| *** jmlowe has quit IRC | 00:15 | |
| *** ramishra has quit IRC | 01:06 | |
| *** sapd1_ has quit IRC | 02:43 | |
| *** sapd1_ has joined #openstack-containers | 02:44 | |
| *** hongbin has joined #openstack-containers | 02:53 | |
| *** hongbin has quit IRC | 03:23 | |
| *** ramishra has joined #openstack-containers | 03:30 | |
| *** udesale has joined #openstack-containers | 04:50 | |
| *** vishalmanchanda has joined #openstack-containers | 04:54 | |
| *** ykarel|afk is now known as ykarel | 05:10 | |
| *** elenalindq has joined #openstack-containers | 06:02 | |
| *** dave-mccowan has joined #openstack-containers | 06:08 | |
| *** hongbin has joined #openstack-containers | 06:23 | |
| *** hongbin has quit IRC | 06:28 | |
| *** openstack has joined #openstack-containers | 07:19 | |
| *** ChanServ sets mode: +o openstack | 07:19 | |
| *** pcaruana has joined #openstack-containers | 07:23 | |
| *** openstack has joined #openstack-containers | 07:43 | |
| *** ChanServ sets mode: +o openstack | 07:43 | |
| *** vishalmanchanda has quit IRC | 08:36 | |
| *** vishalmanchanda has joined #openstack-containers | 08:37 | |
| *** flwang1 has joined #openstack-containers | 08:51 | |
| flwang1 | brtknr: strigazi: around? | 08:51 |
|---|---|---|
| brtknr | flwang1: im here | 08:57 |
| flwang1 | brtknr: i didn't see the response from strigazi | 08:58 |
| brtknr | me neither | 08:59 |
| flwang1 | so i'm going to cancel this one, but we can discuss if you have any topic | 08:59 |
| brtknr | flwang1: nothing urgent this week, will test the patch you emailed about today | 09:01 |
| flwang1 | brtknr: i understand your concern about that patch, another solution is, always passing in a default AZ for the volume | 09:02 |
| brtknr | is availability_zone a label? | 09:02 |
| flwang1 | it is | 09:02 |
| brtknr | yes always passing in default AZ for the volume seems preferable to me | 09:03 |
| flwang1 | the only tricky thing is technically, the AZ between nova and cinder could be different :( | 09:03 |
| brtknr | yes I read about AZ, the whole thing is a bit of a mess | 09:03 |
| flwang1 | but i never heard about that case on a production env because it doesn't make sense and may make the architecture so complicated | 09:04 |
| flwang1 | that's why I end up with current solution | 09:04 |
| brtknr | What if the solution was to pass in default AZ unless its overriden? | 09:05 |
| flwang1 | you mean adding a label for volume AZ just in case? | 09:07 |
| brtknr | so the conductor reads the default AZ for Cinder via Cinder client | 09:08 |
| flwang1 | brtknr: ^ | 09:08 |
| brtknr | and passes that to heat? | 09:08 |
| flwang1 | yep, let me explain it with more details | 09:11 |
| brtknr | ok maybe you could add this to gerrit | 09:11 |
| *** Jeffrey4l has quit IRC | 09:12 | |
| flwang1 | actually, i'd like to involve strigazi to discuss to make sure we all agree with the solution or any solution | 09:13 |
| flwang1 | otherwise, i'm seeing we will easily waste another 1 or 2 weeks | 09:13 |
| *** Jeffrey4l has joined #openstack-containers | 09:13 | |
| brtknr | flwang1: sounds good | 09:14 |
| flwang1 | the patch is sitting there for 2 weeks without any review make me a bit sad | 09:14 |
| brtknr | flwang1: yeah sorry its a tricky one to test for me as ive never used AZ | 09:15 |
| flwang1 | as i said, to test this patch, you don't have to have a multi az | 09:16 |
| flwang1 | anyway, we need to make decision how to solve this issue | 09:20 |
| brtknr | flwang1: do you know what is the name of the az on devstack? | 09:21 |
| flwang1 | i think the goal is avoid user setting the AZ unless they really want to | 09:22 |
| flwang1 | so, we can have two new config options for AZ list for nova and cinder | 09:23 |
| flwang1 | and we can support another label for cinder AZ just in case | 09:24 |
| flwang1 | but with this way, cloud ops has to set the config option when doing deploy or upgrade, it's not backward compatible | 09:25 |
| flwang1 | brtknr: ^ | 09:25 |
| brtknr | hmm flwang1 you're right | 09:25 |
| brtknr | flwang1: im testing it now | 09:26 |
| jakeyip | hiya flwang1 | 09:26 |
| flwang1 | i can't see an api in cinder can get the AZ list | 09:26 |
