20:00:18 <johnsom> #startmeeting Octavia
20:00:20 <openstack> Meeting started Wed May 16 20:00:18 2018 UTC and is due to finish in 60 minutes.  The chair is johnsom. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:00:21 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
20:00:23 <openstack> The meeting name has been set to 'octavia'
20:00:26 <johnsom> Hi folks!
20:00:31 <nmagnezi> o/
20:00:35 <cgoncalves> hi
20:00:51 <AlexStaf> o/
20:00:53 <xgerman_> o/
20:01:01 <johnsom> #topic Announcements
20:01:31 <johnsom> Just a heads up, I am going to cancel next weeks meeting as many of us will be at the OpenStack summit in Vancouver.
20:01:47 <rm_mobile> o/
20:02:17 <johnsom> BTW, who all is going to the summit?  I know that rm_work/mobile, xgerman, and I will be there. Anyone else?
20:02:47 <nmagnezi> not me, sorry
20:02:58 * johnsom looks sad
20:03:26 <johnsom> Ok, well, there will at least be three of us representing Octavia.
20:03:39 <cgoncalves> I'm not
20:03:40 <AlexStaf> nope
20:03:44 <nmagnezi> how about you dayou?
20:03:44 <johnsom> I have two sessions, a project update and onboarding.
20:04:12 <nmagnezi> johnsom, nice! I'll watch those :)
20:04:34 <johnsom> Any other announcements I missed?
20:05:06 <johnsom> #topic Brief progress reports / bugs needing review
20:05:39 <johnsom> I have been focused on getting the provider driver code done.  Everything is now posted for review except the driver library for the call backs.
20:05:54 <johnsom> I have a few documentation things to update as well.
20:06:05 <johnsom> And of course working on slides for the summti
20:06:19 <johnsom> Any other progress updates to share today?
20:06:52 <johnsom> There is some good stuff on dashboard.  One patch posted makes the panels update status automatically.
20:06:59 <johnsom> Very cool to see actually
20:07:25 <johnsom> Adam has also been very busy getting the tempest plugin going.  Great progress there
20:07:45 <nmagnezi> dayou promised this, and delivered :-)
20:07:58 <johnsom> yes, he has been doing great work on dashboard
20:08:03 <xgerman_> +1
20:08:05 <nmagnezi> +1
20:08:21 <johnsom> I also hope to get some reviews in on some client patches as I would like to do a release for that soon too.
20:08:43 <rm_mobile> Ah yeah I have some debt in the client
20:08:56 <johnsom> Yeah, I need to add the provider API
20:09:26 <johnsom> The release folks are reminding me that we should have a client release soon-ish
20:09:46 <rm_mobile> Well the debt I have are for features only in master, so
20:09:51 <rm_mobile> Less urgent
20:10:06 <johnsom> Any other updates?  Rally or Grenade?
20:10:28 <cgoncalves> I think I'm waiting for reviews on grenade
20:10:46 <johnsom> MS2 is the week of June 4th BTW
20:11:12 <xgerman_> yeah, painfully aware
20:11:14 <nmagnezi> nothing new on Rally, we are at the final stretch with tripleO so it got priority
20:11:15 <johnsom> Ok, cool, good to know!  Happen to have the link handy?
20:11:34 <nmagnezi> right afterwards I'll complete the Rally scenario
20:11:38 <nmagnezi> but it's in a good shape
20:11:42 <johnsom> Ok, cool
20:11:54 <nmagnezi> johnsom, yes, both cgoncalves and I can share some URLs to patches
20:12:21 <nmagnezi> #link https://review.openstack.org/#/c/568089/
20:12:27 <nmagnezi> #link https://review.openstack.org/#/c/568138/
20:12:35 <johnsom> Thanks!
