Lorand Jakab [Tue, 20 Jun 2017 11:54:10 +0000 (14:54 +0300)]
Bug 9023: Fix merging of negative prefixes
This happens after a positive mapping removal. Add integration test to
keep from regressing.
Change-Id: I32788e8af5b457bf895197ba11add34e8ae06c57
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Fri, 4 Aug 2017 12:10:24 +0000 (15:10 +0300)]
Bug 8746: Multi-threading improvements
Remove all state from SB handlers
This needs a significant rework of the LispSouthboundServiceTest test
class, moving even more functionality to Mockito, so eventually we can
remove the dependency on the old unittest.tools.
Change-Id: I901c5633cb60563ac5bc38a521796dad2186886a
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Tue, 4 Apr 2017 16:20:14 +0000 (19:20 +0300)]
Bug 7947: Store MappingOrigin in MappingData
As a first step towards separate subscriber storage, based on origin,
this patch adds support for storing the origin of a mapping in the
MappingData object. The origin is defined as a Set, to support the use
case of returning a combined mapping with information from different
origins, such as the NB_AND_SB policy surrently.
This patch doesn't change behavior yet, it just stores the origin.
Change-Id: Id285e829fed794de9ecf40ee9a9cb5e775228e92
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Vina Ermagan [Wed, 16 Aug 2017 04:28:02 +0000 (04:28 +0000)]
Merge "Bug 8746: Multi-threading improvements"
Vina Ermagan [Wed, 16 Aug 2017 03:50:07 +0000 (03:50 +0000)]
Merge "Remove our NETCONF component"
Lorand Jakab [Thu, 10 Aug 2017 08:41:26 +0000 (11:41 +0300)]
Remove our NETCONF component
Our NETCONF connector has not been updated since several releases, and
any attempt to make it work again would likely mean reimplementing from
scratch. It is using the config subsystem too.
Worst case, if we still need to code for whatever reason, we can
resurrect it from old Git revisions.
Change-Id: Ia3e9479a1e83fbdc1f69d946c197e8f39ca9e9a4
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Fri, 31 Mar 2017 11:40:45 +0000 (14:40 +0300)]
Bug 8112: IT: send SMR-invoked Map-Request on SMR
Change-Id: I3458ef941460ad5af91ef097c16f17894f5bbfa9
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Fri, 11 Aug 2017 17:11:20 +0000 (20:11 +0300)]
Bug 8469: Account for losing prefix length in Source EID
Change-Id: I188d3001d887ade55be3a90738f74d0bace48c81
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Thanh Ha [Mon, 14 Aug 2017 17:29:07 +0000 (13:29 -0400)]
Bump versions by x.(y+1).z for next dev cycle
Change-Id: Ib445f2ab3278c9024ca36329a351eeaa8f3e062e
Signed-off-by: Thanh Ha <thanh.ha@linuxfoundation.org>
Lorand Jakab [Mon, 7 Aug 2017 17:34:20 +0000 (20:34 +0300)]
Bug 8746: Multi-threading improvements
Fix southbound Map-Register cache authentication update code
Change-Id: I616d199e1fc6f37357c35ac75a9c60c5884068f7
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Vina Ermagan [Mon, 14 Aug 2017 00:13:54 +0000 (00:13 +0000)]
Merge "Bug 8746: Multi-threading improvements"
Lori Jakab [Thu, 10 Aug 2017 04:43:47 +0000 (04:43 +0000)]
Merge "Bump odlparent 2.0.2 to 2.0.4"
Stephen Kitt [Tue, 8 Aug 2017 16:14:05 +0000 (18:14 +0200)]
Bump odlparent 2.0.2 to 2.0.4
Change-Id: Ie310e026eb3408b236be40dffdea58171a8e166a
Signed-off-by: Stephen Kitt <skitt@redhat.com>
Vina Ermagan [Tue, 8 Aug 2017 18:25:51 +0000 (18:25 +0000)]
Merge "Switch to MD-SAL LISP address model"
Lorand Jakab [Mon, 26 Jun 2017 07:52:35 +0000 (10:52 +0300)]
Bug 8746: Multi-threading improvements
- Remove non-concurrent statistics class
- Make some variables 'volatile'
- Make some methods 'synchronized'
Change-Id: I83471fe7a422e63c2ab90e4c5fbc97b69c187247
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Mon, 31 Jul 2017 15:08:55 +0000 (18:08 +0300)]
Switch to MD-SAL LISP address model
ietf-lisp-address-types.yang was added to MD-SAL in [0] with the
intention to be the canonical place for the model. This patch removes
this model from lispflowmapping and switches to using that model.
