Tuesday, 2017-01-24

*** valw has joined #craton00:09
*** valw_ has joined #craton00:13
*** valw has quit IRC00:14
*** Syed__ has quit IRC01:15
*** valw_ has quit IRC01:55
*** valw has joined #craton02:14
*** lcastell has quit IRC02:26
*** VW has joined #craton02:41
*** harlowja has quit IRC02:44
*** VW has quit IRC02:46
*** lcastell has joined #craton02:57
*** VW has joined #craton03:17
*** lcastell has quit IRC03:26
*** lcastell has joined #craton04:25
*** VW has quit IRC04:40
*** VW has joined #craton04:41
*** VW has quit IRC04:51
*** valw has quit IRC05:01
*** VW has joined #craton05:11
*** VW has quit IRC05:16
*** harlowja has joined #craton06:34
*** harlowja has quit IRC07:26
git-harrysulo: would it be accurate to say that every (endpoint, method) combination should be in scopes in schemas.py?12:18
sulotechnically yes it should be12:25
sulobut i dont think we are using it to check security scoping12:26
suloit was something from swagger that we moved away from12:26
git-harrysulo: yeah, the code doesn't look complete12:26
git-harryThat's why I was checking12:26
suloyeah, lets remove that to make it less confusing12:27
git-harryokay, that simplifies things. I'll do that instead.12:28
git-harryTa12:28
sulothanks12:28
sigmavirussulo: uhm that bug description is wrong13:15
sulowhich?13:15
sigmavirusglobal-requirements does care about python3 support13:15
sigmavirusit uses markers for deps that are py3 only or py2 only13:15
sigmavirusAnd there was a lot of work with sqlalchemy to find a mysql driver that worked well on py313:15
sigmavirusIf you search openstack-dev archives they found one that worked well13:16
sigmavirusI suspect we should use whichever one they found13:16
sigmavirusANd I suspect the one they found works for py2 or py313:16
sulocool13:17
sulowill check13:17
*** valw has joined #craton13:26
*** lcastell has quit IRC13:27
*** acabot has joined #craton13:41
*** valw_ has joined #craton13:44
*** valw has quit IRC13:46
*** valw_ is now known as valw13:59
*** acabot has quit IRC14:08
*** VW has joined #craton14:16
*** VW has quit IRC14:21
*** acabot has joined #craton14:27
*** VW has joined #craton14:46
*** VW has quit IRC14:46
*** VW has joined #craton14:46
*** jovon has joined #craton15:30
*** Syed__ has joined #craton15:33
*** lcastell has joined #craton15:41
*** VW_ has joined #craton15:56
*** VW__ has joined #craton15:59
*** VW has quit IRC16:00
*** VW__ has quit IRC16:01
*** VW has joined #craton16:01
*** VW_ has quit IRC16:01
*** VW has quit IRC16:03
*** VW has joined #craton16:03
sigmavirusjimbaker: we're overlappipng with what sounds like a very interesting tech talk16:45
jimbakersigmavirus, sorry about that16:45
sigmavirusjimbaker: not your fault16:46
sigmavirusI blame Syed__ ;)16:46
Syed__Haha16:46
jimbakersigmavirus, i'm happy to include 5 min on me talking on this very subject in today's meeting. but on second thought, no ;)16:47
Syed__is the tech meeting osic tech talk ?16:47
* jimbaker worked extensively with nosql before going back to sql16:47
sigmavirusSyed__: no, Rackspace16:47
sigmavirusjimbaker: I'm sorry?16:47
jimbakernosql is awesome for certain workloads of course. i am interested in the newsql space, such as cockroachdb and tidb (the latter of which aims to be in a nearly drop-in replacement for mysql, similar to google spanner has done internally at google)16:49
jimbakeri suspect there are various footnotes and other qualifications for the statement of being a "drop-in replacement"16:50
sigmavirusI bet Oracle would disagree =p16:51
jimbakeryeah, i haven't looked at the breadth of offerings that oracle is putting around mysql. certainly some interesting work there as well16:51
jimbakerneed to review the latest in mysql 816:53
jimbakerhttps://etherpad.openstack.org/p/craton-meetings for today's meeting17:00
jimbakersulo, git-harry, jovon - meeting17:00
*** VW has quit IRC17:04
*** VW has joined #craton17:07
*** valw has quit IRC17:30
*** d34dh0r53 is now known as ID-Ten-T17:45
*** ID-Ten-T is now known as blarnath17:48
*** blarnath is now known as d34dh0r5317:48
*** chrisspe- has joined #craton18:34
*** VW has quit IRC18:53
*** VW has joined #craton19:05
sulojimbaker: can you give your views on https://review.openstack.org/#/c/423855/19:10
*** VW has quit IRC19:10
jimbakersulo, will do19:10
jimbakeralso i figured out the foreign key constraint issue with respect to host deletes19:11
sulocool19:11
jimbakerprobably need to do this work more comprehensively19:11
sulowhat was the issue ?19:11
jimbakerjust not setting this up in the alembic migration; the underlying issue for that it seems like sqlalchemy needs some help in setup; or possibly the alembic discovery did not reproduce19:12
