19:04:06 <fungi> #startmeeting infra
19:04:07 <openstack> Meeting started Tue Jan 17 19:04:06 2017 UTC and is due to finish in 60 minutes.  The chair is fungi. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:04:08 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
19:04:10 <openstack> The meeting name has been set to 'infra'
19:04:14 <fungi> #link https://wiki.openstack.org/wiki/Meetings/InfraTeamMeeting#Agenda_for_next_meeting
19:04:18 <fungi> #topic Announcements
19:04:23 <fungi> i don't have any for this week
19:04:25 <fungi> as always, feel free to hit me up with announcements you want included in future meetings
19:04:30 <fungi> #topic Actions from last meeting
19:04:37 <fungi> #link http://eavesdrop.openstack.org/meetings/infra/2017/infra.2017-01-10-19.04.html
19:04:40 <fungi> oof, we have a bunch. some have my name next to them and i know i didn't get them done, so i'll just readd them now
19:04:48 <fungi> #action fungi Obtain docs.openstack.org X.509 certificate.
19:04:50 <fungi> #action fungi Obtain developer.openstack.org X.509 certificate.
19:04:52 <fungi> #action fungi announce the infra ptg pike etherpad to relevant mailing lists.
19:04:59 <fungi> the others have a better change of being done maybe?
19:05:04 <fungi> jeblair mark http://specs.openstack.org/openstack-infra/infra-specs/specs/nodepool-zookeeper-workers.html implemented
19:05:20 <jeblair> #action jeblair mark http://specs.openstack.org/openstack-infra/infra-specs/specs/nodepool-zookeeper-workers.html implemented
19:05:24 <jeblair> :(
19:05:46 <fungi> yeah, i only just looked myself
19:05:57 <fungi> pabelanger Switch DNS for docs.openstack.org from CloudSites to files01.openstack.org.
19:05:59 <fungi> this i know got done
19:06:17 <fungi> thanks pabelanger for making the switch!
19:06:26 <jeblair> that's the trick... 'pabelanger' has to come after "#action' for this to work
19:07:08 <fungi> some additional holes in documentation got filled which the community at large spotted after the cut-over, but last word seemed to be that everything was in ship shape now
19:07:27 <fungi> AJaeger follow up to http://lists.openstack.org/pipermail/openstack-infra/2016-August/004690.html letting them know we're ready for site deletion.
19:07:32 <fungi> that too
19:07:38 <fungi> #link http://lists.openstack.org/pipermail/openstack-infra/2017-January/005023.html
19:07:55 <fungi> i guess we haven't completely told them they can delete it yet
19:08:08 <fungi> we still need to stop publishing to it, but we can discuss that later in the meeting
19:08:09 <jeblair> yeah, i think the actual 'go' has not yet happened
19:08:32 <AJaeger> jeblair: correct, not happened - just the heads-up
19:08:52 <fungi> okay, that concludes the previous action items portion of our show
19:09:24 <fungi> #topic Specs approval: PROPOSED Ethercalc (clarkb)
19:09:37 <fungi> #link https://review.openstack.org/420136 Ethercalc specs proposal
19:09:58 <clarkb> ohai, this is pretty straightforward. Basically PTG organizers (ttx) have requested that we run an ethercalc to do virtual unconference post it note scheduling
19:10:00 <fungi> it's gotten a few reviews, looks straightforward, and we're on a tight timeline to hopefully have it working in time for the ptg
19:10:24 <clarkb> its really similar to etherpad with the biggest difference being redis is required (no mysql option)
19:10:39 <fungi> also i doubt i'm the only one who's been after an excuse to have an ethercalc.openstack.org. it's a nifty service
19:11:08 <fungi> i think jeblair was the one who first brought it to my attention, like a year ago maybe?
19:11:11 <jeblair> yeah, i enjoyed using the project-hosted ethercalc, but it deletes things after not-too-long, so it will be nice to have our own.
19:11:42 <fungi> anyway, seems like a worthwhile experiment
19:12:38 <fungi> anyone object to going ahead with infra council roll call on it over the next couple days?
19:12:42 <clarkb> if this gets the go ahead I will push up the chagne to add the puppet-ethercalc repo and then go from there
19:13:10 <fungi> #info Infra Council voting is open for the "Ethercalc" spec until 19:00 UTC on Thursday, January 19.
19:13:11 <jeblair> sounds good to me, and i think the abbreviated timeline is reasonable considering the similarity to an existing service.
19:13:29 <fungi> does appear we can reuse a lot of our prior art
19:14:18 <fungi> okay, priority efforts...