[0] https://git.opendaylight.org/gerrit/#/c/60528/
Change-Id: Ib731bf9f10efd7f4417d3ea8cb1b73018519062f
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Wed, 28 Jun 2017 12:35:45 +0000 (15:35 +0300)]
Bug 8764: Fix handling of old negative
Change-Id: Ia3295c955ecf7b364865e445cb7f75a0ec80a6bf
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Tom Pantelis [Fri, 4 Aug 2017 20:02:16 +0000 (16:02 -0400)]
Remove config subsystem remnants
lispflowmapping no longer uses the config subsystem so remove
remnants from the pom.
Change-Id: Ia211b670d6c3277f0ebc164fe1676f203c7a79cc
Signed-off-by: Tom Pantelis <tompantelis@gmail.com>
Vina Ermagan [Fri, 21 Jul 2017 05:03:42 +0000 (05:03 +0000)]
Merge "Return negative mapping with 0 TTL for deletion"
Lorand Jakab [Thu, 20 Jul 2017 09:22:09 +0000 (12:22 +0300)]
Return negative mapping with 0 TTL for deletion
When an SMR-invoked Map-Request is asking for a mapping that is
negative, it is most likely an attempt to delete that mapping.
Change-Id: Iec949c5ca9da90845c6634a7d26fabebd65afc81
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Wed, 14 Jun 2017 13:04:09 +0000 (16:04 +0300)]
Make 'keys' CLI output user friendly
Before this patch, the 'keys' CLI command would print out the full
Java objects contained in a map-cache. That's great for in-depth
debugging, but for a quick look it very hard to read. This patch
implements a user friendly CLI output for keys, to improve
readability.
The old full output is still available in Karaf, by adding the '-d' or
'--debug' switch to the CLI command.
Example output:
---8<------------------------------------------------------------------------------
opendaylight-user@root>keys
Instance ID 0
-> EID HMAC Algorithm Shared Key
192.0.2.1/32 HmacSHA1 password
192.0.2.1/32|192.0.2.2/32 HmacSHA1 password
------------------------------------------------------------------------------>8---
Change-Id: Ibe95d0ee4318053df3bc9f30c09c0e01c1d00eba
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Wed, 12 Jul 2017 12:04:59 +0000 (15:04 +0300)]
Fully remove Maven Sites
Change-Id: I8c6ba1fe18e0fff43d0aa47b19c9f11df05ead20
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Wed, 12 Jul 2017 10:53:18 +0000 (13:53 +0300)]
Remove outdated src/site documentation
https://lists.opendaylight.org/pipermail/odlparent-dev/2017-July/001218.html
Content moved to the wiki:
https://wiki.opendaylight.org/view/Southbound_Performance_Cache
Change-Id: I9867417f6f0a9f30cb3166371f4cfe4c5f3cbe14
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lori Jakab [Tue, 11 Jul 2017 11:09:46 +0000 (11:09 +0000)]
Merge "Make negative mapping TTL configurable"
Lorand Jakab [Mon, 10 Jul 2017 13:07:36 +0000 (16:07 +0300)]
Upgrade to odlparent 2.0.2
Build is broken without this upgrade.
Change-Id: Ib9e24eb78d4af103ad8e30febd233ef2ec175963
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Tue, 27 Jun 2017 11:34:24 +0000 (14:34 +0300)]
Make negative mapping TTL configurable
Change-Id: I08c6ad1517f4cc34747a4762dc604b0de87b2bea
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Tue, 27 Jun 2017 08:21:47 +0000 (11:21 +0300)]
Fix build
Change-Id: I5bfc3c58fef1fcb1c20387d840f1307f09fd7990
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Tue, 20 Jun 2017 08:53:15 +0000 (11:53 +0300)]
Bug 8679: Fix widest negative prefix calculation
And add integration tests to avoid regressions.
Change-Id: I6cadaca03d360f475c7c88fd3cb58dd39814d23f
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Wed, 7 Jun 2017 09:36:55 +0000 (12:36 +0300)]
Upgrade to odlparent 2.0.0
Change-Id: I10061eda3522ded1d55adc1ac365ab9d4b22763b
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Wed, 7 Jun 2017 09:36:55 +0000 (12:36 +0300)]
Do not build Karaf 3 features and distro
Disable building the Karaf 3 features and local distribution, since
that's a prerequisite for getting back into autorelease in Nitrogen [0],
apart from the build breaking right now.
For now, just comment them out in the build, we fully remove source code
just before code freeze (in case we need to contrast behaviors between a
Karaf 3 and 4 container).
