08:02:16 <yasufum> #startmeeting tacker
08:02:17 <openstack> Meeting started Tue Jan 12 08:02:16 2021 UTC and is due to finish in 60 minutes.  The chair is yasufum. Information about MeetBot at http://wiki.debian.org/MeetBot.
08:02:18 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
08:02:21 <openstack> The meeting name has been set to 'tacker'
08:02:26 <manpreet> hi
08:04:08 <yasufum> Although there is no topic from etherpad, do you have anything to discuss today?
08:04:16 <yasufum> link: https://etherpad.opendev.org/p/tacker-meeting
08:05:57 <LiangLu> I have 1 topic, would like to hear opinions form you, should I go on?
08:06:28 <yasufum> sure, you can start
08:06:59 <LiangLu> Thanks! https://etherpad.opendev.org/p/tacker-meeting , L620 ~
08:07:35 <LiangLu> This topic is about Zuul FT environment for deploying k8s cluster on vm
08:07:46 <yasufum> #topic ft-for-k8s
08:08:32 <LiangLu> in a WIP patch that implements spec: https://specs.openstack.org/openstack/tacker-specs/specs/wallaby/mgmt-driver-for-k8s-cluster.html
08:09:42 <LiangLu> we need to configure some new FT environment on Zuul, it seems need 2 computer nodes(openstack + tacker) and at least 1 computer node
08:10:12 <LiangLu> might be similar to https://review.opendev.org/c/openstack/tacker/+/766886
08:11:12 <LiangLu> another annoying problem is, it takes a long time to finish our FT, which contains download/deploy vm(ubuntu) and download/deploy k8s cluster
08:12:41 <LiangLu> at least 30 minutes per node(master/worker) is needed in our local test, thus Zuul FT time limit needs extend a lot
08:13:37 <LiangLu> So, I was wondering, should we do this FT on Zuul? or if we should only do UT on Zuul and run FT locally?
08:14:19 <LiangLu> may I ask your opinions for this problem ?...
08:15:57 <yasufum> Frist of all, we have already extended the timeout to upper limit from takahashi-tsc, right?
08:16:22 <takahashi-tsc> Right, current timeout setting is 3hours (10800 seconds)
08:16:40 <takahashi-tsc> https://opendev.org/openstack/tacker/src/branch/master/.zuul.yaml#L46
08:17:03 <yasufum> thanks
08:17:10 <takahashi-tsc> In addition, I cannot extend more...
08:17:34 <yasufum> Second, I am not sure how we can run FT on local.
08:18:02 <yasufum> Do you mean you will provide some scripts for building environment, or so?
08:18:03 <LiangLu> thanks for sharing! I checked current FT time is about in 1h 55m 19s
08:19:22 <LiangLu> still we have 1 hour to use for that FT..if it only have simple components (like 1 master + 1 worker)
08:20:38 <LiangLu> ↑ yes, still we need to learn how to configure zuul.yaml to make a property environment for our FT( maybe we could ask ito-san and horie-san)..
08:21:34 <LiangLu> means, how to build environment on Zuul
08:22:25 <LiangLu> and about run FT locally, currently we are assuming run it on a manual-deployed environment on some specific servers(like our servers)?
08:22:57 <yasufum> Zuul on local env?
08:23:54 <LiangLu> sorry for my misunderstanding words.. local env doesnot use zuul
08:25:38 <LiangLu> we are assuming to run FT on our environment, and share FT code/result to somewhere? I dont know is it good for a patch?
08:28:17 * yasufum My point is we should run FT on zuul without dependencies for any local environments.
08:29:23 <yasufum> In other word, keep it reproducible.
08:30:53 <yasufum> So, I think it is not enough just sharing the result...
08:32:54 <yasufum> I would like to ask you to run somehow your FT on zuul.
08:33:40 <LiangLu> Thanks for opinion.. I agree it is best to run on Zuul.. and now the time-limit is the problem,,
08:33:45 <yasufum> We can consider to remove the number of tests for reducing the total time optionally.
08:36:57 <takahashi-tsc> LiangLu Sorry, can I ask some questions about current situation?
08:36:58 <LiangLu> okay.. also as takahashi-san mentioned, timeout cannot be extended anymore...right
08:37:06 <LiangLu> yes, please
08:38:00 <takahashi-tsc> First, your plan is make new FT env, i.e. new Zuul job. Is it correct?
08:38:25 <takahashi-tsc> for K8S
08:40:19 <LiangLu> yes, we need to deploy some vm from tacker first and install k8s cluster on it by tacker, thus we assuming a new Zuul env is needed,,
08:41:00 <takahashi-tsc> OK, so is your concern is, the time of such new FT on new separated job will exceed 3hours?
