16:03:13 #startmeeting Ansible VMware Working Group Meeting 16:03:13 Meeting started Mon Oct 15 16:03:13 2018 UTC. 16:03:13 This meeting is logged and archived in a public location. 16:03:13 The chair is akasurde. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:03:13 Useful Commands: #action #agreed #halp #info #idea #link #topic. 16:03:13 The meeting name has been set to 'ansible_vmware_working_group_meeting' 16:03:18 #chair ckotte 16:03:18 Current chairs: akasurde ckotte 16:03:20 ckotte, hi 16:03:37 hi, do we have audio for the meeting? 16:04:27 hi 16:04:31 hi 16:04:56 #chair dericcrago 16:04:56 Current chairs: akasurde ckotte dericcrago 16:05:00 #chair moshloop_ 16:05:00 Current chairs: akasurde ckotte dericcrago moshloop_ 16:05:06 hi everyone 16:05:22 ckotte, nope we don't have audio video for this meeting :( 16:06:36 I could host a WebEx next time 16:07:06 or we could just use Skype, etc... 16:07:49 Not sure if that is necessary 16:08:03 ckotte, btw thanks for the new PRs and reviews 16:08:34 I am testing and reviewing stuff so it may take a little more time 16:08:35 Isn't it easier to discuss thing? 16:08:58 I will update several other modules as well 16:09:21 Are their any plans to get an integration testing environment up running vcenter ? 16:09:25 need to focus on the host, switch, and cluster modules first 16:09:42 moshloop_, I am working on the same 16:09:58 once that is stable and robust, will make in public 16:10:21 ckotte, I would recommend to focus on vmware_guest 16:10:52 As per my knowledge, community people uses vmware_guest more than any other modules and we have lot of things pending on that module 16:10:55 @aksurde I need to implement host automation first 16:11:09 should we just submit integration tests marked as unsupported than ? What connection vars do we use? 16:11:22 I know. there are a lot of issues about vmware_guest, but I don't have time for this now 16:11:35 #info http://ansibullbot.eng.ansible.com/ansibot/metadata/byfile_sorted.html 16:11:59 ckotte, If you see vmware_guest is at top number 1 for weeks now 16:12:20 moshloop_, unsupported depends upon vcsim only 16:12:45 going forward once real vCenter is in place then will have conditional to handle all testcase 16:12:53 *testcase 16:12:57 *testcases 16:13:18 Just trying to understand how I can help your testing burden 16:13:53 i.e. Commit an integration test, that you then manually run, until vcenter testing is part of the normal testng cucle 16:14:02 aksurde, it's on top yes, but I need to implement infrastructure automation for my customer first 16:14:15 moshloop_, Wow, thanks for asking. We need real hardware to test all testcases. I have tested most of them but I would love if someone takes another stab at this 16:14:34 how can you "focus" someone. Didn't use IRC since early 2000... 16:15:24 we can start one topic and discuss that first and once that is finish we move on to next one 16:15:35 #topic prioritise vmware_guest 16:15:39 like this 16:15:43 like this "ckotte, If you see vmware_guest is at top number 1 for weeks now" 16:16:18 let us discuss on prioritise vmware_guest issues and PRs 16:17:07 no. I mean. If I want to send a message and "highlight" your username 16:17:51 What is your IRC client ? 16:17:59 LimeChat 16:18:04 just use it the first time 16:18:06 depends on your client ckotte, most of them have autocomplete / highlighting of usernames 16:18:13 sorry for the noob questions :) 16:18:23 ok 16:18:34 if it is xchat then, just type my name and hit tab and your reply 16:18:39 I think there's several people using https://www.irccloud.com/ 16:18:45 limechat follows same 16:18:49 #chair dericcrago 16:18:49 Current chairs: akasurde ckotte dericcrago moshloop_ 16:18:55 dericcrago, welcome 16:19:38 is it possible to use IRC with http/https? IRC is blocked in the company network 16:19:57 ckotte, yes 16:20:00 that's https://www.irccloud.com/ ;) 16:20:14 https://webchat.freenode.net/ 16:20:28 ok. will try next time. thanks 16:22:54 So, whats the plan regarding the vmware_guest modules? 16:23:40 Right now there is no plan as such but I would love to see fix existing PRs and issues 16:23:56 1. Disk management 16:23:58 o/, sorry i'm late 16:24:03 2. Datastore management 16:24:06 #chair pdellaert 16:24:06 Current chairs: akasurde ckotte dericcrago moshloop_ pdellaert 16:24:33 Some PRs are open for a long time 16:24:40 pdellaert, hi, we are discussing about prioriting vmware_guest issues and PRs 16:24:42 e.g. vmware_vswitch 16:24:48 I already have new improvements 16:24:54 ok, i'll catch up 16:25:07 yes, old PRs are waiting on reviews and testing 16:25:57 should we start with the oldest PRs and issues or doesn't it matter? 