15:01:23 <ad_ri3n_> #startmeeting fog-edge-massively-distributed-clouds
15:01:23 <openstack> Meeting started Wed Mar 28 15:01:23 2018 UTC and is due to finish in 60 minutes.  The chair is ad_ri3n_. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:24 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:01:25 <ad_ri3n_> #chair  ad_ri3n_
15:01:27 <openstack> The meeting name has been set to 'fog_edge_massively_distributed_clouds'
15:01:28 <openstack> Current chairs: ad_ri3n_
15:01:32 <ad_ri3n_> #topic roll call
15:01:37 <ad_ri3n_> Hi guys
15:01:44 <ad_ri3n_> Let's see who is  available today
15:01:54 <marie_> Hi, sorry I'm late
15:01:56 <jamemcc> Hi
15:02:08 <ad_ri3n_> Hi jamemcc
15:02:30 <avankemp> o/
15:02:39 <ad_ri3n_> parus ? ansmith ?kgiusti ?
15:03:00 <ad_ri3n_> #info agenda
15:03:00 <ad_ri3n_> #link https://etherpad.openstack.org/p/massivfbely_distributed_ircmeetings_2018 (line 322)
15:03:02 <ansmith> o/
15:03:10 <ad_ri3n_> Hi guys
15:03:23 <kgiusti> o/
15:03:46 <dpertin> Hi o/
15:04:47 <marie_> #link https://etherpad.openstack.org/p/massively_distributed_ircmeetings_2018
15:04:50 <ad_ri3n_> I started to add some news, please feel free to do the same from your side (directly in the pad)
15:06:45 <msimonin> o/
15:07:01 <ad_ri3n_> So only two news from my side as we mainly worked on the paper from our side.
15:07:33 <ad_ri3n_> maybe one last info to share: kgiusti it looks like you will be now officially part of the OpenStack core devs ?
15:07:36 <ad_ri3n_> Congrats !
15:07:50 <ad_ri3n_> or still not yet finalized?
15:07:50 <kgiusti> ad_ri3n_: yes!  Thanks!
15:08:15 <jamemcc> wow kgiusti - thanks
15:08:25 <kgiusti> ad_ri3n_: I'm expecting it to go through
15:08:26 <ad_ri3n_> you are our trojan horse ;)
15:08:48 <msimonin> bravo kgiusti !
15:09:00 <kgiusti> ad_ri3n_: I'd rather be a trojan rabbit...
15:09:00 <marie_> congratulations!
15:09:25 <kgiusti> ad_ri3n_: pardon the obscure monty python reference...
15:09:29 <kgiusti> thanks all
15:09:50 <ad_ri3n_> Ok any other info to share ?
15:09:54 <marie_> don't jump on people's throat
15:10:47 <ad_ri3n_> jdandrea put some information relatd to the MUSIC project
15:11:07 <ad_ri3n_> I thought he will be there to elaborate a bit more but not sure he can attend our meeting this week.
15:11:25 <ad_ri3n_> so let's start on the ongoing/pending actions if there is nothing more to discuss on this topic.
15:12:07 <ad_ri3n_> #topic ongoing-actions-edge-sessions.
15:12:41 <ad_ri3n_> as mentioned the last session has been cancelled due to the F2F meeting in ONS. I cannot attend it but parus did. He will probably give us a brief overview of what has been discussed.
15:12:54 <ad_ri3n_> regarding the meeting before the last one.
15:13:25 <ad_ri3n_> We mainly discuss about possible issues regarding the current approach (i.e. a kind of manager in charge of synchronizing the different OpenStack instances).
15:13:41 * ad_ri3n_ is looking for the etherpad link
15:15:51 * ad_ri3n_ is waiting now for the etherpad to be loaded…
15:17:08 <ad_ri3n_> #link https://etherpad.openstack.org/p/edge-alans-problems
15:17:33 <ad_ri3n_> line 92
15:17:34 <ad_ri3n_> sorry
15:18:06 <ad_ri3n_> so that's all
15:18:11 <ad_ri3n_> sorry for being long…
15:18:46 <ad_ri3n_> and maybe I also missed to announce the new dedicated IRC: Edge Computing Group IRC channel: Freenode: #edge-computing-group
15:18:59 <ad_ri3n_> I didn't connect yet to be honest (I should have but ...)
