#opendaylight-meeting: tsc
Meeting started by colindixon at 17:00:29 UTC
(full logs).
Meeting summary
- roll call and agenda bashing (colindixon, 17:00:37)
- colindixon (colindixon,
17:00:39)
- TSC members please #info in (phrobb1,
17:00:51)
- https://wiki.opendaylight.org/view/TSC:Main#Agenda
the agenda in it’s usual place (colindixon,
17:00:54)
- https://meetings.opendaylight.org/opendaylight-meeting/2015/tsc/opendaylight-meeting-tsc.2015-04-23-17.00.html
last week’s meeting minutes (colindixon,
17:01:44)
- edwarnicke (edwarnicke,
17:02:05)
- gzhao volunteered to work with mlemay and
edwarnicke on the better single feature test (colindixon,
17:02:19)
- we had new committers promoted on both CAPWAP
and GBP on the mailing list (colindixon,
17:02:44)
- jmedved (jmedved,
17:02:50)
- https://wiki.opendaylight.org/view/Weather#Blocking_Bugs_definition
Working definition of blocking bugs, #action of dfarrell07 and
LuisGomez (dfarrell07,
17:02:55)
- dlenrow (dlenrow,
17:03:00)
- LuisGomez (LuisGomez,
17:03:11)
- mohnish anumala (mohnish,
17:03:46)
- ACTION: colindixon to
add the moxy issue to the wiki page with the workaround listed as
being ping Ryan and/or Ed (colindixon,
17:04:09)
- Uri is on the phone (colindixon,
17:04:52)
- Youcef Laribi (Youcef,
17:05:14)
- events (colindixon, 17:06:13)
- http://www.opendaylight.org/news/events/
the main events page (colindixon,
17:06:22)
- NFV World Congress is next week in CA, dev
track will happen (dfarrell07,
17:06:32)
- OpenStack summit will be soon, ODL will have a
presence (dfarrell07,
17:06:53)
- ONS will have an ODL dev track, submit CFPs
ASAP (dfarrell07,
17:07:09)
- ODL Summit CFP has been completed, dev track
talks will start at a later date (dfarrell07,
17:07:28)
- [if you’re looking for more info, these are all
linked in from the events page above (colindixon,
17:07:33)
- http://www.opendaylight.org/news/events
ODL Events (dfarrell07,
17:07:52)
- phrobb1 and dneary are working together to make
it easier for downstream ODL consumers to give us feedback (think
OPNFV) (dfarrell07,
17:08:21)
- lithium and stable/helium updates (dfarrell07, 17:08:48)
- https://wiki.opendaylight.org/view/Weather
the weather page to track any known isssues in Lithium (colindixon,
17:09:14)
- gzhao doesn't have much to report about
Li/stable He updates (dfarrell07,
17:09:15)
- Discussion about stable branch SR release
cutting, that they could be better, colindixon asks for help from LF
admins to make those docs better (dfarrell07,
17:10:52)
- ACTION: zxiiro to
write better description of how to cut your stable/lithium project,
e-mail that out and help projects do that appropriately (colindixon,
17:11:24)
- https://wiki.opendaylight.org/view/Weather#Blocking_Bugs_definition
candidate definition of what would be release blocking bugs that
LuisGomez and dfarrell07 worked on the last week (colindixon,
17:12:17)
- please edit the wiki or reach out to the
communtiy (mailing list thread to be started) if you want to provide
input (colindixon,
17:12:54)
- edwarnicke suggests that spreadsheet used by
gzhao in the past to tracking blocking bugs was useful, discussion
about using that vs wiki (dfarrell07,
17:14:34)
- There isn't currently an Li bug tracking
spreadsheet (dfarrell07,
17:16:09)
- Using the to-be-created spreadsheet above seems
to be (by consensus) the right/authoritative source for blocking bug
info, link to it from wiki (dfarrell07,
17:17:05)
- ACTION: gzhao to
provide a link from the weather page to the release blocking bugs
tab (colindixon,
17:17:49)
- ACTION: gzhao to add
a tab to the release tracking spreadsheet for tracking release
blocking bugs (colindixon,
17:18:10)
- ACTION: gzhao to send
mail to TSC and release noting that the spreadsheet will be the
authoritative list of blocking bugs (colindixon,
17:19:01)
- System Integration and Testing (dfarrell07, 17:19:36)
- LuisGomez has lots of cool perf stuff to
report (dfarrell07,
17:19:56)