jimbakerregardless, we need it in our alembic setup19:13
jimbakeri guess we already knew alembic autodiscovery was weak19:13
jimbakerand that's fine19:13
jimbakerjust one more issue to fix19:14
jimbakeri am not going to explore the possibility that it cannot be generated by sqlalchemy ddl w/o alembic, since clearly that's not pertinent to us19:15
jimbakersulo, did that clarify? ;)19:15
suloyeah19:15
jimbakercould be two problems, masking each other. but it's certainly easy to fix19:15
jimbakeroh, and the fk constraint is just to ensure we do on delete cascade19:16
jimbakerbecause otherwise we are violating the terms of the fk19:16
jimbakerthe other option would have been to do soft deletes, but we early on decided not to do that approach19:17
jimbakersince we have lots of fk constraints, and they are centered on host, we see it show up19:17
jimbakersulo, re your comment on https://review.openstack.org/#/c/423855/2, yes it makes sense to me. so the ip address of the device is the management ip19:20
jimbakerit may have additional interfaces of course19:20
sulocool ... thats what i was going with. thx.19:21
jimbakerwill note on the review19:21
jimbakerthanks for asking me19:21
jimbakersulo, i have convinced myself that probably this is the only case we have for now19:30
jimbakerwith respect to fk constraint issues19:30
sulojimbaker: it is from what i know19:30
jimbakerwe will get more once we add rbac; and workflow/workflow def19:30
jimbakerbut hopefully we will get it right for that modeling19:30
jimbakereasy enough to test19:31
suloyeah i have functional tests thats failing on that .. everything else is passing19:31
jimbakersulo, please add your +2 to https://review.openstack.org/#/c/42385519:31
jimbakeri will add my comment and +219:31
jimbakermight as well do the two reviews in this case19:31
jimbakerand hopefully we will soon have more core reviewers so this becomes more routine practice :)19:32
jimbakeryou can wait for my comment first i guess...19:32
*** VW has joined #craton19:32
suloi already did19:32
jimbakeryep, it's fine :)19:32
sulojimbaker: sigmavirus: so a quick question here https://gist.github.com/sulochan/2a82357f9311b4a964e412cb88de715719:34
sulodo we want to show created_at and updated_at ?19:34
jimbakersulo, yes19:35
sulowe probably do ... which means19:35
sulojimbaker: +119:35
*** VW has quit IRC19:36
*** VW has joined #craton19:36
jimbakersulo, i just hope the convention is to do such reporting from the CLI with respect to UTC. hopefully! from what i can tell, such times are not consistently reported on in openstack CLIs. but it is very useful info19:39
suloi think so .. i cant imagine it not being UTC19:40
*** harlowja has joined #craton19:41
*** valw has joined #craton19:46
jimbakersulo, right, we will do it the sane way :)19:51
*** VW has quit IRC19:57
jimbakersulo, which functional test change that you wrote is failing on host deletion?20:06
suloi havent uploaded it yet .. once you fix that, ill upload the test20:07
*** VW has joined #craton20:08
jimbakersulo, i just wanted to use this as a parent, or otherwise reuse your test20:09
jimbakerbecause laziness ;)20:09
Syed__jimbaker: whats the best time to discuss somethings about brace ?20:10
Syed__rbac20:10
Syed__autocorrect*20:10
jimbakerSyed__, and here i was thinking you were suggesting we use 'import braces'20:10
jimbakerlet me get that right. try this: from __future__ import braces20:11
Syed__cool stuff, didn't knew about that. just found this article http://stackoverflow.com/questions/101268/hidden-features-of-python20:13
*** VW has quit IRC20:13
jimbakerSyed__, re rbac, we can discuss now. last time we discussed, we were looking at the mechanics of registering policy rules in oslo.policy20:30
jimbakerso we can use authorize, not just enforce20:30
Syed__yes so i have been looking to get enforcer authorize working but running into some issues20:31
jimbakerand iirc this is basically on the kinds of the rule, eg, rules like fleet:audit20:31
jimbakerwhere the kind is fleet20:31
Syed__http://docs.openstack.org/developer/oslo.policy/usage.html#how-to-register20:31
suloso, how do we want to display variables though cli ... ?20:31
suloas a json blog ... or list out k:v20:32
jimbakerSyed__, looks straightforward20:32
suloin table20:32
jimbakersulo, well not json by default20:32
jimbakeri wonder if there's some value doing k=v, if this could be consumed as env vars20:33
jimbakeralso k=v is more what i expect from options20:34
jimbakerhold on, worker at my house20:34
sulothe problem with displaying in table with rows of k/v is .. it gets too big20:34
