16:00:10 <gundalow> #startmeeting Network Working Group
16:00:10 <zodbot> Meeting started Wed Nov 22 16:00:10 2017 UTC.  The chair is gundalow. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:10 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:00:10 <zodbot> The meeting name has been set to 'network_working_group'
16:00:50 <gundalow> #chair itdependsnetwork st8less jmcgill298 privateip
16:00:50 <zodbot> Current chairs: gundalow itdependsnetwork jmcgill298 privateip st8less
16:02:34 <caphrim007_> o/
16:02:45 <gundalow> #chair skg-net caphrim007_
16:02:45 <zodbot> Current chairs: caphrim007_ gundalow itdependsnetwork jmcgill298 privateip skg-net st8less
16:02:49 <gundalow> Hello hello!
16:03:08 <caphrim007_> i needed to re-adjust my calendar after daylight savings
16:03:18 <caphrim007_> i've been an hour late the last couple weeks :p
16:03:25 <gundalow> caphrim007_: You should be able to just add by URL the ics
16:04:08 <gundalow> https://raw.githubusercontent.com/ansible/community/master/meetings/ical/network.ics
16:04:21 <caphrim007_> oh goodie. thanks!
16:04:28 <gundalow> #info Agenda https://github.com/ansible/community/labels/network
16:04:46 <gundalow> Add by URL, if you download & add you will not get updates
16:05:19 <gundalow> #chair Anil
16:05:19 <zodbot> Current chairs: Anil caphrim007_ gundalow itdependsnetwork jmcgill298 privateip skg-net st8less
16:05:27 <gundalow> #topic nmap Dynamic inventory
16:05:50 <gundalow> #info nmap dynamic inventory https://github.com/ansible/ansible/pull/32857
16:06:02 <gundalow> #info Maybe of useful to some of you, reviews welcome
16:07:17 <gundalow> #topic Persistent Connection changes in 2.5
16:08:05 <gundalow> #info Now that https://github.com/ansible/ansible/pull/32825 is merged the Persistent Connection socket will be closed at the end of every play. Previously the socket would be left open for $TIMEOUT seconds after an ansible-playbook run
16:09:26 <gundalow> #info We are looking at including the PID of the main ansible process as part of the socket hash.
16:09:51 <gundalow> #info connection: network_cli is in devel, this is a big change, we welcome testing
16:09:56 <gundalow> Any questions on this?
16:10:19 <gundalow> #info We are putting together some draft developer docs on `connection: network_cli`
16:10:42 <Anil> I am testing it already...it works fine
16:10:50 <gundalow> Anil: Excellent
16:11:50 <Anil> off late I got some Authentication (publickey) failed.. any idea on that?
16:12:05 <Anil> while connecting switch
16:12:19 <Anil> I did export ANSIBLE_PARAMIKO_HOST_KEY_AUTO_ADD=True
16:12:56 <gundalow> hum, I don't believe that should have changed
16:13:10 <gundalow> Anil: did you notice that you get better error messages rather than `unable to open shell`
16:13:11 <itdependsnetwork> will that include how non core modules can access ansible_username and password?
16:13:39 <Anil> yes.. unable to open shell is no more there
16:13:55 <gundalow> itdependsnetwork: How do you mean?
16:14:11 <gundalow> oh the dev docs?
16:14:15 <itdependsnetwork> yes, sorry
16:14:16 <gundalow> SO the developer docs will
16:14:25 <itdependsnetwork> thanks
16:14:28 <gundalow> * Explain at a high level what network_cli is
16:15:48 <gundalow> * Include an example of a platform that only supports network_cli (the end goal is to not have top-level nor `provider:` at all) <-- From 2.6 this will be the prefered(/required?) way that new *platforms* must be developed
16:16:14 <gundalow> * Include details on extending provider (connection: local) to also support connection: network_cli
16:17:11 <gundalow> itdependsnetwork: Does that give you what you need?