08:42:51 <LiangLu> yes, actually Zuul new env has no connection with time limit, FT it self needs a lot of time.
08:43:20 <LiangLu> no matter it runs locally or on Zuul
08:45:42 <takahashi-tsc> What is the reason? Only k8s setting + 1 or a few tests exceed 3 hours? That is, we cannot separate K8s tests to 2 sets?
08:48:43 <LiangLu> do you mean we can run tests parallelly on several envs? sorry for a naive question..
08:49:33 <LiangLu> the reasons it takes time are: download vm image, setting k8s cluster for several nodes
08:50:35 <takahashi-tsc> In my understanding, if we separate the jobs, the jobs run in parallel... Sorry if my understanding is incorrect.
08:50:54 <LiangLu> it is decided by complexity of k8s cluster, in our local test it takes about 30min/node
08:51:38 <LiangLu> that helps...if it can run parallelly problem could be solve
08:52:42 <yasufum> You can do that, but should provide scripts for bulding environment.
08:55:48 <LiangLu> thanks! I will go for research how to do that on zuul..
08:56:20 <yasufum> It looks it is already considered in the patch you shared.
08:56:37 <yasufum> #link: https://review.opendev.org/c/openstack/tacker/+/766886
08:57:20 <yasufum> I can find new test “tacker-k8s-functional-devstack-multinode-python3” from the results.
08:57:29 <LiangLu> indeed.. thanks I will ref it
08:58:41 <takahashi-tsc> Yes, and total time seems to exceed 3 hours
08:59:02 <yasufum> Building env is done from additional ansible script in “roles/setup-default-vim/tasks/main.yaml”
09:00:06 <LiangLu> thanks for info!
09:00:52 <yasufum> If your quuestion is “Can we add a FT?” simply, I think it is OK.
09:01:18 <takahashi-tsc> +1
09:02:33 <LiangLu> thanks!
09:03:04 <yasufum> Thanks. If no other comments, go to another topic.
09:04:43 <yasufum> I have two topic, devstack installation failure and patch review.
09:05:09 <yasufum> #topic devstack-failure
09:06:44 <yasufum> In my recent survey, Ubuntu 20.04 on vagrant, devstack installation is failed because for a failure of pip, and can be more…
09:07:06 <yasufum> Someone might have similar problem possibly.
09:07:47 <yasufum> I have tried to fix the problem, or welcome if someone fix it :)
09:08:46 <yasufum> It is a FYI just now.
09:09:33 <yasufum> If no comments, go to my second topic.
09:10:35 <yasufum> #topic patch-review
09:11:26 <yasufum> I have two things for the topic, so would like to discuss shortly.
09:13:04 <yasufum> There are patches of tacker-spec asking reviews and +2 remained. Please give your comment for them.
09:13:23 <yasufum> #link: https://review.opendev.org/q/project:openstack%252Ftacker-specs+status:open
09:13:59 <yasufum> Second one is a fix for lower-constraints from takahashi-tsc.
09:14:03 <yasufum> #link: https://review.opendev.org/c/openstack/tacker/+/769099
09:14:42 <yasufum> It is suggested to make FT ‘voting’ before merging it.
09:16:02 <yasufum> My main concern is that might be related reverting FT because this update changes not only lower-constraints.txt but also requirements.txt
09:16:44 <yasufum> So, I think it is better to revert FT before your patch is merged ideally.
09:17:12 <takahashi-tsc> Understood... OK, I'll try to revert FT first.
09:17:25 <yasufum> But I am not sure reverting FT can be done soon.
09:17:42 <takahashi-tsc> Yes, but anyway, I'll try soon.
09:17:58 <yasufum> Second option, merge your patch before FT, is better if reverting is hard.
09:18:07 <yasufum> This is current my thinking.
09:19:23 <yasufum> Thanks.
09:19:46 <yasufum> Sorry for extend the time.
09:20:10 <yasufum> If no other topics, I would like close this meeting.
09:20:12 <manpreet> Just a humble request to review https://review.opendev.org/c/openstack/tacker/+/759361.
09:21:11 <yasufum> Ahh, sorry for slow reply. I will review it.
09:21:23 <manpreet> Thank you so much :)
09:22:45 <yasufum> Thanks
09:23:14 <yasufum> seems no more comments…
09:23:53 <yasufum> Thank you for joing. Bye!
09:24:10 <manpreet> Thanks, bye
09:24:11 <takahashi-tsc> thanks, bye
09:24:19 <yasufum> #endmeeting