15:19:11 <ad_ri3n_> so I don't know whether it is quite active or not
15:19:41 <ad_ri3n_> so that's all for this topic
15:19:44 <ad_ri3n_> let's move forward
15:20:04 <ad_ri3n_> #topic ongoing-actions-p2p-gapanalysis
15:20:20 <ad_ri3n_> Nothing more on that point than what has been written in the pad
15:20:28 <ad_ri3n_> so if there is no question, let's go on.
15:20:32 <ad_ri3n_> . . .
15:20:34 <ad_ri3n_> . . .
15:20:41 <ad_ri3n_> ok let's go on
15:20:51 <jamemcc> Referencing #edge-computing-group  Thanks. Yeah - it looks active - at least some of the right people are in it
15:21:38 <ad_ri3n_> yes let's try to keep an eye on that indeed.
15:21:40 <ad_ri3n_> #topic ongoing-actions-cockroach
15:21:46 <ad_ri3n_> @marie
15:21:49 <ad_ri3n_> what are the news?
15:22:07 <marie_> we are gathering the results
15:22:50 <marie_> and we are trying to analyse them for now, on replicated keystone
15:23:01 <ad_ri3n_> it seems you will give a look to tidb also?
15:23:49 <marie_> yes, we might look into tidb more seriously because the sysbench test made was great
15:24:24 <msimonin> you mean it performs better than cockroach under sysbench ?
15:24:46 <marie_> yep, and mariadb
15:25:13 <marie_> like 10 times better, so I'm looking into that because it seems suspicious :)
15:25:22 <ad_ri3n_> any news regarding the experimental protocol. Did you make progress on that side?
15:25:42 <marie_> yep, I'm making a draft for the test plan
15:25:43 <ad_ri3n_> or should we wait a bit more :-P
15:25:49 <ad_ri3n_> ok
15:25:51 <ad_ri3n_> great
15:26:18 <msimonin> Will people from cockroach be involved in the test plan ?
15:26:25 <marie_> I'll put it on the pad
15:26:29 <ad_ri3n_> thanks
15:26:37 <ad_ri3n_> So feedbacks/comments welcome
15:27:10 <ad_ri3n_> anything else?
15:27:17 <marie_> I would like to get people from cockroach involved in the test plan, cause cockroach results are not really good and maybe I'm doing something wrong
15:27:48 <marie_> nothing more !
15:28:04 <msimonin> marie_: ack
15:28:53 <ad_ri3n_> cool
15:29:17 <ad_ri3n_> so please do not forget to put your test plan draft on the pad ASAP
15:29:19 <ad_ri3n_> thanks
15:29:30 <ad_ri3n_> #topic ongoing-action-AMQP
15:29:48 <ad_ri3n_> msimonin:  ansmith kgiusti
15:29:49 <ad_ri3n_> ?
15:29:57 <msimonin> yes !
15:30:46 <msimonin> So we've made some progress on the experimentation framework
15:30:54 <kgiusti> nothing much from me this week - investigating some memory issues found by msimonin
15:30:56 * ad_ri3n_ is going through all the interesting links ;)
15:31:14 <msimonin> I've put the most interesting things in the epad
15:31:39 <msimonin> The more noticeable thing is that we now scale to 10K+ RPC agents :)
15:31:49 <ad_ri3n_> :-)
15:31:54 <ad_ri3n_> How many PMs?
15:32:01 <kgiusti> oh yeah!
15:32:33 <msimonin> ad_ri3n_:  The funny thing is that we don't need many Physical Machines
15:32:39 <ad_ri3n_> :-)
15:32:43 <msimonin> we used approx 55 PMs
15:32:52 <msimonin> rpc clients sleeps most of the time
15:32:58 <msimonin> in the workload we have
15:32:59 <ad_ri3n_> and containers/enosStack?
