| *** openstackgerrit has joined #openstack-glance | 00:17 | |
| openstackgerrit | iain MacDonnell proposed openstack/python-glanceclient master: Embed validation data when adding location https://review.openstack.org/602794 | 00:17 |
|---|---|---|
| *** rosmaita has left #openstack-glance | 00:29 | |
| *** Liang__ has joined #openstack-glance | 00:59 | |
| *** imacdonn has quit IRC | 01:23 | |
| *** imacdonn has joined #openstack-glance | 01:24 | |
| *** mhen has quit IRC | 01:42 | |
| *** itlinux has quit IRC | 01:46 | |
| *** mhen has joined #openstack-glance | 01:46 | |
| *** MattMan_ has quit IRC | 01:47 | |
| *** MattMan_ has joined #openstack-glance | 01:47 | |
| Liang__ | smcginnis: sean, about https://review.openstack.org/#/c/612236/, as you said openstack client may don't have equivalent option with -vv. How do you think adding an option e.g. --include-stores? In this way, store id is just an specific info for glance client to show, lower impact with openstack client | 02:01 |
| *** udesale has joined #openstack-glance | 03:50 | |
| *** udesale has quit IRC | 03:50 | |
| *** udesale has joined #openstack-glance | 03:51 | |
| *** abhishekk has joined #openstack-glance | 04:28 | |
| *** brinzhang has joined #openstack-glance | 04:29 | |
| *** ratailor has joined #openstack-glance | 04:57 | |
| *** pvradu has joined #openstack-glance | 05:16 | |
| *** bhagyashris has joined #openstack-glance | 05:29 | |
| *** Luzi has joined #openstack-glance | 06:04 | |
| *** pvradu has quit IRC | 06:18 | |
| *** raginbajin has quit IRC | 06:32 | |
| *** raginbajin has joined #openstack-glance | 06:34 | |
| *** pvradu has joined #openstack-glance | 06:38 | |
| *** mosulica has joined #openstack-glance | 06:42 | |
| *** pvradu has quit IRC | 06:42 | |
| *** pvradu has joined #openstack-glance | 06:43 | |
| *** aniketh has joined #openstack-glance | 06:49 | |
| *** rcernin has quit IRC | 07:03 | |
| *** pcaruana has joined #openstack-glance | 07:05 | |
| *** threestrands has quit IRC | 07:13 | |
| *** pkaralis has joined #openstack-glance | 07:18 | |
| pkaralis | Hi guys, I'm looking for a volunteer for a quick question about a strange behavior of glance? Anyone?? | 07:24 |
| openstackgerrit | Merged openstack/glance master: Refactor periodic "tips" jobs https://review.openstack.org/599837 | 08:00 |
| *** pvradu_ has joined #openstack-glance | 08:08 | |
| *** pvradu has quit IRC | 08:11 | |
| *** brinzhang has quit IRC | 08:24 | |
| *** brinzhang has joined #openstack-glance | 08:24 | |
| *** sayalilunkad has quit IRC | 08:33 | |
| *** sayalilunkad has joined #openstack-glance | 08:33 | |
| *** priteau has joined #openstack-glance | 08:48 | |
| *** bhagyashris has quit IRC | 09:43 | |
| *** Liang__ has quit IRC | 10:30 | |
| *** abhishekk has quit IRC | 10:33 | |
| *** pvradu_ has quit IRC | 10:38 | |
| *** pvradu has joined #openstack-glance | 10:38 | |
| *** udesale has quit IRC | 11:01 | |
| *** lixiaoy1 has quit IRC | 11:05 | |
| *** ratailor has quit IRC | 11:10 | |
| *** pvradu_ has joined #openstack-glance | 12:04 | |
| *** pvradu has quit IRC | 12:08 | |
| *** udesale has joined #openstack-glance | 12:17 | |
| *** udesale has quit IRC | 12:25 | |
| *** brinzhang has quit IRC | 12:27 | |
| *** mvkr has quit IRC | 12:29 | |
| *** rosmaita has joined #openstack-glance | 12:48 | |
| smcginnis | pkaralis: Chances are I can't answer your question, but if you describe your strange behavior then someone might come along here and have an answer. | 12:51 |
| pkaralis | smcginnis, ok I will! thanks | 12:56 |
| rosmaita | jokke_: some glance-spec repo maintenance patches for you to merge (let's clear out the backlog!): | 13:17 |
| rosmaita | https://review.openstack.org/#/c/581549/2 | 13:17 |
| rosmaita | https://review.openstack.org/#/c/579804/1 | 13:17 |
| rosmaita | https://review.openstack.org/#/c/568120/1 | 13:17 |
| rosmaita | jokke_: ^^ | 13:17 |
| *** mvkr has joined #openstack-glance | 13:17 | |
