet show port total numbers, modal shows collapsible list Replaced Tabs with white space"
Madhu Venugopal [Fri, 16 Aug 2013 14:26:28 +0000 (07:26 -0700)]
Migrating caches to TRANSACTIONAL Caches and enabled use1PcForAutoCommitTransactions.
In addition,
- Removed all the unused Caches (especially the save event cache).
- Changed ARP Handler's ICacheUpdateAware processing to not process the cache event
in its own thread (instead of the infinispan's thread).
Change-Id: I608aecd425d0394ba9c4e665424c82790bb874ca
Signed-off-by: Madhu Venugopal <vmadhu@cisco.com>
Ed Warnicke [Fri, 16 Aug 2013 12:41:08 +0000 (12:41 +0000)]
Merge changes I8293baab,I63e248bd
* changes:
Added sample provider and consumer for MD SAL
Clean up of binding broker implementation
Sapan Shah [Fri, 16 Aug 2013 04:45:53 +0000 (21:45 -0700)]
Devices dashlet show port total numbers, modal shows collapsible list
Replaced Tabs with white space
Change-Id: I9a8ede6a761e4d46fb1dcaa90ddffb46822e5e79
Signed-off-by: Sapan Shah <sapshah@cisco.com>
Maurice Qureshi [Thu, 15 Aug 2013 22:43:06 +0000 (15:43 -0700)]
Change HostTracker's ArrayLists to ConcurrentMaps. Add new CLIs
1)Initially some data structures were used which were used ony in main thread. As the code has
evolved, those data structures are being accessed in multiple threads now. Therefore
changed two ArrayList to ConcurrentMaps
2) Changed the name of Caches in consistent with other modules "hosttracker.activeHosts" and "hosttracker.InactiveHosts"
3) Added New CLIs to dump couple of data structures
Change-Id: I06b68e4e559b0193ec502612d6ea29c634bd9a81
Signed-off-by: Maurice Qureshi <maquresh@cisco.com>
lakshya [Thu, 15 Aug 2013 21:06:44 +0000 (14:06 -0700)]
Converting tables in flow and troubleshoot to use datagrid.
Change-Id: I70c69ef9f2e5204ad47caa41e7d0ba3b2603c0f1
Signed-off-by: lakshya <lakshya@cisco.com>
lakshya [Mon, 12 Aug 2013 06:40:34 +0000 (23:40 -0700)]
Devices: NodesLearnt table integration with FuelUx Datagrid.
Removed glyph-icons + added fuelux copyright information.
Change-Id: If0844a9df414d92e01c6d2cd72942c054784f959
Signed-off-by: lakshya <lakshya@cisco.com>
Alessandro Boch [Wed, 14 Aug 2013 17:12:41 +0000 (17:12 +0000)]
Merge "Should not allow nodes to have same node name"
Giovanni Meo [Wed, 14 Aug 2013 13:24:53 +0000 (13:24 +0000)]
Merge "Fixed NullPointerException seen in Hosttracker when fetching the hierarchy of host"
Giovanni Meo [Wed, 14 Aug 2013 12:58:00 +0000 (12:58 +0000)]
Merge "Log all configuration(add/modify/delete) changes to a new log file audit.log Change-Id: Icf16c88dbbd703dafe634f145f9e52f1e425f843 Signed-off-by: Sapan Shah <sapshah@cisco.com>"
Tony Tkacik [Tue, 13 Aug 2013 16:15:17 +0000 (18:15 +0200)]
Added sample provider and consumer for MD SAL
Signed-off-by: Tony Tkacik <ttkacik@cisco.com>
Change-Id: I8293baab178623762b6ce9e1e2ecb582e78d7f4e
Tony Tkacik [Tue, 13 Aug 2013 15:30:07 +0000 (17:30 +0200)]
Clean up of binding broker implementation
Binding Broker implementation changed from java to xtend, which
allows easier extensibility and readability of code without any
performance hit (xtend generates java sources, these are compiled with
normal java compiler).
