ansible_azure_working_group
LOGS
00:00:17 <nitzmahone> #startmeeting Ansible Azure Working Group
00:00:17 <zodbot> Meeting started Thu May 31 00:00:17 2018 UTC.
00:00:17 <zodbot> This meeting is logged and archived in a public location.
00:00:17 <zodbot> The chair is nitzmahone. Information about MeetBot at http://wiki.debian.org/MeetBot.
00:00:17 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
00:00:17 <zodbot> The meeting name has been set to 'ansible_azure_working_group'
00:00:29 <jborean93> howdy
00:00:44 <zikalino> good morning :-)
00:00:45 * nitzmahone will be partially unavailable- trying to debug a build issue
00:01:14 <nitzmahone> get @jborean93 to poke me on slack if needed
00:01:25 <zikalino> i think not much today. i think i am the only one to join today
00:01:42 <nitzmahone> Cool- saw some movement on the unstable test work- thanks!
00:02:07 <zikalino> yes, that's what i am focusing on right now, i wan to close all the issues by the end of this week
00:02:12 <jborean93> yep, always good to see tests getting fixed
00:02:14 <zikalino> however some are tricky
00:02:41 <zikalino> function app fails because that azure api has very low limit of allowed requests per hour.
00:03:02 <zikalino> so i am just trying to figure out what's the reason.
00:03:33 <yungezz> hello, anyone there?
00:03:40 <zikalino> and i have a few sql, and postgresql issues, which don't appear any more.
00:03:40 <jborean93> yep
00:04:05 <zikalino> and vm will require a litle bit more attention.
00:04:09 <yungezz> hello Jordan, i have several prs need your review
00:04:27 <yungezz> will start it after zim
00:05:20 <yungezz> @zikalino, how about the appgateway module?
00:05:28 <zikalino> yeah, i just had quick update on test stability, next i will continue with facts modules. so that's all from me.
00:05:37 <yungezz> cool.
00:05:45 <nitzmahone> #chair jborean93 @zikalino @yungezz
00:05:45 <zodbot> Current chairs: @yungezz @zikalino jborean93 nitzmahone
00:05:46 <zikalino> about app gateway i am creating proper sample code.
00:05:56 <zikalino> i will push it next week
00:06:21 <jborean93> sounds good
00:06:21 <zikalino> @yun
00:06:40 <yungezz> @jborean93  @nitzmahone, pls help to look at web app module after discussion at last week, I updated ux https://github.com/ansible/ansible/pull/40005
00:07:48 <yungezz> https://github.com/ansible/ansible/pull/38084 fixed keyvault bugs, and I added a azure service principal object id lookup plugin, now keyvault tests passed, while some ci test failed to install supervisor package,
00:08:02 <yungezz> no idea how it's related to my change? could you shed any light?
00:08:17 <jborean93> yungezz: so for web app, a windows app can have multiple frameworks while a linux one cant?
00:08:45 <yungezz> yes, you're right
00:08:58 <yungezz> for windows, java_version is mutual exclusive with other framework
00:09:19 <yungezz> you can see sample or tests
00:09:32 <jborean93> is there a reason why linux one's can't have multiple frameworks
00:09:42 <jborean93> is there a chance that will change in the future
00:09:51 <zikalino> probably just design of the service
00:10:26 <yungezz> i guess because linux currently  is container based, no idea whether they're going to change it.
00:10:40 <zikalino> i was actually that there could be a parameter with list of frameworks, and restriction to have only one item for linux.