| pkaralis | I have deployed the os-odl-sfc-ha scenario using the XCI. I have noticed in the glance container of the controller0 that the mount had a timeout and permission denied about the filesystem folder. https://hastebin.com/iyugigobev.http | 13:19 |
| pkaralis | Therefore, when I'm trying to create an instance...this fail with the following error. https://hastebin.com/wiyibanome.rb | 13:20 |
| smcginnis | Yep, I definitely can't answer that one. | 13:21 |
| smcginnis | pkaralis: Is this kolla or something? Container deployment issues might be better answered by them. | 13:22 |
| *** udesale has joined #openstack-glance | 13:22 | |
| pkaralis | Moreover, sometimes the first instance is deployed successfully and the second one fail for the same reason [also I've added a printout after image registration] . https://hastebin.com/yayupazoze.rb | 13:22 |
| smcginnis | Usually when I've seen one request work and the next fail it's been some sort of load balancer issue. | 13:23 |
| rosmaita | pkaralis: when you see that error message in the log about "failed to configure store", that is game over for glance ... it will refuse to use that store until you restart and it tests the store connection again | 13:26 |
| rosmaita | sometimes we see this when NFS is being used as the backing for the filesystem store, if glance comes up before NFS is ready for writing, you get this failure | 13:26 |
| rosmaita | could be what's happening here | 13:26 |
| pkaralis | rosmaita, do you believe that this is configuration problem? | 13:28 |
| pkaralis | rosmaita, also this is happening only for controller0. The controller01 and controller02 has not any permission problem | 13:29 |
| rosmaita | i think it's worth looking into, give me a minute, we had a bug about this with some info about how to configure | 13:29 |
| pkaralis | rosmaita, ok thx | 13:29 |
| pkaralis | rosmaita, also if you see on the bottom on the link https://hastebin.com/yayupazoze.rb | 13:29 |
| pkaralis | you will see the show printout after the image registration....if you see, int the the path there is the 'v2' subfolder. But the filesystem has not this subfolder into the filesystem path | 13:31 |
| pkaralis | rosmaita, do you believe that this is something related to? I'm trying to change the glance role supporting the 'v2'. Am I going to the right direction? | 13:32 |
| pkaralis | rosmaita, it seems that something is goig wrong with the mount directory or something and I'm not sure what is the wrong on this... | 13:33 |
| rosmaita | possibly ... check your glance-api.conf for 'filesystem_store_datadir' ... make sure it's what you expect | 13:35 |
| rosmaita | but glance will create that directory (if it can write to the filesystem) when it starts up | 13:36 |
| pkaralis | rosmaita, I will! Thanks! If you faced any similar situation or there is any patch in progress for something similar or a ticket or something, please let me know! | 13:40 |
| pkaralis | rosmaita, I also believe that my problem is related to the NFS server and the mount accesses or something like that... | 13:41 |
| rosmaita | pkaralis: still looking for that reference for you | 13:43 |
| pkaralis | rosmaita, ok, thanks again! :) | 13:50 |
| *** KeithMnemonic has joined #openstack-glance | 13:51 | |
| rosmaita | pkaralis: finally found it! there's some discussion of it here: http://eavesdrop.openstack.org/meetings/glance/2017/glance.2017-11-02-14.01.log.html#l-104 | 13:51 |
| rosmaita | and i think this is what you'll want to look at: https://serverfault.com/questions/682553/systemd-start-a-unit-after-another-unit-really-starts | 13:52 |
| pkaralis | rosmaita, great! I will check it! | 13:54 |
| rosmaita | pkaralis: good luck! | 13:54 |
| pkaralis | rosmaita, thx! | 13:55 |
| *** udesale has quit IRC | 14:07 | |
| *** pvradu_ has quit IRC | 14:14 | |
| *** pvradu has joined #openstack-glance | 14:14 | |
| openstackgerrit | Matt Riedemann proposed openstack/glance master: Document os_shutdown_timeout image property https://review.openstack.org/613041 | 14:17 |