- Integration team has been doing a lot of work
to add perf/scale tests to CI (dfarrell07,
17:20:38)
- during the hackfest and since then the
integration team has been working to get scalability and performance
tests into CI (colindixon,
17:20:43)
- so far, this has been focused on
openflowplugin, but in the future the goal is to test more projects
and have projects provide these tests not just integration
(colindixon,
17:21:34)
- Integration is hoping in future releases,
projects will be able to build on the examples we created during Li
to build their own tests (dfarrell07,
17:21:46)
- LuisGomez shows OF PktIn test result graphs in
CI (dfarrell07,
17:22:08)
- https://jenkins.opendaylight.org/releng/view/openflowplugin/
the jobs can be see in the openflowplugin for now (colindixon,
17:22:37)
- NB REST API add/del tests also added to CI,
LuisGomez shows plot (dfarrell07,
17:23:33)
- https://jenkins.opendaylight.org/releng/view/openflowplugin/job/openflowplugin-csit-1node-cds-cbench-performance-only-master/plot/
plots of cbench performance (colindixon,
17:23:43)
- https://jenkins.opendaylight.org/releng/view/openflowplugin/job/openflowplugin-csit-1node-cds-config-performance-only-master/plot/Config%20Performance/
plots of RESTCONF performance for add and delete (colindixon,
17:25:06)
- jmedved says that he needed 8 python clients to
get the max RESTCONF performance and he was finding something like
2,000 RESTCONF ops/sec (colindixon,
17:26:36)
- The current NB REST tests in CI may be blocked
on threads dedicated to Python client, Integration will address that
feedback and adjust test out-of-band (dfarrell07,
17:27:25)
- colindixon asks how we’re persisting this data,
LuisGomez says not any differently right now (colindixon,
17:27:56)
- ACTION: LuisGomez and
tykeal to work out retaining the performance metrics for as long as
we can (colindixon,
17:28:22)
- Data from plots isn't currently long-term
persistent, there have been discussions about that via zxiiro and
LuisGomez but there's still uncertainty about how to do so
(dfarrell07,
17:28:37)
- LuisGomez demos long-term CBench (so OF PktIn)
test, jamoluhrsen gives details, same test but 1 hour long
daily (dfarrell07,
17:30:03)
- https://jenkins.opendaylight.org/releng/view/openflowplugin/job/openflowplugin-csit-periodic-1node-cds-cbench-daily-only-master/plot/
this is the same cbench numbers but instead of running for 4
minutes, it runs for an hour once per day to get longer time (colindixon,
17:30:17)
- jamoluhrsen says that we can store the graphs
forever (or a whole release) if we want (colindixon,
17:30:45)
- jamoluhrsen reports that there seems to be
support for storing results long-term in the Jenkins Plot
Plugin (dfarrell07,
17:30:49)
- LuisGomez shows max-number-of-switches scale
test in CI (dfarrell07,
17:31:03)
- This test is really cool, basically ups number
of switches until controller breaks and plots how high we get
(dfarrell07,
17:31:28)
- https://jenkins.opendaylight.org/releng/view/openflowplugin/job/openflowplugin-csit-periodic-1node-cds-scalability-daily-only-master/plot/Inventory%20Scalability/
scalabilty testing the max number of switches in a linear topology:
right now we hit an OOM at 400 mininet switches (colindixon,
17:32:25)
- Switch # scale test uses Mininet (someone
asked) (dfarrell07,
17:33:20)
- mohnish asks about details around how this is
done (colindixon,
17:33:33)
- Discussion about other types of scale
(Integration is well aware that we want *all the scale
tests*) (dfarrell07,
17:34:33)
- mohnish notes that we need to be careful about
what we report in terms of # of switches, people might take that to
mean that we can handle that number of physical switches, or the
number of vswitches, or the number of hosts (colindixon,
17:34:48)
- LuisGomez says yes, we definitely want to test
more kinds of scale and we would love help getting them up and
going (colindixon,
17:35:15)
- LuisGomez also notes that right now the goal
