| *** gb21 has quit IRC | 00:55 | |
| *** buzztroll has quit IRC | 00:56 | |
| *** buzztroll has joined #openstack-glance | 01:03 | |
| *** mtanino has quit IRC | 01:05 | |
| *** btully has joined #openstack-glance | 01:05 | |
| *** buzztroll has quit IRC | 01:05 | |
| *** gb21 has joined #openstack-glance | 01:07 | |
| *** mine0901 has quit IRC | 01:10 | |
| *** ozialien10 has quit IRC | 01:12 | |
| *** sdake has joined #openstack-glance | 01:22 | |
| *** gb21 has quit IRC | 01:28 | |
| *** dims has quit IRC | 01:30 | |
| *** dims has joined #openstack-glance | 01:35 | |
| *** gb21 has joined #openstack-glance | 01:43 | |
| *** julim has joined #openstack-glance | 01:45 | |
| *** ducttape_ has quit IRC | 02:05 | |
| *** sdake_ has joined #openstack-glance | 02:08 | |
| *** sdake has quit IRC | 02:11 | |
| *** ducttape_ has joined #openstack-glance | 02:17 | |
| *** buzztroll has joined #openstack-glance | 02:18 | |
| *** sdake_ has quit IRC | 02:27 | |
| *** pushkaru has quit IRC | 02:36 | |
| *** pushkaru has joined #openstack-glance | 02:37 | |
| *** pushkaru has quit IRC | 02:39 | |
| *** pumarani__ has joined #openstack-glance | 02:39 | |
| *** pumarani__ has quit IRC | 02:44 | |
| openstackgerrit | wangxiyuan proposed openstack/glance: Remove DB downgrade https://review.openstack.org/229220 | 02:45 |
|---|---|---|
| *** pushkaru has joined #openstack-glance | 02:46 | |
| *** buzztroll has quit IRC | 02:47 | |
| *** ducttape_ has quit IRC | 02:50 | |
| *** gb21 has quit IRC | 02:52 | |
| *** pushkaru has quit IRC | 02:53 | |
| *** cdelatte has quit IRC | 03:03 | |
| *** gb21 has joined #openstack-glance | 03:04 | |
| nikhil | flwang: hello sir. around by any chance? | 03:31 |
| flwang | nikhil: yes, boss | 03:31 |
| nikhil | flwang: ah, so glad 😃 | 03:32 |
| flwang | nikhil: how can i help? | 03:32 |
| nikhil | flwang: wanted to discuss the location metadata weigth stuff if you've time now | 03:32 |
| flwang | nikhil: oh, yes | 03:32 |
| flwang | nikhil: i'm really happy you can give some sunshine on this corner | 03:33 |
| nikhil | flwang: LOL | 03:34 |
| nikhil | flwang: sorry about the delay, summit stuff has been crazy | 03:34 |
| flwang | nikhil: i see. and i'm sorry i missed this one | 03:35 |
| nikhil | and it's hard to find you online at the same time & when 5 people aren't pinging me at the same time 😃 | 03:35 |
| flwang | and we missed to have a f2f again | 03:35 |
| nikhil | flwang: indeed. I was thinking the same thing! | 03:35 |
| nikhil | flwang: take care of your little one, that's more imp now | 03:36 |
| *** pt_15 has quit IRC | 03:36 | |
| flwang | nikhil: sure thanks | 03:36 |
| nikhil | to be little one* | 03:36 |
| flwang | nikhil: so what's the question for the location weight? | 03:36 |
| nikhil | flwang: first, I'm trying to find the patch. is it just the python-glanceclient or there's a server side change as well? | 03:37 |
| flwang | https://review.openstack.org/#/c/268865 | 03:38 |
| flwang | here you are | 03:38 |
| flwang | no change for the client side | 03:38 |
| nikhil | darn, I swear I had it open | 03:39 |
| nikhil | but couldn't find it under your owner list :( | 03:40 |
| nikhil | flwang: I see, so this was the old style lite-spec. I wasn't aware of that. I just glanced at the code last night when you linked it here.. | 03:41 |
| flwang | nikhil: yep, it's the old lite-spec | 03:41 |
| flwang | anything else i need to do? | 03:42 |
| nikhil | flwang: ah, it's Jake's so doesn't apprear in your filters. ok. | 03:43 |
| flwang | based on the new style, seems we still need to submit a spec for this, right? | 03:43 |
| flwang | nikhil: yes | 03:43 |
| nikhil | flwang: no, we agreed to not ask the ones that were approved | 03:43 |
| flwang | ok, cool | 03:43 |
| nikhil | I can move that stuff if needed, but nothing from your end | 03:43 |
| flwang | awesome | 03:44 |
| nikhil | flwang: mind me asking if you are using locations like that? the backend stores are being passively hidden behind locations? | 03:44 |
| flwang | nikhil: now, we're not using cell | 03:46 |
| flwang | and we're only using ceph for now, no swift in our env | 03:46 |
| nikhil | flwang: gotcha. so you're mostly proxying stuff for Jake? | 03:47 |
| nikhil | flwang: the reason I ask is I'm very skeptical when people talk about multiple locations and copy-on-write being done w/o glance knowledge | 03:48 |
| flwang | nikhil: yep, i think it's a good user case | 03:48 |
| flwang | and given community is moving to cell very much | 03:48 |
| nikhil | flwang: and a bit scared about the way it's done too. there's no gurantee to the data consistency and makes it ephemeral -- if you see what I'm saying | 03:48 |
| flwang | so i'm pretty sure this could be a very useful feature for those | 03:49 |
| nikhil | ok, useful for sure. how is important here.. anyway I'm reading the bug closely again | 03:49 |
| flwang | nikhil: it can improve the data transfer performance | 03:50 |
| flwang | what do you mean the data consistency? | 03:50 |
| nikhil | flwang: on the data transfer performance -- I've a book to write 😃 | 03:53 |
| flwang | nikhil: btw, except Glare, that's your focus in Newton for glance? | 03:53 |
| nikhil | I mean it | 03:53 |
| flwang | you can answer my above question later :) | 03:54 |
| nikhil | what is _that_? | 03:54 |
| flwang | s/that/what | 03:54 |
| flwang | sorry | 03:54 |
| nikhil | I am not supposed to have a focus so I don't have a focus | 03:54 |
| flwang | wahaha | 03:55 |
| nikhil | I am supposed to enable people to have a focus | 03:55 |
| flwang | ok, then enable me first :) | 03:55 |