[0] https://lists.opendaylight.org/pipermail/odlparent-dev/2017-June/001036.html
Change-Id: Id7f9a82e4b9868ad57141d3ef7e5a0cc2ef74946
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Mon, 19 Jun 2017 06:57:04 +0000 (09:57 +0300)]
Upgrade V3PO model version
Else build fails
Change-Id: I57439066304d70c633dd565873d9401870238b13
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Vina Ermagan [Thu, 15 Jun 2017 01:08:27 +0000 (01:08 +0000)]
Merge "Bug 8503: Remove empty structures in DAO"
Thanh Ha [Tue, 13 Jun 2017 20:20:20 +0000 (16:20 -0400)]
Migrate to odlparent 1.9.0
Change-Id: I9de99604521440580045e6bedcff4606389561a2
Signed-off-by: Thanh Ha <thanh.ha@linuxfoundation.org>
Lorand Jakab [Thu, 25 May 2017 06:29:15 +0000 (09:29 +0300)]
Bug 8503: Remove empty structures in DAO
Change-Id: I52bb47edbb2e34c0b72d874b1ceab33125156f42
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Robert Varga [Tue, 6 Jun 2017 20:41:58 +0000 (22:41 +0200)]
BUG-6709: migrate karaf-empty
No longer provided by controller.
Change-Id: I11fa1db7b9aaf2a15be3c8979c744d23d5c77d62
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>
Thanh Ha [Sat, 10 Jun 2017 05:31:21 +0000 (01:31 -0400)]
Migrate to odlparent 1.8.0-Carbon
Per request of odlparent project we are downgrading all Nitrogen
projects to use the released odlparent 1.8.0-Carbon to allow for the
odlparent project to start performing semver style releases.
Change-Id: I38f3b068e85b2e9aa81fb6fc8a48e862c249bf95
Jira: RELENG-159
RT: 41406
Signed-off-by: Thanh Ha <thanh.ha@linuxfoundation.org>
Lorand Jakab [Fri, 2 Jun 2017 22:19:24 +0000 (15:19 -0700)]
Fix configuration file parsing
Change-Id: Ifd747d1b5d5b39da80920bc088c7c9c0e5cb7111
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Thu, 1 Jun 2017 22:20:49 +0000 (15:20 -0700)]
Add knob to disable authentication
Change-Id: I87a2311be0d6b3821fa276feb84cd208d8e8dd30
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Thu, 25 May 2017 05:32:39 +0000 (08:32 +0300)]
Cancel ongoing SMR futures on new event
The SMR scheduler keeps retrying the sending of SMRs a few (configurable
number) of times with a given spacing between the retries until an
SMR-invoked Map-Request is received for the particular EID. When that
happens, the retrying is cancelled for the subscriber that sent it.
However, up until now, when a new set of SMRs was scheduled for a given
EID, the old set, if still retrying, was left untouched. This patch
fixes that, and starts with a clean slate for the given EID.
Change-Id: Ic64ce0980b39d780433ac2c4e96608af12302392
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Wed, 24 May 2017 17:19:57 +0000 (20:19 +0300)]
Reverse SmrScheduler Map order Subscriber <--> Eid
Change-Id: Ic4b5614c2825f0e03b637896fae5b93f21bdbe32
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Mon, 22 May 2017 22:38:37 +0000 (01:38 +0300)]
Replace Rloc with Subscriber for SMR key
Change-Id: I362c6037707efc091d58c6d1aa4e4e16351523d3
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Thu, 18 May 2017 16:19:02 +0000 (19:19 +0300)]
Add cluster testing Postman collection
Change-Id: I190b9d0e952c361db0fe3f204ff8dd7fe43c81fc
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Vina Ermagan [Tue, 16 May 2017 18:29:04 +0000 (18:29 +0000)]
Merge "Add CSR1Kv configs"
Lorand Jakab [Thu, 11 May 2017 08:37:48 +0000 (11:37 +0300)]
More detailed subscriber event logging
Change-Id: I21de8a93d0c4973a29b210dcf857f36ba08a84db
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Thu, 11 May 2017 10:21:24 +0000 (13:21 +0300)]
Bug 8429: Fix SMR handling concurrency issue
Change-Id: I1944a82600a3956c053edd9134541a6eab776e0a
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Tue, 16 May 2017 08:07:06 +0000 (11:07 +0300)]
Add CSR1Kv configs
Add CSR1Kv example configs from the tutorial, so they can be linked to.
