18:07:01 <rluethi> #startmeeting training_labs
18:07:02 <openstack> Meeting started Mon Aug 29 18:07:01 2016 UTC and is due to finish in 60 minutes.  The chair is rluethi. Information about MeetBot at http://wiki.debian.org/MeetBot.
18:07:03 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
18:07:05 <openstack> The meeting name has been set to 'training_labs'
18:07:26 <rluethi> anyone here?
18:07:33 <rluethi> julen? dguitarbite?
18:08:17 <rluethi> yeah, thought so.
18:08:23 <rluethi> #topic python-port
18:09:16 <rluethi> The Python port of osbash has now overtaken its predecessor.
18:09:46 <rluethi> It can now do a PXE install on KVM which osbash cannot.
18:10:04 <rluethi> So, it's pretty much feature complete (for now).
18:10:04 <dguitarbite> hello
18:10:12 <rluethi> hey dguitarbite
18:10:15 <rluethi> welcome
18:10:19 <dguitarbite> my apologizes for being late, tried to run back home asap.
18:10:36 <rluethi> No problem, I had a nice conversation with myself.
18:10:59 <dguitarbite> wow
18:11:08 <dguitarbite> you beat me at that too!
18:11:24 <rluethi> at what? conversation with myself?
18:11:32 <dguitarbite> yeah!
18:11:41 <rluethi> That's just how good I am.
18:12:37 <rluethi> Anyhow, as I was saying, since the Python port does now everything osbash does and some more, I think I may look at Newton soonish.
18:12:46 * dguitarbite jel;ly jelly
18:13:06 <dguitarbite> rluethi: That is awesome, I read an interesting article]
18:13:56 <dguitarbite> which I conclude as to having your way of things bottom up rather than a fine new architecture and top down rewrite from scratch is not ideal for our situation.
18:14:35 <rluethi> Can you elaborate?
18:15:12 <dguitarbite> So we should just focus a bit on the architecture for migrating to what we want ideally with realistic baby steps. Basically start merging the current code with little refactoring and then work on improving it to get it to the ideal point.
18:15:36 <dguitarbite> Do you understand me? AFAIK and what I would love to do is to work on the python port you created and get it merged.
18:17:22 <rluethi> I have no idea how to put that into practice.
18:17:29 <dguitarbite> Ofcourse add the libvirt kvm backend to it too.
18:18:10 <dguitarbite> rluethi: Give me one week to finalize the parser, it is near to be done and then we start working on the python port. I could ideally push myself harder and work more but its something I do not wish to promise.
18:18:39 <dguitarbite> And there is always that doomed day job!
18:19:00 <rluethi> I already have a KVM backend that works. What does a libvirt KVM backend buy us?
18:19:13 <dguitarbite> in that case nothing.
18:19:48 <rluethi> I will try to push the current code soonish.
18:20:00 <rluethi> It has improved quite a bit.
18:20:20 <dguitarbite> Please push it to github fork :)
18:21:01 <dguitarbite> Then we could refactor it a bit more and finally merge it upstream.
18:21:07 <rluethi> Will do. I wanted to take care of some TODO and FIXME in the code, but I may push the code with them in it.
18:21:22 <dguitarbite> Yes :)
18:21:36 <dguitarbite> How long before you push it?
18:21:50 <dguitarbite> So what I would be looking at first is how to refactor and rearrange them
18:22:09 <dguitarbite> Then once I have some conclusions there I will move on the run the code and make it deploy the cluster.
18:22:16 <rluethi> Well, you are busy with the parser anyway, so I guess I'll push some time later this week.
18:22:26 <dguitarbite> I should potentially finish my python POC for running the same code with distro switch logic
18:24:10 <rluethi> are we done talking python port?
18:24:24 <dguitarbite> Yes
18:24:30 <rluethi> #topic barcelona
18:24:49 <rluethi> I'd like to book the hotel soon.
18:25:00 <rluethi> Rates won't get cheaper if we wait.
18:25:15 <dguitarbite> Is it possible to wait for next week?
18:25:18 <rluethi> Sure.
18:25:25 <rluethi> What's next week?
18:25:41 <dguitarbite> My manager is back from vacation next week so I would have some clarity on the summit travel.
18:25:53 <rluethi> Of course. Next week it is.
18:26:50 <dguitarbite> Awesome, then next week in the evening :)
18:27:00 <rluethi> #topic parser
18:27:25 <dguitarbite> :D
18:27:28 <rluethi> so, what's the plan for the parser this week?