| nikhil | flwang: I searched for it and "wahaha" is some song 😃 | 03:55 |
| flwang | i will translate 'wahaha' with 'lol' | 03:56 |
| nikhil | flwang: are you aware of the location discussions we have been having? | 03:56 |
| flwang | nikhil: not really, but i suppose it's related to security which mike and flaper87 and i discussed a lot | 03:57 |
| flwang | nikhil: is there is an etherpad/link? | 03:57 |
| nikhil | flwang: there's no link | 03:59 |
| flwang | O:-) | 04:00 |
| nikhil | flwang: can you help me understand the "ceph clone abilities" and "each instance booting will re-upload image to ceph" ? | 04:00 |
| nikhil | I am asking as people told me that you guys use ceph | 04:00 |
| nikhil | feel free to say no | 04:00 |
| *** buzztroll has joined #openstack-glance | 04:02 | |
| flwang | i think the 'ceph clone abilities' Jake was talking about is the CoW of ceph | 04:03 |
| flwang | to be more accurate | 04:04 |
| flwang | ceph will create a copy of the image(in ceph is a volume), the copy/clone is most like a link | 04:04 |
| flwang | that said, ceph is not really creating a new copy, but a link | 04:04 |
| flwang | so it could be very fast | 04:05 |
| *** buzztroll has quit IRC | 04:05 | |
| nikhil | I see, I still need to wrap my head around what exactly is happening. is this during snapshot or is this COW done live while the instance is running or is the image directly used as volume by the instance while it's in ceph.. 😃 | 04:07 |
| flwang | http://docs.ceph.com/docs/master/rbd/rbd-snapshot/#layering | 04:08 |
| nikhil | flwang: thanks for the link 😃 it also means I take one more day to internalize everything! | 04:09 |
| flwang | based on my understanding, it's a 'clone' based on a snapshot | 04:09 |
| flwang | nikhil: sorry, i think a doc link can be more clear since i'm not really confident to be a storage expert :) | 04:10 |
| nikhil | flwang: makes sense. It would be nice to get these answers from Jake too unless you know him personally and are proxying 😃 | 04:10 |
| nikhil | flwang: np sire! the way I figured at the summit was bunch of people are hacking around glance as ceph+glance can be bad combination for such redundant uploads | 04:11 |
| nikhil | flwang: so, let me ask the same questions on Jake so that he can answer what they are doing in their deployment precisely and we both are not guessing it. but thanks for the link and I will read it. | 04:11 |
| nikhil | flwang: btw, I was curious why this can't be done outside of the server? | 04:12 |
| flwang | outside glance? | 04:12 |
| nikhil | flwang: to be more precise you add a config here (line 29) https://review.openstack.org/#/c/268865/6/glance/common/location_strategy/metadata_weight.py | 04:13 |
| nikhil | flwang: I am curious if that is meant to standardize the metadata key for locations ? | 04:13 |
| nikhil | flwang: why can't each deployer specify their metadata keys as they wish | 04:14 |
| nikhil | flwang: and the computation of the location weights are done on the client side | 04:14 |
| nikhil | if the locations are known then we can safely assume operator have exposed them already | 04:15 |
| flwang | nikhil: then you have to define 'client' | 04:16 |
| nikhil | flwang: for now, I am assuming the virt driver is the client | 04:17 |
| flwang | generally, user just want to use horizon or nova cli to create a new instance | 04:17 |
| flwang | instead of a private python script to talk with nova client to do it. i think that that's the normal case | 04:17 |
| *** sdake has joined #openstack-glance | 04:18 | |
| nikhil | flwang: right but the proposal is for instance to know the location to boot from. so they are just asking for cell level computes to know of this information and the locations are opaque from user's perspective. | 04:19 |
| nikhil | here -> "If this is made possible, each cell can return the closest location for each glance image, and also do copy-on-write for local stores like RBD." | 04:19 |
| nikhil | I am getting more and more convinced that this should live in the RBD (libvirt) driver if they want it | 04:20 |
| flwang | yep, so you mean this could be done in nova? | 04:20 |
| nikhil | flwang: exactly | 04:20 |
| flwang | or more lower level? | 04:20 |
| flwang | not only related to RBD for this case | 04:21 |
| flwang | if you have several different backend | 04:21 |
| nikhil | flwang: because all the assumptions are that they want to use ceph or they want to use swift+ceph etc. It looks like a corner case scenario for specific virt drivers they are using - if that made sense. | 04:21 |
| flwang | this feature can help get the one you want | 04:21 |
| nikhil | sure, but how many store type do you think you would want to optimize this way | 04:22 |
| nikhil | there's one more complexity to this use case | 04:22 |
| flwang | nikhil: at least 2 i think | 04:22 |
| nikhil | how are these images getting sync | 04:22 |
| nikhil | flwang: which ones? | 04:22 |
| flwang | nikhil: that's the question for all the multi locations scenarios | 04:22 |
| nikhil | correct | 04:23 |
| flwang | nikhil: not only for this | 04:23 |
| nikhil | and it'd be left for the operator to determine judiciously if they need to use multi locations | 04:23 |
| nikhil | So, the way I look at it is: | 04:24 |
| nikhil | multiple locations are an advanced configuration scenarios | 04:24 |
| nikhil | more so designed to optimize data pipe in specific cases | 04:24 |
| nikhil | like disaster recovery | 04:24 |
| nikhil | (includes data replication, data preservation, etc) | 04:24 |
| *** buzztroll has joined #openstack-glance | 04:25 | |