Change-Id: I28d65d9bb65a769770b3d446634b77a6ec01a4fb
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Fri, 5 May 2017 19:02:53 +0000 (22:02 +0300)]
Trace-level improved IT logging
Change-Id: I29bc887ff4efd81d11629c0028a95beaf4cf2d68
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Wed, 3 May 2017 11:37:49 +0000 (14:37 +0300)]
Add IPv6 default key for addkeys CLI command
The addkeys Karaf CLI command adds a default authentication key
"password" of type one for the whole IPv4 EID space to ease southbound
testing. This patch extends the command to add a default authentication
key for the whole IPv6 EID space as well.
Change-Id: I93b84ab03699592cd5722f42ca63fe66437eb6d5
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Mon, 24 Apr 2017 18:55:50 +0000 (21:55 +0300)]
Bug 8248: Don't access DSBE from different threads
With Epoll we may have more than one LispSouthboundHandler thread, and
they may access the same DataStoreBackEnd (DSBE) instance concurrently.
This patch moves DSBE access to the single threaded
LispSouthboundPlugin. Additionally, it implements some suggestions from
the following thread:
https://lists.opendaylight.org/pipermail/release/2017-April/010273.html
Change-Id: I32f48b6dcb58b241a4a679f7145bd36b5f410861
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Thu, 13 Apr 2017 12:43:02 +0000 (15:43 +0300)]
IT: Before each checkSMR(), print map-caches to log
Change-Id: I64ff1f3a10ec0b0ddcc1605703fbd5ce95325aa6
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Thu, 22 Dec 2016 16:54:38 +0000 (18:54 +0200)]
Make 'mappings' CLI output user friendly
Before this patch, the mappings CLI command would print out the full
Java objects contained in a map-cache. That's great for in-depth
debugging, but for a quick look it very hard to read. This patch
implements a user friendly CLI output for mappings, to improve
readability.
The old full output is still available in Karaf, but is by adding the
'-d' or '--debug' switch to the CLI command.
Example output:
---8<--------------------------------------------------------------------
opendaylight-user@root>mappings
Policy map-cache
----------------
Instance ID 1
[1] 192.0.2.1/32|192.0.2.2/32, TTL: 1440
-> Locator State Pri/Wgt
10.10.10.10 no-route 1/1
Southbound map-cache
--------------------
Instance ID 0
192.0.2.1/32, TTL: 1440
-> Locator State Pri/Wgt
192.168.16.21 up 1/1
fdab:cc19:b80e::21 up 1/1
-----------------------------------------------------------------
-> Subscriber RLOC Subscriber EID
192.168.16.21 No Address Present
127.0.0.2 No Address Present
-------------------------------------------------------------------->8---
Change-Id: Id3c304ac9eda77a0cc36bff1cdeb18d61608a796
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Wed, 19 Apr 2017 18:20:13 +0000 (21:20 +0300)]
Prepare for checkstyle version upgrade
See https://git.opendaylight.org/gerrit/#/c/54981/
Change-Id: I451e9016a2545b0455d31e057191cb9df1990aaf
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Anil Belur [Tue, 11 Apr 2017 01:45:41 +0000 (11:45 +1000)]
Bump versions by x.(y+1).z for next dev cycle
Change-Id: Ibf7642de95c0c4bd04d683ba2de53549ecd7c8f4
Signed-off-by: Anil Belur <abelur@linuxfoundation.org>
Lorand Jakab [Tue, 21 Feb 2017 12:58:59 +0000 (14:58 +0200)]
Bug 7272: Fully remove mappings
When removing a mapping, remove all subkeys, and the radix trie entry,
to avoid false lookup results.
Also, instead of sending a notification w/o a subscriber list, expecting
the listeners to look up the subscribers from the mapping system,
include the subscriber list with the notification. That way all subkeys
can be deleted when the change is a mapping removal.
Change-Id: Ic922b82fab42f533f9fa8ec828e4d21f6fa35198
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Wed, 22 Feb 2017 21:07:31 +0000 (23:07 +0200)]
Migrate IT to use the Karaf 4 distribution
Change-Id: Iecef5c6745bcee3853810303815199c12b6957a5
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Vina Ermagan [Wed, 22 Mar 2017 23:04:16 +0000 (23:04 +0000)]
Merge "Use Optional to avoid null return value"
Lorand Jakab [Tue, 21 Mar 2017 14:34:29 +0000 (16:34 +0200)]
Use Optional to avoid null return value
Change-Id: I390f182341fffa36da989fb0e9daaa72473ec466
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Shakib Ahmed [Tue, 21 Mar 2017 23:18:19 +0000 (16:18 -0700)]
Update port tests with proper implementation
New implementation of Neutron PortDataProcessor left the old unit tests
useless. Those tests are being updated with validated ones.
