*** tetsuro has quit IRC | 00:20 | |
*** tetsuro has joined #openstack-tc | 00:21 | |
*** tetsuro has quit IRC | 00:21 | |
*** ianychoi__ is now known as ianychoi | 00:44 | |
openstackgerrit | Merged openstack/election master: Correct TC Election event in template_emails.py https://review.opendev.org/752113 | 00:53 |
---|---|---|
*** evrardjp has quit IRC | 04:33 | |
*** evrardjp has joined #openstack-tc | 04:33 | |
*** diablo_rojo has quit IRC | 04:44 | |
*** tetsuro has joined #openstack-tc | 05:03 | |
*** tetsuro has quit IRC | 05:22 | |
*** tetsuro has joined #openstack-tc | 05:23 | |
*** tetsuro has quit IRC | 05:26 | |
*** tetsuro has joined #openstack-tc | 05:35 | |
*** tetsuro has quit IRC | 05:59 | |
*** mrunge_ is now known as mrunge | 06:00 | |
*** ralonsoh has joined #openstack-tc | 06:03 | |
*** lpetrut has joined #openstack-tc | 06:15 | |
*** e0ne has joined #openstack-tc | 06:24 | |
*** tetsuro has joined #openstack-tc | 06:40 | |
*** tetsuro has quit IRC | 06:41 | |
*** slaweq has joined #openstack-tc | 06:48 | |
*** ralonsoh has quit IRC | 07:17 | |
*** camelCaser has quit IRC | 07:30 | |
*** camelCaser has joined #openstack-tc | 07:31 | |
*** tosky has joined #openstack-tc | 07:43 | |
*** tetsuro has joined #openstack-tc | 07:53 | |
*** tetsuro has quit IRC | 08:12 | |
*** tetsuro has joined #openstack-tc | 08:18 | |
*** ralonsoh has joined #openstack-tc | 08:20 | |
*** ralonsoh has quit IRC | 08:24 | |
*** tetsuro has quit IRC | 08:24 | |
yoctozepto | morning TC; I couldn't get in touch with our Masakari PTL so we have got no timeslot for Masakari PTG - can it be somehow amended? (I know the deadlines have been missed) | 08:30 |
*** ralonsoh has joined #openstack-tc | 09:16 | |
*** slaweq has quit IRC | 10:41 | |
*** slaweq has joined #openstack-tc | 10:54 | |
*** slaweq has quit IRC | 12:01 | |
*** slaweq has joined #openstack-tc | 12:30 | |
fungi | yoctozepto: maybe check with diablo_rojo_phon in a few hours and see if she has suggestions | 12:39 |
*** ccamel has joined #openstack-tc | 12:53 | |
*** camelCaser has quit IRC | 12:54 | |
*** ricolin has quit IRC | 13:20 | |
fungi | the tripleo team are exploring the new distributed leadership model: http://lists.openstack.org/pipermail/openstack-discuss/2020-September/017317.html | 13:32 |
*** ricolin has joined #openstack-tc | 13:45 | |
fungi | though it seems at least some of them are confused by the inclusion of the traditional "ptl with liaisons" description being included in that resolution | 14:11 |
gmann | o/ | 14:24 |
*** lpetrut has quit IRC | 14:38 | |
knikolla | o/ | 14:48 |
openstackgerrit | Thierry Carrez proposed openstack/governance master: Define TC-approved release in a resolution https://review.opendev.org/752256 | 14:49 |
gmann | Office hour time | 15:01 |
fungi | since this is probably the active office hour these days, quick reminder: nominations for tc and ptl open tuesday next week just before utc midnight | 15:02 |
fungi | six tc seats are expiring, four seats will be up for election | 15:02 |
gmann | gnoochi issue for Mysql on Focal is merged now but not yet release -https://github.com/gnocchixyz/gnocchi/issues/1069 (https://github.com/gnocchixyz/gnocchi/pull/1047) | 15:02 |
gmann | this is blocking many other projects to migrate integration testing to Focal | 15:03 |
mnaser | hi folks | 15:04 |
mnaser | fungi: thanks for that remidner | 15:04 |
gmann | another one blocking is barbican which is not yet fixed. | 15:04 |
gmann | after these two we will be good to move devstack base jobs to run on Focal | 15:05 |
mnaser | gmann: forgive me but how come it will break everyone if we switch to focal? wouldn't it break barbican and gnocchi consumers only? | 15:05 |
fungi | integration testing | 15:06 |
gmann | mnaser: yeah their consumer but many projects has their integrated jobs. gor gnoochi it is via ceilometer integration testing | 15:06 |
mnaser | i see | 15:06 |
gmann | *for | 15:06 |
mnaser | maybe a silly question | 15:07 |
gmann | sahara disable ceilometer and proceed but i am not sure if that is possible for other project too, like tacker cannot do i think | 15:07 |
