08:01:21 <huzhj> #startmeeting daisycloud
08:01:22 <openstack> Meeting started Fri Mar 31 08:01:21 2017 UTC and is due to finish in 60 minutes.  The chair is huzhj. Information about MeetBot at http://wiki.debian.org/MeetBot.
08:01:23 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
08:01:25 <openstack> The meeting name has been set to 'daisycloud'
08:01:36 <luyao1> #info luyao
08:01:37 <huzhj> #topic Roll Call
08:01:43 <huzhj> #info Zhijiang
08:02:27 <alexyang> #info alexyang
08:03:27 <huzhj> Today's topic
08:04:00 <huzhj> 1) Roll Call
08:04:00 <huzhj> 2) OPNFV: VM/BM Deployment & Functest Integration
08:04:00 <huzhj> 3) Escalator Support
08:04:00 <huzhj> 4) Cinder/Ceph support
08:04:00 <huzhj> 5) Daisy as Kolla Image Version Manager
08:04:01 <huzhj> 6) Multicast Image Distribution
08:04:01 <huzhj> 7) AoB
08:05:13 <huzhj> before those topic I would like to metion I can not manage to contact to other core members of Daisy4NFV to proceed this PS https://gerrit.opnfv.org/gerrit/#/c/31965/
08:05:13 <zhouya> #info zhouya
08:05:39 <huzhj> So  I will try to merge it directly
08:06:34 <huzhj> #topic OPNFV: VM/BM Deployment & Functest Integration
08:06:47 <alexyang> hi zhijiang, the build jobs failed for several days
08:06:58 <alexyang> so there is no new version to use
08:07:58 <huzhj> like this ? #link https://build.opnfv.org/ci/job/daisy4nfv-verify-master/482/console
08:08:37 <zhouya1> luyao and my commit havn't merged
08:08:52 <alexyang> yes
08:09:53 <Julien-zte> #info Julien
08:10:05 <zhouya1> till the commit https://review.openstack.org/#/c/447969/ and https://review.openstack.org/#/c/450056/
08:10:17 <huzhj> Seem exceed the node capabilities... https://build.opnfv.org/ci/computer/lf-build1/
08:10:26 <zhouya1> https://review.openstack.org/#/c/450056/ this commit is the bug fix of the deploy
08:11:41 <zhouya1> I have see the fail of bm deploy
08:12:08 <Julien-zte> huzhj, ohh, I submmit a path, which means that the issue is not resolved
08:12:23 <Julien-zte> I submit the patch to block with each other
08:12:52 <huzhj> but https://build.opnfv.org/ci/computer/lf-build1/ shows there is still one slot that not used
08:13:13 <Julien-zte> https://gerrit.opnfv.org/gerrit/#/c/31517/
08:13:40 <Julien-zte> more work will be done on this, I will troubleshooting this issue
08:13:45 <Julien-zte> I will cancel some of them
08:13:46 <huzhj> Oh, Julien-zte you mean , they block by each other so event there is slot left but they can not run ?
08:14:08 <Julien-zte> I mean, each master block the other jobs run in this node
08:14:18 <Julien-zte> to prevent dead-lock for new excutors
08:14:46 <zhouya1> first of all:seems the deploy.sh do not call the 'daisy install $cluster_id' command
08:15:06 <huzhj> #info Julien will solve the job dead lock issue
08:15:07 <huzhj> Thanks Julien-zte
08:15:11 <zhouya1> second the bug of luyao's haven't merged
08:16:42 <huzhj> The upstream daisy CI was broken for several days  but now is solved
08:17:07 <huzhj> So zhouya1, if there is any PS to merge please recheck and Let me know
08:17:40 <zhouya1> ok,I will recheck till the ci is ok
08:18:40 <Julien-zte> from the CI side, no virtual deployment successful!