Change-Id: I6b419aa06cd03721606c2e3cf7a6cac31b8cea4d
Signed-off-by: Shakib Ahmed <sheikahm@cisco.com>
Shakib Ahmed [Sat, 18 Feb 2017 00:12:56 +0000 (16:12 -0800)]
Create mapping record with proper VNI
Change-Id: I63f94bb9b87b7256d8c12375bfbd10defd937280
Signed-off-by: Shakib Ahmed <sheikahm@cisco.com>
Shakib Ahmed [Wed, 8 Feb 2017 22:20:11 +0000 (14:20 -0800)]
Update mapping records with port data update
Neutron port update can provide a new set of fixed ips. With new fixed
ips, we need to removed old set of mapping recordss in case there were
already old fixed ips and we have mapping records in mapcaches related
to those fixed ips. And then we need to add the new mapping records
in the mapcache.
Change-Id: Id2eca598fcb073169e6d63b72dd14830a35a4f5b
Signed-off-by: Shakib Ahmed <sheikahm@cisco.com>
Shakib Ahmed [Mon, 6 Feb 2017 00:14:17 +0000 (16:14 -0800)]
Add mapping record with port data add with fixed ip
When a neutron port create comes with a fixed ip, we consider the fixed
ip as EID. We need to save this EIDs specific to host id, so that when
we have host id to locator mapping available we can create a mapping
record.
This patch adds hostid to all eid mappings keyed by port uuid. And also
attempts to create a mapping record if hostid to locator mapping is
available.
Change-Id: Ia6ae4643e1ecf5d16c592bcd9a52170b28f38777
Signed-off-by: Shakib Ahmed <sheikahm@cisco.com>
Shakib Ahmed [Tue, 10 Jan 2017 22:17:54 +0000 (14:17 -0800)]
Add Neutron host-id to RLOC mapping
Neutron port create call provides us informations like port uuid, host-
id, sometimes fixed-ips but we don't know the interface VPP is going to
use for sending traffic from one host to another. That interface is
going to be the locator for the mapping record we will be putting in
the map-cache. For that, we need to know the locator interface of a
host. Just to clarify, for OpenStack environment is going to be the
tenant network interface.
This patch adds the functionality of listening to VbridgeTopology for
topology updates in Virtual Bridge Domain and pre-populate host to
locator information base.
Change-Id: Ic4c0844ccfcb972e46122df875b628c789f8c837
Signed-off-by: Shakib Ahmed <sheikahm@cisco.com>
Vina Ermagan [Thu, 16 Mar 2017 22:41:14 +0000 (22:41 +0000)]
Merge "Add timeout to subscriber data"
Vina Ermagan [Thu, 16 Mar 2017 18:22:18 +0000 (18:22 +0000)]
Merge "Provide Karaf 4 features"
Lorand Jakab [Mon, 20 Feb 2017 22:10:29 +0000 (00:10 +0200)]
Add timeout to subscriber data
The timeout of a subscriber is immutable too, and needs to be
transmitted in a MappingChanged notification using the YANG modeled
subscriber data.
Change-Id: Id62485c79b92ba35c6cd682c78a888ed14234059
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Mon, 5 Dec 2016 12:15:51 +0000 (14:15 +0200)]
SMR parent prefix continued
This patch completes work started in
https://git.opendaylight.org/gerrit/#/c/47120/
First, it handles the insertion of a positive mapping that overlaps
a negative one, by removing the negative mapping and generating SMRs for
its subscribers. This leads to pushing the positive mapping to the right
subscribers and generating only the needed negative mappings.
Second, it merges negative mappings when a positive mapping is removed
(reversing what was done in the first point above).
Additionally, it creates a variable for the action considered the
default for negative mappings, and uses it in the right places instead
of harcoding Action.NativelyForward.
Change-Id: I6a8799d3af23c20a9797840b3e64f632a3e2ee91
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Fri, 6 Jan 2017 19:06:55 +0000 (21:06 +0200)]
Provide Karaf 4 features
This patch provides Karaf 4 features for lispflowmapping. The neutron
upstream doesn't have its Karaf 4 patch finalized and merged yet, so our
odl-lispflowmapping-neutron feature for Karaf 4 is not enabled yet (but
it is migrated and ready to be enabled).
Bug: 7628
Change-Id: Icf8d404f4d46c71557b1f6f088626c2fb75200c8
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Daniel Malachovsky [Fri, 10 Mar 2017 11:52:47 +0000 (12:52 +0100)]
GUI - Add default redirect on app start
Due Dlux split we removed Topology app from odl-dlux-core feature.