mnaser | but couldn't we change the nodeset for their specific jobs | 15:07 |
mnaser | and then that way we don't have to block the world for them and then retroactively 'clean up' after? | 15:08 |
gmann | mnaser: yeah that is one option. | 15:08 |
gmann | let me check how many such jobs are | 15:08 |
mnaser | gmann: i don't know if we're working with a specific timeline but maybe we can have a cutoff time | 15:09 |
gmann | mnaser: deadlines are already passed but i can work on making the blocked jobs on bionic and rest other on focal. | 15:09 |
mnaser | that way no one merges something that breaks focal while we wait for others to move to focal, though that's mostly unlikely | 15:09 |
mnaser | oh, if we're past it, then maybe that seems like the best approach for now | 15:09 |
gmann | agree. | 15:10 |
fungi | the gnocchi error looks like it prevents gnocchi from running with mysql 8, so if a devstack job needs to run ceilometer which needs to run gnocchi then it can't use focal | 15:10 |
mnaser | fungi: right yep, and i guess by extension "if anyone needs to run ceilometer on focal, they cant" | 15:11 |
gmann | devstack base job does not need but few projects specific jobs does | 15:11 |
fungi | mnaser: which in turn means ceilometer can't release as part of victoria, i guess, since it doesn't meet our tested runtime expectations | 15:11 |
mnaser | i guess we might have the tooling to find devstack jobs that have ceilometer in its required projects or something | 15:11 |
mnaser | fungi: i think that's another discussion, we don't really test for most of our runtime expectations to be honest | 15:11 |
gmann | we have testing patches up so we know what all jobs or projects | 15:12 |
mnaser | i think opensuse is there and i dont think i see it running on all integrated services, nor centos (but at least tripleo is a 'signal') | 15:12 |
fungi | mnaser: well, if it doesn't run on *any* of our tested runtimes though... | 15:12 |
mnaser | fungi: did we remove 18.04 from a tested runtime? i cant remember | 15:12 |
* mnaser opens a new tab | 15:12 | |
fungi | yes | 15:13 |
gmann | yes | 15:13 |
mnaser | ill take your words for it then :) | 15:13 |
mnaser | ok yes, that is problematic then | 15:13 |
fungi | ubuntu 20.04 lts, centos 8, opensuse 15 | 15:13 |
mnaser | https://governance.openstack.org/tc/reference/runtimes/victoria.html yep | 15:13 |
gmann | but it is not like that they do nt run on any of testing runtime. unit and functional testing on Focal for them | 15:13 |
fungi | maybe if one of those still ships mysql<8... | 15:14 |
fungi | (but that seems unlikely) | 15:14 |
smcginnis | We probably should have more tests run on centos 8 and opensuse 15. | 15:14 |
mnaser | fungi: but that introduces another discussion of "validated packages" | 15:14 |
mnaser | i.e. maybe it works with ubuntu uca but not in stock packages | 15:14 |
* mnaser doesn't know that for a fact but just tossing discussion | 15:15 | |
mnaser | smcginnis: i think that would be a reasnoable thing to do. i think we run devstack against all runtimes in the devstack gates, though not sure how often that runs | 15:15 |
gmann | added that discussion in PTG etherpad | 15:15 |
gmann | smcginnis: +1, or we should distribute it in more balance way | 15:15 |
fungi | right, there's the fact that ussuri was packaged for focal months ago, so no idea how or if their packaged gnocchi/ceilometer are working | 15:15 |
mnaser | yeah i guess | 15:16 |
mnaser | packaging doesn't mean working :p or maybe uca doesnt use mysql 8 | 15:16 |
gmann | exactly and barbican too | 15:16 |
fungi | i don't think anybody has reached out to jamespage or corecb to ask how they got on | 15:16 |
mnaser | perhaps the uca has a different set of 'infra' packages | 15:16 |
mnaser | or maybe it's a specific option we need to set in mysql to allow for that behaviour? | 15:17 |
mnaser | (speaking for gnocchi) | 15:17 |
fungi | just confirmed, the centos 8 appstream also only includes mysql 8, to get 5.7 you need third-party packages apparently | 15:17 |
mnaser | o_O | 15:17 |
mnaser | i think tripleo ships barbican | 15:17 |