19:14:27 <fungi> #topic Nodepool: Use Zookeeper for Workers (jeblair)
19:14:47 <fungi> i guess we can cover "nodepool v3" here
19:15:06 <jeblair> mordred sent an email about a branch of nodepool for the v3 shim
19:15:08 <fungi> or is this more about the Zuul v3 spec?
19:15:15 <fungi> ahh, that yes
19:15:40 <fungi> #undo
19:15:41 <openstack> Removing item from minutes: #topic Nodepool: Use Zookeeper for Workers (jeblair)
19:15:44 <fungi> #topic Priority Efforts: Zuul v3 (jeblair)
19:16:21 <jeblair> my read is that while most of us would prefer the situation be different, no one strongly objects to the 'branch nodepool to create a temporary shim program' approach
19:16:26 <fungi> #link http://lists.openstack.org/pipermail/openstack-infra/2017-January/005048.html "Nodepool v3 shim for talking to v2 zuul"
19:16:27 * mordred waves
19:16:47 <jeblair> and i do think it will be the easiest way to get where we're going (and an appropriate amount of effort for a temporary component)
19:17:15 <jeblair> does that sound right?  should we go ahead and proceed with that?
19:17:22 <fungi> probably far less effort than the temporary bits of zuul v2.5
19:17:41 <fungi> the plan seemed sound to me
19:17:44 <jeblair> fungi: yep
19:17:56 <mordred> if there are no objections, I'll make the branch for it today
19:18:05 <clarkb> ya I didn't have objections once I groked the plan
19:18:10 <mordred> \o/
19:18:21 <clarkb> maybe I should've been more explicit in my email
19:18:40 <fungi> #agreed The Nodepool v3 shim plan seems sound.
19:18:46 <clarkb> basically "these two options seem viable but haven't spent enough time to say one way or another which would be better so I defer to you"
19:19:12 <fungi> #action mordred make the temporary branch for the Nodepool v3 shim.
19:20:27 <fungi> SpamapS: if you're around, did jeblair's reply address your concerns about that plan?
19:21:39 * mordred is about to send a quick response to that thread with slightly more info
19:21:57 <fungi> so be it
19:22:00 <fungi> thanks mordred!
19:22:05 <fungi> and jeblair!
19:22:13 <fungi> #topic Priority Efforts: Docs Publishing via AFS (AJaeger)
19:22:49 <AJaeger> docs.o.o works fine, thanks everybody.
19:22:55 <AJaeger> We can now stop publishing to it https://review.openstack.org/#/c/420135/ and then declare the spec implemented https://review.openstack.org/#/c/420138/
19:22:56 <mordred> \o/
19:23:00 <fungi> #link http://lists.openstack.org/pipermail/openstack-docs/2017-January/009479.html docs.o.o looks great - doesn't it?
19:23:32 <fungi> and i guess also follow up with the cloudsites admins to let them know they're free to delete
19:23:33 * AJaeger saw two or three bugreports against outdated pages and we missed initially some infra content - all fixed or commented on
19:23:57 <AJaeger> Yes, will do that once 420135 is merged and we do not publish there anymore
19:23:58 <mordred> I love that this works:  ls /afs/openstack.org/docs
19:24:19 <jeblair> mordred: ++
19:24:29 <fungi> or lynx /afs/openstack.org/docs/index.html ;)
19:24:35 <AJaeger> ;)
19:24:48 <jeblair> i think that could be a really good debugging tool for the docs team
19:24:57 <AJaeger> So, nothing more to add from my side - just waiting for those to merge...
19:25:06 <AJaeger> grepping over the tree ;)
19:25:07 <jeblair> i don't know how many folks have installed afs clients, but we do have some install instructions
19:25:18 <clarkb> I just did zypper search afs and found no afs
19:25:18 <SpamapS> fungi: Yes I think so
19:25:22 <fungi> i approved 420135 just now
19:25:29 <fungi> thanks SpamapS!
19:25:41 <jeblair> if folks wanted to add instructions for other os's, that'd be fine i think
19:26:09 <jeblair> #link afs client install http://docs.openstack.org/infra/system-config/afs.html#client-configuration
19:26:16 <AJaeger> clarkb: yeah ;( There're some packages in the build service but haven't tested those
19:26:39 <mordred> clarkb, AJaeger: there is an old thread here: https://forums.opensuse.org/showthread.php/442068-how-to-install-openafs-on-11-3 that does not shed much light
19:26:46 <SpamapS> fungi: np. I also realized just how much I've avoided learning about nodepool by reading jeblair's reply. :)
19:27:08 <fungi> hah
19:27:20 <jeblair> SpamapS: as you can see, we have good reasons for changing all the things we are in v3 ;)
19:27:37 <fungi> nodepool as a passive gearman sniffer
19:27:43 <SpamapS> (BTW, as the current steward of gearmand, that use of the admin protocol horrifies me. ;) Anyway, carry on.