isn’t the numbers per say but to make sure we look for any
regressions (colindixon,
17:35:47)
- Super important: we're not trying to report max
numbers with these tests, we're looking for regressions. Don't quote
these numbers as our max values! (dfarrell07,
17:36:02)
- along those lines, this is running things in a
VM without super amounts of tuning, so these results are likely well
below what actual max performance/scale numbers would be
(colindixon,
17:37:46)
- edwarnicke notes that mininet tends to flap a
lot (e.g., switches disconnecting and reconnecting) when get beyond
about 100 switches depending on the environment, we should make sure
we’re testing our controller not the test tools (colindixon,
17:39:05)
- colindixon has run out of adjectives to express
his joy (edwarnicke,
17:40:24)
- colindixon notes that Uri is jumping with joy
over the testing metrics shown by LuisGomez (phrobb1,
17:41:08)
- Future TODOs around putting all this in a
central dashboard-like place, writing docs to describe all of
these (dfarrell07,
17:42:03)
- ACTION: LuisGomez (or
somebody else) to send out an e-mail with links to the tests and a
few sentence descriptions of what they are (colindixon,
17:42:28)
- ACTION: snoble and
LuisGomez to connect and talk about trying to do some Ixia testing
for Lithium (colindixon,
17:42:54)
- Committer promotions (dfarrell07, 17:46:05)
- No one here from OVSDB, mailing lists
(dfarrell07,
17:46:19)
- Beryllium Planning (dfarrell07, 17:46:29)
- Phill will send out info to lists, please
review (dfarrell07,
17:46:42)
- introductions (colindixon, 17:46:49)
- ODL TSC welcomes odlcasey (dfarrell07,
17:47:32)
- Casey Cain is new to the project with the LF
folks and will be helping out with overall project management and
community management, on IRC he’s odl-casey (colindixon,
17:47:57)
- cookies (colindixon, 17:50:09)
Meeting ended at 17:50:12 UTC
(full logs).
Action items
- colindixon to add the moxy issue to the wiki page with the workaround listed as being ping Ryan and/or Ed
- zxiiro to write better description of how to cut your stable/lithium project, e-mail that out and help projects do that appropriately
- gzhao to provide a link from the weather page to the release blocking bugs tab
- gzhao to add a tab to the release tracking spreadsheet for tracking release blocking bugs
- gzhao to send mail to TSC and release noting that the spreadsheet will be the authoritative list of blocking bugs
- LuisGomez and tykeal to work out retaining the performance metrics for as long as we can
- LuisGomez (or somebody else) to send out an e-mail with links to the tests and a few sentence descriptions of what they are
- snoble and LuisGomez to connect and talk about trying to do some Ixia testing for Lithium
Action items, by person
- colindixon
- colindixon to add the moxy issue to the wiki page with the workaround listed as being ping Ryan and/or Ed
- gzhao
- gzhao to provide a link from the weather page to the release blocking bugs tab
- gzhao to add a tab to the release tracking spreadsheet for tracking release blocking bugs
- gzhao to send mail to TSC and release noting that the spreadsheet will be the authoritative list of blocking bugs
- LuisGomez
- LuisGomez and tykeal to work out retaining the performance metrics for as long as we can
- LuisGomez (or somebody else) to send out an e-mail with links to the tests and a few sentence descriptions of what they are
- snoble and LuisGomez to connect and talk about trying to do some Ixia testing for Lithium
- snoble
- snoble and LuisGomez to connect and talk about trying to do some Ixia testing for Lithium
People present (lines said)
- dfarrell07 (54)
- colindixon (52)
- odl_meetbot (10)
- dneary (5)
- phrobb1 (5)
- edwarnicke (4)
- ChrisPriceAB (3)
- ttkacik1 (1)
- dlenrow (1)
- jmedved (1)
- mohnish (1)
- icbts (1)
- snoble (1)
- LuisGomez (1)
- hideyuki (1)
- Youcef (1)
- evanz (1)
- gzhao (1)
- jamoluhrsen (1)
- tbachman (0)
Generated by MeetBot 0.1.4.