mnaser | oh, right, but gnocchi is the issue for mysql | 15:18 |
mnaser | oooh this is interesting | 15:19 |
mnaser | https://bugs.launchpad.net/barbican/+bug/1734329 | 15:19 |
openstack | Launchpad bug 1734329 in Barbican "CHECK constraint bug in db migration on Mariadb 10.2" [Undecided,New] | 15:19 |
fungi | yeah, same issue | 15:20 |
mnaser | i'm curious, why do we use mysql? why not mariadb? | 15:20 |
fungi | i found that when digging on the gnocchi error | 15:20 |
mnaser | maybe if we ran mariadb it'll be "new enough" to not be affected by this issue | 15:20 |
jungleboyj | o/ | 15:20 |
mnaser | btw i recently learned about how mariadb and mysql are different in ways that surprise you | 15:20 |
mnaser | they have different REGEX patterns. it's wild | 15:20 |
mnaser | don't tell me how i found out. | 15:20 |
gmann | there was discussion on ML about using mariodb in past right? | 15:21 |
mnaser | aaah | 15:21 |
mnaser | fungi, gmann: https://github.com/openstack/devstack/blob/13dac66885ce26ed72e4065be6b8d96fca661c8f/lib/databases/mysql#L18-L29 | 15:22 |
mnaser | we pretty much run mariadb everywhere | 15:22 |
mnaser | except for ubunt | 15:22 |
gmann | ah | 15:22 |
mnaser | thats probably how it works inside ubuntu and centos (uca/tripleo) -- i think they all mostly deploy mariadb | 15:22 |
fungi | ahh, yep, some distros replaced mysql with mariadb but kept calling the package mysql | 15:22 |
fungi | just to add to the confusion | 15:23 |
mnaser | yep | 15:23 |
gmann | humm | 15:23 |
mnaser | i'm not on the devstack team but i would be +1 on a change that updates ubuntu to use 'mariadb' | 15:23 |
fungi | (at the time they hadn't diverged much at all) | 15:23 |
mnaser | and that might solve the gnocchi issue at least | 15:23 |
mnaser | https://packages.ubuntu.com/focal-updates/mysql-server vs https://packages.ubuntu.com/focal/mariadb-server | 15:24 |
fungi | gmann: we had heated debates on whether we test with postgresql, and that led to the creation of the "support" definition, but i don't recall much in the way of mariadb vs mysql discussion | 15:24 |
mnaser | fungi: it seems that ubuntu is nice enough to actually ship both | 15:24 |
fungi | yup | 15:24 |
mnaser | gmann: before we go about switching it to everyone -- perhaps we should try a ceilometer job where we set MYSQL_SERVICE_NAME=mariadb | 15:24 |
gmann | for long term i think that is best way onwards. | 15:24 |
mnaser | and then maybe make a change to devstack base with MYSQL_SERVICE_NAME=mariadb and see how it feels | 15:25 |
mnaser | i agree. plus, it's more OSS-y. but don't quote me, i don't know much about licenses :) | 15:25 |
fungi | this is the closest we got to saying what databases we "support" but it was several years ago now: https://governance.openstack.org/tc/resolutions/20170613-postgresql-status.html | 15:25 |
mnaser | i know there are users that are happily running postgresql which i consult for | 15:25 |
gmann | but i will say to do in Wallaby not as part of Focal migration as we are on the edge of V release | 15:25 |
fungi | and in the footnotes for that resolution, we included "The term MySQL is used as a reference to the MySQL database and the close variants MariaDB and Galera." | 15:26 |
mnaser | they are unfortunately not s oclose anymore | 15:26 |
mnaser | gmann: i mean, there's two schools of thoughts -- we're already doing something so last minute by switching to focal now -- so maybe do it all at once | 15:26 |
mnaser | but there is the other one that "we don't know what else might break" | 15:26 |
mnaser | which usually means "everything we didn't expect" :p | 15:27 |
gmann | yeah that is my worry about what else can break | 15:27 |
mnaser | you would be surprised. like i said, they don't even share teh same regex patterns | 15:27 |
mnaser | so that really blew my mind when i found that out | 15:27 |
mnaser | gmann: is ceilometer 'plugin' inside openstack/devstack or in their repo? | 15:27 |
gmann | may be we can move forwards with keeping those jobs on bionic and move rest of other on focal | 15:27 |
gmann | mnaser: in their repo | 15:28 |