16:17:26 <gundalow> Obviously difficult to tell to you see the words :)
16:18:15 <gundalow> Plan for this is we will use the wiki to quickly get some docs together, and between us improve it when ever there are question. Once it's in a reasonable shape we will turn it into a formal doc RST under dev_guide in the main repo
16:19:03 <itdependsnetwork> I'm going to guess that will be enough
16:19:05 <gundalow> Any other questions on that?
16:19:31 <skg-net> gundalow: that will help.. can it have migration steps for other platforms which is supported by community
16:19:32 <itdependsnetwork> Still a little abstract for me, but I think this will be the place to document it one way or anther
16:19:43 <gundalow> skg-net: Yup
16:19:49 <gundalow> itdependsnetwork: Yup, I agree
16:21:53 <gundalow> #topic Open Floor
16:21:58 <gundalow> #chair
16:21:58 <zodbot> Current chairs: Anil caphrim007_ gundalow itdependsnetwork jmcgill298 privateip skg-net st8less
16:22:03 <gundalow> Anyone got anything else?
16:22:12 <itdependsnetwork> Can I get recommendation for naming on this: https://github.com/ansible/ansible/pull/28446
16:22:28 <gundalow> #info Holiday in USA Thu & Friday, so limited response over the next two days
16:22:47 <gundalow> #topic Type manipulation #28446
16:22:56 <gundalow> #info https://github.com/ansible/ansible/pull/28446
16:23:50 <gundalow> itdependsnetwork: did you have any thoughts?
16:24:02 <itdependsnetwork> list_to_dict :)
16:24:06 <gundalow> hum, i though there were some ideas suggested before, though I don't see them
16:24:16 <itdependsnetwork> intent_dict I think was one
16:25:00 <itdependsnetwork> cast_list_to_dict
16:25:02 <st8less> I'm trademarking "intent"!  :P
16:25:25 <st8less> I like the cast name.
16:25:55 <gundalow> If two people here can agree on a name then I'll merge it
16:26:06 <itdependsnetwork> +1 from me too :)
16:26:28 <gundalow> As we know naming things (Declaritive Intent, Aggregates/Collections) is really difficult
16:26:46 <st8less> The second hardest problem in CS!
16:26:56 <jmcgill298> what is the hardest?
16:27:03 <gundalow> 2) counting
16:27:11 <jmcgill298> +1
16:27:12 <gundalow> 0) out by one errors
16:27:22 <gundalow> (aka zero index)
16:28:13 <gundalow> itdependsnetwork: cool update and ping me on IRC and I'll merge it
16:28:21 <itdependsnetwork> thank you sir
16:28:32 <gundalow> nah, thank you
16:28:39 <gundalow> #topic Open Floor
16:28:44 <gundalow> Any more for anymore?
16:29:01 <itdependsnetwork> There was one more
16:29:05 <gundalow> Sure
16:29:28 <itdependsnetwork> legacy facts
16:29:50 <gundalow> #topic Legacy Facts
16:30:07 <gundalow> itdependsnetwork: Is that related to https://github.com/ansible/ansible/pull/31783
16:30:46 <itdependsnetwork> don't think so, vlans was moved to legacy, trying to understand what the direction of facts is
16:32:12 <itdependsnetwork> https://meetbot.fedoraproject.org/ansible-network/2017-11-01/network_working_group.2017-11-01-16.00.html
16:32:23 <jmcgill298> is the goal to use gather_facts: yes for network modules?
16:32:23 <itdependsnetwork> part 5 on that ^^
16:33:13 <gundalow> hum, looks like that gundalow fellow isn't doing his actions
16:33:21 <itdependsnetwork> lol
16:33:25 <itdependsnetwork> next week?
16:34:01 <gundalow> itdependsnetwork: right, it's on the Internal Network agenda
16:34:22 <itdependsnetwork> thanks
16:34:35 <gundalow> #chair rcarrillocruz
16:34:35 <zodbot> Current chairs: Anil caphrim007_ gundalow itdependsnetwork jmcgill298 privateip rcarrillocruz skg-net st8less
16:36:08 <gundalow> 1) is the goal to use gather_facts: yes for network modules?