| nikhil | and for operator defined local storages | 04:25 |
| nikhil | for example | 04:25 |
| *** btully has quit IRC | 04:26 | |
| nikhil | for each compute node in a cell, one dedicated glance-api is deployed for faster boot times | 04:26 |
| nikhil | what stores are being used then? | 04:26 |
| nikhil | swift, ceph, filesystem? | 04:26 |
| nikhil | is glance cache going to be the faster way for boot | 04:26 |
| nikhil | or is the operator going to run some off the band job to pull data from safer storage like swift to local filesystem store | 04:27 |
| nikhil | or in Jake's they use ceph | 04:27 |
| nikhil | in each of these cases, operator are going into long length to optimize their data transfer and know what virt drivers they want to use | 04:27 |
| flwang | Jake is the glance operator in this case ;) | 04:28 |
| nikhil | so, I want to safely assume that they will know where the location weight is going to be used to pull from closer location | 04:28 |
| nikhil | where == which part of nova | 04:28 |
| flwang | nikhil: hmm... | 04:28 |
| flwang | nikhil: i won't say it's not correct | 04:29 |
| flwang | since both glance and nova are the part of the booting process | 04:29 |
| flwang | technically, you do try to get the best image at either side | 04:30 |
| *** btully has joined #openstack-glance | 04:30 | |
| flwang | but personally, i think the location strategy is used to for this kind of scenario | 04:30 |
| flwang | that said, i can't see why we can't do this in glance | 04:30 |
| flwang | does that make any sense? | 04:30 |
| *** buzztroll has quit IRC | 04:31 | |
| nikhil | flwang: to be fair, multiple locations were introduced in glance for faster boot times to help _operators_ (as per the original use case) | 04:31 |
| nikhil | flwang: but the management of the locations for that wasn't supposed to be part of glance | 04:32 |
| nikhil | flwang: my problem is that we do not want to assume all locations are correct | 04:32 |
| nikhil | flwang: because they can be out of sync | 04:32 |
| nikhil | flwang: and without knowledge of glance | 04:32 |
| nikhil | flwang: so any assumption that leads for glance to believe all locations are equal in terms of data consistency is false | 04:33 |
| nikhil | and I proposed that to the team during the summit and got no push back (yet) | 04:33 |
| flwang | sorry? what do you mean 'got no push back'? | 04:34 |
| nikhil | moreover any location that the user/operator sets without the data going through glance may have different checksum | 04:34 |
| nikhil | flwang: sorry, actually scratch that. I just meant the point was clear and people seemed to agree and we ran out of time. | 04:35 |
| nikhil | flwang: a bit irrelevant to our discussion but I was trying to be explicit about it | 04:35 |
| nikhil | I hope my points are clear. | 04:36 |
| flwang | nikhil: i see. but as i said above, the images sync is not related to this feature, IMHO | 04:37 |
| flwang | since it's general issue for multi location | 04:37 |
| *** buzztroll has joined #openstack-glance | 04:37 | |
| flwang | which flaper87 and i have discussed this in a security bug | 04:37 |
| flwang | and yes, i agree it would be nice if glance can check the checksum for each locations | 04:38 |
| flwang | but it's hard to do that for some locations, like http | 04:38 |
| flwang | you have to download it :( | 04:38 |
| nikhil | 😃 | 04:38 |
| nikhil | flwang: out of sync is related to the bug. how can you determine the weight of a location if the different locations are not guranteed to be giving same data from the point of view of glance. | 04:40 |
| nikhil | flwang: where as the locations are being updated by nova/virt drivers so that information is local to them. they are in a much better position to say what's correct. | 04:41 |
| flwang | nikhil: no, my point is out of sync is a general issue, even without this feature, we do still have the issue, right? | 04:41 |
| flwang | nikhil: i can work on the multi location sync issue if that's your concern :) | 04:42 |
| flwang | seems you're trying to figure out a focus for me in newton ;D | 04:42 |
| nikhil | flwang: haha, indeed 😃 | 04:44 |
| nikhil | flwang: but we've ton of work ahead of us man! | 04:44 |
| nikhil | flwang: we need your help with quota stuff out there in the delimiter library. | 04:45 |
| flwang | nikhil: so what's the dirty job i can help to make you happy? | 04:45 |
| nikhil | LOL | 04:45 |
| flwang | nikhil: ah, i see. i'm happy to be counted | 04:45 |
| nikhil | dirty job, why would you want a dirty job :P | 04:45 |
| nikhil | flwang: LOL. you're always in. just disappeared after a bit | 04:45 |
| flwang | nikhil: because dirty job is always can make boss happy :) | 04:46 |
| nikhil | flwang: nina's case was super important so we had to figure out the right way to do things! | 04:46 |
| flwang | nikhil: i just need to get some excited job, TBH | 04:46 |
| nikhil | flwang: LOL, who's the boss here! we're all serving OpenStack 😃 | 04:46 |
| flwang | nikhil: so what's the conclusion for glance quota? | 04:47 |
| flwang | do nested quota since the first step? | 04:47 |
| *** sdake has quit IRC | 04:47 | |
| flwang | or go for the delimiter and then support it in glance? | 04:47 |
| *** sdake has joined #openstack-glance | 04:48 | |
| nikhil | flwang: no, what we need is a way for all the quota types to exist from step#1 so that we can avoid migrations | 04:48 |
| nikhil | I was able to identify 4 types of quotas in openstack as of today | 04:48 |
| flwang | so nested quota from day1? | 04:49 |
| *** buzztroll has quit IRC | 04:49 | |