fungi | anyway, i think it's entirely in keeping with the 20170613 resolution to test with mariadb, but we didn't really consider what happens when something can no longer run on mysql though can still run on a variant. i think we expected that we would do our best to keep software working on mysql and all the common mysql variants (including mariadb and galera) | 15:28 |
mnaser | gmann: so how about we try this -- update their jobs to focal and in their plugin have something that sets MYSQL_SERVICE_NAME=mariadb | 15:28 |
mnaser | personally, i think its better that we have everyone on focal with one project using a variation of different db backend, rather than a different underlying OS | 15:28 |
mnaser | and honestly in my experience most deployments are using mariadb.. | 15:29 |
fungi | it also worries me how we would communicate this at release, so say "openstack works on mysql, oh except if you're using ceilometer and then you need mariadb instead" | 15:29 |
mnaser | fungi: a release note inside ceilometer? | 15:30 |
fungi | i guess that's doable, but not very operator-friendly | 15:30 |
mnaser | if we really want to be extreme | 15:30 |
mnaser | we can build a multinode job where mariadb is used for gnocchi | 15:30 |
fungi | and implies that if mariadb is actually the thing all our services can run with, then we probably need to change our messaging to prefer mariadb over mysql | 15:30 |
mnaser | fungi: i am in favour of that for the next release | 15:31 |
gmann | +1 | 15:31 |
mnaser | the messaging ain't great for operators but also it's not really our fault either | 15:31 |
mnaser | cause its not openstack that is incompatible with mysql, its gnocchi | 15:31 |
gmann | i think with reno it can be best effort to communicate, and if it get fixed before V release we will be good | 15:31 |
mnaser | so a known issue could be "if you are using gnocchi, please be aware of bug x.y.z and make sure you run version 1.2.3 to avoid it OR cherry-pick patch abc" | 15:32 |
mnaser | assuming gnocchi makes a release that eventually fixes this | 15:32 |
mnaser | there are "clean" solutions but i dont know if we have the time to do them | 15:32 |
gmann | hope so, fix merged almost after a year so am hoping release also | 15:33 |
fungi | i guess you could say the problem isn't actually gnocchi, but a regression in mysql 8 | 15:33 |
mnaser | fungi: well we technically don't know that yet ;) | 15:33 |
fungi | great point | 15:33 |
gmann | ok, let me try that today and see how it goes. if not well then running them on bionic is last option we can go with | 15:33 |
mnaser | gmann: i think maybe it might be good to run focal job for ceilometer with MYSQL_SERVICE_NAME=mariadb and see what happens | 15:33 |
gmann | yeah | 15:34 |
mnaser | if it doesn't work either, we go back to drawing board -- if it does, even better, and lets try and plan to move towards mariadb cause that's what most deployment tools use in prod | 15:34 |
gmann | k | 15:34 |
mnaser | what about barbican | 15:35 |
gmann | frickler and stephenfin has some fixes up but not merge yet | 15:35 |
gmann | https://review.opendev.org/#/c/731281/ or https://review.opendev.org/#/c/746281/ | 15:35 |
gmann | i pinged frickler on barbican channel also about it | 15:36 |
mnaser | at least the nice thing with this is we have somewhat control over its 'destiny' | 15:36 |
redrobot | 👀👀👀 | 15:36 |
mnaser | oh hi there | 15:37 |
mnaser | it seems like another mysql-y thing | 15:37 |
gmann | yeah | 15:37 |
mnaser | databases | 15:39 |
mnaser | why do we need those | 15:39 |
gmann | redrobot: did you check which fix can be merged soon so that we can proceed further on switching testing to Focal | 15:40 |
*** e0ne has quit IRC | 15:42 | |
redrobot | gmann, not yet... got meeting hell this AM. Should be able to look later today. | 15:42 |
gmann | redrobot: k, thanks | 15:43 |
gmann | mnaser: no PTG slot, did you decided what all slot we need ? | 15:44 |
mnaser | gmann: i think diablo_rojo_phon was waiting for me to chime in and i didn't get a chance to say "yeah, anything works" | 15:44 |
gmann | I am planning to book one for RBAC popup team and trying to avoid conflicts as much as i can | 15:44 |