| imacdonn | jokke_: around? | 14:35 |
| *** Liang__ has joined #openstack-glance | 14:52 | |
| *** mosulica has quit IRC | 14:56 | |
| *** Luzi has quit IRC | 15:02 | |
| *** gyee has joined #openstack-glance | 15:15 | |
| openstackgerrit | Liang Fang proposed openstack/glance-specs master: Spec Lite: Show Store Info https://review.openstack.org/612236 | 15:15 |
| jokke_ | imacdonn: now I am | 15:30 |
| imacdonn | jokke_: hi ... maybe we can discuss some of the open questions, if you have some time? hopefully rosmaita is around too | 15:31 |
| * rosmaita hides | 15:31 | |
| imacdonn | heh | 15:31 |
| jokke_ | imacdonn: if you can, please give me few min | 15:31 |
| imacdonn | I'll be here all day ;) | 15:32 |
| imacdonn | I dumped a few questions at the bottom of https://etherpad.openstack.org/p/glance-spec-lite-stein-locations-with-checksum | 15:38 |
| jokke_ | cool ... I'm having quick chat with Lance and get some coffee and will get into that :) | 15:42 |
| imacdonn | thanks | 15:42 |
| *** KeithMnemonic has quit IRC | 15:42 | |
| *** KeithMnemonic has joined #openstack-glance | 15:45 | |
| *** KeithMnemonic has quit IRC | 15:49 | |
| *** pcaruana has quit IRC | 15:49 | |
| *** itlinux has joined #openstack-glance | 16:18 | |
| *** pvradu has quit IRC | 16:26 | |
| jokke_ | imacdonn: thanks for digging into the checksum replacement | 16:26 |
| *** pvradu has joined #openstack-glance | 16:26 | |
| jokke_ | imacdonn: like mentioned in my comment I assumed it was one of these fields that were protected in multiple layers of the domain model, good to hear that it was not the case | 16:27 |
| imacdonn | Seems that rosmaita would prefer it it was :) See etherpad discussion if you're not already there | 16:27 |
| jokke_ | yeah I was looking through that as well | 16:28 |
| jokke_ | I do agree that changing it will get messy ... I just don't like the domain model and that's one less thing to worry about if we ever get to refactor that stuff out of glance | 16:30 |
| jokke_ | like API consistency wise it would make sense not to accept update on the checksums, ever. Once it has been set anything but the same values should be rejected | 16:31 |
| *** pvradu has quit IRC | 16:31 | |
| imacdonn | OK, I can live with that | 16:32 |
| imacdonn | The three top-level questions are still open | 16:33 |
| jokke_ | Also what comes to the mandatory 3 fields. I do agree that it's problematic non of the consumers utilizing anything else than the checksum atm. I'm just not sure if that should be something mandated (lets remember currently there is no values/validation on any of these). I guess my big question is do we want to handhold the user by making all the fields mandatory on a cost of making yet another AP | 16:35 |
| jokke_ | I change when we don't want to have the checksum populated anymore | 16:35 |
| jokke_ | And is the multihash reasonable being mandatory, when it's not consumed by anyone | 16:35 |
| rosmaita | consumers will be using multihash with the next glanceclient release | 16:38 |
| rosmaita | that will include nova and cinder | 16:38 |
| *** jmlowe has quit IRC | 16:39 | |
| *** imacdonn has quit IRC | 16:40 | |
| jokke_ | rosmaita: and it just falls back to checksum (if there's one) in case the multihash doesn't exist? | 16:40 |
| rosmaita | right | 16:40 |
| rosmaita | does that automatically | 16:41 |
| jokke_ | So I think we should deprecate checksum after that release and not demand it (and likely stop calculating it in Glance at T) | 16:41 |
| rosmaita | that's up to you, the feedback i got when we proposed multihash was please do not ever remove the checksum | 16:42 |
| jokke_ | I think it's ok to let it be set by this change (for example that might make testing easier) | 16:42 |
| jokke_ | rosmaita: yeah, we won't be removing the property nor the fallback to keep support for the old images | 16:42 |
| *** imacdonn has joined #openstack-glance | 16:43 | |
| jokke_ | I just don't see point of wasting the compute resources to keep calculating it upon upload | 16:43 |