| nikhil | not necessarily | 04:49 |
| nikhil | if we adopt delimiter (when ready), it will allow operator to choose simple or nested or floating etc quotas for their tables | 04:50 |
| nikhil | so if we adopt the library, operator can choose which type they want to have in their deployment | 04:50 |
| flwang | i see, but the question is when it could be ready | 04:51 |
| flwang | and what should we do now? especially in newton for glance | 04:51 |
| nikhil | and for that we all need to pitch in 😃 | 04:51 |
| nikhil | ok, for newton in glance we still have a ton of work | 04:51 |
| nikhil | depends on what you are looking for | 04:51 |
| nikhil | 1. if you are interested in quota, delimiter is the most exciting path. small team lots to do, get to know different cases. | 04:52 |
| nikhil | 2. there's import that stuart is wokring on | 04:52 |
| nikhil | 3. mike is working on nova | 04:52 |
| nikhil | 4. flavio is working on microversions | 04:52 |
| nikhil | 5. brian is working on docs and rolling upgrades | 04:53 |
| nikhil | 6. hemanth is working on config improvements | 04:53 |
| nikhil | and then we've glare! | 04:53 |
| flwang | nikhil: cool, seems quota is the one i can take | 04:54 |
| nikhil | oh yes, there's community/inherited images support too | 04:54 |
| nikhil | #8 😃 | 04:54 |
| flwang | Community image visibility Su Zhang | 04:55 |
| nikhil | we need reviews on all of these, help (depending on how you can collaborate) on all of these! | 04:55 |
| nikhil | timothy from symantec has taken over that spec | 04:55 |
| flwang | ok, cool | 04:55 |
| nikhil | we are going to discuss that tomorrow during the meeting | 04:55 |
| flwang | nikhil: so now the meeting time is still 14:00 UTC, right? | 04:56 |
| nikhil | so, there are ton of things to do and people won't be able to get them done themselves in one cycle! | 04:56 |
| nikhil | flwang: yes | 04:56 |
| flwang | :( | 04:56 |
| flwang | 2:00am for NZ | 04:56 |
| nikhil | there is some metadefs stuff that we need to fix | 04:56 |
| flwang | so as for quota, we have decide to brace delimiter, so that means until it's ready, we basically can't do anything in glance, right? | 04:57 |
| *** tshefi has joined #openstack-glance | 04:57 | |
| nikhil | flwang: what does your work timing look like? | 04:57 |
| nikhil | flwang: yeah :/ | 04:57 |
| flwang | nikhil: ok, i see | 04:57 |
| nikhil | flwang: are you against working on non-incubated projects? | 04:57 |
| flwang | no | 04:58 |
| nikhil | ok | 04:58 |
| flwang | i just want to figure out anything i can do in glance | 04:58 |
| nikhil | flwang: just curious about your interest in that library | 04:58 |
| nikhil | flwang: my feeling is that we need some expertise in that lib for us to adopt in glance | 04:58 |
| flwang | nikhil: i'm interested anything can make glance better :) | 04:58 |
| nikhil | I can do that if no one else is interested | 04:59 |
| nikhil | but then I'm not supposed to do that and let people focus on stuff! :) | 04:59 |
| nikhil | flwang: haha, that's a great spirit 😃 | 04:59 |
| flwang | ok, feel free take it if you want | 04:59 |
| nikhil | LOL | 04:59 |
| flwang | then i will find another part i can grab | 04:59 |
| flwang | sorry, man,i have to run to pick up my boy | 05:00 |
| nikhil | flwang: I was saying that you'd else I will have to 😃 | 05:00 |
| flwang | i will catch up with you later as for the focus | 05:00 |
| nikhil | flwang: ok, np. take care. I just meant feel free to pick up what you want/need. we can figure out the logistics later. | 05:00 |
| *** pdeore has joined #openstack-glance | 05:01 | |
| nikhil | cioa | 05:01 |
| flwang | quota is my plate now, see you later | 05:01 |
| *** gb21 has quit IRC | 05:01 | |
| *** janonymous has joined #openstack-glance | 05:38 | |
| *** buzztroll has joined #openstack-glance | 05:48 | |
| janonymous | Changelog in https://github.com/openstack/glance/blob/master/MANIFEST.in#L1 , https://github.com/openstack/glance/blob/master/MANIFEST.in#L8 means same right ? | 05:48 |
| *** mosulica has joined #openstack-glance | 06:00 | |
| *** openstackgerrit has quit IRC | 06:03 | |
| *** openstackgerrit has joined #openstack-glance | 06:03 | |
| *** mingdang1 has joined #openstack-glance | 06:10 | |
| *** e0ne has joined #openstack-glance | 06:32 | |
| *** sdake has quit IRC | 06:32 | |
| *** e0ne has quit IRC | 06:43 | |
| *** tesseract has joined #openstack-glance | 06:44 | |
| *** tesseract is now known as Guest21288 | 06:45 | |
| *** mingdang1 has quit IRC | 07:01 | |
| *** btully has quit IRC | 07:34 | |
| *** mvk_ has quit IRC | 07:44 | |
| *** dmk0202 has joined #openstack-glance | 07:57 | |
| *** btully has joined #openstack-glance | 07:58 | |
| *** btully has quit IRC | 08:03 | |
| *** dmk0202 has quit IRC | 08:03 | |
| *** e0ne has joined #openstack-glance | 08:06 | |
| *** dmk0202 has joined #openstack-glance | 08:11 | |
| *** dshakhray has joined #openstack-glance | 08:15 | |
| *** jistr has joined #openstack-glance | 08:34 | |
| *** mvk_ has joined #openstack-glance | 08:39 | |
| *** buzztroll has quit IRC | 08:45 | |
| *** e0ne has quit IRC | 08:46 | |
| *** e0ne has joined #openstack-glance | 08:49 | |
| *** sdake has joined #openstack-glance | 09:24 | |
| *** buzztroll has joined #openstack-glance | 09:41 | |
| *** buzztroll has quit IRC | 09:46 | |
| *** e0ne has quit IRC | 09:48 | |
| *** e0ne has joined #openstack-glance | 09:49 | |
| *** e0ne has quit IRC | 10:05 | |
| *** groen692 has joined #openstack-glance | 10:14 | |
| *** sdake has quit IRC | 10:29 | |
| *** MattMan has quit IRC | 10:39 | |
