07:59:44 <eranrom> #startmeeting storlets
07:59:45 <openstack> Meeting started Tue Jan 10 07:59:44 2017 UTC and is due to finish in 60 minutes.  The chair is eranrom. Information about MeetBot at http://wiki.debian.org/MeetBot.
07:59:46 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
07:59:48 <openstack> The meeting name has been set to 'storlets'
07:59:58 <eranrom> Hi
08:01:13 <kota_> hello
08:01:30 <eranrom> Hi kota_
08:01:37 <kota_> eranrom: happy new year
08:01:59 <eranrom> Happy new year & apologies for last week's cacellation
08:02:05 <akihito> Hi! happy new year!
08:02:08 <kota_> eranrom: np
08:02:15 <kota_> akihito: hi happy new year
08:02:18 <eranrom> akihito: Hi
08:02:20 <takashi> hi
08:02:24 <sagara> hi
08:02:27 <eranrom> takashi: Hi
08:02:28 <kota_> takashi: hi!
08:02:34 <eranrom> sagara: Hi. Welcome
08:02:37 <kota_> sagara: welcome!
08:02:42 <sagara> thanks
08:02:47 <takashi> eranrom, kota_: Hi :-)
08:03:00 <eranrom> takashi: Hi
08:03:14 <eranrom> ok. Lets start our very first meeting for 2017 :-)
08:03:23 <kota_> yey
08:03:45 <takashi> yes! :-)
08:03:46 <eranrom> I have a short agenda here: https://wiki.openstack.org/wiki/Meetings/Storlets#Agenda:
08:04:01 <eranrom> but do feel free to add items.
08:04:40 <eranrom> You do not need to edit, just type whatever subject you have.
08:04:52 <eranrom> That is type it here in IRC :-)
08:04:57 <kota_> sure
08:05:10 <takashi> ok
08:05:15 <takashi> thanks for adding my items :-)
08:05:34 <eranrom> :-)
08:06:01 <eranrom> The list is not all the items, just those few that I have picked from the bottom going up
08:06:12 <eranrom> The idea is to look at all the list and prioritize
08:06:43 <eranrom> Here is the query to see all the list https://review.openstack.org/#/q/project:openstack/storlets+status:open
08:07:53 <kota_> eranrom: that's same with me which is usually used.
08:08:19 <eranrom> kota_: right.
08:09:29 <takashi> Personally I would like to land https://review.openstack.org/#/c/406620/ first, because it fixes broken unittests, which caused by my packaging work maybe
08:09:37 <takashi> not first, but asap
08:10:53 <kota_> takashi: so you're waiting on me :P
08:11:17 <eranrom> ok from me this is https://review.openstack.org/#/c/395152/
08:12:08 <eranrom> I kinda want to land this before I continue with taking out the storlets code from the container and complete the migration to devstack
08:12:43 <takashi> eranrom: I like that idea, but my only concern is that I'm not sure that (having process management script) is consistent with devstack-ish way
08:13:36 <kota_> eranrom: cool, let me know what can be available on `s2aio.sh start` and `s2aio.sh stop`?
08:13:39 <eranrom> takashi: right. The reason I have used it is that stack/unstack takes a long time
08:14:03 <takashi> when using devstack, we use unstack.sh or rejoin-stack.sh for stop/restart
08:14:21 <kota_> eranrom: something like cleanup/reinstall? or restarting daemon?
08:14:41 <kota_> not yet looking at the actual code though
08:14:49 <takashi> kota_: just for stopping/restarting daemon, AFAIK
08:14:56 <kota_> takashi: thx
08:14:56 <eranrom> kota_: what it does is basically stop/start Keystone & swift_init all restart
08:15:27 <eranrom> Takashi is right that there is the devstack way of doing so, but as far as I recall it is so slow
08:15:52 <eranrom> and I need something quick to allow me test changes to the storlets code when running functional tests
08:16:27 <takashi> eranrom: I see
08:16:54 <eranrom> takashi: I think that the unstack code is more of a cleanup code - is that right?
08:17:16 <takashi> AFAIK, they have different code for cleanup, clean.sh
08:17:20 <takashi> eranrom ^^^
08:17:26 <eranrom> takashi: I see.
08:17:35 <takashi> eranrom: and unstack.sh is juse stopping all services running in devstack env
08:17:41 <takashi> s/juse/just/g
08:17:58 <eranrom> takashi: ok. gotcha
08:19:01 <takashi> Currently we are doing a little bit different way for process running. I mean, in devstack we totally use virtualenv for each services, but in our devstack usage we don't run swift in virtualenv.
08:19:49 <takashi> So I notice that it can be difficult for us to use unstack/rejoin-stack in current state...
08:20:12 <eranrom> takashi: ok, 2 comments:
08:20:13 <takashi> but when we totally adopt to devstack, I think it is better to use ordinal things (unstack,resoin-stack).
08:20:27 <takashi> eranrom: ok
08:20:53 <eranrom> 1. I have made a mistake and used stack.sh instead of rejoin-stack and this is why I thought it takes forever
08:21:58 <eranrom> 2. According to your last comment, I should first run swift the devstack way, and probably abandon the start/stop patch
08:22:39 <takashi> eranrom: I think so
08:22:48 <eranrom> takashi: ok. thanks
08:23:01 <eranrom> Anything else on the pending patches?
08:23:08 <takashi> eranrom: but at the same time, having temporal script (s2aio.sh start/stop) also makes sense to me, because it is surely useful
08:23:38 <takashi> does it make sense to land your patch first, and record todo to remove that script?