| flwang1 | brtknr: https://www.redhat.com/en/blog/openshift-openstack-introduction-availability-zones this blog series is really good | 09:26 |
| flwang1 | it's about openshift on openstack when doing az, but it also works for k8s on openstack | 09:27 |
| flwang1 | jakeyip: hello | 09:27 |
| flwang1 | jakeyip: it would you appreciate if you can provide your comments on this https://review.opendev.org/#/c/705592/ | 09:27 |
| jakeyip | have you seen this? https://www.mirantis.com/blog/the-first-and-final-word-on-openstack-availability-zones/ | 09:27 |
| jakeyip | scroll down to OpenStack Project availability zone Comparison Summary | 09:28 |
| brtknr | jakeyip: hehe i read the mirantis blog | 09:29 |
| jakeyip | this says cinder doesn't accept None. However the cinder cli seens to say it defaults to None if not specified | 09:29 |
| brtknr | when i was trying to learn about AZ | 09:29 |
| flwang1 | jakeyip: i can confirm cinder doesn't accept None | 09:30 |
| flwang1 | that's the thing i don't like openstack | 09:31 |
| flwang1 | different projects implement same concept different :( | 09:31 |
| flwang1 | that's really frustrating | 09:32 |
| brtknr | OpenStack needs some UX people :) | 09:32 |
| jakeyip | flwang1: I am a bit confused about your patch | 09:32 |
| brtknr | that work across projects | 09:32 |
| flwang1 | jakeyip: what confuse you? | 09:33 |
| jakeyip | if AZ isn't set then you'll be creating a `Magnum::Optional::Cinder::Volume` without availability_zone param? isn't this same as None ? | 09:34 |
| flwang1 | jakeyip: if it's not pass in, heat seems handle it with a different way | 09:35 |
| flwang1 | i haven't checked the heat code, but i do tested it with None | 09:35 |
| flwang1 | jakeyip: it would be nice if you can test it | 09:35 |
| flwang1 | I would be happy if i'm wrong | 09:36 |
| jakeyip | how do you pass in a None value in Heat? | 09:37 |
| jakeyip | ey that's the thing, we (Nectar) enforces AZ to be mandatory, so I can't test None easily. Wondering how everyone's cloud work? | 09:37 |
| flwang1 | I tried None and none | 09:38 |
| flwang1 | and null, i can't exactly remember | 09:38 |
| flwang1 | it's about 3 weeks ago | 09:38 |
| jakeyip | to be honest, I think the easy way is another label | 09:39 |
| flwang1 | another label for volume? | 09:40 |
| brtknr | would be nice if Cinder's default config was respected | 09:40 |
| flwang1 | volume AZ? | 09:41 |
| brtknr | instead of having a magnum default for Cinder volumes | 09:41 |
| flwang1 | brtknr: i don't really understand | 09:41 |
| brtknr | flwang1: we need to take another look at what heat is doing when az is not specified | 09:42 |
| jakeyip | if nova AZ != cinder AZ we are enforcing them to be same when they shouldn't be...? | 09:43 |
| flwang1 | jakeyip: obviously, we can't | 09:43 |
| flwang1 | that's why i recommend above blog from redhat | 09:44 |
| brtknr | hmm volume_availability_zone label seems like the best option | 09:44 |
| flwang1 | there are 2 scenarios, nova multi az= cinder multi az and nova has multi az and cinder has only one az | 09:45 |
| flwang1 | jakeyip: do you also like the idea to create a new label for volume az? | 09:46 |
| jakeyip | flwang1: yes I would think so | 09:46 |
| *** ykarel|lunch is now known as ykarel | 09:47 | |
| jakeyip | just treat them differently. we won't want to be too strict | 09:47 |
| jakeyip | nova can also attach to cinder az of different names - https://docs.openstack.org/nova/latest/configuration/config.html#cinder.cross_az_attach | 09:47 |
| flwang1 | jakeyip: i know that. that's why we have to support this, because some env doesn't support cross az attach | 09:48 |
| flwang1 | otherwise, it won't be a problem | 09:48 |
| flwang1 | i understand a new label like volume_availability_zone can fix current issue | 09:49 |
| flwang1 | i'm just thinking further, how to avoid setting the label in the template | 09:50 |
| flwang1 | otherwise, we have to always set that label for any template | 09:50 |
| jakeyip | flwang1: I think there will be 2 scenerios if we don't set a vol az label - either it defaults to cinder default, or it uses AZ label. | 09:51 |
| flwang1 | jakeyip: do you mean we should get the default cinder/volume az from cinder if there is no label or config option? | 09:52 |