00:10:44 <jborean93> yea
00:10:58 <jborean93> Was just about to say instead of having `linux_framework` and `windows_framework`
00:11:09 <jborean93> just have `framework` which takes a list of dicts containing name and version
00:11:11 <yungezz> then it's hard to explain to users, user need to know the complext logic and different
00:11:36 <jborean93> Not really just put in the description "for linux apps you can only have 1 framework"
00:11:49 <yungezz> with one framework parameter, user will need have deep knowledge on what to set, and list of supported framework in windows is different with linux
00:11:54 <jborean93> right now we have 2 options that do similar things with a different option structure
00:12:21 <yungezz> windows support java, php, python, node, netframwork, linux supports java, python, ruby, node i think
00:12:44 <yungezz> it's clear to user to what to set, for windows and for linux
00:12:44 <jborean93> You can always put that in the description and on examples
00:13:13 <jborean93> Right now for Windows you are setting an option for each framework. When more are added that's another version
00:13:25 <yungezz> if one paremter, somethings it's a list, sometimes it's must contains one and only one
00:13:35 <yungezz> yes
00:14:21 <jborean93> https://www.irccloud.com/pastebin/kcC0woWL/
00:14:47 <jborean93> That's how I would see it
00:15:00 <jborean93> both ways are the same, the name of the framework may differ
00:15:51 <jborean93> Also I thought the plan options were changing, looks like you just removed the suboptions definitions but the options are still there
00:15:57 <jborean93> I can't remember where we landed with that
00:16:08 <yungezz> 2 considerations, the paramter, sometimes it's a linux, some times it's limited to one element only. and windows and linux support different framework
00:16:37 <jborean93> yep, we can always add a check in there, when linux make sure there is only 1 entry
00:16:41 <jborean93> for windows, go for gold
00:17:11 <yungezz> you also need to check value under each  platform, and they're different list
00:17:29 <jborean93> That's what the descriptions are for
00:17:30 <yungezz> to get platform, you'll need make a call to azure to determine it
00:17:43 <jborean93> Isn't that specified in the plan?
00:17:54 <jborean93> how expensive is that call?
00:18:13 <yungezz> for existing plan, you'll need make the call
00:18:39 <yungezz> i think it's a normal azure get call
00:19:14 <jborean93> that doesn't sound too bad to me
00:20:26 <jborean93> if you really wanted to, there's a potentially to skip that if a user specified a windows or linux only app but that's not a scenario that will happen all the time
00:22:08 <jborean93> then there's also a question whether the java container settings should be it's own option or just part of the framework definition
00:22:19 <yungezz> with one parameter, user will need understand the linux/windows logic by themselves
00:22:24 <yungezz> to fill this paramter right
00:23:00 <jborean93> they need to know if it is Windows/Linux anyway. I don't see how it is different from how the linux_frameworks is being done right now anyway
00:24:09 <yungezz> for now, if linux,usre just choose from a choice, that's all. for windows, user will see multiple suboptions to fill in,it's straigtway
00:24:28 <jborean93> as more and more apps are added to Windows, that means more version options to add to the windows frameworks
00:25:00 <yungezz> yes. if it's one parameter, user will fill in them one by one anway
00:25:25 <yungezz> by findout what's framework name for windows, what's framework name for linux
00:25:52 <jborean93> We list those in the documentation
00:26:12 <jborean93> and hopefully give a link to the Microsoft doc that defines them as well
00:26:38 <jborean93> aka what you have currently done for linux_framework
00:26:52 <yungezz> yes, documentation will require user need read it before, and understand what's it meanings. with 2 parameters, it's naturally to get sense what to fill in
00:27:22 <jborean93> I disagree, people shouldn't be blindly using options without reading what they are for beforehand
00:27:34 <jborean93> they need to read it anyway to find out what subkeys the windows_framework one does
00:27:41 <yungezz> user no need to take care of understanding when to fill multiple options, when i have to specify one option only
00:27:44 <jborean93> so why not just list it in the same place
00:28:40 <yungezz> how about let me re-think over the framework thing and go back to you?
00:28:58 <yungezz> i understand your main concern,
00:29:11 <yungezz> from meaning, they seems one thing
00:29:12 <zikalino82> i think in most of the cases they will try to use existing samples, so i think samples are most important. then they can try to modify options.
00:30:14 <jborean93> yes having good examples are a really good thing
00:30:22 <yungezz> agree
00:30:34 <jborean93> if a user tries to have multiple frameworks for linux then we should give them good error messages that tell them that isn't allowed
00:30:41 <yungezz> and i've have samples
00:31:25 <yungezz> yes, that's one option, fail it and let user know it.
