#opendaylight-ovsdb: ovsdb_weekly_call

Meeting started by tbachman at 19:59:20 UTC (full logs).

Meeting summary

    1. https://meetings.opendaylight.org/opendaylight-ovsdb/2015/ovsdb_weekly/opendaylight-ovsdb-ovsdb_weekly.2015-02-17-20.00.html Previous week’s meeting minutes (tbachman, 19:59:48)

  2. Agenda (tbachman, 19:59:53)
    1. Agenda item: Trello Board (tbachman, 20:01:31)
    2. Agenda item: Live-coding of port addition from edwarnicke (tbachman, 20:03:47)

  3. Trello Board (tbachman, 20:03:52)
    1. the model migration has been pushed through (tbachman, 20:04:16)
    2. The neutron work: the project’s been renamed, the artifact’s been renamed; the features tests are running (tbachman, 20:04:35)
    3. https://trello.com/b/ddIvDQE0/ovs-openstack OVSDB Trello Board (tbachman, 20:04:49)
    4. https://trello.com/b/lKnNc7PB/ovsdb-mdsal-migration MD-SAL migration Trello board (tbachman, 20:05:01)
    5. https://trello.com/b/ctunVAEr/ovsdb-mdsal-southbound MD-SAL southbound Trello board (tbachman, 20:05:10)
    6. flaviof says with edwarnicke’s change, we can use a different feature than odl-nsf-all, which enables using neutron feature from new project instead of the one from odl-nsf-all (tbachman, 20:06:23)
    7. There’s an M1/M2 release plan from regXboi for the Neutron project (tbachman, 20:07:49)
    8. flaviof worked with mestery on the Jenkins Job Builder migration; currently getting keystone failures. flaviof is going to ping someone who can help with this issue (tbachman, 20:08:30)
    9. https://gist.github.com/anonymous/03d764b7796741489e58 stack issues while using openstack gerrit. Work in progress. (flaviof, 20:09:34)
    10. vishnoianil is working on work to support VTN’s needs for Lithium (tbachman, 20:10:19)
    11. plaurin is making progress on setting up their environment; (tbachman, 20:11:03)
    12. repenno pushed code to parse the bridges; working on code to parse ports, relationship of bridge to ports etc; will commit result to same data store as the node, which would enable interested dataChangeListeners (tbachman, 20:13:05)
    13. snackewm asks if edwarnicke is working on a port augmentation (tbachman, 20:14:06)
    14. edwarnicke says that’s the piece that we’re going to live-code in today’s call (tbachman, 20:14:35)
    15. snackewm says he put up a patch yesterday for this work (tbachman, 20:15:06)
    16. repenno says that edwarnicke’s patch doesn’t conflict with his — will probably use some of the stuff in edwarnicke’s patch (tbachman, 20:15:36)
    17. shague asks if vishnoianil’s patch covered the 3 cards he had (tbachman, 20:16:12)
    18. vishnoianil says it covers the “report existing bridge” card (tbachman, 20:16:22)

  4. Live Coding by edwarnicke (tbachman, 20:16:30)
    1. edwarnicke says writes to data store are transaction; with multiple OVSDB monitors, this can cause issues with the transactions (tbachman, 20:19:15)
    2. edwarnicke says the MD-SAL supports transaction chains — allows ordering effect across transactions, even if the transactions are from multiple contexts (tbachman, 20:19:53)
    3. colindixon asks if that’s what transaction chains were for — thought they were just a way of batching transactions, where batching stops at point where an operation in the chain fails (tbachman, 20:20:35)
    4. edwarnicke says yes, it does that as well (tbachman, 20:21:18)
    5. flaviof asks if there are rollbacks (tbachman, 20:22:05)
    6. edwarnicke says the transactions that worked go through — it’s only the ones after the failure that aren’t tried (tbachman, 20:23:08)
    7. edwarnicke says that if a transaction in a chain fails, then the right behavior is to retry starting with the failed transaction (tbachman, 20:23:33)
    8. shague says if you failed the first time, aren’t you likely to fail the second time? (tbachman, 20:23:46)
    9. edwarnicke says not necesarilly — failure could be b/c someone wrote to the same element of the subtree before you could (tbachman, 20:24:14)
    10. colindixon says this is made easier if you use WriteOnlyTransaction — should just succeed, and blows away whatever was there before (tbachman, 20:25:13)
    11. colindixon says a ReadWriteTransaction is more likely to fail (tbachman, 20:25:22)
    12. edwarnicke suspects that’s not true, b/c of the case of deletions — someone deletes when someone is trying to do an update (tbachman, 20:25:43)
    13. colindixon says there are cases where it will fail, but it will fail a lot less often (tbachman, 20:25:55)
    14. edwarnicke talks about the command pattern (tbachman, 20:26:27)
    15. please refer to the recording at about 20-25 mins in for walk-thru of this code — too challenging to type the code and the folks talking about it :( (tbachman, 20:39:05)
    16. https://www.dropbox.com/s/l9bah3vby46wcqn/Screenshot%202015-02-24%2015.48.31.png?dl=0 topological pic of bridges and ports/etc (flaviof, 20:49:28)
    17. http://openvswitch.org/ovs-vswitchd.conf.db.5.pdf where it came from ^^ :) (flaviof, 20:49:51)
    18. http://blog.scottlowe.org/2012/10/19/link-aggregation-and-lacp-with-open-vswitch/ blog describing how to set up channel bonding (tbachman, 21:05:15)


Meeting ended at 21:06:42 UTC (full logs).

Action items

  1. (none)


People present (lines said)

  1. tbachman (63)
  2. odl_meetbot (5)
  3. flaviof (3)
  4. mlemay (2)
  5. shague (1)
  6. grmontpetit (1)
  7. plaurin_ (1)
  8. plaurin (1)


Generated by MeetBot 0.1.4.