15:33:20 <ad_ri3n_> how do you emulate independency between agents? maybe you have not to ?
15:33:20 <msimonin> that makes 10K-ish containers
15:33:26 <ad_ri3n_> ok
15:33:48 <ad_ri3n_> maybe it would make sense to give a look at solutions such as Distem
15:34:03 <ad_ri3n_> (or at least to better understand the positioning w.r.t enosStack)
15:34:12 <ad_ri3n_> to conduct such experiments
15:34:19 <ad_ri3n_> but anyway: welldone guys
15:34:44 <ad_ri3n_> still awesome what you succeed to achieve from the experiment viewpoint !
15:34:44 <msimonin> Javier is adding the latency constraints in the framework
15:35:40 <ad_ri3n_> ok
15:35:54 <msimonin> We have big plans for the coming weeks :)
15:36:06 <ad_ri3n_> \o/
15:38:13 <ad_ri3n_> ok nothing more
15:38:13 <ad_ri3n_> ?
15:38:19 <msimonin> nope
15:38:20 <msimonin> sorry
15:38:21 <ad_ri3n_> #topic open-discussions
15:38:37 <ad_ri3n_> so we have two points to discuss at least:
15:38:53 <ad_ri3n_> opportunities to see/collaborate with more folks from RedHat
15:38:58 <ad_ri3n_> kgiusti:  ansmith?
15:39:18 <ad_ri3n_> any news? in particular from the ones that are working on evaluating OpenStack WANWide?
15:39:24 <ansmith> not many folks were around this week due to ONS, internal meetings
15:39:26 <kgiusti> this has been a tough nut to crack.
15:40:22 <ansmith> will double back on this next week
15:40:23 <ad_ri3n_> od
15:40:26 <ad_ri3n_> s/od/ok
15:40:43 <ad_ri3n_> some changes to see someone joining soon? It looks definitely relevant
15:41:11 <ad_ri3n_> Actually It is a suprise to see so many people attending events such as the F2F informal meeting yesterday at ONS
15:41:20 <ad_ri3n_> and only the best guys on this IRC .)
15:41:21 <ad_ri3n_> ;-)
15:41:47 <ad_ri3n_> s/changes/chances
15:42:17 <ad_ri3n_> I don't know how we can attract more people and federate all the ongoing actions
15:42:25 <ad_ri3n_> in particular the ones that focus OpenStack
15:42:35 <kgiusti> I'm not aware of anything specific - ansmith?
15:43:08 <ad_ri3n_> This looks to be a nonsense to work, each,  on our side
15:43:21 <ansmith> there is some focus around multi-site, just need to track down those folks
15:43:22 <ad_ri3n_> kgiusti:  ansmith I pointed those guys during the PTG
15:43:37 <ad_ri3n_> you said that they are probably taking part to the OpenStack group @redhat
15:44:26 <ad_ri3n_> I do not remind the name unfortunately but I can find it if needed
15:45:00 <ansmith> I believe it was Azhar and/or Sanjay, we had reached out to them before and will do so again
15:45:28 <ad_ri3n_> thanks ansmith for trying… highly appreciated ;-)
15:45:52 <ad_ri3n_> I know also that people from IWinStack were interested to take part but we should try to identify concrete actions with them
15:46:07 <ad_ri3n_> any ideas of current actions we can start and that make sense for our working group
15:46:33 <ad_ri3n_> we talked a lot about evaluating cell V2 concept but it looks like the cell abstraction will only be related to nova in the future
15:46:58 <ad_ri3n_> so not sure whether we should bet on that
15:47:10 <ad_ri3n_> @dpertin will start to dive a bit more on Nova
15:47:25 <ad_ri3n_> to see what are the missing mechanisms/capabilities we should add in Nova
15:47:30 <ad_ri3n_> in terms of edge requirements
15:47:33 <msimonin> We'd like to evaluate multisite + qdr in the future
15:47:40 <msimonin> qdr = qpid-dispatch-router
15:47:50 <ad_ri3n_> @msimonin great : can you add a note in the pad
15:48:17 <ad_ri3n_> my idea is to identify a few actions that make sense and to propose to iWinSTACK to see whether
15:48:30 <msimonin> since you defer the routing to the messaging layer you could have site(cells) without actual cells
15:48:32 <ad_ri3n_> they might be interested to dig into
15:49:25 <ansmith> any pointers to iWinSTACK? not finding via googling
15:49:50 <ad_ri3n_> Chineese company
15:50:26 <ad_ri3n_> https://www.openstack.org/foundation/board-of-directors/
15:50:32 <ad_ri3n_> and looks for Joseph Wang
15:50:45 <ad_ri3n_> http://www.inwinstack.com/en/home/
15:51:04 <ansmith> thanks!