| imacdonn | lost network connectivity for a few mins .. sigh | 16:43 |
| imacdonn | missed anything since ~1635 | 16:44 |
| jokke_ | imacdonn: ok TL;DR multihash will be consumed from next client release as default and checksum will be just fallback, so IMO we should not demand setting it and likely stop populating that field when image is uploaded to glance somewhere around T | 16:45 |
| imacdonn | OK | 16:46 |
| *** pcaruana has joined #openstack-glance | 16:46 | |
| jokke_ | the property nor the fallback mechanism will not be removed to maintain support for all the images created prior | 16:46 |
| jokke_ | and it might make the testing life easier in some cases to use that instead of multihash | 16:47 |
| jokke_ | but really it should be just deprecated, never removed property in the image | 16:47 |
| imacdonn | rosmaita: are you ok if with "if validation_data is provided, multihash must be included, but checksum is optional" ? | 16:48 |
| jokke_ | the reason I'd like to make it officially deprecated it so send clear signal "We have better options now, don't use md5 checksums anymore" | 16:48 |
| rosmaita | i personally do not like that | 16:49 |
| rosmaita | if we're never going to remove it, just leave it there | 16:49 |
| imacdonn | "leave it there" meaning "make it required"? doesn't seem like the same thing | 16:50 |
| rosmaita | i mean, the point is that these images should be just like all the others | 16:50 |
| rosmaita | so they should all have all 3 properties populated | 16:50 |
| imacdonn | for me, the point is not exactly that | 16:50 |
| imacdonn | for me, the point is "there needs to be a way to enable integrity checking on image contents" | 16:51 |
| jokke_ | rosmaita: but we do not have all 3 properties populated anyways | 16:51 |
| rosmaita | yes, but we do from Rocky onward | 16:52 |
| jokke_ | rosmaita: we still have the old ones that does not have multihas, we have all the images that has none of the properties populated and then we have the ones that are created with the data flowing through Glance that has the 3 | 16:52 |
| rosmaita | yeah, but we don't want to make policy based on pathological cases ... the "typical" image from rocky on should have all 3 populated | 16:53 |
| openstackgerrit | Liang Fang proposed openstack/python-glanceclient master: Show the backend store info https://review.openstack.org/605014 | 16:54 |
| jokke_ | mandating the 3 on this call will just make it "Either there is none or there is all 3 or there is just checksum" which will not improve the security at all vs. making it possible to have only multihash which is consumed by default anyways | 16:54 |
| jokke_ | rosmaita: the Nova & Glance using common Ceph use case will still have neither | 16:55 |
| *** jmlowe has joined #openstack-glance | 16:55 | |
| jokke_ | on snapshots that is | 16:55 |
| rosmaita | well, my opinion is all 3 for this particular case | 16:57 |
| *** Liang__ has quit IRC | 16:57 | |
| jokke_ | I do not see the point for that, but tbh it will be extremely unlikely for me to ever use this anyways so imacdonn if you're fine with that as a consumer of this API I'm fine with that ... we can direct any external complaints to rosmaita :D | 17:01 |
| imacdonn | I'm OK with requiring both until we get to the point where checksum is no longer of value - i.e. when cinder uses multihash to validate contents when creating a volume from an image | 17:03 |
| imacdonn | when we're at that point, it seems like unnecessary burden to require the checksum, if it's not going to be used for anything | 17:04 |
| *** jdillaman1 has quit IRC | 17:04 | |
| jokke_ | that's the thing, we will reach that point when this feature is released ... in Stein which is also the point when Cinder and Nova will for sure use the client that prefers multihash ;) | 17:06 |
| imacdonn | if that's true, then I really don't think checksum should be required .. I'm almost leaning towards removing it entirely | 17:07 |
