01:01:58 <joehuang> #startmeeting tricircle
01:01:58 <openstack> Meeting started Wed Jul 26 01:01:58 2017 UTC and is due to finish in 60 minutes.  The chair is joehuang. Information about MeetBot at http://wiki.debian.org/MeetBot.
01:01:59 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
01:02:01 <openstack> The meeting name has been set to 'tricircle'
01:02:07 <joehuang> hello
01:02:27 <RongHui11> hi
01:02:30 <zhiyuan> hi
01:02:30 <joehuang> #topic roll call
01:02:34 <joehuang> #info joehuang
01:02:40 <RongHui11> #info ronghui
01:02:53 <dongfeng> #info dongfeng
01:03:19 <zhiyuan> #info zhiyuan
01:03:26 <Yipei> #info Yipei
01:03:46 <joehuang> #topic pike-3 release
01:04:10 <joehuang> according to our plan, we'd like to have pike-3 release this Friday
01:04:32 <joehuang> let's check whether there is risk
01:05:11 <joehuang> one by one for each feature?
01:06:31 <zhiyuan> there are two smoke tests(for trunk and service function chaining) waiting to be merged, but I think there's no risk
01:06:58 <Yipei> load balancing across instances residing in multiple network works, but have two issues
01:07:04 <joehuang> yes, for smoke test, it could be merged quickly if no bugs found
01:07:32 <joehuang> good , yipei, what's the issue
01:07:59 <Yipei> 1) it breaks bridge network environment
01:08:29 <Yipei> before adding a member in another region, instances in two regions are pingable
01:08:41 <Yipei> once an instance is added, they do not
01:09:09 <Yipei> once an instance is added as a member, they do not
01:10:01 <joehuang> east-west networking doesn't work?
01:10:07 <Yipei> yes
01:10:35 <joehuang> what's your networking topology?
01:11:11 <Yipei> the networking topology is created following https://github.com/openstack/tricircle/blob/master/doc/source/multi-pod-installation-devstack.rst
01:11:43 <Yipei> two instances in subnet1, two in subnet2
01:11:50 <Yipei> lb is in region two
01:12:48 <joehuang> have you update the code for get_subnet or get_subnets?
01:16:06 <Yipei> no, that's the second issue, i did use octavia api, instead of neutron-lbaas api
01:16:16 <joehuang> ok
01:16:39 <Yipei> still need to test neutron-lbaas api
01:17:12 <Yipei> i try to use neutron-lbaas api
01:17:22 <Yipei> following this doc https://wiki.openstack.org/wiki/Neutron/LBaaS/API_2.0#Add_a_New_Member_to_a_Pool
01:17:32 <joehuang> I found that the document for net creation should be updated with availability zone
01:17:46 <joehuang> because currently the default network is not local network
01:18:50 <joehuang> #info update network creation with availability zone in installation guide to create local network
01:19:23 <joehuang> because net1 and net2 are two cross neutron L2 network, I think there may be some issue
01:20:56 <joehuang> there is still some check test for QoS patch, this will be in great risk
01:21:19 <Yipei> should the url be "http://REGION_TWO_IP:9696/lbaas/pools/b23ac8e0-5349-4309-b96a-564c5bd04534/members", if using neutron-lbaas api?
01:22:19 <joehuang> if you install lbass in region two.
01:22:22 <zhiyuan> I reviewed the QoS code yesterday, policy module is imported in central plugin and cause the smoke test failure
01:23:59 <joehuang> all patches releated to tricircle client have been merged now
01:24:10 <dongfeng> 1) doc migration patch is under review. there maybe a risk. 2) job pagination patch is updated just now. it needs review. 3) I will write doc and release note for job. I think it can be merged in time.
01:24:29 <RongHui11> yes and we try to fix it and pleased give us some help
01:24:50 <joehuang> a patch for job pagination support in tricircle client is needed
01:25:00 <dongfeng> yes
01:25:22 <joehuang> for the patch can not land in time, need to ask for exception in mail-list
01:26:03 <joehuang> only if RFE(request for exception) is approved after discussion, then the feature could be merged in pike
01:26:37 <joehuang> because pike-3 is feature freeze milestone, after pike-3, usually no big feature should be merged for stability concerns
01:27:23 <joehuang> I also noticed that QoS not register a blueprint yet
01:27:39 <RongHui11> yes
01:27:53 <joehuang> though spec has been approved, it'll be good to register one blueprint for this feature
01:28:10 <joehuang> and for other features, please update the status of your blueprint
01:28:19 <RongHui11> ok
01:28:36 <joehuang> #link https://blueprints.launchpad.net/tricircle
01:29:21 <joehuang> all patches in review queue is here https://review.openstack.org/#/q/project:openstack/tricircle
01:29:52 <joehuang> doc migration should be merged in pike-3
01:30:01 <dongfeng> en
01:30:03 <joehuang> this is community wide task
01:30:29 <dongfeng> I update the doc migration patch, but I an not quite famliar with doc
01:31:26 <dongfeng> after your review I will make further modification
01:31:30 <zhiyuan> I notice that the doc test has been passed
01:31:36 <dongfeng> en
01:31:39 <joehuang> you can click the link gate-tricircle-docs-ubuntu-xenial
01:31:44 <joehuang> to view the result
01:31:50 <joehuang> the result seems ok
01:32:09 <dongfeng> :) zhiyuan pointed the problem
01:32:21 <dongfeng> I fixed last night.