16:26:04 pdellaert, tcernomax and dericcrago has done a great job of review PRs 16:26:44 ckotte, it does not matter, depends on severity and number of people hit 16:26:50 with the bug 16:27:46 this is the one thing we kind of miss in github issues: priorities, or are there labels that can highlight priorities? 16:29:40 yes there are labels, but those are tightly attached to ansible release cycle 16:29:46 like P1 and P2 etc. 16:29:53 "depends on severity and number of people hit": where do you see this information? 16:30:32 If I add P1 then it block Ansible release something like that 16:30:40 P1 is highest priority 16:30:59 we can use Projects - https://github.com/ansible/ansible/projects 16:31:30 more specific like https://github.com/ansible/ansible/projects/20 16:31:42 Network automation team using it for their tracking 16:31:51 akasurde: i like that idea! 16:32:26 I like it too 16:32:48 there is a drawback to GitHub Project though - only People under Ansible can move the tickets 16:34:26 pdellaert, I can create columns with priority and start working on the classification on issues 16:38:45 Are we agree on this ? 16:41:27 You can create a project, and then we can see if it is useful as intended? 16:43:58 Yes sure 16:44:27 #action Akasurde To create vmware related project in GitHub project and decide priorities of issues and PRs 16:48:22 coming back to the vcenter integration testing 16:48:30 akasurde++ 16:49:08 Does RedHat have a vcenter cluster / hardware available or is this something we need to get VMWare or another company to sponsor ? 16:49:35 #topic vcenter integration testing 16:49:47 moshloop_: the problem with that is that we need a full vCenter with multiple DC's, clusters, hosts, datastores, ... etc. Which is expensive, and which is unlikely to be available, additionally, testing all kind of setup combinations is going to be really hard 16:50:29 a vCenter standard license doesn't cost that much ;) 16:50:33 moshloop_, RedHat has limited hardware and no licenses so I rely on Eval copy 16:50:49 We are open to hardware sponsers 16:51:05 I doubt VMware will be eager to sponsor/provide such a setup entirely, they are supporting Ansible a bit, but not significantly enough to sponsor hw and sw 16:51:07 We just need test status 16:51:10 at one point in time, there were some people from VMware that joined this meeting, I'm not sure what happened with that 16:51:47 did you ask VMware? I guess they want to push vRealize Automation and Orchestrator or PowerCLI instead of Ansible 16:51:52 ckotte, as a community project we can not afford it 16:51:58 We are a decent enough size Redhat and VMWare customer, so I will try push on that sude 16:52:37 VRA / VRO are integrating into ansible these days - so I don't see why they wouldn't want to support it 16:52:43 ckotte: a standard vcenter licens is not that expensive, but eventually you'd need about 9 ESXi licenses as well, if you want to test dvSwitches, you also end up with Enterprise Plus licenses :) 16:52:43 maybe more people will use Ansible and later buy support or use Ansible Tower if the vmware modules are better 16:52:44 dericcrago, I met bunch of VMware folks in both VMworld and AnsibleFest, they are planning to support HW 16:52:47 let us see 16:53:08 ah, that's good news :) 16:53:21 pdellaert, agree 16:53:37 here at my customer, the TAM wants to push vRA/vRO and nowadays also Lifecycle Manager.... 16:53:43 ckotte, Thats the target 16:54:24 Unfortunetly, lots of (new) VMware products are buggy. With Ansible, I can fix it by myself. With VMware, the ticket disappears in the "void" 16:55:03 I think we just need to apply enough pressure as customers to the bean counters at vmware 16:56:21 moshloop_, Do you have infrastructure which can be utilize to run existing and new testcase then we are happy to talk 16:56:39 aksurde What do you mean with "they are planning to support HW". HW = hardware?? 16:57:41 We do have some capacity now - let me see what I can do 16:58:06 What type of memory allocation do you think we will need ? 16:58:42 I have some capacity too for integration tests, but we cannot connect from outside.. 16:59:32 We have 30++ HP G8 blades doing nothing 16:59:47 in one site... plus more in other sites 17:00:31 for anyone eager to test something, I just submitted a PR for a couple of vmware_tools connection plugins 17:00:36 In theory we don't need inbound ? Only outbound to poll and comment on the PR's ? 