15:51:21 <ad_ri3n_> they have implemented a few mechanisms from the OEC group
15:51:29 <ad_ri3n_> so they are familiar with the edge challenges
15:51:54 <ad_ri3n_> they attended the SF event last september. I'm pretty sure they are good candidates for contributing to the SiG
15:53:03 <ad_ri3n_> ok
15:53:27 <ad_ri3n_> so I open a new action in the epad. Please feel free to add any actions you would like to discuss/make progress on
15:53:37 <ad_ri3n_> nothing more from my side.
15:53:40 <ad_ri3n_> anything else?
15:54:02 <msimonin> nothing from my side
15:54:08 <ad_ri3n_> maybe one question
15:54:14 <ad_ri3n_> Guys are you evaluating ZMQ also?
15:54:17 <dpertin> shall we discuss something regarding vancouver summit forum/f2f/...?
15:54:18 <ad_ri3n_> or it is definitely dead?
15:54:37 <ad_ri3n_> @dpertin we should indeed open a new bullet for that but we can probably discuss it next time.
15:54:43 <kgiusti> ad_ri3n_: a moment of silence would be appropriate.
15:54:46 <dpertin> ok
15:55:20 <ad_ri3n_> kgiusti:  :-)
15:55:22 <kgiusti> I've emailed both the operator's and the dev mailing list asking for volunteers to support it
15:55:31 <kgiusti> so far no takers.
15:55:35 <ad_ri3n_> msimonin:  ?
15:55:42 <ad_ri3n_> it is also your opinon?
15:55:47 <ad_ri3n_> opinion?
15:56:05 <ad_ri3n_> ZMQ no way to test it?
15:56:21 <ad_ri3n_> it was a fully P2P approach so conceptually speaking
15:56:30 <ad_ri3n_> does it make sense to give it a look?
15:56:33 <kgiusti> At oslo PTG it was decided to move it to deprecated.  Currently we removed all zmq unit tests as they've been broken for months now.
15:56:49 <ad_ri3n_> or what can be done with ZQM can also be done with qpid?
15:56:59 <ad_ri3n_> ok
15:57:03 <ad_ri3n_> thanks kgiusti
15:57:06 <ad_ri3n_> good to know
15:57:07 <kgiusti> functionally yes - both support RPC
15:57:45 <ad_ri3n_> I was thinking in terms of overlay/P2P concepts. but ok if it is dead…. RIP
15:58:10 <kgiusti> unless some brave soul(s) come forward, yes RIP
15:58:11 <ad_ri3n_> ok two minutes left
15:58:17 <ad_ri3n_> :-)
15:58:21 <ad_ri3n_> inwinstack :D
15:58:44 <ad_ri3n_> stupid joke
15:58:46 <ad_ri3n_> ok
15:58:53 <ad_ri3n_> so if nothing more… enjoy the rest of the day
15:58:53 <msimonin> ad_ri3n_: yes zmq is not a good bet
15:59:10 <ad_ri3n_> and talk to you in two weeks
15:59:15 <ad_ri3n_> thanks guys ! have fun !
15:59:17 <msimonin> actually I think this didn't scale well
15:59:23 <msimonin> (the implementation)
15:59:26 <ad_ri3n_> #endmeeting