| jakeyip | not really. I feel that whichever option you choose, some cloud will be different and then they need to override it | 09:53 |
| jakeyip | we could default to AZ if not set and still allow override. I think this is the safest | 09:53 |
| flwang1 | 1. add a config option volume_az 2. try to get the volume az from label 3. otherwise, get it from cinder service list api | 09:54 |
| flwang1 | jakeyip: ^ | 09:57 |
| jakeyip | how to get default? | 09:57 |
| jakeyip | I assume (3) is get the default AZ ? | 09:57 |
| flwang1 | yes | 09:58 |
| jakeyip | is it possible to geet it from API? | 09:58 |
| flwang1 | feilong@feilong-pc:~$ openstack volume service list | 09:58 |
| flwang1 | +------------------+------------------------+------+---------+-------+----------------------------+ | 09:58 |
| flwang1 | | Binary | Host | Zone | Status | State | Updated At | | 09:58 |
| flwang1 | +------------------+------------------------+------+---------+-------+----------------------------+ | 09:58 |
| flwang1 | | cinder-scheduler | feilong-pc | nova | enabled | up | 2020-02-19T09:29:42.000000 | | 09:58 |
| flwang1 | | cinder-volume | feilong-pc@lvmdriver-1 | nova | enabled | up | 2020-02-19T09:29:41.000000 | | 09:58 |
| flwang1 | +------------------+------------------------+------+---------+-------+----------------------------+ | 09:58 |
| flwang1 | should be | 09:58 |
| jakeyip | AFAIK it's a config option https://github.com/openstack/cinder/blob/master/cinder/common/config.py#L95 | 09:59 |
| jakeyip | what if there is >1 and the operator has set a default config option :P | 09:59 |
| flwang1 | check the cinder-volume binary and get the az | 09:59 |
| flwang1 | for that case, the cloud operator should do that in the config option | 10:00 |
| flwang1 | otherwise, Magnum can't know | 10:00 |
| jakeyip | hmm that's an extra API call. you ok? | 10:02 |
| *** yoctozepto has joined #openstack-containers | 10:02 | |
| flwang1 | we only need that api call if both config option and the label are not set | 10:02 |
| jakeyip | btw we have multiple cinder-volumes | 10:04 |
| flwang1 | i don't think that matters, is it? | 10:04 |
| jakeyip | can you run me through your decision tree again please? | 10:05 |
| jakeyip | if label=availability_zone is not set, and if it is set ? | 10:06 |
| flwang1 | 1. if the config option default_volume_az=nz-1 is set, use it, otherwise go next | 10:06 |
| flwang1 | 2. check if the label default_volume_az is set or not, if it is, use it, otherwise go next | 10:07 |
| flwang1 | 3. try to get it by call cinder service list and filter the cinder-volume to get the az list and randomly select one | 10:07 |
| jakeyip | (1) is introducing a new config option in magnum ? | 10:08 |
| flwang1 | yes | 10:08 |
| jakeyip | what about label=availability_zone? | 10:09 |
| flwang1 | add a new label default_volume_az | 10:10 |
| flwang1 | the existing label availability_zone is for nova | 10:10 |
| flwang1 | and it's only used by nova at this moment | 10:10 |
| jakeyip | Ithink you're on to something | 10:11 |
| flwang1 | what do you mean on to something? | 10:11 |
| jakeyip | I would say 1. try label volume_az first | 10:11 |
| jakeyip | 2. fallback to default in config | 10:11 |
| jakeyip | I don't think 3 is necessary... randomly select one won't work | 10:11 |
| flwang1 | i'm happy with that | 10:12 |
| flwang1 | and i kind of like to drop the #3 | 10:12 |
| jakeyip | nice and clean :) | 10:12 |
| flwang1 | brtknr: thoughts? | 10:12 |
| brtknr | flwang1: that sounds preferable but may be worth chatting with strigazi about this first as they are the other az users afaik | 10:13 |
| jakeyip | if config isn't set what happens... | 10:13 |
| flwang1 | jakeyip: like now, failed i think | 10:13 |
| jakeyip | flwang1: in your cloud can you create cinder volume without passing in az ? | 10:14 |
| flwang1 | we can | 10:15 |
| flwang1 | because there is default az in cinder | 10:15 |
| flwang1 | i assume you're talking about cinder only case | 10:16 |
| flwang1 | not in magnum context, right? | 10:16 |
| jakeyip | yes | 10:16 |
| jakeyip | can you do a --debug on that cinder call to see if AZ is passed in ? | 10:16 |
| flwang1 | are you talking about a multi az env? | 10:17 |