16:36:18 <gundalow> What other things would you like me to find out?
16:37:11 <itdependsnetwork> me?
16:37:27 <gundalow> me=gundalow
16:37:48 <itdependsnetwork> lol, is question directed to itdependsnetwork
16:37:54 <gundalow> yes
16:38:02 <gundalow> It's been a long day
16:38:16 <itdependsnetwork> https://meetbot.fedoraproject.org/ansible-network/2017-11-01/network_working_group.2017-11-01-16.00.log.html
16:38:47 <itdependsnetwork> same as there, 16:26 -> 16:37
16:39:37 <itdependsnetwork> but short of it, seemed like facts were actively being pulled out, but when I suggested a get method, was told to use facts
16:39:37 <skg-net> https://github.com/ansible/ansible/commit/751eab187f51cdc58a4c112abcb4265b7112ef1b#diff-83a6bf7294d0c5ac59ee458de130f795R355
16:39:48 <itdependsnetwork> these two seem in contrast with each other
16:40:33 <itdependsnetwork> but again, that is my interpretation of what legacy means
16:42:14 <gundalow> itdependsnetwork: That's great. want to make sure I ask the right questions
16:42:42 <itdependsnetwork> good deal
16:42:44 <gundalow> skg-net: Ah, good link, thanks
16:42:50 <gundalow> Anything else on facts?
16:44:02 <samerd_> Hi, I made a pull request, and it seems to be failing on docker exec
16:44:10 <samerd_> can anyone help me with it?
16:44:40 <samerd_> https://github.com/ansible/ansible/pull/33121
16:45:03 <gundalow> samerd_: Hi welcome. This channel is about using Ansible to configure network devices. Could I direct you to #ansible-devel they will be able to help
16:45:08 <gundalow> #topic Open Floor
16:45:12 <gundalow> #chair
16:45:12 <zodbot> Current chairs: Anil caphrim007_ gundalow itdependsnetwork jmcgill298 privateip rcarrillocruz skg-net st8less
16:45:25 <gundalow> Anyone got anything else?
16:45:25 <jtanner> gundalow: it's a networking PR
16:45:32 <gundalow> jtanner: oh
16:45:37 <gundalow> samerd_: appologies
16:45:42 <jtanner> mellanox switch config
16:46:04 <gundalow> samerd_: sorry, I though it was a docker PR I can help you with that
16:46:46 <jtanner> "2017-11-21 09:26:58 ERROR: Timeout waiting for vyos/1.1.0 instance 3aef8a0a-915f-44bc-baa5-9b49d817b093."
16:47:10 <gundalow> + yum issue, I've hit rerun
16:47:54 <gundalow> #info There is an issue with VyOS 1.1.7 AWS image causing Shippable TEST=network issues to fail with a ssh key error. This is an AWS/VyOS bug. Fix is in flight
16:49:14 <samerd_> gundalow: so the tests will be re-run again now?
16:49:29 <gundalow> yup
16:49:54 <samerd_> Thanks, will track it and see
16:50:38 <samerd_> Actually this is my first PR in ansible, could you let me know about the process of the PR?
16:51:15 <samerd_> I mean after the test passes
16:55:17 <gundalow> samerd_: We will review and once happy merge. Just as a heads up we are part way though changing the connection framework in 2.5 so we are discussing internally about how new modules should work moving forward, so we may be in touch with some other details
16:55:35 <gundalow> Though if we do, we will guide you through that process
16:56:30 <jmcgill298> gundalow: I know a lot of modules use ssh, but I wanna make sure any differences in http is also included
16:56:54 <gundalow> Good question, let me go into some more detail
16:57:03 <gundalow> #topic 2.5 connection: network_cli
16:57:22 <gundalow> #info Right now in 2.5 you can do `connection: network_cli` for the Core maintained network modules
16:58:42 <jmcgill298> so no planned changes for http connections at this point?