20:12:47 <cgoncalves> https://review.openstack.org/#/q/(owner:%22Carlos+Goncalves+%253Ccgoncalves%2540redhat.com%253E%22)+OR+(owner:%22Nir+Magnezi+%253Cnmagnezi%2540redhat.com%253E%22+)
20:13:03 <nmagnezi> i guess that work too :)
20:13:16 <rm_mobile> I'm stuck in summit prep and travel prep for other stuff too, so I'm basically just trying to get through the reviews I have up already for the rest of this week
20:13:45 <nmagnezi> #link https://review.openstack.org/#/c/568395/
20:14:10 <nmagnezi> #link https://review.openstack.org/#/c/568801/
20:14:15 <rm_mobile> We've got a few large chunks of stuff just waiting because everyone is working hard on getting stuff done, but now there is so much done that it's a giant pile of reviews
20:14:29 <nmagnezi> johnsom, we have bunch more..
20:14:53 <johnsom> Yep, ok, thanks!
20:14:58 <rm_mobile> So we may want to do a hardcore review spike during summit off-hours if possible <_<
20:15:10 <rm_mobile> I'll be prodding people about that
20:15:13 <xgerman_> I hope there are off hours…
20:15:18 <rm_mobile> Lol yeah...
20:15:20 <johnsom> Yeah, I have been trying to mix some in.
20:15:28 <rm_mobile> I feel like usually we have a bit of time
20:15:46 <johnsom> #topic We need to decide on HTTP status codes for driver issues
20:16:06 <johnsom> So I would like some input on the status codes we want to return for provider driver issues.
20:16:11 <cgoncalves> rm_mobile, off-hours as in at 2 am? :)
20:16:13 <johnsom> #link https://review.openstack.org/#/c/563795/12/octavia/common/exceptions.py
20:16:22 <rm_mobile> Lol if need be
20:16:27 <johnsom> Ha, his off ours is like 8am
20:16:37 <cgoncalves> lol
20:16:52 <johnsom> So I initially picked 5xx for some of those
20:17:07 <rm_mobile> I think 501 for basically all of them
20:17:11 <johnsom> The bummer is that it would be nice if we never return a 500 on purpose
20:17:34 <johnsom> Well, ProviderNotEnabled is clearly user error and valid with 400
20:17:36 <rm_mobile> 501 != 500
20:17:42 <rm_mobile> Right
20:17:46 <rm_mobile> But the rest, 501
20:17:49 <johnsom> Basically they didn't enter a valid driver name
20:18:40 <openstackgerrit> Merged openstack/octavia-dashboard master: Imported Translations from Zanata  https://review.openstack.org/567789
20:19:02 <johnsom> 501 is "Not Implemented"
20:19:25 <cgoncalves> 406 Not Acceptable for ProviderUnsupportedOptionError?
20:19:35 <johnsom> Which seems wrong for "ProviderDriverError" which is basically the driver raised some unknown exception
20:19:51 <cgoncalves> I think we should use more of 4xx at a quick glance
20:20:22 <johnsom> 406 is typically header stuff I think.
20:20:28 <cgoncalves> 405 Method Not Allowed for ProviderUnsupportedOptionError
20:20:33 <rm_mobile|> 4xx is all classified as client errrors
20:20:42 <johnsom> Both not implemented and unsupported option could be 400 as well
20:20:47 <rm_mobile|> Which is why I don't like it
20:20:52 <xgerman_> https://http.cat/418
20:21:27 <cgoncalves> 4xx "is intended for situations in which the error seems to have been caused by the client"
20:22:04 <rm_mobile|> Which in most cases here is not at all true
20:22:22 <rm_mobile|> We need to stick with 5xx for server config fails
20:22:33 <johnsom> Depends on if you expect a user asking for a feature knows if the driver supports it or not. lol
20:22:46 <rm_mobile|> Lol
20:22:57 <xgerman_> johnsom: +1
20:23:02 <rm_mobile|> Ok time to figure out user intent programmatically
20:23:16 <johnsom> So, maybe switch ProviderNotImplementedError and ProviderUnsupportedOptionError to 400?
20:23:25 <rm_mobile|> No thanks IMO
20:23:30 <xgerman_> yep, if it’s a RTFM type situation or the user is left intentionally in the dark about whihc providers are installed and support waht
20:24:09 <johnsom> Well there is an API that tells them which drivers are enabled for use. That is why ProviderNotEnabled is clearly a 400
20:24:39 <cgoncalves> +1 on 400
20:25:03 <johnsom> Yeah, looking again, 501 not implemented is probably right for those.