| *** MattMan has joined #openstack-glance | 10:40 | |
| *** groen692 has quit IRC | 10:41 | |
| *** e0ne has joined #openstack-glance | 10:51 | |
| *** links has joined #openstack-glance | 11:05 | |
| *** buzztroll has joined #openstack-glance | 11:30 | |
| *** buzztroll has quit IRC | 11:34 | |
| *** pdeore has quit IRC | 11:40 | |
| *** smatzek has joined #openstack-glance | 11:45 | |
| *** someara2 has joined #openstack-glance | 11:53 | |
| *** links has quit IRC | 11:53 | |
| *** someara2_ has joined #openstack-glance | 11:57 | |
| *** someara2 has quit IRC | 12:01 | |
| *** ekarlso has quit IRC | 12:06 | |
| *** ekarlso has joined #openstack-glance | 12:06 | |
| *** MVenesio has joined #openstack-glance | 12:06 | |
| *** ducttape_ has joined #openstack-glance | 12:07 | |
| openstackgerrit | Bhagyashri Shewale proposed openstack/python-glanceclient: WIP:Fix 'UnicodeEncodeError' for unicode values in url https://review.openstack.org/312915 | 12:08 |
| *** btully has joined #openstack-glance | 12:10 | |
| *** mingdang1 has joined #openstack-glance | 12:20 | |
| *** ducttape_ has quit IRC | 12:24 | |
| *** burgerk has joined #openstack-glance | 12:33 | |
| *** someara2_ has left #openstack-glance | 12:33 | |
| *** ninag has joined #openstack-glance | 12:37 | |
| *** bapalm has joined #openstack-glance | 12:54 | |
| *** tsymanczyk has joined #openstack-glance | 13:02 | |
| *** ducttape_ has joined #openstack-glance | 13:11 | |
| *** buzztroll has joined #openstack-glance | 13:18 | |
| *** buzztroll has quit IRC | 13:22 | |
| *** ducttape_ has quit IRC | 13:36 | |
| *** mtanino has joined #openstack-glance | 13:41 | |
| *** ducttape_ has joined #openstack-glance | 13:44 | |
| *** sigmavirus24_awa is now known as sigmavirus24 | 13:45 | |
| *** openstackgerrit has quit IRC | 13:47 | |
| *** openstackgerrit has joined #openstack-glance | 13:47 | |
| *** tshefi has quit IRC | 13:50 | |
| *** ametts has joined #openstack-glance | 13:55 | |
| openstackgerrit | Stuart McLaren proposed openstack/glance: [WIP] Add image import schema / schema validation https://review.openstack.org/312972 | 13:58 |
| nikhil | Courtesy meeting reminder: ativelkov, cpallares, flaper87, flwang1, hemanthm, jokke_, kragniz, lakshmiS, mclaren, mfedosin, nikhil_k, Nikolay_St, Olena, pennerc, rosmaita, sigmavirus24, sabari, TravT, ajayaa, GB21, bpoulos, harshs, abhishek, bunting, dshakhray, wxy, dhellmann, kairat | 13:59 |
| *** MVenesio has quit IRC | 14:00 | |
| *** pushkaru has joined #openstack-glance | 14:03 | |
| *** links has joined #openstack-glance | 14:07 | |
| *** burgerk has quit IRC | 14:12 | |
| *** links has quit IRC | 14:16 | |
| *** MVenesio has joined #openstack-glance | 14:17 | |
| *** jaypipes has joined #openstack-glance | 14:25 | |
| *** mingdang1 has quit IRC | 14:26 | |
| *** krotscheck has quit IRC | 14:27 | |
| *** jistr has quit IRC | 14:29 | |
| *** krotscheck has joined #openstack-glance | 14:33 | |
| *** djkonro has joined #openstack-glance | 14:44 | |
| *** cdelatte has joined #openstack-glance | 14:48 | |
| *** sdake has joined #openstack-glance | 14:50 | |
| *** jistr has joined #openstack-glance | 14:51 | |
| *** ayoung has quit IRC | 14:59 | |
| *** mclaren has joined #openstack-glance | 15:01 | |
| nikhil | rosmaita: did you have more input on the location strategy stuff? | 15:01 |
| nikhil | sorry we ran out of time there | 15:01 |
| rosmaita | i'll have to look at that patch | 15:01 |
| nikhil | ok | 15:01 |
| rosmaita | i thought zhi had designed it so htat you could plug in your own strategy | 15:01 |
| nikhil | mclaren: o/ | 15:02 |
| rosmaita | "you" == deployer | 15:02 |
| *** burgerk has joined #openstack-glance | 15:02 | |
| mclaren | hey | 15:03 |
| nikhil | we're just chatting on locations | 15:03 |
| nikhil | rosmaita: yeah, you're right! it's in glance/common/location_strategy | 15:04 |
| nikhil | I've never had a chance to touch that code | 15:04 |
| nikhil | who uses it!? | 15:04 |
| rosmaita | i think there are 2 options by default | 15:04 |
| rosmaita | zhi had a use case when he was at ibm | 15:04 |
| rosmaita | not by default | 15:05 |
| rosmaita | i mean 2 strategies exist that can be applied | 15:05 |
| rosmaita | (but that's by memory, i don'thave the code in front of me) | 15:05 |
| *** buzztroll has joined #openstack-glance | 15:06 | |
| rosmaita | nikhil: if you want, i'll put a minus on the spec saying please explain why the curent stuff won't work for you | 15:06 |
| nikhil | rosmaita: yes please! | 15:06 |
| rosmaita | nikhil: do you have the link handy? | 15:06 |
| nikhil | rosmaita: I read through it and memory is coming back. they wanted to implement os-brick like stuff for glance | 15:07 |
| nikhil | rosmaita: that makes me uncomfortable. I do not know if we've tests around these! | 15:07 |
| rosmaita | TIL there is an os-brick project | 15:07 |
| nikhil | rosmaita: here's the link https://github.com/openstack/glance/tree/master/glance/common/location_strategy | 15:07 |
| rosmaita | nikhil: thanks | 15:08 |
| rosmaita | nikhil: i meant to the bug, but i see it on the meeting agenda | 15:08 |
| nikhil | cool | 15:09 |
| nikhil | mclaren: did you want to discuss import stuff now or just prefer it on the review? | 15:09 |
| mclaren | on the review is good I think | 15:10 |
| nikhil | cool | 15:10 |
| *** buzztroll has quit IRC | 15:10 | |
| *** TravT has quit IRC | 15:19 | |
| *** TravT has joined #openstack-glance | 15:20 | |
| *** julim has quit IRC | 15:22 | |
| *** julim has joined #openstack-glance | 15:24 | |
| *** ayoung has joined #openstack-glance | 15:32 | |
| *** ducttape_ has quit IRC | 15:37 | |
