========================================================== #opendaylight-integration: Integration Team Weekly Meeting ========================================================== Meeting started by dfarrell07 at 15:02:13 UTC. The full logs are available at http://meetings.opendaylight.org/opendaylight-integration/2015/integration_team_weekly_meeting/opendaylight-integration-integration_team_weekly_meeting.2015-07-02-15.02.log.html . Meeting summary --------------- * jamoluhrsen running the meeting for Luis today (dfarrell07, 15:02:36) * LINK: https://trello.com/b/ACYMpTVD/opendaylight-integration Trello (dfarrell07, 15:04:32) * LINK: https://wiki.opendaylight.org/view/CrossProject:Integration_Group:Controller-Cluster_Deployment New clustering wiki to review/edit (dfarrell07, 15:09:47) * Everyone should raise tasks that they think should be done in Be via Trello (dfarrell07, 15:11:55) * Build boxes will be upgraded to Cent7 in the next few weeks (huzzah!) (dfarrell07, 15:13:47) * LINK: https://jenkins.opendaylight.org/sandbox/job/ovsdb-csit-1node-cds-netvirt-only-master/7/robot/report/log.html (grmontpetit, 19:54:17) * LINK: https://wiki.opendaylight.org/view/RelEng/Builder/Jenkins#Patch_Test_Job (LuisGomez, 18:06:10) * jenkins had issues last night b/c of left over java and mininet processes (tbachman, 15:03:22) * zxiiro cleaned that up this morning (tbachman, 15:03:29) * zxiiro says one of the problems with the jclouds plugin is that it doesn’t have a concept of state — it doesn’t know how many VMs are spun up, and just uses the rackspace API to spin up VMs (tbachman, 15:03:57) * zxiiro says nodepool has a concept of state, to track which VMs are active, so it doesn’t need to keep calling the rackspace APIs (tbachman, 15:04:16) * zxiiro says tykeal is on PTO, but is going to look into this when he gets back (tbachman, 15:04:32) * Note that this is the Integration Team's weekly meeting, "TSC" in the title was a silly mistake on my part (dfarrell07, 15:04:45) * LuisGomez asks about whether or not the LF has brought a resource online outside of North American time zones (tbachman, 15:06:02) * zxiiro says there is someone who’s coming online soon (tbachman, 15:06:12) * News (tbachman, 15:08:59) * To submit tickets to the helpdesk (for stuff like Jenkins outages) email helpdesk@opendaylight.org (dfarrell07, 15:09:03) * LuisGomez asks if anyone has gotten feedback on using Lithium (tbachman, 15:09:18) * dfarrell07 says there’s all kinds of activity on twitter about people using it (tbachman, 15:09:35) * LuisGomez says docs got negative feedback for the Helium release, and asked if anyone has had the chance to look at documentation for Lithium (tbachman, 15:10:12) * Integration should expect a continued flow of patches into Integration as projects catch up on test code (dfarrell07, 15:11:17) * jamoluhrsen asks how much longer we want to keep the stable/helium and stable/lithium jobs around (tbachman, 15:12:39) * LuisGomez says stable/helium will be put in an End of Support (EOS) state now that Lithium out, though there are SR4 and SR5 releases scheduled (tbachman, 15:13:30) * Note that the TSC will vote to officially EOS/EOL Helium in a few months. At that point, we can remove stable/helium jobs. (dfarrell07, 15:15:56) * LuisGomez says the distribution code for stable/lithium is frozen for now; only changes supported now are in master (tbachman, 15:18:01) * Integration committers should cherry-pick *testing* code from master to Lithium for now, to keep max testing there. Once that becomes too complex (subjective, expect emails and other discussion), we can stop cherry-picking to Lithium. Distribution code should only go to master. (dfarrell07, 15:18:36) * LuisGomez notes that we’ve had to remove projects from a stable/ release in the past due to projects being abandoned; in these cases, the TSC has to make this decision (tbachman, 15:19:45) * vratko asks about what if a projects breaks something just before an SR but is slow in responding/fixing the breakage (tbachman, 15:20:33) * Vratko asks if it's possible to rename jobs, and if it can be automated. zxiiro says JJB doesn't support renaming jobs. A discussion (generally around branching/jobs) continues. (dfarrell07, 15:23:25) * Vratko clarifies that it's totally possible to change Jenkins names via the UI (not supported by JJB, as noted above) (dfarrell07, 15:26:15) * Trello updates (dfarrell07, 15:26:31) * LINK: https://trello.com/b/ACYMpTVD/opendaylight-integration Trello board (dfarrell07, 15:26:46) * LuisGomez is going to push a job for stability testing sometime today — one day longevity test for openflow (tbachman, 15:27:09) * LuisGomez says that job will probably run on the weekend — maybe Saturdays (tbachman, 15:27:28) * PhilShea has his last fix for clustering tests, and is waiting for that gerrit/job to be merged (tbachman, 15:28:26) * PhilShea has been working hard on cluster tests, he reports that they are ready for review/merge (/me is excited) (dfarrell07, 15:28:37) * We have a newly open TODO around the DSBenchmark test (Datastore Benchmark, via Jan and other Cisco test ppl I think), the person working on it is no longer working on ODL Integration (dfarrell07, 15:30:54) * Radhika is interested in working on the DSBenchmark test (dfarrell07, 15:31:47) * LuisGomez recommends running the DSBenchmark test as a single-node instance before trying a multi-node test (tbachman, 15:32:44) * LINK: https://git.opendaylight.org/gerrit/#/c/23928/ (PhilShea, 15:36:53) * Veryx demo of PktBlaster (tbachman, 15:38:24) * PktBlaster is a software based solutino that runs on x86, with support for OpenFlow 1.0/1.3, REST API support for automation (tbachman, 15:39:01) * It can do performance testing and network emulation (tbachman, 15:39:12) * A test setup is shown using an Intel i7 server, 8GB of RAM connected to a DELL R420 2 CPU Xeon with 16GB RAM, which implements the server (tbachman, 15:40:08) * jamoluhrsen asks if just REST calls can be used (i.e. not TCL) (tbachman, 15:40:38) * The REST calls are wrapped into TCL scripts (tbachman, 15:40:49) * the demo requires 1) setting the controller and configuration; 2) select network configuration; 3) select the test type, duration, and number of iterations (tbachman, 15:41:58) * LuisGomez asks what’s expected if an errored packet is received (tbachman, 15:45:42) * The controller should be (silently) dropping the errored packet (tbachman, 15:45:52) Meeting ended at 16:19:26 UTC. People present (lines said) --------------------------- * odp-gerritbot (105) * LuisGomez (64) * tbachman (44) * ChUsama (39) * dfarrell07 (33) * grmontpetit (29) * jamoluhrsen (21) * PhilShea (18) * icbts (14) * raphaelamorim (13) * odl_meetbot (7) * catohornet (4) * tykeal (2) * mgkwill (2) * zxiiro (2) * kailash (1) Generated by `MeetBot`_ 0.1.4