Syed__jimbaker: https://gist.github.com/ahsan518/3d843212c4a596f4a26a1d5c21898d0820:34
sigmavirusSyed__: I'l just say that "pth" files are the devil20:35
sulosigmavirus: you got any thoughts on displaying variables ^20:35
* sigmavirus scrolls up20:36
*** VW has joined #craton20:36
jimbakersulo, so let's assume we are displaying a table like so https://gist.github.com/sulochan/2a82357f9311b4a964e412cb88de715720:36
sigmavirusI think I'd prefer it if we could display a table inside the column20:37
jimbakeri would assume that k, v would just follow that model; except that the value could potentially be arbitrarily long/problematic20:37
sigmaviruse.g. variables | <print new table here>20:37
sigmavirusjimbaker: right20:37
sigmavirusI would almost say we should have a separate call to get variables20:38
sigmavirusor a flag to control their retrieval/printing20:38
suloyeah, but that doesnt help too much i think ...since variable: {very long var chain}20:38
*** VW has quit IRC20:38
*** VW has joined #craton20:38
sigmavirussulo: I mean a table similar to what's there20:38
sigmavirusjust gives you a two column table inside the second column with that info20:39
sigmavirusbecause I'm an awful person and like to make people's lives harder when they try to sed/awk output20:39
jimbakersigmavirus, but why nest? effectively we are just adding properties here20:39
sulosigmavirus: ah ok .. so not worry about displaying indiviual vars right20:39
jimbakeri agree that sigmavirus is suggesting that he's evil, or something like that ;)20:39
sigmavirusso if those vars are displayed as if they're *real* attributes, you are going to confuse the life out of someone20:40
suloyeah agree20:40
jimbakersigmavirus, there's that20:40
sigmaviruswe could do a ruby-ish thing of that second column printing "var => value"20:41
sigmavirusBut I really think it should be something else20:41
sigmaviruscraton region-get --id 1 --variables20:41
sigmaviruscraton variables-for-region 120:41
sigmaviruscraton variables-for-user 2020:41
sulosigmavirus: +120:42
sigmaviruscraton variables-for-host 124664320:42
sigmavirusreads like a thing too20:42
jimbakerwhich can just use a standard table output20:42
jimbakerno ambiguity20:42
sigmavirusIdk what kind of thing, but a thing!20:42
sigmavirusjimbaker: right20:42
sigmavirusour formats should allow for JSON outputting too though20:42
jimbakeryes there's an extra strep20:42
jimbakerstep20:42
sigmavirusso people can use the stdout values and parse them20:42
jimbakerbut they can avoid with the python client20:42
sigmavirusjimbaker: not everyone writes things in python =P20:43
jimbakersigmavirus, hence the CLI20:43
sigmavirusthere are people who would pipe that to jq and do stuff20:43
sigmaviruscraton variables-for-host 1235 | bash_script_using_jq.sh20:43
jimbakeranyway the important thing is to make the cli reasonable, as opposed to maximum efficiency (which reqs the python client and presumed caching support, etc)20:44
sulocraton host-variables 1234 --json | bash_script.sh20:44
suloand craton host-variables 123 = tabled format20:44
jimbakerjq is new to me, but looks reasonable (any idea on how widely used jq is?)20:45
jimbakerhttps://stedolan.github.io/jq/20:45
jimbakersulo, maybe --json is a synonym for --format=json20:46
sulojimbaker: sure that works too20:47
jimbakeri can also see --format=env to output in env var form20:47
jimbakerpresumably with some sort of prefix. maybe selectable. ahh, the endless bikeshedding possibilities of defining a CLI :)20:48
jimbaker(btw, having used a bikeshed in sweden, i can see why they are amenable to such improvements. because they seem to end up being so terrible in practice)20:49
sigmavirussulo: --format json20:49
sigmavirusnot --json20:49
sigmavirusnever --json20:49
sulosigmavirus: 10-420:49
jimbakerworks for me20:50
sigmavirusand having a hackable/plugin-able --format flag has been wonderful for Flake820:51
sigmavirusSo I was going to steal my own code for cratonclient20:51
jimbaker+120:51
*** valw has quit IRC20:57
*** VW has quit IRC21:40
*** VW has joined #craton21:47
*** valw has joined #craton21:53
*** VW has quit IRC22:12
*** VW has joined #craton22:13
*** valw has quit IRC22:14
*** valw has joined #craton22:30
*** jovon has quit IRC22:40
*** david-lyle has quit IRC22:50
*** valw has quit IRC22:51
*** david-lyle has joined #craton22:53
*** VW_ has joined #craton23:04
*** VW has quit IRC23:08
*** harlowja has quit IRC23:08
*** VW_ has quit IRC23:08
*** VW has joined #craton23:33
*** VW has quit IRC23:38
*** harlowja has joined #craton23:58

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!