08:18:47 <Julien-zte> it's ergent for this job
08:19:22 <alexyang> because serena's PS not merged https://review.openstack.org/#/c/451191/
08:19:53 <Julien-zte> for baremetal: os in installing, the progress of each node is 66 66 66 66 66%
08:20:07 <Julien-zte> this means that time is not enough for deployment?
08:20:23 <Julien-zte> refer https://build.opnfv.org/ci/job/daisy-deploy-daily-master/40/console
08:20:25 <zhouya1> no
08:20:58 <Julien-zte> alexyang, the patch from Serena is merged
08:21:01 <huzhj> https://review.openstack.org/#/c/451191/   merged
08:21:06 <zhouya1> I have execute 'daisy install 'command mannally on #40  this afternoon
08:21:11 <Julien-zte> ok, merged just now
08:21:12 <huzhj> I have just merged https://review.openstack.org/#/c/451191/
08:21:30 <Julien-zte> good
08:21:49 <huzhj> Sorry for delay, working on multicast spec all day
08:22:11 <zhouya1> so seems the tempest.py file install_os_for_bm_oneshot function haven't  work ?
08:22:39 <huzhj> Julien-zte, I can not manage to contact other core to proceed this: https://gerrit.opnfv.org/gerrit/#/c/31965/
08:23:04 <huzhj> Can I and Eric and merge it ?
08:23:17 <Julien-zte> https://gerrit.opnfv.org/gerrit/#/c/31965/
08:23:34 <Julien-zte> this patch is only for Committers!, don't +1 in this patch except committers
08:24:23 <Julien-zte> huzhj, you can send email to ray to express current situation
08:24:36 <huzhj> OK, contributors +1 has been removed
08:24:38 <Julien-zte> I propose to add 2 committers in this project.
08:24:45 <Julien-zte> then we can pass 1/2
08:24:46 <huzhj> OK, I will send email to him
08:25:28 <huzhj> when to add the other one? also in this PS?
08:25:31 <Julien-zte> you can send email to ask them whether they step donw or not?
08:25:52 <huzhj> I have sent email to them, but no one reply...
08:25:56 <Julien-zte> another patch give the alike things
08:26:03 <huzhj> OK
08:26:41 <Julien-zte> nominate zhouya as another committer
08:26:43 <huzhj> yes 2 is not enough, we need a tie breaker :)
08:26:52 <huzhj> Agree
08:27:25 <huzhj> I will send another PS to proceed this
08:27:53 <Julien-zte> and before this, please send email to all the committers and Ray
08:27:58 <Julien-zte> cc to ray, please
08:28:20 <huzhj> Got it
08:29:08 <Julien-zte> suggest to keep other committers if they willing to continue work in this project
08:29:33 <huzhj> #action huzhj to amend the upstream CI issue and merge critical PS as quick as possible to solve the virtual deployment failure
08:30:59 <huzhj> Let's move to next topic
08:31:12 <huzhj> #topic Escalator Support
08:31:18 <huzhj> I see guomin is online
08:31:41 <huzhj> Hi guomin , any thing updated for this topic?
08:32:03 <huzhj> guomin is working on impl. some interface between escalator and daisy
08:32:28 <huzhj> such as get version list/ do upgrade
08:32:37 <zhouya1> great
08:32:59 <guomin> daisy need add a new interface for update assigned hosts
08:33:32 <huzhj> you mean update specific host?
08:33:40 <guomin> yes
08:34:09 <huzhj> It is the most complex part
08:34:18 <huzhj> hope Kolla to support it
08:35:04 <huzhj> hope Kolla DO support sort of partial cluster upgrading ,
08:35:28 <huzhj> kind of we can impl a dummy API first
08:36:48 <guomin> i try it
08:37:13 <huzhj> Great, also please see if Kolla support it yet
08:37:35 <guomin> ok
08:37:44 <huzhj> anything else for this topic?
08:38:06 <guomin> no
08:38:12 <huzhj> OK
08:38:15 <huzhj> #topic Cinder/Ceph support
08:38:56 <zhouya1> I have tested the cinder support on virtual env and physical env.