18:27:37 <dguitarbite> To make it run and deploy OpenStack :D
18:27:55 <rluethi> I was more thinking about _how_ :)
18:27:55 <dguitarbite> Just a few simple tricks and it should be ready to be tested with the training-labs system/.
18:28:10 <dguitarbite> I had a chat with Andreas today.
18:28:40 <dguitarbite> So only two things remain from my side apart from some minor bugs.
18:28:47 <dguitarbite> 1. Distro switch logic
18:29:05 <dguitarbite> 2. Jijna templates to hook it with all the training-labs format and exceptions like for neutron and nova etc.
18:29:34 <rluethi> I think you are underestimating "some minor bugs".
18:29:51 <rluethi> There's plenty of stuff to consider and fix, we talked about some of them before.
18:30:01 <dguitarbite> I did take a look at it today and the minor bugs are just some parsing issues. Yes!
18:30:13 <dguitarbite> I hoope I could address them all sooner than later.
18:30:20 <rluethi> So what are you going to do about variable?
18:30:52 <rluethi> Are we going to drop our system that lets users change network addresses in a configuration file?
18:31:10 <dguitarbite> I thought about it and the simple solution is to introduce some changes to install-guides.
18:31:29 <rluethi> Heh. Simple for us.
18:31:39 <dguitarbite> We create and maintain a bash script which does all the variable population which the install guide users would wget/download and run while using it manually :D
18:32:03 <dguitarbite> I thought about potential blockers/objections, there should be none to very minimal. Its easy to solve :).
18:32:53 <dguitarbite> The other way is to have some sort of NLP with ML to improve the system over time, otherwise its almost impossible to make any logical sense out of the same because things change, API\s are different for different project and commands and tasks and so on
18:33:08 <dguitarbite> So writing a simple ubiquitous logic is near to impossible,.
18:33:16 <dguitarbite> It is a np incomplete problem.
18:33:35 <rluethi> I know it's hard :).
18:33:57 <dguitarbite> So let's KISS it ;)
18:34:13 <rluethi> The question is rather: how useful can a system that is not too hard to implement be?
18:34:20 <rluethi> Time will tell, I guess.
18:35:16 <dguitarbite> I don't need to butter my ego or pride with the parser, I simply want us to focus on less mundane tasks and also in turn help the community while we help ourself, so is the parser simple or complicated, I honestly dont know and care.
18:35:33 <dguitarbite> But it is definitely simpler compared to other parsers for sure.
18:36:02 <dguitarbite> Maybe over time the perfectionist in me would make it much more design principle oriented, although it already is quite so.
18:37:19 <rluethi> For now, my focus is on seeing that it can work. I haven't seen that yet.
18:37:43 <rluethi> It's good to see it make progress :).
18:38:06 <dguitarbite> rluethi: Come on what you saw was already magnificent is it not? The last 1% is the highest amount of work :D ... but now I am in the zone so I am faster and more efficient for the same to be done quite soon enough.,
18:38:21 <rluethi> If you make updates to the rst2bash repo, please let me know so I can check it out.
18:39:11 <dguitarbite> Yes, my focus is to get it working as per your wishes by Wednesday because I want to present it to AJ on Thrus and finalize some other decisions like where would the parser lie etc.
18:39:31 <dguitarbite> s/lie/home|repo location ;)
18:39:33 <rluethi> I was quite happy to finally see something tangible, but I am afraid there is a lot work ahead of us before it does what it is supposed to od.
18:40:31 <rluethi> hah, that's a good question.
18:40:53 <dguitarbite> rluethi: Yes, but I am not doing anything else till I finish this and I will do it by Wednesday for sure. Further improvements to the design and datastructures would be a high priority on Friday and I want to get this ready by next week with all the decisions required for the same and be ready to start merging the code and run it as a non-compulsory jenkins test very soon.
18:41:44 <rluethi> can't wait to see it.
18:42:04 <rluethi> anything else we need to discuss?
18:43:16 <rluethi> dguitarbite?
18:43:41 <dguitarbite> nope
18:43:52 <dguitarbite> I think I am all done, lets call it a day :)
18:44:01 <rluethi> then I wish you best of luck with the parser :).
18:44:11 <rluethi> ttyl
18:44:13 <dguitarbite> Thank you. I will work hard :)
18:44:17 <dguitarbite> TTYL. Thanks and bye,
18:44:20 <rluethi> #endmeeting