16:58:57 <gundalow> #info HTP(s) transports such as EAPI and NXAPI have not changed *yet*. It's likely that in 2.5 that they will not change, so you will still need `provider:\n  transport: eapi`
16:59:23 <jmcgill298> ya, I was thinking about ACI, but same deal :)
16:59:35 <gundalow> #info The aim is in 2.6 to add support for `connection: network_eapi` `connection: network_nxapi`
16:59:39 <gundalow> etc
17:00:08 <jmcgill298> hmm, is there enough difference for connection handlers to not have network_http?
17:00:19 <gundalow> #info when using `connection: network_cli` you must use `become:` & `become_method`
17:00:44 <gundalow> jmcgill298: oh, and the module does the switching, interesting idea
17:00:53 <gundalow> s/module/connection plugin/
17:01:08 <jmcgill298> just seems to make it simpler to use for the end users
17:01:22 <gundalow> Aye, that's a good idea, not though of that
17:02:32 <gundalow> We haven't fully designed the HTTP interfaces yet, for example, we are not sure how validate_certs would be specified
17:03:37 <gundalow> #info User facing docs for network_cli are in https://github.com/ansible/ansible/pull/31807 which also details `become`
17:03:55 <jmcgill298> how are other groups dealing with it? seems like that should be part of inventory like ansible_host
17:04:17 <gundalow> "other groups"?
17:04:38 <jmcgill298> windows, cloud, etc.
17:05:52 <gundalow> Well for cloud you are always talking to a valid https server
17:06:08 <jmcgill298> ya, I guess only network peeps are insecure
17:06:32 <rcarrillocruz> well, some clouds you can make it with self-signed
17:06:41 <rcarrillocruz> openstack modules there's a validate_certs no param
17:06:50 <gundalow> jmcgill298: slow steps away from telnet
17:07:08 <gundalow> rcarrillocruz: ah, good point. I wasn't thinking about on-prem cloud
17:07:37 <privateip> sorry i got pull away
17:07:45 <privateip> yes the idea is to have a single network connection for api
17:07:48 <rcarrillocruz> openstack is not just on-prem, but yeah, i don't think there's any openstack public cloud out there that is not serious enough to have self-signed certs
17:08:12 <privateip> most likely that will be a post 2.5 implementation
17:08:31 <gundalow> privateip: Have we thought about how validate_certs would be specified?
17:08:46 <privateip> with the new config stuff we will be able to add that as a property
17:09:56 <privateip> was there anything else i missed?
17:10:07 <gundalow> #topic Legacy Role
17:10:11 <gundalow> itdependsnetwork: You still there
17:10:18 <gundalow> privateip: will paste some lines in
17:10:33 <gundalow> what is the status on facts? Based on this: https://github.com/ansible/ansible/commit/751eab187f51cdc58a4c112abcb4265b7112ef1b#diff-83a6bf7294d0c5ac59ee458de130f795R355
17:10:41 <gundalow> Is the goal to use gather-facts: yes for network modules?
17:10:51 <gundalow> seemed like facts were actively being pulled out, but when I suggested a get method, was told to use facts
17:10:57 <gundalow> these two seem in contrast with each other
17:11:06 <gundalow> but again, that is my interpretation of what legacy means
17:11:10 <privateip> we are re-writing the entire facts system for network devices
17:11:11 <gundalow> ^ From ken
17:11:24 <privateip> that will be a 2.6 project
17:11:31 <privateip> parts are going into 2.5 though
17:11:45 <privateip> the underpinnings is getting the module_utils work done
17:12:17 <privateip> but to answer the question, yes the goal is to support gather_facts: yes for network modules
17:16:31 <gundalow> itdependsnetwork: jmcgill298 Nor sure if you are still around, though hopefully the above answers your questions
17:21:43 <gundalow> Ok, will end now
17:21:48 <gundalow> Thanks everybody
17:21:51 <gundalow> #endmeeting