| jokke_ | removing it entirely won't happen | 17:07 |
| imacdonn | I mean from this smuggling mechanism | 17:08 |
| *** irclogbot_2 has quit IRC | 17:08 | |
| jokke_ | I see | 17:10 |
| imacdonn | but O' | 17:10 |
| imacdonn | I'm OK with including it, if it's optional | 17:10 |
| jokke_ | I think rosmaita's current stand is if you provide any, you provide them all | 17:12 |
| imacdonn | right | 17:12 |
| imacdonn | we have to change his mind ;) | 17:12 |
| jokke_ | which, like said I'm fine with as it won't affect my life really. And if that's what's needed to get consensus to move this forward, someone else can spend the time fighting about that @D | 17:13 |
| jokke_ | :D | 17:13 |
| imacdonn | tell me about it ... I've been chasing this for 2 months now ... every time I seem to be getting close, we take a big jump back .. or sideways .. or something | 17:14 |
| *** jmlowe has quit IRC | 17:18 | |
| *** pvradu has joined #openstack-glance | 17:23 | |
| *** pvradu has quit IRC | 17:27 | |
| *** mvkr has quit IRC | 17:34 | |
| *** rosmaita has left #openstack-glance | 17:41 | |
| *** jdillaman has joined #openstack-glance | 18:06 | |
| *** mvkr has joined #openstack-glance | 18:17 | |
| *** jistr_ has joined #openstack-glance | 18:42 | |
| *** itlinux_ has joined #openstack-glance | 18:43 | |
| *** tridde has joined #openstack-glance | 18:43 | |
| *** FrankZhang has quit IRC | 18:48 | |
| *** itlinux has quit IRC | 18:48 | |
| *** priteau has quit IRC | 18:48 | |
| *** sayalilunkad has quit IRC | 18:48 | |
| *** raginbajin has quit IRC | 18:48 | |
| *** zzzeek has quit IRC | 18:48 | |
| *** pkaralis has quit IRC | 18:48 | |
| *** trident has quit IRC | 18:48 | |
| *** zul has quit IRC | 18:48 | |
| *** jistr has quit IRC | 18:48 | |
| *** spotz has quit IRC | 18:48 | |
| *** sayalilunkad has joined #openstack-glance | 18:50 | |
| *** spotz has joined #openstack-glance | 18:52 | |
| *** FrankZhang has joined #openstack-glance | 18:54 | |
| *** FrankZhang has joined #openstack-glance | 18:54 | |
| *** jmlowe has joined #openstack-glance | 19:00 | |
| *** openstackgerrit has quit IRC | 19:06 | |
| *** irclogbot_2 has joined #openstack-glance | 19:26 | |
| *** irclogbot_2 has quit IRC | 19:27 | |
| *** irclogbot_2 has joined #openstack-glance | 19:27 | |
| *** irclogbot_2 has quit IRC | 19:27 | |
| *** irclogbot_2 has joined #openstack-glance | 19:34 | |
| *** cranges_mcbasket has joined #openstack-glance | 19:35 | |
| cranges_mcbasket | Getting error in Rocky trying to create image: | 19:36 |
| cranges_mcbasket | ERROR glance.common.wsgi AttributeError: 'module' object has no attribute 'add_to_backend_with_multihash' | 19:36 |
| cranges_mcbasket | http://paste.openstack.org/raw/732979/ | 19:36 |
| cranges_mcbasket | ideas where to look? | 19:36 |
| *** jmlowe has quit IRC | 19:50 | |
| *** cranges_mcbasket has quit IRC | 19:57 | |
| *** jmlowe has joined #openstack-glance | 20:11 | |
| *** irclogbot_2 has quit IRC | 20:21 | |
| *** KeithMnemonic has joined #openstack-glance | 20:24 | |
| *** pcaruana has quit IRC | 20:50 | |
| *** openstack has quit IRC | 21:03 | |
| *** openstack has joined #openstack-glance | 21:04 | |
| *** ChanServ sets mode: +o openstack | 21:04 | |
| *** openstackgerrit has joined #openstack-glance | 21:23 | |
| openstackgerrit | iain MacDonnell proposed openstack/glance-specs master: spec-lite: Embed validation data in locations https://review.openstack.org/597648 | 21:23 |
| *** itlinux_ has quit IRC | 21:33 | |
| *** sdinescu has quit IRC | 21:51 | |
| *** sdinescu has joined #openstack-glance | 21:51 | |
| *** KeithMnemonic has quit IRC | 21:53 | |
| openstackgerrit | iain MacDonnell proposed openstack/glance master: Embed validation data in locations https://review.openstack.org/597368 | 22:21 |
| *** itlinux has joined #openstack-glance | 23:04 | |
| *** rcernin has joined #openstack-glance | 23:05 | |
| *** aniketh has quit IRC | 23:11 | |
| *** openstackgerrit has quit IRC | 23:20 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!