- Bugfix in RpcDirectProxy generation
- Added helper methods to easier configuring a delegate to proxy
- Added check for multiple subsequent registrations
Signed-off-by: Tony Tkacik <ttkacik@cisco.com>
Change-Id: I63e248bda99051e5b21ac0a5cbf57bd935ded1bc
Madhu Venugopal [Wed, 14 Aug 2013 04:21:45 +0000 (21:21 -0700)]
Upgrading & configuring Infinispan 5.2.3 -> 5.3.0 for better 2-phase commit support.
This follows the recent change from DummyTransactionManager to JBossTransactionManager.
ConnectionManager depends on the Consistency guarantee honored by the Clustering Services in order to
provide consistent Connection Management Services. Existing Infinispan 5.2.3 carries a few issues
(Refer : ISPN-3366 and ISPN-3357) which makes putIfAbsent not atomic even for TRANSACTIONAL caches.
Also, Since we are using NON-TRANSACTIONAL Caches, Infinispan doesnt provide any Atomicity guarantee
for the Cache Syncs. This is the first commit in the controller repo to convert an existing cache
from NON-TRASACTIONAL to TRANSACTIONAL. Few more caches that needs Atomicity guarantee will be
converted soon.
In addition, Infinispan configurations must be tweaked to make the transaction commits synchronous
(Currently it is inadvertently configured as Async).
With these changes, the Connection manager NodeConnections cache is made TRANSACTIONAL and the
cache syncs are guaranteed to be atomic.
NOTE: There are still few issues with Infinispan cache sync (ISPN-2719)
Change-Id: I97037ac7d6413f5b988496a03fb05bd88bbbce3f
Signed-off-by: Madhu Venugopal <vmadhu@cisco.com>
Alessandro Boch [Wed, 14 Aug 2013 00:15:28 +0000 (17:15 -0700)]
Fix NPE in ICMP.computeChecksum()
ISSUE:
ICMP.computeChecksum() incorrectly assumes rawPayload is always present
while determining the lenght of the ICMP packet byte stream on which
computing the checksum.
Change-Id: I41c1e0509f21a2bbb0a5ec6df038ed18a81f327a
Signed-off-by: Alessandro Boch <aboch@cisco.com>
Pramila Singh [Tue, 13 Aug 2013 22:30:03 +0000 (15:30 -0700)]
Should not allow nodes to have same node name
Change-Id: Ia658826c85dfd9a9fa5e4b82731c6e26d52e5b40
Signed-off-by: Pramila Singh <pramisin@cisco.com>
Diti Bhatia [Tue, 13 Aug 2013 16:48:55 +0000 (09:48 -0700)]
Fixed NullPointerException seen in Hosttracker when fetching the hierarchy of host
Change-Id: I0ae84707df9aad0551758a9b70ebc54631dab2c5
Signed-off-by: Diti Bhatia <dibhatia@cisco.com>
Ubuntu [Tue, 13 Aug 2013 19:55:20 +0000 (12:55 -0700)]
Log all configuration(add/modify/delete) changes to a new log file audit.log
Change-Id: Icf16c88dbbd703dafe634f145f9e52f1e425f843
Signed-off-by: Sapan Shah <sapshah@cisco.com>
Giovanni Meo [Tue, 13 Aug 2013 08:51:37 +0000 (08:51 +0000)]
Merge "Add container context debug osgi cli command to Switch Manager"
Alessandro Boch [Mon, 12 Aug 2013 23:25:48 +0000 (16:25 -0700)]
Add container context debug osgi cli command to Switch Manager
CHANGES:
Available osgi comamnds implemented in the service classes
do not allow to access the desired container instance of service.
This change, following example in Gerrit #831, allow to dump the
nodes and node connectors database contained in Switch Manager
for the desired container.
Change-Id: I4cf959763e06facd135a1a5c3ebca59ee85bcf22
Signed-off-by: Alessandro Boch <aboch@cisco.com>
Giovanni Meo [Mon, 12 Aug 2013 16:10:37 +0000 (18:10 +0200)]
Add standalone JBOSSTM-JTA component
- Introduce a proper transaction manager, and make sure infinispan
uses it.
