#opendaylight-meeting: MD-SAL interest call

Meeting summary

1. agenda bashing (colindixon, 16:06:32)
   1. https://wiki.opendaylight.org/view/MD-SAL_Weekly_Call#Agenda the details of the agenda and call (as always) (colindixon, 16:07:14)
   
   
2. new functionality (colindixon, 16:09:46)
   1. tony says that the enew write out codec to restconf is ready and waiting on verify before it getst merged (colindixon, 16:10:14)
   
   
3. questions (colindixon, 16:10:21)
   1. https://lists.opendaylight.org/pipermail/controller-dev/2014-August/006371.html this mailing list item hasn’t gotten the response we’d like (colindixon, 16:10:39)
   2. the question is how to get the parent of a node in the MD-SAL tree (colindixon, 16:22:48)
   3. the answer is to use InstanceIdenfier.firstIdentifierOf() and pass the class of the expected parent to get an instance identifier of the parent, which you can then use to read the parent (colindixon, 16:23:47)
   4. ramkumar asks if the same data change listener be used for multiple change registrations? (colindixon, 16:25:09)
   5. answer is that, yes, you can register twice, but there’s no promise about the ordering of events between the two registrations (colindixon, 16:26:17)
   
   
4. topics for the design forum (colindixon, 16:28:57)
   1. jan medved wants to hear things about clustering (both how it’s built and how it can be used) (colindixon, 16:29:36)
   2. brainstorming about binding specification (colindixon, 16:30:57)
   3. https://wiki.opendaylight.org/view/YANG_Tools:Design:Binding_Specification_v2_Analysis this page has been aggregating some ideas (colindixon, 16:31:06)
   4. https://bugs.opendaylight.org/show_bug.cgi?id=1411 this bug is the v2 spec which has all other bugs that require v2 spec to solve (colindixon, 16:32:19)
   5. colindixon suggests the config subsytem: (i) introductuion to basic use, (ii) how to extend with own classes, (iii) how to find what already exists vs. what I need to register again, (iv) how to make it easier to use [others adding the sub bullets] (colindixon, 16:33:59)
   6. how to model and represent hetergeous network topology (layers, etc.) (colindixon, 16:36:08)
   7. there’s a lot of discussion from ramkumar, rovarga, colindixon, and others about how we currrently do things (colindixon, 16:39:15)
   8. devinavery says that modeling best practices in general probably needs to be a topic either here or at the design forum (colindixon, 16:40:10)
   9. around modeling Uyen points out that there’s a lot of modeling silos right now, figuring out how to drag (at least some) models out to application-independent ways (colindixon, 16:43:16)
   
   
5. JMX stats for MD-SAL (colindixon, 16:43:45)
   1. see recording for the demo (colindixon, 16:44:53)
   2. tom pantelis shows getting status out of JMX, including things about data-store-executor, notification-executor, e.g., the active number of threads, completed tasks, etc. (colindixon, 16:48:01)
   3. jan medved asks about commit stats, TomP shows commit stats including average, longest, and shortest commit times, as well as total number of commits (colindixon, 16:51:35)
   4. TomP notes that the clustering stuff has more info too, but that would have to wait until later for a demo (colindixon, 16:58:23)

Action items

1. (none)

People present (lines said)

1. colindixon (28)
2. odl_meetbot (3)