mnaser | i will try and sync up with diablo_rojo_phon tomorrow unless we do it earlier and just end up putting up a time, i think she made a fair amount of progress on consensus and i am not fully up to date on that | 15:45 |
mnaser | i've been a little swamped to say the least | 15:45 |
diablo_rojo_phon | I booked rooms :) | 15:45 |
jungleboyj | Cool. | 15:45 |
diablo_rojo_phon | I just hadnt announced them yet. | 15:45 |
diablo_rojo_phon | You can look at the sign up ethercalc but I put them in the otg planning etherpad too. | 15:46 |
diablo_rojo_phon | *ptg | 15:46 |
diablo_rojo_phon | It's like 2 hours Tuesday and 4 Friday or something like that. | 15:46 |
gmann | great, just saw, i was searching 'TC' :) | 15:47 |
gmann | thanks | 15:47 |
diablo_rojo_phon | Ahh similar to my problem looking for you in the list of poll responses the other day :) | 15:48 |
gmann | heh | 15:48 |
*** slaweq has quit IRC | 15:48 | |
*** slaweq has joined #openstack-tc | 15:50 | |
gmann | diablo_rojo_phon: just in case if it is possible, i requested for policy popup PTG slot in ML. I can fill survey and book slot accordingly. | 15:51 |
diablo_rojo_phon | gmann: Yeah if you get it in today we can still get it on the official schedule and list of teams. Worst case whatever is left in the schedule is just reserveable space. | 15:53 |
gmann | cool, doing it. thanks | 15:53 |
mnaser | diablo_rojo_phon: thank you for taking care of it | 15:54 |
mnaser | btw, since we merged the change for meetings, i am going to have a doodle up to circulate for availabl etimes | 15:54 |
mnaser | so we don't lose the office hours | 15:54 |
* jungleboyj has the meetings on my calendar now. :-) | 15:54 | |
gmann | mnaser: +1 | 15:55 |
mnaser | or we could use the same slot time | 15:55 |
gmann | i think checking time again will be good as we do not have all members in office hour so may be this is not suitable time? | 15:57 |
mnaser | i think we had like 9 or 10 people marking that they'd be available at this time | 15:58 |
gmann | yeah | 15:58 |
gmann | 10 | 15:58 |
* mnaser shrugs | 15:59 | |
mnaser | we'll get to it i am hoping | 15:59 |
gmann | ok. anyways after TC election we may need to check time again so doing in this time until then is fine. | 16:00 |
mnaser | yup, that's true! | 16:00 |
gmann | next week Wed office hour time is our first weekly meeting? | 16:02 |
mnaser | i think the 'community' decided that we have both office hours and meetings | 16:02 |
mnaser | so we're going to have to come up with a time to meet, outside office hours | 16:02 |
mnaser | unless people want to replace this office hour by a meeting | 16:04 |
mnaser | i wouldn't mind that :) | 16:04 |
gmann | my vote is to replace :) | 16:05 |
jungleboyj | That would be ok with me. | 16:06 |
njohnston | +1 | 16:06 |
jungleboyj | Wait what time is this office hour supposed to be at? | 16:06 |
gmann | its 15 UTC | 16:09 |
jungleboyj | Ah, that is good. I had it at 13 UTC. So, yeah, this time works well for me. | 16:10 |
*** tosky has quit IRC | 16:11 | |
*** tonyb has quit IRC | 16:41 | |
*** tonyb has joined #openstack-tc | 16:42 | |
*** tobberydberg has quit IRC | 16:51 | |
*** tobberydberg_ has joined #openstack-tc | 16:51 | |
*** tonyb has quit IRC | 17:01 | |
*** ralonsoh has quit IRC | 17:26 | |
*** ralonsoh has joined #openstack-tc | 17:58 | |
*** dklyle has joined #openstack-tc | 18:09 | |
*** tosky has joined #openstack-tc | 18:51 | |
*** ralonsoh has quit IRC | 19:20 | |
*** ralonsoh has joined #openstack-tc | 19:21 | |
*** ralonsoh has quit IRC | 19:28 | |
*** slaweq has quit IRC | 21:14 | |
*** slaweq has joined #openstack-tc | 21:15 | |
*** slaweq has quit IRC | 21:27 | |
*** slaweq has joined #openstack-tc | 21:33 | |
*** slaweq has quit IRC | 21:54 | |
*** tosky_ has joined #openstack-tc | 21:56 | |
*** tosky has quit IRC | 21:59 | |
*** tosky_ is now known as tosky | 22:00 | |
*** slaweq has joined #openstack-tc | 22:01 | |
*** slaweq_ has joined #openstack-tc | 22:10 | |
*** slaweq has quit IRC | 22:13 | |
*** slaweq_ has quit IRC | 22:19 | |
*** tosky has quit IRC | 22:59 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!