| *** dmk0202 has quit IRC | 15:38 | |
| *** djkonro has quit IRC | 15:45 | |
| openstackgerrit | Niall Bunting proposed openstack/glance: Fix glance-cache-prefetcher https://review.openstack.org/275322 | 15:45 |
| *** ayoung has quit IRC | 15:46 | |
| *** Guest21288 has quit IRC | 15:46 | |
| *** ducttape_ has joined #openstack-glance | 15:48 | |
| *** buzztroll has joined #openstack-glance | 15:55 | |
| *** davideagnello has joined #openstack-glance | 16:05 | |
| *** ducttape_ has quit IRC | 16:06 | |
| *** ducttape_ has joined #openstack-glance | 16:11 | |
| *** ducttape_ has quit IRC | 16:12 | |
| *** sdake_ has joined #openstack-glance | 16:29 | |
| *** sdake has quit IRC | 16:31 | |
| openstackgerrit | Ian Cordasco proposed openstack/glance: Remove unnecessary executable privilge of unit test file https://review.openstack.org/310991 | 16:33 |
| *** djkonro has joined #openstack-glance | 16:35 | |
| bunting | If anyone has any free time to review, this is a patch for glance_store swift functional tests. https://review.openstack.org/#/c/302374/ | 16:40 |
| *** ducttape_ has joined #openstack-glance | 16:46 | |
| *** djkonro has quit IRC | 16:58 | |
| *** mosulica has quit IRC | 16:58 | |
| *** sdake_ has quit IRC | 17:00 | |
| nikhil | bunting: thanks, will take a look in a bit | 17:04 |
| *** mine0901 has joined #openstack-glance | 17:11 | |
| *** jistr has quit IRC | 17:17 | |
| *** cebreidian has quit IRC | 17:24 | |
| *** nikhil has quit IRC | 17:31 | |
| *** nikhil has joined #openstack-glance | 17:32 | |
| *** ayoung has joined #openstack-glance | 17:33 | |
| rosmaita | nikhil: left a comment on bug 1528453 | 17:44 |
| openstack | bug 1528453 in Glance "Provide a ranking mechanism for glance-api to order locations" [Wishlist,Triaged] https://launchpad.net/bugs/1528453 - Assigned to Jake Yip (waipengyip) | 17:44 |
| *** ducttape_ has quit IRC | 17:45 | |
| nikhil | rosmaita: ack, thanks. looking. | 17:46 |
| *** tsymanczyk has quit IRC | 17:46 | |
| *** dshakhray has quit IRC | 17:50 | |
| *** TravT has quit IRC | 17:50 | |
| *** tsymanczyk has joined #openstack-glance | 17:51 | |
| *** tsymanczyk is now known as Guest29252 | 17:51 | |
| *** itlinux has joined #openstack-glance | 17:52 | |
| *** pushkaru has quit IRC | 18:10 | |
| *** burgerk_ has joined #openstack-glance | 18:18 | |
| *** burgerk has quit IRC | 18:22 | |
| *** ducttape_ has joined #openstack-glance | 18:27 | |
| *** ninag has quit IRC | 18:30 | |
| *** ninag has joined #openstack-glance | 18:30 | |
| *** e0ne has quit IRC | 18:32 | |
| *** ninag_ has joined #openstack-glance | 18:32 | |
| *** ninag_ has quit IRC | 18:32 | |
| *** ninag_ has joined #openstack-glance | 18:33 | |
| *** ninag has quit IRC | 18:35 | |
| *** ninag_ has quit IRC | 18:38 | |
| *** Guest29252 is now known as tsymanczyk | 18:42 | |
| nikhil | thanks rosmaita for the excellent research! | 18:45 |
| nikhil | I added an additional comment relating to my comment on the bug. | 18:45 |
| *** sdake has joined #openstack-glance | 18:48 | |
| *** btully has quit IRC | 18:49 | |
| *** ninag has joined #openstack-glance | 18:53 | |
| *** mvk_ has quit IRC | 18:54 | |
| *** pushkaru has joined #openstack-glance | 18:54 | |
| *** sdake_ has joined #openstack-glance | 18:55 | |
| *** ninag has quit IRC | 18:56 | |
| *** sdake has quit IRC | 18:57 | |
| openstackgerrit | Brian Rosmaita proposed openstack/glance-specs: Update image import refactor spec https://review.openstack.org/311871 | 18:58 |
| tsymanczyk | i just noticed the "core reviewers" section in the community image spec came copy paste from when kragniz owned it. it currently lists brian-rosmaita and stuart-mclaren . should i change this? | 19:07 |
| kragniz | if they're not the people going to review it, probably | 19:09 |
| kragniz | I'm sure rosmaita will remain | 19:09 |
| tsymanczyk | he seems interested, yeah. i guess remove stuart for now and wait to be told who to add? | 19:10 |
| kragniz | sure, do that | 19:10 |
| nikhil | tsymanczyk: you can add my lp id "nikhil-komawar" | 19:11 |
| tsymanczyk | cool, done. | 19:12 |
| *** smatzek has quit IRC | 19:15 | |
| *** pushkaru has quit IRC | 19:15 | |
| -openstackstatus- NOTICE: Gerrit is restarting to address performance issues related to a suspected memory leak | 19:21 | |
| nikhil | bunting: around by chance? | 19:27 |
| nikhil | I wanted to discuss https://review.openstack.org/#/c/275322/ | 19:27 |
| *** e0ne has joined #openstack-glance | 19:27 | |
| nikhil | you say in the commit message creds are not getting passed and in the bug that it works for keystone v1 (somethign as per openstack community doesn't exist) | 19:28 |
| *** ninag has joined #openstack-glance | 19:29 | |
| *** mvk_ has joined #openstack-glance | 19:32 | |
| nikhil | I added comment to review, let's discuss there | 19:33 |
| *** ninag has quit IRC | 19:36 | |
| *** ninag has joined #openstack-glance | 19:37 | |
| *** btully has joined #openstack-glance | 19:40 | |
| *** avarner has joined #openstack-glance | 19:44 | |
| bunting | nikhil: I am | 19:44 |
| *** avarner is now known as MistySkies | 19:45 | |
| *** MistySkies has left #openstack-glance | 19:45 | |
| *** flwang1 has joined #openstack-glance | 19:45 | |
| flwang1 | rosmaita: ping re member id | 19:45 |
| nikhil | bunting: o/ | 19:46 |
| flwang1 | nikhil: can you remind me why we don't want to verify the member id by talking with keystone? | 19:46 |
| rosmaita | flwang1: just because it takes time | 19:46 |
| rosmaita | at least that's my memory | 19:46 |
| nikhil | verify as in verify if it exists? | 19:46 |