08:24:02 <takashi> if it takes some time to totally adopt to devstack-way.
08:24:13 <takashi> (and I expect so)
08:24:24 <eranrom> takashi: IMO would be good, and then we can fix it so that stop==unstack, start==rejoin
08:24:33 <takashi> eranrom: yes
08:24:42 <eranrom> takashi: ok great. thanks
08:25:06 <eranrom> anything else on the patches? I think there are some long pending python functional tests patches
08:25:14 <akihito> My functional test items are currently being fixed.
08:25:20 <akihito> Sorry for the delay in fix..
08:25:47 <eranrom> akihito: no worries.
08:25:49 <kota_> akihito: np
08:25:50 <akihito> I will fix it this week.
08:25:56 <kota_> nice
08:26:09 <eranrom> great. thanks
08:26:36 <eranrom> next topic?
08:26:42 <takashi> eranrom: yes
08:26:52 <eranrom> #topic PTG
08:27:23 <eranrom> I guess it would be good time to start planning.
08:27:55 <eranrom> I will create an Etherpad and post the URL in our channel
08:29:07 <eranrom> If there is nothing else on PTG, we can probably move to open discussion
08:29:27 <kota_> eranrom: https://etherpad.openstack.org/p/storlets-otaca-design-summit <- this one?
08:30:01 <eranrom> kota_: Do you want to reuse it?
08:30:04 <takashi> kota_: maybe we need 'pike' version
08:30:09 <kota_> oops
08:30:22 <kota_> i picked it up from today's agenda
08:30:45 <eranrom> kota_: Good that you have mentioned it, we can look for a reference at all the things we still want to do
08:31:00 <eranrom> kota_: I will update the meeting page with the new one :)
08:31:08 <kota_> eranrom: thx!
08:31:12 <eranrom> once I create it.
08:31:21 <eranrom> ok. open discussion?
08:31:42 <kota_> i might create the brand new page at etherpad
08:31:48 <kota_> https://etherpad.openstack.org/p/storlets-pike-design-summit
08:31:57 <eranrom> kota_: thanks, that was fast
08:31:59 <kota_> was attempting to check the existence
08:32:27 <kota_> nothing added yet though
08:32:43 <eranrom> kota_: no worries. It will get filled :)
08:33:04 <kota_> eranrom: yes please ;-)
08:33:32 <eranrom> kota_: BTW do you want to bring IPython integration to Swift?
08:34:47 <kota_> eranrom: not sure for now, it could be nice but swift may be less interested in the ipython extension because swift is not designed to store coding cells iirc
08:35:06 <kota_> designed to storle blob rather than code
08:35:17 <kota_> s/storle/store
08:35:56 <eranrom> kota_: right. I was mainly thinking about GET/PUT
08:36:14 <eranrom> but worse case we can add it in ours if we find it useful
08:36:19 <kota_> i may introduce the feasibility in the design session but i think just related project would be good for swift.
08:36:30 <eranrom> kota_: sure.
08:37:10 <kota_> thinking of storlet, i think it should be in upstream repo because it's useful to deploy storlet apps into swift
08:37:45 <kota_> eranrom: and I'm waiting you to resolve your get line extension ;-)
08:37:54 <kota_> resolve merge conflict
08:38:11 <eranrom> kota_: np. sure, I will resolve + make some changes
08:38:52 <eranrom> its on the top of my list :-)
08:38:54 <kota_> eranrom: yeah, the point that we can get the response content as a variable is super useful.
08:39:02 <eranrom> kota_: indeed.
08:39:08 <takashi> kota_, eranrom: For me, it is very good chance to learn about iPython. :-)
08:39:26 <takashi> and I love the idea
08:39:38 <kota_> takashi: me too. I just was starting to learn in the last Dec.
08:39:57 <eranrom> I am working on machine learning integration with storlets, and IPython is tool many data sciuentists use
08:40:07 <eranrom> scientists
08:40:21 <eranrom> Kota's patch came exactly at the right time for me
08:40:24 <eranrom> :-)
08:40:30 <takashi> eranrom: great!
08:40:57 <kota_> eranrom: yeah, to be honest, I had a chance to talk with some data scientists too.
08:41:10 <eranrom> kota_: any interesting insights?
08:41:21 <kota_> eranrom: who uses ipython/jupyter notebook ;-)
08:41:25 <kota_> ah
08:41:52 <eranrom> kota_: Was this the reason you have added it?
08:41:54 <kota_> eranrom: they said something like dry-run should be useful to test their programs before deploying
08:42:01 <kota_> eranrom: sure
08:42:07 <eranrom> a-ha! nice!
08:43:09 <eranrom> ok, so if we talk about it, I think the next steps could be making life easier by creating templates that automat the loading of the dataset
08:43:55 <kota_> sounds cool
08:44:00 <eranrom> e.g. instead of the data scientist needing to take the in_stream and load it, add some magic code that adds this for her
08:44:03 <eranrom> or him
08:44:32 <eranrom> I will post this sometime...
08:46:03 <eranrom> anything else for today?
08:46:55 <akihito> I'm nothing.
08:47:09 <kota_> that's full
08:47:40 <takashi> nothing from my side
08:48:05 <eranrom> sagara: ?
08:48:35 <sagara> nothing
08:48:48 <eranrom> ok all so thank very much for joining!
08:48:57 <takashi> thanks!
08:49:01 <sagara> thx
08:49:01 <eranrom> talk to you later!
08:49:10 <kota_> thanks eranrom leading the meeting!
08:49:12 <akihito> thank you!
08:49:18 <eranrom> #endmeeting