==================================
#opendaylight-meeting: MD-SAL call
==================================


Meeting started by edwarnicke at 16:05:49 UTC.  The full logs are
available at
http://meetings.opendaylight.org/opendaylight-meeting/2014/md_sal_call/opendaylight-meeting-md_sal_call.2014-05-20-16.05.log.html
.



Meeting summary
---------------

* What was done last week  (edwarnicke, 16:06:34)
  * Big memory reductions  (edwarnicke, 16:06:44)
  * stability improvements  (edwarnicke, 16:06:51)
  * LINK:
    https://jenkins.opendaylight.org/integration/job/integration-csit-base-of13/
    - all OF13 integration tests passing  (edwarnicke, 16:07:33)
  * LINK:
    https://docs.google.com/spreadsheets/d/1O5PxEM1eLszMuWf9M11AD5w7siT9vHOkzR85gjCB-6c/edit#gid=322436364
    edwarnicke provided a spreadsheet to track these items  (tbachman,
    16:08:14)

* What will be don (what's planned)  (tbachman, 16:08:42)
  * bug 1010, Return from Restconf the errors as definied in the IETF
    draft  (tbachman, 16:09:03)
  * /modules and /streams failing, fix should be coming as well
    (tbachman, 16:09:25)
  * Netconf connector - return the netconf errors back in the RPC result
    (allow the RPC errors to be passed through). waiting on a parent
    patch.  (tbachman, 16:11:40)
  * Ask question needs some focus:
    https://ask.opendaylight.org/question/99/how-does-request-routing-works/
    (devinavery, 16:12:36)
  * Tony - is writing down the explanation  (devinavery, 16:12:55)
  * LINK:
    https://ask.opendaylight.org/question/99/how-does-request-routing-works/
    - question no how request routing works  (edwarnicke, 16:13:36)
  * bug 990 and one other, where union types aren't getting translated
    in binding independent and binding aware correctly  (tbachman,
    16:13:37)
  * ACTION: Tony to answer
    https://ask.opendaylight.org/question/99/how-does-request-routing-works/
    (edwarnicke, 16:13:42)
  * ttkacik merged fix for 990  (tbachman, 16:15:52)
  * rovarga has about 10 patches to address the data store  (tbachman,
    16:18:00)
  * devinavery et. al. have stuff in the works to add mount points, etc.
    to all plugin of 3rd party protocols  (tbachman, 16:18:31)

* Other Topics upcoming  (tbachman, 16:21:21)
  * clustering in the MD-SAL  (tbachman, 16:21:32)
  * there are discussions happening on the discuss list about this now
    (colindixon, 16:23:34)
  * MD_SAL Status call minutes have been sent to the controller-dev and
    yangtools-dev mailing lists  (tbachman, 16:25:20)
  * edwarnicke recommends using balance of time to cover the email that
    tkacik was prreparing  (tbachman, 16:27:11)
  * clustered data store discussion moved to next week's MD-SAL meeting
    (tbachman, 16:27:28)
  * tkacik covers How does request routing works  (tbachman, 16:28:20)
  * Question was how does request routing work on a per node/host basis
    (tbachman, 16:28:39)
  * How does ODL invoke the right plugin for the right node?  (tbachman,
    16:30:02)
  * Confusion comes from looking at AD-SAL vs. MD-SAL  (tbachman,
    16:30:27)
  * MD-SAL doesn't have the notion of node type.  (tbachman, 16:30:51)
  * MD-SAL instead allows you to register an RPC in the context of a
    node.  (tbachman, 16:31:21)
  * For example, the OpenFlow plugin registers RPCs for flow programming
    for a given node.  (tbachman, 16:31:43)
  * That causes flow programming for that node to be routed to the
    OpenFlow plugin  (tbachman, 16:32:10)
  * edwarnicke asks what is the problem that is being solved  (tbachman,
    16:34:16)
  * Solaris doesn't support OVS, so they want to provide their own
    plugin so that it handles Solaris nodes differently  (tbachman,
    16:34:53)
  * Using the example of configuring a VLAN, edwarnicke recommends
    creating a set-vlan RPC in the NB model, and the solaris plugin
    would register itself to create VLAN RPCs for nodes discovered by
    the solaris plugin.  (tbachman, 16:36:42)
  * Can mutliple SB plugins register for the same property change on the
    same node?  (tbachman, 16:39:26)
  * Only one SB can register for the implementation per RPC.  (tbachman,
    16:39:55)
  * devinavery notes that this is exactly what they're putting together
    in their example.  (tbachman, 16:40:25)
  * New example we are working on is tracked with bug 1050  (devinavery,
    16:44:06)
  * rovargo notes that there needs to be a unified view of the topology.
    (tbachman, 16:46:20)



Meeting ended at 16:58:02 UTC.



People present (lines said)
---------------------------

* tbachman (37)
* edwarnicke (14)
* odl_meetbot (5)
* devinavery (3)
* colindixon (3)
* icbts (2)
* Madhu (1)
* alagalah (1)
* CASP3R (0)



Generated by `MeetBot`_ 0.1.4