17:00:37 use at your own risk ;) 17:01:02 https://github.com/ansible/ansible/pull/47072 17:01:24 ckotte, HW == hardware 17:01:34 moshloop_, that would be great 17:01:50 I have enough free hardware 17:02:04 I wanted to implement integration testing for my stuff anyway 17:02:24 moshloop_, ckotte we just need info if tests passed or not 17:02:35 I also planned to spin up ESXi and vCenter servers after every commit in my Ansible repo 17:02:48 and not breaking anyhting which is available already there 17:02:54 dericcrago, bravo 17:02:56 haven't have time yet to implement CD/CI 17:03:32 ckotte, gitlab CI is nice to start with - easy and very intuitive 17:03:51 aksurde sure, but time is the problem 17:04:32 ckotte, ok 17:05:36 So we will try and setup some infra, publish the test results somewhere and just manually comment on the PR's for now ? 17:05:54 If there is HW available, internal or external, with some CD, you could easily track the existing ansible repo, reformat the tests a little bit in the repo, and execute them on real HW instead of against vcsim. But there will be some significant work to be done in automating all of that ... All for it, but let's be realistic that accept for akasurde, we are all vounteers with a full time day job 17:06:00 (well, some of us more ... 17:06:02 ... than full time, i think) 17:06:28 moshloop_: that would be good, indeed. But it is also important to document how the infra is set up and how the tests get executed :) 17:07:05 What is the mechanism that is used for AWS to inject credentials ? 17:07:18 s/accept/except (previous statement) 17:07:20 I am not sure about AWS 17:09:32 https://github.com/lamw/vghetto-vsphere-automated-lab-deployment Seems like a good start ? 17:09:39 but ask 17:11:23 i have to run, sorry. akasurde if you have a project setup, it would be great, i can/am willing to help out in the project mamagement if you need/want some help. (First suggestions for lists/collumns in the project tracker: Have a list of open issues (Todo), a list of high priority issues and a list of open PRs to be reviewed (in review). Other collumns can be those to indicate which need 17:11:29 backporting to different releases ... 17:11:32 ... and maybe a 'working on' list so people now somebody has commited on working on those) 17:12:38 backporting to older Ansible releases? Is that really necessary? there's so much todo... 17:12:56 generally backporting is on demand 17:13:03 people ask for backprot 17:15:13 because they can't use the devel version on their servers? 17:15:51 yes, another reason being people use tower 17:15:51 I use the CentOS Ansible version and link the new modules to my repo in the library folder 17:16:21 ckotte: a lot of professional deployments really do not allow you to run any kind of dev version 17:16:29 ok. want to try the opensource version of tower soon. not sure if I need to rewrite my Ansible playbooks etc. pp.... 17:16:47 i know places where they are running 2.3 because they do not want to go to a newer release 17:17:13 with the current state of the modules... sorry to say that.. I don't think you can do a professional deployment 17:17:23 the mentality of 'if it is not broken, do not try to fix it' is very much engraved in many stable professional deployments 17:17:35 you need to wait for Ansible 2.8 with lots of fixes. maybe then it's enterprise ready 17:17:47 ckotte: i know places where they use them, very happily 17:17:55 ok 17:18:12 I found lots of issues when I started to replace existing PowerCLI automation 17:18:15 it fits their use cases, they just don't want to jump through every release, because they would need to change their playbooks with every release :) 17:18:52 ok. lets see where we need backporting 17:20:43 ckotte, you can file an issue if you find it because it will create public record and people might want to work on it 17:21:30 aksurde Thats why I fix all modules I (want) to use 17:21:35 cannot wait for someone else 17:21:52 (want to) 17:22:05 ckotte, sure you are always welcome to fix them 17:22:15 thats how community grows 17:22:44 people help each other to solve their common problem and share their knowledge with each other 17:23:03 Speaking of fixes - I have a couple PR's that are still waiting to be merged 17:23:33 44885 and 44832 17:24:59 We are little over meeting time, but we can still discuss after this meeting 17:25:10 moshloop_, let me check 17:28:15 #endmeeting