00:31:26 <jborean93> In cases where the plan is being created, do the users have to specify `is_linux` right now, does it default to True/False if they omit it
00:31:33 <yungezz> no
00:31:38 <yungezz> they omit it
00:31:45 <jborean93> So what happens then
00:31:49 <jborean93> is a Windows or Linux app created
00:32:11 <yungezz> that depends on platform of the exsiting plan
00:32:21 <yungezz> if the plan is linux, then linux, otherwise windows
00:32:24 <jborean93> but if that plan doesn't exist (it is being created)
00:32:44 <jborean93> or will this module not create the plan
00:33:03 <yungezz> oh, then is_linux is required if user want to create a new plan, is_linux default false
00:33:21 <yungezz> the module will create the plan
00:33:27 <jborean93> So in all cases we know (or can get) the info as to whether the platform will be linux or not
00:33:37 <jborean93> so we can easily check the number of entries in frameworks and fail accordingly
00:34:12 <jborean93> as well as fail if the framework name is not valid for that platform
00:34:17 <jborean93> if we really wanted to do that as well
00:35:17 <yungezz> yes, we can , so one option is fail user to let them know something, or another option let them not fill in wrong value
00:35:52 <jborean93> yes, make sure we are proactive in checking and failing with a nice error message telling them what is wrong
00:36:06 <jborean93> we should also make sure the examples and option description makes this quite clear
00:36:27 <yungezz> jborean93, i'll think over the webapp framework thing. i think we can move on to other prs
00:36:35 <jborean93> ok
00:36:48 <yungezz> i have a keyvalut bug fixing pr https://github.com/ansible/ansible/pull/38084
00:37:11 <yungezz> one of ci job failed to install some supervisor package, no idea how it's related to my change
00:37:23 <yungezz> not sure do you have any idea?
00:37:41 <yungezz> yuwei, do you have any pr need Jordan's review?
00:38:39 <jborean93> for https://github.com/ansible/ansible/pull/38084/files, having the lookup done as a separate PR is probably better
00:39:12 <yungezz> o, ok, got it, will seperate them out then, thanks
00:39:28 <yungezz> i think that's all my part
00:39:37 <yungezz> @yuwei do you have any thing wnat to talk
00:39:51 <yungezz> btw, @jborean93, what's next 2.7 release plan?
00:40:01 <yuwei> nothing
00:40:06 <jborean93> I don't think any dates have been announced
00:40:18 <jborean93> 2.6.0 is still going on
00:40:25 <jborean93> we are just in bugfix mode now
00:40:46 <yungezz> got it.
00:40:54 <yungezz> thanks Jordan
00:41:05 <yuwei> I find current version of devel is 2.7
00:41:32 <jborean93> yep, the stable-2.6 branch has been created and the devel version bumped to 2.7
00:41:48 <jborean93> so any merges into devel are going to be for 2.7
00:42:01 <jborean93> any 2.6/2.5 bugfixes now need to be backported to those respected stable branched
00:42:34 <yuwei> I see. Thanks
00:42:38 <jborean93> The final 2.6.0 release should be coming out soon, we just need to go through some betas and release candidates before making the final release
00:43:17 <yuwei> maybe there are some new bit need to backport to 2.6 in the past few days
00:44:28 <jborean93> if they have been recently merged then potentially. I'm not sure when stable-2.6 was cut but it definitely was recently
00:45:03 <jborean93> looks like 5 days ago
00:45:06 <yuwei> okay I'll check it
00:45:51 <yungezz> @zikalino, do you have anything wnat to talk?
00:46:25 <yungezz> zikalino82
00:46:57 <yungezz> i think we're done then
00:47:06 <jborean93> cool
00:47:15 <yungezz> thank you Jordan
00:47:16 <jborean93> thanks everyone
00:47:19 <zikalino82> no, nothing more today
00:47:23 <zikalino82> thank you all!
00:47:27 <yungezz> will ping you in github then :)
00:47:30 <jborean93> #endmeeting