08:38:59 <huzhj> https://review.openstack.org/#/c/447969
08:39:37 <huzhj> Seems CI lost contact again
08:40:07 <Julien-zte> is there any issue in the patch?
08:40:07 <zhouya1> and we can see cinder-backup cinder-volume cinder-scheduler cinder-api: extra docker container .
08:40:11 <Julien-zte> thanks zhouya1
08:40:58 <Julien-zte> according to cinder, I want to know any other components missing in daisy besides cinder?
08:41:40 <huzhj> Julien-zte, can only know it when we know we need it....
08:42:02 <huzhj> most default component was there
08:42:15 <SerenaFeng> zhouya1, have you tested create a vm with volume?
08:42:22 <zhouya1> no
08:42:47 <SerenaFeng> can you test it first before I continue the integration with Functest
08:42:48 <zhouya1> I just execute 'openstack volume create --size 1 steak_volume'
08:42:49 <Julien-zte> I think nova, neutron, keystone, horizon, glance is already there
08:42:58 <zhouya1> OK
08:43:18 <Julien-zte> great
08:43:23 <zhouya1> Then I will test this
08:43:26 <SerenaFeng> thanks
08:43:47 <Julien-zte> shall we begin to deploy on POD2 using upstream jobs
08:44:14 <Julien-zte> I think zhouya only test baremetal not virtual
08:44:23 <Julien-zte> zhouya1 do you test virtual env?
08:44:26 <zhouya1> yes
08:44:39 <zhouya1> I also test virtual env
08:44:57 <Julien-zte> supported now?
08:45:01 <zhouya1> yes
08:45:25 <zhouya1> the difference between bm(multinode) and vm(all-in-one)
08:45:27 <zhouya1> is
08:45:43 <Julien-zte> great!
08:45:50 <Julien-zte> and Serena can continue to test
08:45:53 <Julien-zte> good news
08:46:15 <SerenaFeng> after zhouya1 test it works with vm I will continue integrating
08:46:35 <zhouya1> all in one is (tgtd and iscsid container) multinode is (ceph-osd and ceph-mon container)
08:46:48 <huzhj> If vm is multinode, can it also use Ceph instead of LVM? zhouya1
08:47:29 <Julien-zte> still not suggest in VM, huzhj
08:47:40 <Julien-zte> it is useless for the performance of Ceph is not acceptable
08:47:46 <zhouya1> I think so, but we should spare an extra disk like /dev/sdb or someting like to use as a ceph bakend for cinder
08:48:12 <Julien-zte> Virtual Env is only to perform test, not for production
08:48:17 <huzhj> Julien-zte but acording to upstream , using LVM in a multinode cluster is unstable
08:48:27 <huzhj> Oh, Got it
08:49:19 <huzhj> But the question , if vm is multinode and we still use LVM, which node is used for creating the cinder_vg?
08:49:56 <Julien-zte> in baremetal env, which node is chosen?
08:50:01 <Julien-zte> if using LVM?
08:50:30 <huzhj> no idea.
08:50:58 <huzhj> now for BM, is Ceph only.
08:51:07 <Julien-zte> before
08:51:08 <zhouya1> yes
08:51:35 <Julien-zte> I think in baremetal, lvm and ceph, both can be supported, just ceph is the default choose
08:52:18 <huzhj> Julien-zte , please see this bug https://launchpad.net/bugs/1571211
08:52:18 <openstack> Launchpad bug 1571211 in Cinder "In multinode setup VM fails to launch due to cinder not checking with all glance api servers" [Medium,In progress] - Assigned to John Griffith (john-griffith)
08:52:18 <Julien-zte> ok, let's speed up
08:52:45 <Julien-zte> that's future requirement, let's support bm+ceph, and allinone+lvm
08:52:49 <huzhj> The Cinder community has closed a bug as WontFix which makes it impossible for LVM to be used at all in a multi-controller setup. The only option for multi-controller storage to work correctly at present is via a Ceph deployment.