There were also default redirect to that app, so it was landing page, when
DLUX was started.
This patch adds default route to GUI in this project.
More info:
https://wiki.opendaylight.org/view/Weather#Rename_some_odl-dlux-.2A_features_to_odl-dluxapps-.2A
Also fixed errors caused by refactor in YangUI and Yangutils
Change-Id: I125ebc851b749a244b834627326833a71e8079ea
Signed-off-by: Daniel Malachovsky <dmalacho@cisco.com>
Lorand Jakab [Mon, 6 Mar 2017 10:10:18 +0000 (12:10 +0200)]
Bug 7818: xTR-ID timestamp updated only when merging
A warning was shown on the Map-Register fast path with default setting,
saying somthing "Could not update timestamp for EID 192.0.2.0/24 xTR-ID
E073DF027BF20605DC9B9518BA753345, no mapping found". Default settings
mean that mapping merge is not on and xTR-ID mappings are not stored, so
the warning is not warranted. And it's not just the warning, the extra
lookup at EVERY fast path Map-Register timestamp upadte has a
performance cost, so make sure it only happens when merge is on.
Change-Id: Id7a66ff8c3309bbcb4ef0bf014f2987149f02fe7
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Florin Coras [Fri, 3 Mar 2017 09:40:21 +0000 (01:40 -0800)]
Fix radix trie last byte prefix comparison
Change-Id: Id360576c7a6e59fc556d484d9f869985e8f1714a
Signed-off-by: Florin Coras <fcoras@cisco.com>
Signed-off-by: Shakib Ahmed <sheikahm@cisco.com>
Lorand Jakab [Mon, 27 Feb 2017 11:21:44 +0000 (13:21 +0200)]
Allow Address for getString()
For now getString only takes LispAddress arguments (and XtrId). Some
times we only have access to the Address inside LispAddress, and we want
to be able to print that too in a user friendly way.
Change-Id: I9568ec9b08dba5c78572140b467d982e32f592ab
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Shakib Ahmed [Fri, 3 Mar 2017 12:28:34 +0000 (04:28 -0800)]
Fix NPE related to mapping record in IT
Change-Id: Ica7dbdf7e054729f1586c7867d5a40a80ad2eab5
Signed-off-by: Shakib Ahmed <sheikahm@cisco.com>
Anil Belur [Fri, 3 Mar 2017 10:02:25 +0000 (20:02 +1000)]
Add missing <name> fields for pom.xml files
This is used by autorelease scripts to automatically parse which project
is failing a build and report to the mailing list automatically. We need
names in the format:
ODL :: <groupId> :: <artifactId>
This patch formats in the same format as found in the startup archetypes
patch found here: https://git.opendaylight.org/gerrit/52522
Change-Id: I9c822509f8cb035cd656f5b2ff8c427c85cf7176
Signed-off-by: Anil Belur <abelur@linuxfoundation.org>
Lorand Jakab [Mon, 27 Feb 2017 11:29:00 +0000 (13:29 +0200)]
Fix a NPE in SmrScheduler
Change-Id: Iae5bd84361d19b5a6f30f881b2fd5eced26eb171
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Michael Vorburger [Tue, 28 Feb 2017 22:54:59 +0000 (23:54 +0100)]
Bug 7182 related: Remove M2E lifecycle mapping
These should never be in individual projects anymore now, we handle this
centrally, either in odlparent, or
https://github.com/vorburger/opendaylight-eclipse-setup, or by
appropriate lifecycle-mapping-metadata.xml in a Maven plugin (that's
what Bug 7182 does for the yang-maven-plugin; this removes what makes
that not work for these projects).
Change-Id: Ic3e7f81560e54882a8c17bece500148637aa1087
Signed-off-by: Michael Vorburger <vorburger@redhat.com>
Lorand Jakab [Fri, 24 Feb 2017 09:46:26 +0000 (11:46 +0200)]
Improve OOR server config in tutorial resources
Change-Id: Ic39d735a78d5b07e83ce71546f9c96f6ba5b2f64
Reported-by: Rashmi Pujar <rashmi.c.pujar@gmail.com>
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Vina Ermagan [Tue, 21 Feb 2017 06:44:03 +0000 (06:44 +0000)]
Merge "Time Bucket DS for handeling SB mapping timeout"
Lorand Jakab [Fri, 17 Feb 2017 21:33:55 +0000 (23:33 +0200)]
Bug 6071: Fix fast path Map-Notify auth data
Change-Id: I40884696576be4b311930312e8e09148f15810bb
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Shakib Ahmed [Thu, 1 Dec 2016 19:19:28 +0000 (11:19 -0800)]
Time Bucket DS for handeling SB mapping timeout
Mapping records in map caches have expiration time. When
mapping records surpass their expiration time, they are
supposed to be expired, meaning, removed from the map cache.