Change-Id: I13cb4f69be095ac50e2d3ff32fe85c8fd8acdb27
Signed-off-by: Giovanni Meo <gmeo@cisco.com>
Asad Ahmed [Fri, 9 Aug 2013 21:56:27 +0000 (14:56 -0700)]
Adding code to handle multiple hosts per node connector
Change-Id: I32b18fe019a2806f0bb2778321f0a3ac6c6b23a2
Signed-off-by: Asad Ahmed <asaahmed@cisco.com>
Madhu Venugopal [Fri, 9 Aug 2013 19:32:32 +0000 (19:32 +0000)]
Merge "Implement cluster wide topology notifications and let routing use it"
Giovanni Meo [Thu, 8 Aug 2013 17:43:38 +0000 (19:43 +0200)]
Implement cluster wide topology notifications and let routing use it
- Made TopologyManager to generate ClusterWide updates along with the
local ones.
- Implemented a mechanism in topology manager to batch the topology
updates being synched via clustering services. This would save on
unnecessary churn in recalculations under massive topology updates.
- Modified routing Dijkstra to be an ITopologyManagerClusterWideAware
client rather than on of ITopologyManagerAware.
- Modified Dijkstra implementation to generate routing updates
notifications only on the coordinator, in fact given Dijkstra will
have the same view on all the controllers node in the cluster, then
it's pointless to have all the clients of IRoutingAware to
recalculate, they would lead to the same result.
- Dijkstra edgeUpdate logic was considering the CHANGED topology event
as DELETE, that is wrong and this patch fix it.
- Added gogo shell CLI to get a route from the
DijkstraImplementation.java
- Enhanced "TopoEdgeUpdate" class to has the isLocal flag to
distinguish between the local updates and the remotes one. Also remove
non-key field from equal and hashCode calculation.
- Remove unnecessary CONFIGSAVEEVENT in TopologyManagerImpl because
now the configuration service provides a cluster wide trigger.
Change-Id: Ia74d9d1ec0731e1f5815a69edc25bbb5b4c1f531
Signed-off-by: Giovanni Meo <gmeo@cisco.com>
Yevgeny Khodorkovsky [Fri, 9 Aug 2013 04:54:44 +0000 (21:54 -0700)]
UI: Remove unnecessary exception throws
When trying to register/remove a UI session, but User Manager service is
unavailable, log a warning instead of throwing exception.
Change-Id: Ia9c161f0c68b5edadad02e701949f0bfda69c4e1
Signed-off-by: Yevgeny Khodorkovsky <ykhodork@cisco.com>
Pramila Singh [Thu, 8 Aug 2013 20:59:25 +0000 (13:59 -0700)]
Fix to set/delete Switch Configuration using Northbound API
Change-Id: Ib3d7184a64bba4341637052ae96ae104073a64bc
Signed-off-by: Pramila Singh <pramisin@cisco.com>
Alessandro Boch [Thu, 8 Aug 2013 18:05:38 +0000 (18:05 +0000)]
Merge "Logging Bridge need to show the source of the OSGi Log events"
Alessandro Boch [Thu, 8 Aug 2013 17:28:11 +0000 (17:28 +0000)]
Merge "Fix race condition when registering services"
Giovanni Meo [Thu, 8 Aug 2013 09:48:24 +0000 (11:48 +0200)]
StatisticsManager "started" method may fail
"started" method in the
org.opendaylight.controller.statisticsmanager.internal.StatisticsManager
occasionally can fail to complete, this is because the method tries to
"put" data in clustered caches without checking if the value being
inserted is null or no. Infinispan caches in fact don't support null
value insertion and hence an exception is thrown preventing subsequent
calls to continue. The fix is make sure we don't insert NULL values.
Change-Id: I0037931d73e7e7df9535ef83509e770b45e81c72
Signed-off-by: Giovanni Meo <gmeo@cisco.com>
Giovanni Meo [Thu, 8 Aug 2013 09:25:40 +0000 (11:25 +0200)]
Logging Bridge need to show the source of the OSGi Log events
Logging bridge is in charge of bridging the OSGi log events toward
SLF4J but even when it does, like because an Apache Dependency Manager
callback didn't complete, it doesn't give the information on who is
the culprit and why, making debugging more difficult, this enhace the
logging.