20:25:05 <rm_mobile|> Yes that one only
20:25:16 <johnsom> So it's really just ProviderDriverError the catch all
20:25:20 <rm_mobile|> The rest need to be 5xx
20:25:39 <johnsom> Do we leave it 500 or give it something else.
20:26:00 <rm_mobile|> 501?
20:26:21 <johnsom> If people don't read the error, 500 gets interpreted as "octavia is broken". But....  in a way that driver is broken
20:27:20 <cgoncalves> ProviderDriverError 500. driver was expected to fulfill request but failed
20:27:31 <johnsom> I guess my vote is to leave it 500
20:27:46 <johnsom> It's the most accurate technically
20:28:43 <johnsom> Any other votes?  I think rm_mobile is 501 on that
20:30:32 <xgerman_> I like to blame the user…
20:30:55 <rm_mobile> Yep
20:31:14 <rm_mobile> 500 I never even read, I just assume the server screwed up
20:31:17 <xgerman_> so lean towards 4xx but not strong enought for a vote
20:31:23 <rm_mobile> Usually it's uncaught explosions
20:31:47 <johnsom> Yeah, that is what this is, uncaught driver explosions
20:32:05 <rm_mobile> Well, we catch it
20:32:13 <rm_mobile> And we know kinda what's up
20:32:48 <xgerman_> how do other driver based projects do it? neutron?
20:33:01 <johnsom> Ok, well, it doesn't sound like we have strong opinions so I'll leave it to the reviewers votes...
20:33:13 <johnsom> I just thought it would be good to discuss as a team
20:33:28 <rm_mobile|> Yeah
20:33:32 <xgerman_> +1
20:33:44 <johnsom> #topic Open Discussion
20:33:51 <johnsom> Any other topics for this week?
20:34:40 <xgerman_> oh, I need some OSP tech support
20:35:04 <xgerman_> especially why Octavia doesn’t work in OSP 12…
20:35:06 <johnsom> Ha
20:35:25 <cgoncalves> xgerman_, it's not supported (tm)
20:35:37 <cgoncalves> what's the status code for that? 4xx or 5xx? :P
20:35:38 <nmagnezi> that's because.. we don't actually ship it in OSP12 as a supported component :)
20:35:40 <johnsom> Very happy to see the OSP 13-beta release notes with Octavia fully supported
20:35:50 <xgerman_> +1
20:35:53 <nmagnezi> johnsom, indeed :)
20:36:15 <xgerman_> anywho, the Tech Preview in 12 doesn’t install the SSL certs — hope that’s fixed in 13
20:36:31 <cgoncalves> yep!
20:36:46 <cgoncalves> either use can provide certs or let OSP generate
20:37:01 <cgoncalves> https://bugzilla.redhat.com/show_bug.cgi?id=1462670
20:37:03 <openstack> bugzilla.redhat.com bug 1462670 in openstack-tripleo-common "Octavia TripleO support: allow auto-generated or user-provided certificates when configuring octavia" [High,Verified] - Assigned to beagles
20:37:08 <nmagnezi> yup. and anyhow for OSP13 we'll be happy to get any feedback and fix what's needed
20:37:17 <cgoncalves> OSP Octavia tracker: https://bugzilla.redhat.com/show_bug.cgi?id=1433523
20:37:19 <openstack> bugzilla.redhat.com bug 1433523 in openstack-octavia "[Tracker] [Octavia] Full support" [High,Assigned] - Assigned to nmagnezi
20:37:24 <xgerman_> somehow my org wants to use OSP12
20:37:31 <nmagnezi> "Assigned to nmagnezi"
20:37:32 * nmagnezi runs
20:37:46 <cgoncalves> nmagnezi, you're the squad lead :P
20:37:58 <nmagnezi> cgoncalves, that was a secret..
20:37:59 <cgoncalves> xgerman_, bad! OSP13 is LTS
20:38:06 <nmagnezi> :)
20:38:39 <xgerman_> I am juts an engineer (TM)
20:39:55 <johnsom> Anything else today?
20:40:50 <johnsom> Ok, thanks folks!  Enjoy the quiet while we are at the Summit.  Great time for reviewing.....  grin
20:41:04 <johnsom> #endmeeting