Right now the mapping records are expired in a lazy manner,
meaning, we query and handle the validity of the mapping only
when the mapping are beign asked for which introduces the
possibility of mapping records being in map cache indefinitly
if they are not requested for.
Time Bucket data structure maintains mapping records and
delete them upon expiration in a non-lazy manner.
Change-Id: I664249cc5c5d14cb54759369c6f982aef89a0ca3
Signed-off-by: Shakib Ahmed <sheikahm@cisco.com>
Shakib Ahmed [Sat, 11 Feb 2017 00:17:16 +0000 (16:17 -0800)]
Simplify and centralize sb mapping removal
Whenever we remove a mapping from smc, we also need to remove the
necessary data from data store. Now all the mapping removal are in
seperate places in mapping system making it hard to track whether all
the mapping removal call removing all the data properly.
The goal of the patch is to simplify mapping removal and centralize
all the mapping removal calls in few methods which will remove the
mapping properly from everywhere.
Change-Id: Ia279c019a186e3cdc349eb912abd499bde3e3802
Signed-off-by: Shakib Ahmed <sheikahm@cisco.com>
Lorand Jakab [Tue, 14 Feb 2017 11:27:24 +0000 (13:27 +0200)]
Bug 7789: Don't break DLUX when loading LISP UI
This simple patch just fixes the breakage caused by loading the LISP UI,
which results in an empty page, and the inability to access any of the
already loaded DLUX apps. However, making the LISP UI functional again
will require more work.
Change-Id: I8b2a08c93eeb1a77acc5496cc26a67f99f5aa825
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Shakib Ahmed [Mon, 13 Feb 2017 22:41:40 +0000 (14:41 -0800)]
Fix default mapping data for MappingSystemTest
Change-Id: I3193fc66474f78dedad7a3034326168d3520265c
Signed-off-by: Shakib Ahmed <sheikahm@cisco.com>
Lorand Jakab [Fri, 10 Feb 2017 16:38:32 +0000 (18:38 +0200)]
Checkstyle: fix ParenPad violations
Enforcing no space(s) after a left parenthesis and before a right
parenthesis is a proposed change [0] which can only be merged when all
violations are fixed in projects enforcing checkstyle. This patch fixes
those violations.
[0] https://git.opendaylight.org/gerrit/#/c/51316/
Change-Id: Ib4c793428f67e465a3caa6ccbf49364c3ca3c905
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Vina Ermagan [Fri, 10 Feb 2017 03:41:39 +0000 (03:41 +0000)]
Merge "Make part of the Subscribers object modeled"
Michael Vorburger [Thu, 9 Feb 2017 00:07:52 +0000 (01:07 +0100)]
Use org.osgi:org.osgi.core instead of org.eclipse.tycho's osgi (Equinox)
see https://git.opendaylight.org/gerrit/#/c/51520/
I'll like to block use of dependency tycho ASAP.
Change-Id: Iacb6b5efe337eed156a44dc32528445eda99fd06
Signed-off-by: Michael Vorburger <vorburger@redhat.com>
Lorand Jakab [Tue, 31 Jan 2017 16:33:03 +0000 (18:33 +0200)]
Make part of the Subscribers object modeled
This is a first step in solving bug 7272 and propagating subscriber data
to slave nodes in a cluster. This patch makes the YANG additions,
renames the SubscriberRLOC class to Subscriber, which makes more sense
(originally the SubscriberRLOC object only held the RLOC, now it holds
the source EID too), and uses the modeled SubscriberAddresses inside
Subscriber to store the source RLOC and EID.
Change-Id: I2e3ae71e7c9948d962921553b00cfecd76f445f1
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Vina Ermagan [Tue, 31 Jan 2017 02:04:53 +0000 (02:04 +0000)]
Merge "Remove xtr id mappings on mapping overwrite"
Miroslav Toth [Wed, 19 Oct 2016 11:22:00 +0000 (13:22 +0200)]
SMR parent prefix
Change-Id: I54ee2460012a0ac08a2cc0333e2ce0d9ae98953b
Signed-off-by: Miroslav Toth <mirtoth@cisco.com>
Shakib Ahmed [Mon, 23 Jan 2017 21:43:29 +0000 (13:43 -0800)]
Remove xtr id mappings on mapping overwrite
If there are 2 xtrs with different merge mapping configuration
registers same eid, we may or may not have merged mapping for
the eid in Map Cache but we will have xtr id specific mapping
in XTRID_RECORDS all the time. When we have xtr id specific
mapping from the xtr with merge configuration, but overwritten
mapping record in RECORD, then the xtr id specific mappings
may become untracked and hence not removed ever.