Change-Id: I388516b3bc40d94552f49d6928b3b33c8349b687
Signed-off-by: Giovanni Meo <gmeo@cisco.com>
Yevgeny Khodorkovsky [Thu, 8 Aug 2013 04:31:48 +0000 (21:31 -0700)]
Use Set for Subnet's NodeConnectors
- Use Set instead of List in Subnet's aggreagated NodeConnectors
to avoid handling dpulicates
- Style fixes
Change-Id: I037764503b5bc01791716a9f515825a706451153
Signed-off-by: Yevgeny Khodorkovsky <ykhodork@cisco.com>
Alessandro Boch [Wed, 7 Aug 2013 19:35:52 +0000 (19:35 +0000)]
Merge "Fixed minimum values for TP_SRC/TP_DST match fields."
Hideyuki Tai [Wed, 7 Aug 2013 16:53:28 +0000 (12:53 -0400)]
Fixed minimum values for TP_SRC/TP_DST match fields.
In OpenFlow 1.0, the tp_src and tp_dst match fields are used for transport source/destination ports or ICMP type/code.
Therefore, the range of the fields must include 0.
Change-Id: I7011109ced925e198ec2785f5ed9dc9a8e81623c
Signed-off-by: Hideyuki Tai <h-tai@cd.jp.nec.com>
Giovanni Meo [Wed, 7 Aug 2013 09:35:29 +0000 (11:35 +0200)]
Fix race condition when registering services
Dependency Manager is able to notify every component when a new
service tracked is added/removed, this kind of operation is usually
done by allocating a synchronized set or an concurrent map to store
the new service reference or remove it.
This works fine, just it's necessary to be aware that the services may
be learnt by the Component before the "init" method of the component
it's called. In fact the init method of a Dependency Manager component
is called after all the required dependency have been met so that
means that the services may have already been registered. Now if a
component allocate the data structure needed to hold the service
reference during the init time, it can run in the condition where the
services already registered would be wiped out.
The fix is to make sure all the Dependency Manager component init
method don't allocate the data structure needed to hold the services,
but they should be created or during constructor time or during field
initialization. The second approach is prefered here.
Change-Id: I9057351830c09736c5ad7b6777330ef14b39fe9c
Signed-off-by: Giovanni Meo <gmeo@cisco.com>
Madhu Venugopal [Wed, 7 Aug 2013 05:27:20 +0000 (22:27 -0700)]
Process host pending list on a Cache sync update.
In a Clustered Controller environment, with switches connected to different controllers, ARP messages originated from
a Controller may not see the ARP reply on the same controller. The ARP response from the Host might end up in another
controller. Hence it is not appropriate to expect responses from host to terminate in the originating controller.
discoverHost is one such method that has this wrong assumption. This fix is to listen to Active Cache update and
trigger a Pending list processing (In addition to existing code that just waits on a Future object for ARP response
on the local controller).
Change-Id: I4223db66ff75bbf4d60011c9581e4dff594c0a9c
Signed-off-by: Madhu Venugopal <vmadhu@cisco.com>
Giovanni Meo [Wed, 7 Aug 2013 07:41:40 +0000 (07:41 +0000)]
Merge "HA - ARPHandler Event sync"
Yevgeny Khodorkovsky [Fri, 2 Aug 2013 23:38:14 +0000 (16:38 -0700)]
HA - ARPHandler Event sync
- Replay ARP request/reply events across the cluster using cluster
allocated cache. Then the appropriate controller handles sending the
ARP.
- Add isUnicastMAC() to NetUtils
Change-Id: I2d2b60348341710ffbd3ffc749ff52869b9c0b3f
Signed-off-by: Yevgeny Khodorkovsky <ykhodork@cisco.com>
Alessandro Boch [Tue, 6 Aug 2013 22:27:46 +0000 (22:27 +0000)]
Merge "Speed up builds"