| flwang1 | i don't think there is az being passed in, because there is no where the client side can get it | 10:17 |
| flwang1 | i have to go offline, sorry | 10:18 |
| jakeyip | ok | 10:18 |
| flwang1 | strigazi: could you please take a look above discussion and leave your comments here or on the patch | 10:18 |
| flwang1 | jakeyip: thank you for your good thoughts and input | 10:18 |
| flwang1 | i appreciate that | 10:19 |
| flwang1 | brtknr: are you going to join the Vancouver PTG? | 10:19 |
| flwang1 | jakeyip: ^ | 10:19 |
| flwang1 | strigazi: ^ | 10:19 |
| *** ivve has joined #openstack-containers | 10:20 | |
| jakeyip | no I am not. might have someone from Nectar still | 10:20 |
| flwang1 | ok | 10:21 |
| flwang1 | jakeyip: ttyl | 10:22 |
| jakeyip | flwang1: seey | 10:22 |
| jakeyip | flwang1: seeya | 10:22 |
| brtknr | flwang1: sadly not going to be traveling anywhere for a little while even if I was allowed to while the twins are still small | 10:26 |
| flwang1 | brtknr: make sense :) | 10:27 |
| openstackgerrit | Pierre Riteau proposed openstack/magnum master: Fix typo in docs https://review.opendev.org/708626 | 11:56 |
| *** mgoddard has quit IRC | 12:16 | |
| *** mgoddard has joined #openstack-containers | 12:22 | |
| *** iokiwi has quit IRC | 12:32 | |
| *** iokiwi has joined #openstack-containers | 12:32 | |
| *** ykarel is now known as ykarel|afk | 12:35 | |
| *** rcernin has quit IRC | 12:37 | |
| *** udesale_ has joined #openstack-containers | 12:49 | |
| *** udesale has quit IRC | 12:52 | |
| *** ykarel|afk is now known as ykarel | 13:49 | |
| *** jmlowe has joined #openstack-containers | 14:48 | |
| *** jmlowe has quit IRC | 14:56 | |
| *** irclogbot_3 has quit IRC | 14:56 | |
| *** irclogbot_0 has joined #openstack-containers | 15:01 | |
| *** pcaruana has quit IRC | 15:02 | |
| *** pcaruana has joined #openstack-containers | 15:03 | |
| cosmicsound | Status Reason | 15:07 |
| cosmicsound | default-master failed ,default-worker failed | 15:07 |
| cosmicsound | Killer error | 15:07 |
| cosmicsound | https://mdb.uhlhost.net/uploads/085b86b041b50cc5/image.png anyone any idea what is causing these timeouts | 15:08 |
| cosmicsound | Would be helpfull to get some hints in the error what to seek into | 15:08 |
| *** dave-mccowan has quit IRC | 15:08 | |
| *** lpetrut has quit IRC | 15:11 | |
| *** jmlowe has joined #openstack-containers | 15:20 | |
| *** johanssone has quit IRC | 15:24 | |
| *** johanssone has joined #openstack-containers | 15:27 | |
| *** ramishra has quit IRC | 16:31 | |
| *** ivve has quit IRC | 16:41 | |
| *** udesale_ has quit IRC | 16:56 | |
| *** ykarel is now known as ykarel|away | 17:06 | |
| *** jmlowe has quit IRC | 17:11 | |
| *** jmlowe has joined #openstack-containers | 17:12 | |
| *** jmlowe has quit IRC | 17:24 | |
| *** jmlowe has joined #openstack-containers | 17:37 | |
| *** jmlowe has quit IRC | 17:43 | |
| *** jmlowe has joined #openstack-containers | 18:00 | |
| *** dioguerra has quit IRC | 18:00 | |
| *** ivve has joined #openstack-containers | 18:32 | |
| *** jmlowe has quit IRC | 18:33 | |
| *** jmlowe has joined #openstack-containers | 18:36 | |
| *** vishalmanchanda has quit IRC | 19:11 | |
| *** flwang1 has quit IRC | 19:29 | |
| *** jmlowe has quit IRC | 20:24 | |
| *** jmlowe has joined #openstack-containers | 20:44 | |
| *** jmlowe has quit IRC | 20:56 | |
| *** mgariepy has quit IRC | 21:02 | |
| *** jmlowe has joined #openstack-containers | 21:09 | |
| openstackgerrit | Simon Merrick proposed openstack/magnum-ui master: Dynamicly restrict cluster action based on state https://review.opendev.org/699893 | 21:10 |
| *** elenalindq has quit IRC | 21:12 | |
| *** elenalindq has joined #openstack-containers | 21:12 | |
| *** elenalindq has quit IRC | 21:13 | |
| *** jmlowe has quit IRC | 21:16 | |
| *** yolanda has quit IRC | 21:38 | |
| *** yolanda has joined #openstack-containers | 21:52 | |
| *** yolanda has quit IRC | 22:06 | |
| *** jmlowe has joined #openstack-containers | 22:32 | |
| *** jmlowe has quit IRC | 22:51 | |
| *** jmlowe has joined #openstack-containers | 22:59 | |
| *** jmlowe has quit IRC | 23:00 | |
| *** rcernin has joined #openstack-containers | 23:04 | |
| *** jmlowe has joined #openstack-containers | 23:11 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!