This patch removes previous xtrid specific mappings too when
mapping records are being overwritten.
Change-Id: Iaf8686367fb881b5794a3bf232e6c2e12d5ec6f8
Signed-off-by: Shakib Ahmed <sheikahm@cisco.com>
Lorand Jakab [Fri, 20 Jan 2017 14:56:17 +0000 (16:56 +0200)]
Add virtual parent sibling prefix lookup methods
Change-Id: I6fa317cbb7cdd0bcef2531b761e1c2dc19d1ce6d
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Fri, 20 Jan 2017 18:13:03 +0000 (20:13 +0200)]
Bug 7586: Make Extended SingleFeatureTest pass
We missed the fact that southbound is a dependency for mappingservice,
because of the RPCs to send reply packets. This patch corrects the
feature dependency and allows the new extended SingleFeatureTest to
pass, so we can be removed from the feature blacklist.
Change-Id: I5aadb06cdb2c9ce1bd9c1c3b4efdade53e4a1bcc
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Vina Ermagan [Fri, 20 Jan 2017 01:42:00 +0000 (01:42 +0000)]
Merge "Fix removal of expired mappings when merge on"
Vina Ermagan [Fri, 20 Jan 2017 01:40:57 +0000 (01:40 +0000)]
Merge "Remove SimpleMapCache#getXtrIdTable()"
Lorand Jakab [Wed, 21 Dec 2016 11:11:13 +0000 (13:11 +0200)]
Cleaning operational should clean map-cache too
The recently added remove-all-operational-content RPC removes the
operational datastore content, but the in-memory southbound map-cache is
not cleaned on master nodes, since southbound operations go to the cache
first. This patch explicitly clears the southbound map-cache on master
nodes when the RPC is called.
Change-Id: I3c1b899de592f5499779c55fb880481a60f9a64a
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Tue, 17 Jan 2017 21:12:14 +0000 (23:12 +0200)]
Fix removal of expired mappings when merge on
Change-Id: Ie1f1752471f72400788d28e81fd7f41ede6273f7
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Tue, 17 Jan 2017 19:43:09 +0000 (21:43 +0200)]
Remove SimpleMapCache#getXtrIdTable()
Change-Id: I53394480d41a5f0eb6356dfed24e293fbc5dfc19
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Wed, 21 Dec 2016 15:50:09 +0000 (17:50 +0200)]
Separate authentication key in-memory storage
Authentication key operations (add/get/remove) are part of the IMapCache
interface, so all map-caches implement them, even though we only really
use them in SimpleMapCache. This makes map-caches less "general", in the
sense that they need to be aware of the AuthenticationKey type, and we
have unused code.
This patch creates a new IAuthKeyDb interface and the AuthKeyDb class
implementing it, which contains only the authentication key operations.
The MappingSystem and the LispSouthBoundPlugin are switched to the new
implementation.
This is a prerequisite for a full solution to bug 7272, by making it
possible to remove all sub-keys when removing a mapping, leaving the
RadixTrie in a consistent state. It also makes a proper implementation
for the newly introduced remove-all-operational-content RPC, since
deleting all operational content should delete all southbound mappings,
but not authentication keys, which are now in the same in-memory
structures.
Checked system tests, all passing.
Change-Id: I9536dbccb06d486e205f8a15e9e8ac3fa1c53fcc
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Lorand Jakab [Tue, 17 Jan 2017 11:13:53 +0000 (13:13 +0200)]
Remove the last remnants of the CSS
We don't use the config sub-system (CSS) anywhere in our project, so
remove the dependecy completely.
Change-Id: Ic2d0de34907ef4273339d50796aca594c77d8b05
Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Vina Ermagan [Mon, 16 Jan 2017 23:40:05 +0000 (23:40 +0000)]
Merge "Account for merging on mapping removal"
Lorand Jakab [Wed, 4 Jan 2017 21:34:10 +0000 (23:34 +0200)]
Fix NPE in SimpleMapCache
This was causing a serious performance regression too, introduced with
Gerrit c/46778, "Make SimpleMapCache simple again".
Change-Id: I531a5f103322f5389b20197895a381aceef79fa5
Signed-off-by: Lorand Jakab <lojakab@cisco.com>