| flwang1 | rosmaita: i think the case is only about image share, right? | 19:47 |
| nikhil | I think the user who wants to share that image should be careful about that | 19:47 |
| rosmaita | flwang1: probably, i can't think of anything else atm | 19:47 |
| flwang1 | rosmaita: and given image share is not a very frequent action, i can't be convinced to worry about the performance | 19:47 |
| rosmaita | nikhil: the problem is if you make a typo, you get no feedback, your sharing just doesn't work | 19:47 |
| rosmaita | flwang1: there was a summit session about keystone adding a call for this | 19:48 |
| rosmaita | some other project needs it, too | 19:48 |
| nikhil | rosmaita: well :) we'd check if member is of type uuid at max | 19:48 |
| rosmaita | nikhil: not guaranteed to be a uuid | 19:48 |
| rosmaita | tenant_id is just a string | 19:48 |
| nikhil | ah | 19:49 |
| rosmaita | maybe only alphanumeric, not sure about that, though | 19:49 |
| nikhil | I think the DB should take care of that | 19:49 |
| nikhil | "the column type" | 19:49 |
| flwang1 | rosmaita: nikhil: i would like propose verify member id with keystone | 19:50 |
| nikhil | isn't it insecure to make a call to keystone to see if a member id exists? | 19:50 |
| rosmaita | still, if you're off by a digit, the person you shared with can't use the image, and you're left scratching your head | 19:50 |
| nikhil | rosmaita: why would someone want to expose that info | 19:50 |
| flwang1 | would you mind me submitting a spec for this? | 19:50 |
| nikhil | rosmaita: for example, we return a "404" for an image that doesn't belong to a tenant | 19:51 |
| tsymanczyk | insecure in what regard? | 19:51 |
| rosmaita | flwang1: someone has a spec up about making image member_id alphanumeric only | 19:51 |
| nikhil | we do not return 403 | 19:51 |
| nikhil | flwang1: was that directed to me? | 19:51 |
| flwang1 | rosmaita: but that's not the solution | 19:51 |
| flwang1 | to verify if it's a 'valid' tenant id | 19:51 |
| rosmaita | no, but we should not do both | 19:52 |
| bunting | nikhil: It seems that part of the commit message refers to ps1 | 19:52 |
| nikhil | I don't think so we need to validate with keystone | 19:52 |
| nikhil | look at my 404 vs 403 analogy above | 19:52 |
| nikhil | to put it other way: I'm not convinced that we'd | 19:53 |
| rosmaita | flwang1: gimme a sec to find the info about htat summut session | 19:53 |
| nikhil | the problem of one off by digit exists even in case of GET image/id | 19:53 |
| rosmaita | nikhil: suppose i share with tenant_id "1234l434" | 19:53 |
| flwang1 | rosmaita: cool,thx | 19:54 |
| rosmaita | it looks to me like i shared with 12341434 | 19:54 |
| rosmaita | but that guy contacts me to say, where is the image? | 19:54 |
| rosmaita | then i have to fix it | 19:54 |
| rosmaita | or, i tell him i shared it, he still can't see it, and gives up | 19:54 |
| nikhil | bunting: gotcha, because the code did not seem to pass any cred info. I'm assuming you are going to update stuff ? | 19:54 |
| *** e0ne has quit IRC | 19:55 | |
| nikhil | rosmaita: sorry, but that looks like a edge case :) | 19:55 |
| rosmaita | nikhil: maybe | 19:56 |
| nikhil | we're adding validation logic for each call for an edge case? | 19:56 |
| rosmaita | hey, don't blame me, it's flwang1 's idea! | 19:56 |
| nikhil | this is similar to the string problem with my keyboard on my phone | 19:56 |
| nikhil | It might be worth thinking of a validation list on top of glance | 19:57 |
| nikhil | like a spell corrector for keyboard | 19:57 |
| nikhil | rather than annoying auto correct? | 19:57 |
| nikhil | in this case the auto correct is super expensive | 19:57 |
| bunting | nikhil: Doing it now | 19:58 |
| rosmaita | yeah, because even if you know the thing is a valid tenant_id, you still *don't* know if it's the right one | 19:58 |
| nikhil | yeah | 19:58 |
| nikhil | what if both the tenants exist above? | 19:59 |
| openstackgerrit | Niall Bunting proposed openstack/glance: Fix glance-cache-prefetcher https://review.openstack.org/275322 | 19:59 |
| nikhil | 12341434 and 1234l434 | 19:59 |
| nikhil | ? | 19:59 |
| rosmaita | ok, you have convinced me that this isn't so bad | 20:00 |
| rosmaita | flwang1: Project ID validation with Keystone - https://etherpad.openstack.org/p/newton-nova-keystone | 20:00 |
| flwang1 | rosmaita: thanks | 20:00 |
| flwang1 | nikhil: if both exist, user may accept a wrong image | 20:01 |
| *** smatzek has joined #openstack-glance | 20:01 | |
| rosmaita | flwang1: yes, but even if we validate the tenant_id, that will still happen | 20:02 |
| rosmaita | because you're not proposing to return "You are proposing to share with tenant 123345 whose name is xxxxx and who lives at xxxxxx and whose credit card # is xxxxxxx" | 20:02 |
| rosmaita | :) | 20:02 |
| flwang1 | correct | 20:02 |
| flwang1 | ok, i'm going to withdraw my member verification idea again :D | 20:03 |
| rosmaita | no, stick with it! | 20:03 |
| flwang1 | rosmaita: haha | 20:04 |
| rosmaita | i guess we need to figure out how much of an edge case this is | 20:04 |
| flwang1 | rosmaita: agree | 20:04 |
| nikhil | agree | 20:04 |
| flwang1 | that's why i said maybe we need a spec | 20:04 |
| rosmaita | because it's possible to put usernames in there now and wonder why sharing isn't working | 20:04 |
| flwang1 | so that we have a place to track and get more feedback from others | 20:05 |
| rosmaita | (that actually happened during testing in our cloud) | 20:05 |
| flwang1 | rosmaita: oh, yes, tenant name and tenant id | 20:05 |
| rosmaita | i agree that it's worth discussing | 20:05 |
