16:01:01 <akasurde> #startmeeting Ansible VMware Working Group Meeting
16:01:01 <zodbot> Meeting started Mon Oct  9 16:01:01 2017 UTC.  The chair is akasurde. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:01:01 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:01:01 <zodbot> The meeting name has been set to 'ansible_vmware_working_group_meeting'
16:02:39 <akasurde> jtanner, dag are you around ?
16:04:42 <garbled> Hi, new to this meeting, just wanted to introduce myself.  Tim, fiddling with vmware stuff and adding a mess of PR's. :)
16:07:43 <akasurde> #chair garbled
16:07:43 <zodbot> Current chairs: akasurde garbled
16:07:52 <akasurde> garbled, hi
16:08:03 <akasurde> welcome to meeting and VMware working group
16:08:22 <akasurde> hope to have great collaboration with you.
16:08:44 <garbled> Thanks, I hope so too.  Been doing alot of heavy automation with it lately
16:08:46 <akasurde> We generally discuss PRs and blocker here. Also, roadmap of VMware working group
16:09:42 <akasurde> garbled, Our current blocker is actual VMware hardware, on which we can test stuff
16:10:07 <garbled> So most of my code is written on actual hardware, and then I go back and test against your vcsim infra
16:10:38 <garbled> but I can't test every odd combination, because it's a real system, and I can't arbitrarily add stuff to it
16:11:32 <akasurde> yes I know
16:11:44 <akasurde> but atleast you have something real :)
16:12:10 <garbled> true.  and alot of my recent pr's are because frankly, alot of the needed features for doing full build automation were just missing
16:12:33 <garbled> I'm especially interested in the PR that adds cdrom to vmware_guest
16:12:50 <akasurde> you can explore more things in current modules and expand your automation on top of that
16:13:22 <akasurde> yes, we were going to discuss CD-ROM pr but no one is there
16:13:28 <garbled> ha, yeah
16:14:06 <akasurde> Let us, I will push of merge of that PR
16:14:12 <akasurde> for*
16:16:18 <garbled> a small nit, I've noticed there is no way to move a vm to a different folder
16:17:18 <akasurde> you mean in vmware_guest ?
16:17:27 <garbled> I'm not sure if vmware_guest was supposed to be able to do that, but it doesn't actually work
16:19:24 <akasurde> we can include a module for it then, if it is that pressing issue
16:20:10 <garbled> originally I thought vcenter_folder would do it, but that seems to just add/delete folders.  not sure if it belongs in there, because that's vcenter only?
16:20:28 <garbled> https://github.com/ansible/ansible/pull/27005
16:25:19 <garbled> The other problem, I wasn't able to work around easily, was datastore selection
16:26:40 <akasurde> Ok
16:26:42 <garbled> My solution was writing a filter_plugin, that takes minfree, spacerequest, and a name regex, and feeding it the output of datastore_facts
16:27:03 <garbled> essentially, if you ever fill a datastore, it's the end of the planet
16:27:18 <akasurde> Sounds good idea
16:27:24 <garbled> so it's vital that before I even try, I know I have not just enough space, but extra space
16:28:34 <jtanner> sorry, i've been attending to my dad
16:28:37 <garbled> I could submit this filter, but I wasn't we wanted vmware specific filters in the general filter space?
16:30:34 <akasurde> #chair jtanner
16:30:34 <zodbot> Current chairs: akasurde garbled jtanner
16:31:53 <jtanner> garbled: we don't really have anything that approaches the lifecycle of VMs other than to created/delete/poweron+off
16:31:54 <akasurde> I am sure about filter_plugin in general space
16:33:04 <garbled> jtanner: that's mostly the stuff i'm writing now, more general lifecycle management tools
16:34:01 <jtanner> i like to pretend people are using vmware as pets instead of cattle
16:34:12 <garbled> ha
16:34:13 <jtanner> or vice versa
16:35:19 <garbled> so, If it's acceptable, I'd like to discuss my proposal in https://github.com/ansible/ansible/pull/31257 regarding resource pool selection
16:36:35 <jtanner> makes sense to me
16:36:43 <jtanner> i naively thought we were already doing that
16:37:05 <jtanner> will need tests for sure
16:37:08 <garbled> what we currently do boggles me entirely.  :)  I had to hack the module to build vm's in my environment at all
16:37:22 <jtanner> yep, we're on the 80% rule
16:37:44 <akasurde> yes
16:38:08 <akasurde> that is because it is not tested on actually scenarios
16:38:38 <garbled> OK, so I'll test my code out against my env, I've already got tests for it in vcsim, and will update my PR
16:38:54 <garbled> it should fix all three pr's on that in one go
16:38:55 <akasurde> thats nice
16:39:42 <garbled> ok, thanks
16:40:18 <akasurde> anything we would like discuss ?
16:41:34 <garbled> Do any of my other PR's need anything from your view?  (assuming you don't have anything else to discuss)
16:42:24 <garbled> I know the vsphere copy/fetch ones are waiting on dag to discuss
16:42:45 <jtanner> ah, dougm has some thoughts on https://github.com/vmware/govmomi/issues/863#issuecomment-335212287
16:44:37 * akasurde reading
16:45:00 <akasurde> jtanner, bingo
16:47:42 <jtanner> yeah, anfernee's claim seemed strange to me
16:48:37 <garbled> vcsim does some odd things with certain data that real hw does not do as far as I can tell
16:49:34 <garbled> for example, if you keep doing parent.parent in a loop on real hw, eventually you get null, but on vcsim sometimes (not all the time) you get invalidaccessor
16:50:12 <akasurde> garbled, I seen that only in case of python3
16:50:37 <garbled> I got it in 2.7 on my vmware facts code I've been working on
16:50:47 <jtanner> make sure you file bugs on vcsim for those types of differences
16:50:49 <garbled> trying to climb the folder tree to build a foldr pathname
16:51:04 <garbled> yeah.. I need to see if I can build a testcase
16:51:50 <jtanner> could probably reverse vmware_walk.py and have it go up the tree instead of down
16:52:01 <jtanner> parents instead of children
16:52:13 <akasurde> yup
16:52:19 <garbled> ahh.. ok
16:52:21 <jtanner> standalone script might be easier/smaller though
16:55:39 <akasurde> jtanner, I am thinking why we can tell people to chain vmware_guest_find and vmware_guest ?
16:56:01 <akasurde> vmware_guest_find will give folder and that can be used in vmware_guest
16:57:10 <jtanner> that was one of the ideas ...
16:57:18 <garbled> if you have time, I'd like to see 30815, 31137 and 31441 merged, or if you have revision suggestions, more than open to them.
16:57:24 <jtanner> primary idea was vmware_guest_find can be accurate AND slow
16:57:41 <jtanner> vmware_guest will continuously have debate about speed
16:57:48 <garbled> so slow.   I gave up running it on my env.
16:57:55 <garbled> (find)
16:59:24 <garbled> I don't find vmware_guest to be slow though
17:04:44 <akasurde> I will endmeeting as we are out of time
17:04:47 <akasurde> #endmeeting