11:00:25 <oneswig> #startmeeting scientific-sig
11:00:26 <openstack> Meeting started Wed Mar 11 11:00:25 2020 UTC and is due to finish in 60 minutes.  The chair is oneswig. Information about MeetBot at http://wiki.debian.org/MeetBot.
11:00:27 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
11:00:30 <openstack> The meeting name has been set to 'scientific_sig'
11:01:02 <oneswig> Good morning.
11:01:21 <martial_> hello Stig :)
11:01:28 <oneswig> Apologies, no agenda was set today, had too much on.
11:01:35 <oneswig> Up early today martial_!
11:01:49 <oneswig> #chair martial_
11:01:50 <openstack> Current chairs: martial_ oneswig
11:02:36 <oneswig> COVID-19 is on our doorstep.  How about you?
11:03:02 <belmoreira> working from home!
11:03:37 <b1airo> hey hey
11:03:45 <oneswig> belmoreira: I'm due to fly to Geneva next weekend for a wedding in Chamonix.
11:04:18 <martial_> well we got the time change last week end, so a little later (7am here). Still on school bus duty so mostly in listen mode
11:04:19 <belmoreira> oneswig good luck... should be fine though :)
11:04:36 <oneswig> All public gatherings are now banned in that department.  Weddings as private gatherings - OK.
11:04:41 <oneswig> for now
11:04:50 <b1airo> ah, i see i have joined for thee COVID-19 travel disruption agony aunt hour :-)
11:04:54 <oneswig> Hey b1airo
11:05:03 <oneswig> #chair b1airo
11:05:04 <openstack> Current chairs: b1airo martial_ oneswig
11:05:11 <b1airo> how goes it?
11:05:19 <oneswig> Yep!  Good to get it out of the way :-)
11:05:43 <oneswig> Anyone taking bets for ISC?
11:06:03 <b1airo> 10:1 it's cancelled
11:06:09 <verdurin> Morning.
11:06:15 <oneswig> Hi verdurin, morning.
11:06:37 <b1airo> or at least trimmed down and moved online
11:06:50 <oneswig> Coronavirus is an opportunity to perfect our remote teleconferencing skills
11:07:29 <verdurin> It's more a test of our patience during the 10 minutes at the start when it doesn't work.
11:07:42 <belmoreira> hope this finished by then
11:07:43 <oneswig> Hangouts not so good for a family holiday, alas.
11:08:15 <b1airo> yeah, but the beaches are quiet!
11:08:59 <b1airo> i spend the majority of my day on Zoom already, so hasn't had any real impact yet
11:09:49 <oneswig> I think we are lucky to work in a sector that requires so little travel.  Although you guys do seem to get around...
11:10:38 <oneswig> belmoreira: is working from home a CERN policy or self-imposed?
11:11:41 <belmoreira> during these weeks we can chose to work from home
11:11:49 <b1airo> i think "requires" is the key work there oneswig - there are a lot of people suddenly realising they can still work and do business without being physically colocated, and being circumspect about when real physical meetings/workshops are required is a good thing from an environmental perspective
11:11:56 <b1airo> i think "requires" is the key word there oneswig - there are a lot of people suddenly realising they can still work and do business without being physically colocated, and being circumspect about when real physical meetings/workshops are required is a good thing from an environmental perspective
11:13:49 <oneswig> Certainly takes the shine off sitting in a plane, that's for sure.
11:13:51 <b1airo> i'm kind of morbidly excited about the wealth of data that (assuming it gets collected and standardised properly) could be hugely interesting across so many fields
11:14:10 <oneswig> you mean epidemiological data?
11:15:12 <b1airo> no, all the secondary and ternary effects, e.g., finance, travel activities, consumption, social impacts/changes, etc
11:15:53 <oneswig> Interesting point.
11:16:07 <b1airo> but having good epidemiological data will be important to quantify impacts on other areas
11:17:08 <oneswig> We should probably talk some SIG-specific things, although I've been so busy recently I haven't given it much thought, alas.
11:17:56 <b1airo> anyway, we got any OpenStack things to actually talk about. i don't really, other than wondering whether it's possible to rebuild a server originally booted from image into one booted from volume...
11:19:10 <priteau> b1airo: snapshot the vm, create volume from snapshot?
11:19:14 <b1airo> also curious to hear any Bright OpenStack success stories... trying to figure out whether it is worth investing further efforts in or ...