| rosmaita | and i think you have a pretty good idea of the concerns people will have! | 20:06 |
| flwang1 | rosmaita: nikhil: so does it worth a spec or just propose it to our weekly meeting? | 20:06 |
| nikhil | makes sense on the spec | 20:06 |
| nikhil | flwang1: may be let's start with a lite-spec | 20:06 |
| nikhil | do propose your problem statement though | 20:07 |
| rosmaita | +1 on lite-spec, in a meeting we'll have the same conversation we just had, only the names will be different | 20:07 |
| nikhil | otherwise people will get confused like I did just now | 20:07 |
| nikhil | rosmaita is on a roll today | 20:07 |
| nikhil | names will be different was hilarious | 20:08 |
| flwang1 | nikhil: rosmaita: cool, thanks for the feedback | 20:09 |
| flwang1 | i can't remember how many times i propose to verify the member id :D | 20:09 |
| nikhil | flwang1: sorry to hear that! | 20:10 |
| *** ayoung has quit IRC | 20:10 | |
| nikhil | flwang1: it'd be easier to propose things | 20:10 |
| *** TravT has joined #openstack-glance | 20:10 | |
| flwang1 | nikhil: cool | 20:12 |
| flwang1 | running to office now | 20:12 |
| flwang1 | ttyl | 20:12 |
| nikhil | literally? | 20:12 |
| nikhil | ok ttyl | 20:12 |
| nikhil | :) | 20:12 |
| rosmaita | flwang1: have a good run | 20:12 |
| flwang1 | only 12km | 20:12 |
| flwang1 | basically no traffic | 20:12 |
| flwang1 | hopefully :D | 20:12 |
| nikhil | wow | 20:12 |
| nikhil | 7.5 miles | 20:13 |
| rosmaita | that's a serious run | 20:14 |
| nikhil | bunting: I optimistically gave a +2 https://review.openstack.org/#/c/275322/ as you just changed the commit message | 20:16 |
| *** buzztroll has quit IRC | 20:16 | |
| *** buzztroll has joined #openstack-glance | 20:16 | |
| nikhil | this is something that hemanthm and rosmaita might be using so will defer the other +2 to them | 20:16 |
| openstackgerrit | Timothy Symanczyk proposed openstack/glance-specs: Community image visibility BP https://review.openstack.org/271019 | 20:16 |
| openstackgerrit | Hemanth Makkapati proposed openstack/glance: Get rid of redundant config opts in scrubber https://review.openstack.org/313144 | 20:17 |
| *** flwang1 has quit IRC | 20:17 | |
| *** smatzek has quit IRC | 20:20 | |
| *** pushkaru has joined #openstack-glance | 20:24 | |
| -openstackstatus- NOTICE: Gerrit is restarting to revert incorrect changes to test result displays | 20:29 | |
| *** dmk0202 has joined #openstack-glance | 20:29 | |
| openstackgerrit | Timothy Symanczyk proposed openstack/glance-specs: Community image visibility BP https://review.openstack.org/271019 | 20:31 |
| *** cdelatte has quit IRC | 20:34 | |
| openstackgerrit | Hemanth Makkapati proposed openstack/glance: Improve help text of scrubber opts https://review.openstack.org/312720 | 20:35 |
| *** sdake has joined #openstack-glance | 20:35 | |
| *** sdake_ has quit IRC | 20:36 | |
| openstackgerrit | Merged openstack/glance: Remove unnecessary executable privilge of unit test file https://review.openstack.org/310991 | 20:50 |
| bunting | nikhil: Sound | 20:56 |
| *** rcernin has joined #openstack-glance | 21:05 | |
| *** smatzek has joined #openstack-glance | 21:13 | |
| *** MVenesio has quit IRC | 21:16 | |
| *** cdelatte has joined #openstack-glance | 21:24 | |
| *** mingdang1 has joined #openstack-glance | 21:26 | |
| *** pushkaru has quit IRC | 21:26 | |
| *** flwang1 has joined #openstack-glance | 21:31 | |
| *** ametts has quit IRC | 21:34 | |
| *** MVenesio has joined #openstack-glance | 21:42 | |
| *** MVenesio has quit IRC | 21:44 | |
| *** sdake has quit IRC | 21:50 | |
| *** sdake has joined #openstack-glance | 21:51 | |
| *** sdake has quit IRC | 21:56 | |
| *** ametts has joined #openstack-glance | 21:56 | |
| *** ayoung has joined #openstack-glance | 22:03 | |
| *** ninag has quit IRC | 22:05 | |
| *** ninag has joined #openstack-glance | 22:06 | |
| *** ducttape_ has quit IRC | 22:06 | |
| *** ninag has quit IRC | 22:10 | |
| *** sigmavirus24 is now known as sigmavirus24_awa | 22:14 | |
| *** cdelatte has quit IRC | 22:22 | |
| *** rcernin has quit IRC | 22:29 | |
| *** ninag has joined #openstack-glance | 22:35 | |
| *** ametts has quit IRC | 22:35 | |
| *** ninag has quit IRC | 22:39 | |
| *** TravT has quit IRC | 22:51 | |
| *** cdelatte has joined #openstack-glance | 22:51 | |
| *** david-lyle has quit IRC | 22:51 | |
| *** david-lyle has joined #openstack-glance | 22:53 | |
| *** ducttape_ has joined #openstack-glance | 22:54 | |
| openstackgerrit | Timothy Symanczyk proposed openstack/glance-specs: Community image visibility BP https://review.openstack.org/271019 | 22:55 |
| *** david-lyle has quit IRC | 22:58 | |
| *** davideag_ has joined #openstack-glance | 23:04 | |
| *** ayoung has quit IRC | 23:07 | |
| *** davideagnello has quit IRC | 23:07 | |
| *** GB21 has joined #openstack-glance | 23:10 | |
| *** jaypipes has quit IRC | 23:16 | |
| *** sgotliv has quit IRC | 23:21 | |
| *** mingdang1 has quit IRC | 23:27 | |
| *** GB21 has quit IRC | 23:28 | |
| *** mtanino has quit IRC | 23:30 | |
| *** mtanino has joined #openstack-glance | 23:30 | |
| *** krotscheck has quit IRC | 23:31 | |
| *** krotscheck has joined #openstack-glance | 23:31 | |
| *** GB21 has joined #openstack-glance | 23:32 | |
| *** cdelatte has quit IRC | 23:34 | |
| *** dmk0202 has quit IRC | 23:42 | |
| *** MVenesio has joined #openstack-glance | 23:45 | |
| *** davideag_ has quit IRC | 23:45 | |
| *** MVenesio has quit IRC | 23:50 | |
| *** ayoung has joined #openstack-glance | 23:53 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!