19:59:58 <nitzmahone> #startmeeting Ansible Windows Working Group
19:59:58 <zodbot> Meeting started Tue Apr 14 19:59:58 2020 UTC.
19:59:58 <zodbot> This meeting is logged and archived in a public location.
19:59:58 <zodbot> The chair is nitzmahone. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:59:58 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
19:59:58 <zodbot> The meeting name has been set to 'ansible_windows_working_group'
20:00:10 <nitzmahone> Bah, 2s early
20:01:20 <jborean93> yo
20:01:20 <nitzmahone> Nothing new on the agenda, so
20:01:29 <nitzmahone> #topic open floor
20:01:39 <nitzmahone> #chair jborean93
20:01:39 <zodbot> Current chairs: jborean93 nitzmahone
20:02:14 <briantist> 👋hi everyone, working on another new module for configuring powershell remoting session configurations, should have a PR up in the next week or so
20:02:14 <jborean93> I think the docs team have decided on a changelog process for collections. Just waiting on the final decision and details to come out before migrating the existing 2.10 changelog fragments to each collection
20:02:39 <nitzmahone> briantist: noice, looking forward to seeing that
20:03:05 <jborean93> interested to see how you plan on bypassing the WinRM service restart requirement there
20:03:10 <briantist> the action plugin that lets it avoid being killed by the winrm restart is nifty if I do say so myself
20:03:16 <jborean93> ah
20:03:18 <jborean93> cool
20:06:19 <nitzmahone> Welp, if nothing pressing, we'll close up in 2m.
20:06:24 <briantist> jborean93: I ran into another instance of an async task having a weird failure related to registry unloading, put the details in that one ticket but not sure if you saw it
20:06:34 <briantist> (^ that's not pressing, doesn't need to hold the meeting)
20:06:46 <jborean93> yea I did see that, haven't had time to look into it yet sorry
20:06:55 <briantist> no worries
20:07:26 * nitzmahone still wants to get a 100% reliable repro on that one
20:07:56 <jborean93> it being related to roaming profiles or something would make sense. It was similar to what I saw when testing against Azure hosts in CI where it would sometimes fail until I logged on interactively to find out more what happened
20:08:19 <nitzmahone> If we can get something really easy, can probably engage MS to hopefully suggest a workaround
20:08:25 <briantist> I can say it's not related to roaming profiles in my case as we don't use them
20:09:16 <nitzmahone> It seems to happen more on cloud hosts, which makes me suspect it's a race that's exacerbated on slow machines
20:09:38 <jborean93> especially with TiWorker thrashing those machines on first startup
20:09:43 <nitzmahone> Yep
20:09:49 <nitzmahone> (grr)
20:10:27 <briantist> some of my previous repros were on long lived machines. Recent one is recent startup, but it only happens with 1 of the 2 users tried
20:10:38 <briantist> and that same user, doesn't repro on other  hosts
20:10:41 <briantist> it reallyis baffling
20:10:55 <nitzmahone> I'm still betting there's some internal/pseudo-documented API we can use to add another ref to the profile handle
20:13:52 <nitzmahone> jborean93: can't remember, did you try this in your messings around on that? https://support.microsoft.com/en-us/help/2287297/a-com-application-may-stop-working-in-windows-when-a-user-logs-off
20:14:28 <nitzmahone> (the policy change to stop forced profile unloading)
20:14:47 <jborean93> no I don't think I did, it was quite some time ago
20:15:23 <briantist> did we ever look into using powershell scheduled jobs as a mechanism for async? I remember seeing some code maybe in the windows update module that was using them, but wondering if it might make for a good general async solution
20:15:30 <jborean93> Weirdly I also remember a weird issue were using become on 2008 and 08 R2 would have the user's registry hive never unload which seems to be the opposite of the issue with async
20:15:59 <nitzmahone> briantist: you might want to try that if you've got some places it happens pretty regularly
20:16:00 <jborean93> nitzmahone wrote the original async using CreateProcess and the job breakaway flag
20:16:19 <jborean93> I wrote the current WMI based one due to some connections not allowing the process to breakaway from the job
20:16:54 <jborean93> plus become + async never worked until the WMI stuff on 2008 and 08 R2
20:17:12 <nitzmahone> I suspect this problem might also be lessened if/when we do persistent interpreter and/or connection
20:17:59 <briantist> nitzmahone did you mean try the policy change? my latest repro was last week, I can see if it's still happening and see if the policy changes that
20:18:47 <nitzmahone> Anyway, wrapping up for today- till next week!
20:18:52 <nitzmahone> #endmeeting