11:19:33 <oneswig> Interesting question b1airo.  It would involve moving pools in ceph,
11:20:27 <b1airo> priteau: i was wanting to do a rebuild so i keep the IP address (flat network) and don't give up my hypervisor slot to a race condition of some other little instance crowding me out
11:20:54 <oneswig> b1airo: Is Piotr still leading that for them?  Perhaps we should invite him to a SIG session to update on Bright OpenStack.
11:20:55 <priteau> Rebuild in the OpenStack sense
11:21:00 <oneswig> What are you running?
11:22:30 <b1airo> priteau: yeah, as in a `openstack server rebuild`. i suspect it is possible but maybe requires some arcane bdm magic
11:24:24 <b1airo> we're on something pretty old now, perhaps BOS 7.2 or 7.3
11:25:39 <oneswig> The "teapot" proposal somewhat overlaps with their world.  But in a k8s-centric way
11:27:11 <b1airo> yeah, that looks ... interesting
11:27:23 <b1airo> would like to know how it came about
11:28:08 <oneswig> Do you recall the demo Red Hat did in the keynote at Vancouver a while back?  Perhaps it is evolved from that, but with more k8s
11:31:22 <oneswig> How do "ideas" like that get filed?  Looks like Zane put a lot of work into it.
11:31:28 <b1airo> "more k8s", it truely is the modern IT panacea
11:34:42 <b1airo> does anyone here have much experience trying to take a VM image based application stack deployment and shove it into a k8s hosted service (with state). i haven't played much in this space yet so am wondering what the state of play is towards k8s being a reasonable generic hosting option for anything that can be mangled into a container...
11:35:05 <priteau> These "ideas" can be contributed to https://opendev.org/openstack/ideas via Gerrit, just like specs
11:35:13 <priteau> Here's the change for Teapot: https://review.opendev.org/#/c/710173/
11:38:09 <b1airo> ah ok, hadn't noticed the "openstack idea" concept come to fruition - makes sense, we used to do this stuff through specs, but always seemed kind of a forced way to manage bigger conversations
11:38:36 <priteau> This is especially useful for cross-projects ideas or completely new projects
11:39:32 <martial_> If I had to guess it has a lot to do with the OpenShift integration
11:39:53 <oneswig> Only the one idea so far it seems
11:40:18 <b1airo> martial_: you're talking about my VM question?
11:40:54 <martial_> I am thinking about the teapot concept too
11:41:39 <b1airo> oh ok, that makes more sense - getting rather late here :-)
11:44:14 <martial_> I am wondering how it will be reflected during the next PTG
11:44:56 <martial_> (which so far seems to still be happening in person)
11:47:05 <martial_> ready for the official AOB?
11:47:57 <martial_> #topic AOB
11:49:01 <martial_> related to VM to microservices conversation, Istio and Envoy are at the center of this conversation
11:50:35 <martial_> participated at a conference at NIST recently on microservices and a lot of istio talks
11:50:47 <martial_> a read on it here https://thenewstack.io/microservicing-with-envoy-istio-and-kubernetes/
11:51:50 <b1airo> thanks martial_ , but in this case i don't want microservices, i'm talking about a case where the application stack is a black box
11:52:27 <martial_> without endpoints?
11:52:52 <b1airo> just a webserver for example
11:53:28 <b1airo> but then has some heavy processing it might do
11:54:19 <martial_> not as sure then
11:55:17 <b1airo> this is a case of not wanting to get into a let's rearchitect and build this whole application "the right way (tm)" conversation for every researcher that comes along looking to host their custom new doodad that they want to share with their community
11:56:26 <martial_> understood, and it was not built using vagrant I guess?
11:58:36 <martial_> mostly wondering as it would have an idea of the setup and such
11:58:47 <b1airo> but i also don't want fleets of highly spec'd (cause e.g. it's a genomics widget) but very low utilisation VMs sitting about waiting for the odd request to come in - just a waste of resources (and management + security headache). feels like a reasonable middle-ground is putting the effort into defining the thing as a Dockerfile or similar, then setting it up as an on-demand service
12:00:15 <martial_> agreed, this would be easier and more efficient
12:00:16 <oneswig> Ah, we are on the hour
12:00:18 <b1airo> but the Dockerfile bit might even be a bridge too far in many cases - like pulling teeth
12:00:50 <b1airo> i've managed to talk everyone to sleep it seems oneswig
12:00:59 <martial_> if you can encapsulate the tool part vs the system part, might be feasible
12:01:10 <b1airo> good to "see" you all :-)
12:01:11 <martial_> fun challenge
12:01:24 <oneswig> OK, time to close.  Now go and wash your hands :-)
12:01:29 <oneswig> #endmeeting