19:27:52 <mordred> SpamapS: it horrifies us too
19:28:10 <SpamapS> For some reason I thought it was just a small part of it. Glad we're on the same page now. :)
19:28:11 <jeblair> yup
19:28:30 <SpamapS> Now.. how do I get you to also decide to use rustygear..... ;-)
19:29:23 <fungi> AJaeger: okay, so once 420135 merges (~3 minutes more, zuul willing) i'll approve the change to mark the spec implemented and you or i can follow up one more time to the cloudsites admins thread
19:29:44 <AJaeger> fungi: I'll followup privately and CC you - ok?
19:30:03 <fungi> AJaeger: that's fine, though there's also that thread on the infra ml we could use
19:30:12 <fungi> not sure what does or doesn't need to be private
19:30:20 <AJaeger> let's do it public...
19:30:34 <fungi> it's my default preference ;)
19:31:13 <fungi> thanks everyone for working on the afs docs implementation! it's one of the awesomest things we have
19:31:22 <jeblair> also, like our oldest spec?
19:31:39 <fungi> quite likely, especially if you count back to the prior swift-based plan
19:31:46 <fungi> a milestone!
19:31:59 <jeblair> it starts out with a link to "Juno summit session"
19:32:07 * fungi shudders
19:32:09 <mordred> I kindof want t-shirts that say "OpenStack Infra: Horrifying the world by solving problems using proven old technology instead of chasying shiny"
19:32:31 <fungi> what's old is new again
19:33:45 <fungi> #topic Pike Cycle signing key ready for attestation (fungi)
19:33:51 <fungi> #link https://sks-keyservers.net/pks/lookup?op=vindex&search=0xc96bfb160752606daa0de2fa05eb5792c876df9a&fingerprint=on Pike Cycle signing key
19:33:55 <fungi> #link http://docs.openstack.org/infra/system-config/signing.html#attestation attestation process
19:33:59 <mordred> fungi: I feel like I just did this
19:34:02 <fungi> i went ahead and did this early
19:34:11 <fungi> especially since we have a shorter cycle
19:34:30 <fungi> the idea is to normally give everyone plenty of time to do this rather than scrambling after the release
19:35:20 <fungi> i also talked to dhellmann and got a reminder for it on the release process/timeline to coincide with when the release managers propose the custom gerrit acls for the upcoming stable branch
19:36:40 <fungi> one tricky bit is that this is happening before there is actually a release date set for pike. as such i set the key expiration for 12 months just to make sure it's viable past whatever release date is chosen (we'll likely stop using it at least a couple months before it expires, but doesn't hurt for it to be a little longer)
19:37:23 <clarkb> fungi: so we don't need tkeys to be valid during the entire stable support period?
19:37:35 <mordred> fungi: done. thank you for documenting an easy to follow process
19:37:52 <fungi> clarkb: sort of. the key remains "valid" just no longer in use (and so expired)
19:38:10 <fungi> expiration can always be extended if deemed necessary
19:38:14 <clarkb> ok
19:38:18 <fungi> revocation, on the other hand, is forever ;)
19:38:51 <fungi> we would only revoke keys if we thought they were compromised (as someone could reuse them for nefarious purposes in that case)
19:39:30 <fungi> but an expired key that was not expired at the time it signed something doesn't make whatever it signed untrusted just because you're in the future now
19:40:43 <fungi> we're more relying on the creation/expiration dates to provide a rough window for when the keys were used to create signatures (and then we also provide much more precise date ranges at https://releases.openstack.org/#cryptographic-signatures for those who need them)
19:41:33 <fungi> anybody have any questions about this before we move on to open discussion?
19:43:58 <fungi> #topic Open discussion
19:44:52 <fungi> AJaeger: i've approved 420138 now and will take the Docs Publishing via AFS topic out of priority efforts in our agenda after the meeting. thanks!
19:45:26 <AJaeger> thanks, fungi!
19:45:49 <AJaeger> fungi, mail sent and tony answered already...
19:45:50 <fungi> looks like i've also forgotten to remove "Common OpenStack CI Solution" from the agenda after it was implemented so will clean that up as well
19:46:25 <fungi> approving his held post through moderation now
19:47:59 <fungi> #link http://lists.openstack.org/pipermail/openstack-infra/2017-January/005073.html docs.openstack.com excessive INODE consumption
19:48:12 <fungi> awesome wrap-up!
19:49:38 <fungi> since nobody seems to have anything for open discussion, i grant you all 10 minutes of recess!
19:49:44 <fungi> thanks everyone
19:49:56 <fungi> #endmeeting