08:52:49 <huzhj> OK next topic
08:53:05 <huzhj> #topic Daisy as Kolla Image Version Manager
08:53:18 <huzhj> #info this is almost done from luyao
08:53:32 <huzhj> but we encountered this BUG https://github.com/docker/docker/issues/22260
08:53:54 <huzhj> which will result in upgrading failed
08:54:13 <huzhj> in a intermittent basis
08:54:21 <Julien-zte> I think maybe the docker verison is too old
08:54:41 <Julien-zte> in 1.12 there is a issue which block the operation of docker service
08:54:48 <Julien-zte> or before 1.12
08:54:49 <huzhj> actually the newer the worse
08:55:13 <Julien-zte> do you have a env to troubleshooting?
08:55:23 <huzhj> luyao has one
08:55:33 <Julien-zte> I let a member from docker to support you
08:56:00 <huzhj> Great, seems we can not reach luyao currently, I will contact her offline
08:56:08 <Julien-zte> send a email to me and I will allocate a task
08:56:33 <huzhj> OK, thanks:)
08:57:03 <huzhj> next topic
08:57:12 <huzhj> #topic Multicast Image Distribution
08:57:20 <Julien-zte> how many tasks left?
08:57:33 <Julien-zte> time is limited
08:57:34 <huzhj> only one topic left
08:57:35 <huzhj> https://gerrit.opnfv.org/gerrit/#/c/32709/
08:57:47 <huzhj> I have finished the spec
08:58:06 <huzhj> and let discuss on gerrit
08:58:30 <huzhj> yes, running out of time
08:58:32 <Julien-zte> great
08:58:40 <Julien-zte> add one new topic
08:59:00 <Julien-zte> a new installer project Armada submitted by AT&T
08:59:01 <huzhj> yes please
08:59:07 <Julien-zte> do you know about this
08:59:26 <huzhj> no, sound like a installer for ARM? :P
08:59:35 <Julien-zte> functest pre-ptl suggest only support 5 installers in Functest in next release
08:59:45 <Julien-zte> not at all, just a name, not arm:)
09:00:03 <huzhj> approved?
09:00:04 <SerenaFeng> Armada a new installer
09:00:09 <Julien-zte> let's speed up
09:00:21 <Julien-zte> discussed in TC, and will submit to TSC
09:00:23 <SerenaFeng> approved by tech-group
09:00:28 <SerenaFeng> is proposed to tsc now
09:00:46 <Julien-zte> time is ergent for us.
09:01:02 <Julien-zte> Armada is a project open sourced by AT&T, in git
09:01:05 <Julien-zte> in github
09:01:32 <Julien-zte> ok for this, and I will send a email for this
09:01:34 <SerenaFeng> and Compass just reply with a positive
09:02:11 <SerenaFeng> Testperf are discussing impl Darwinian election to installers
09:02:11 <Julien-zte> SerenaFeng what' progress of you unitest task
09:02:26 <huzhj> diversity is good:)
09:02:26 <SerenaFeng> jjb is submitted
09:02:49 <SerenaFeng> huzhi no, because resources is limited
09:03:08 <SerenaFeng> they are talking about limit the number of installers
09:03:39 <Julien-zte> discuss offline for the new installer
09:03:50 <SerenaFeng> ok
09:03:57 <huzhj> I think they should limit the scennario of a install instead
09:03:58 <Julien-zte> any other reports ?
09:04:05 <huzhj> no
09:04:11 <huzhj> let's wrap up
09:04:19 <Julien-zte> alexyang?
09:04:26 <huzhj> thanks all for the great work!
09:04:32 <alexyang> no
09:04:37 <Julien-zte> thanks all
09:04:43 <huzhj> bye
09:04:47 <SerenaFeng> good weekend
09:04:51 <huzhj> #endmeeting