01:32:28 <joehuang> for example, for the current compilation result : http://docs-draft.openstack.org/49/486849/6/check/gate-tricircle-docs-ubuntu-xenial/50b2845//doc/build/html/
01:32:46 <zhiyuan> once a test fails, you can click the link to check the log
01:33:00 <RongHui11> zhiyuan, for the QoS test problem, do you have some good way to fix yhis problem
01:33:48 <dongfeng> to zhiyuan, got it.
01:35:39 <zhiyuan> to ronghui, debug is needed. according to the log, central neutron server fails to access "keys" atttribute from a None object
01:36:23 <joehuang> #topic pike-rc
01:36:30 <zhiyuan> I add the link for log in my previous comment
01:36:57 <joehuang> ok, let's discuss next milestone, for qos, we can discuss in #openstack-tricircle after the meeting
01:37:09 <RongHui11> ok
01:37:34 <joehuang> according to community schedule: https://releases.openstack.org/pike/schedule.html
01:38:16 <joehuang> final release of pike is Aug 21 ~ 25
01:39:46 <dongfeng> after jul 28, can the tricircleclient and tricircle take new patches?
01:40:01 <joehuang> I suggest that we have pike-rc (release candidate) on Aug 22, if no critical and high bugs, then this release will be the final release
01:40:41 <joehuang> otherwise, we may have to quick fix critical and high, and release a new one on Aug.25
01:40:57 <joehuang> to Dongfeng, yes, new patch is ok
01:41:22 <joehuang> but big feature is not allowed only if RFE is approved
01:41:32 <dongfeng> got it~
01:41:38 <joehuang> RFE is "request for exception"
01:42:28 <zhiyuan> to joe, I think the release time is ok
01:42:49 <joehuang> after feature freeze milestone, we have to make the release as stable as much, so after pike-3, regression test and bug fix is the major activities
01:43:28 <joehuang> so please manually do regression test for your features merged in pike, and fix it ASAP.
01:44:13 <joehuang> how about the schedule for pike-rc? Aug 22?
01:44:18 <zhiyuan> or you can write tests for smoke test :)
01:44:39 <zhiyuan> just put your manual test steps into a yaml file
01:45:04 <joehuang> good proposal
01:45:48 <joehuang> we need more coverage for smoke test, and add your test script in yaml file, and make it a part of smoke test
01:46:17 <zhiyuan> example like: https://github.com/openstack/tricircle/blob/master/tricircle/tempestplugin/single_gw_topology_test.yaml
01:46:54 <joehuang> to zhiyuan, do we need a document for how to write smoke test yaml file?
01:47:39 <zhiyuan> the spec already discuss the schema, do you think it's clear enough?
01:48:32 <zhiyuan> https://github.com/openstack/tricircle/blob/master/specs/pike/smoke-test-engine.rst
01:48:39 <zhiyuan> the link for smoke test spec
01:49:59 <joehuang> may someone write a yaml file for smoke test according to the spec to see if it's good enough?
01:50:39 <dongfeng> oh, I haven't used it.
01:50:46 <joehuang> smoke test failure https://review.openstack.org/#/c/480326/
01:50:56 <joehuang> is it a random failure in smoke test?
01:51:35 <dongfeng> for patch 480326, I only changed one line
01:51:50 <dongfeng> It shouldn't fail
01:52:29 <joehuang> ok, please check the failure after the meeting
01:52:35 <dongfeng> ok
01:52:38 <zhiyuan> py35 also fails
01:52:39 <joehuang> #topic open discussion
01:52:56 <joehuang> any other topic?
01:53:02 <zhiyuan> yes
01:53:22 <zhiyuan> I notice that sometime smoke test fails with async job still running
01:53:32 <zhiyuan> do we need to increase the wait time?
01:53:40 <zhiyuan> currently is 20s * 10
01:53:49 <zhiyuan> wait 10 times, each 20 seconds
01:54:19 <zhiyuan> https://review.openstack.org/#/c/480326/ fails like this
01:55:22 <joehuang> yes, I think so, the waiting time is too short lead to random failure
01:55:34 <zhiyuan> oh sorry, it's 10s * 10
01:55:59 <zhiyuan> in my trunk patch, I change the time to 20s * 10
01:56:59 <joehuang> ok
01:57:03 <joehuang> other topic?
01:57:25 <dongfeng> no for me
01:58:01 <Yipei> no from me
01:58:30 <RongHui11> no for me
01:58:44 <joehuang> ok, thank you for attending the meeting
01:58:55 <joehuang> #endmeeting