| *** dpawlik has joined #openstack-glance | 00:02 | |
| *** dpawlik has quit IRC | 00:06 | |
| *** LiangFang has joined #openstack-glance | 00:27 | |
| *** brinzhang has joined #openstack-glance | 00:37 | |
| *** lixiaoy1 has joined #openstack-glance | 00:48 | |
| *** Dinesh_Bhor has joined #openstack-glance | 01:11 | |
| *** Dinesh_Bhor has quit IRC | 01:31 | |
| *** Dinesh_Bhor has joined #openstack-glance | 01:37 | |
| *** Dinesh_Bhor has quit IRC | 02:22 | |
| *** Dinesh_Bhor has joined #openstack-glance | 02:26 | |
| *** imacdonn has quit IRC | 02:50 | |
| *** markvoelker has joined #openstack-glance | 02:50 | |
| *** imacdonn has joined #openstack-glance | 02:51 | |
| *** abhishekk has joined #openstack-glance | 03:27 | |
| *** openstackgerrit has joined #openstack-glance | 03:36 | |
| openstackgerrit | Abhishek Kekane proposed openstack/glance master: Multiple backend support for scrubber and cache-prefetcher https://review.openstack.org/604971 | 03:36 |
|---|---|---|
| *** rcernin_ has quit IRC | 03:42 | |
| *** rcernin has joined #openstack-glance | 03:43 | |
| *** dpawlik has joined #openstack-glance | 04:02 | |
| *** dpawlik has quit IRC | 04:07 | |
| *** udesale has joined #openstack-glance | 04:18 | |
| openstackgerrit | Abhishek Kekane proposed openstack/glance-specs master: Spec Lite: Rethinking our filesystem access https://review.openstack.org/605584 | 04:47 |
| *** abhishekk has quit IRC | 04:51 | |
| *** abhishekk has joined #openstack-glance | 04:54 | |
| *** aniketh has joined #openstack-glance | 05:13 | |
| *** ratailor has joined #openstack-glance | 05:52 | |
| *** dpawlik has joined #openstack-glance | 06:23 | |
| *** Dinesh_Bhor has quit IRC | 06:24 | |
| *** pcaruana has joined #openstack-glance | 06:33 | |
| *** abhishekk has quit IRC | 06:43 | |
| *** Dinesh_Bhor has joined #openstack-glance | 06:43 | |
| *** abhishekk has joined #openstack-glance | 06:48 | |
| *** mosulica has joined #openstack-glance | 06:50 | |
| *** Luzi has joined #openstack-glance | 07:06 | |
| *** rcernin has quit IRC | 07:12 | |
| *** MattMan_1 has quit IRC | 07:16 | |
| *** MattMan_1 has joined #openstack-glance | 07:16 | |
| *** mhen has joined #openstack-glance | 07:27 | |
| *** aniketh has quit IRC | 08:09 | |
| *** Dinesh_Bhor has quit IRC | 08:16 | |
| *** alexchadin has joined #openstack-glance | 08:21 | |
| *** pbielak has quit IRC | 08:53 | |
| *** lixiaoy1 has quit IRC | 08:55 | |
| *** Dinesh_Bhor has joined #openstack-glance | 08:55 | |
| *** a-pugachev has joined #openstack-glance | 09:01 | |
| *** Dinesh_Bhor has quit IRC | 09:29 | |
| *** alexchadin has quit IRC | 09:42 | |
| *** Dinesh_Bhor has joined #openstack-glance | 10:04 | |
| *** brinzhang has quit IRC | 10:22 | |
| *** Dinesh_Bhor has quit IRC | 10:29 | |
| *** LiangFang has quit IRC | 10:31 | |
| *** udesale has quit IRC | 10:54 | |
| *** alexchadin has joined #openstack-glance | 11:22 | |
| *** ratailor has quit IRC | 11:40 | |
| *** pcaruana has quit IRC | 11:50 | |
| *** pcaruana has joined #openstack-glance | 12:39 | |
| *** alexchadin has quit IRC | 12:43 | |
| *** rosmaita has joined #openstack-glance | 13:21 | |
| *** rosmaita has left #openstack-glance | 13:25 | |
| *** rosmaita has joined #openstack-glance | 13:25 | |
| *** itlinux has quit IRC | 13:59 | |
| *** LiangFang has joined #openstack-glance | 14:07 | |
| *** jistr is now known as jistr|call | 14:31 | |
| *** abhishekk has quit IRC | 14:34 | |
| *** mosulica has quit IRC | 14:36 | |
| *** evrardjp has quit IRC | 14:45 | |
| *** evrardjp has joined #openstack-glance | 14:47 | |
| *** itlinux has joined #openstack-glance | 14:58 | |
| *** evrardjp has quit IRC | 14:58 | |
| smcginnis | rosmaita, jokke_: FYI - http://lists.openstack.org/pipermail/openstack-dev/2018-September/134911.html | 14:59 |
| rosmaita | smcginnis: ty | 14:59 |
| jokke_ | thnx | 15:01 |
| rosmaita | jokke_: what do you think about a backport of https://review.openstack.org/#/c/594769/ to stable/rocky (glanceclient)? | 15:02 |
| jokke_ | ohh crap ... ref the ML | 15:02 |
| rosmaita | time for new email filters! | 15:03 |
| jokke_ | some people really seems to have a need to make major changes for the sake of a change to justify their jobs :( | 15:05 |
| jokke_ | rosmaita: backporting that would likely make sense ... let me read through that change and see the scope of it first 'though | 15:07 |
| *** ratailor has joined #openstack-glance | 15:11 | |
| jokke_ | rosmaita: so on that change the default is not to allow md5 fallback? | 15:11 |
| jokke_ | and we do this validation by default? | 15:12 |
| *** Luzi has quit IRC | 15:13 | |
| *** LiangFang has quit IRC | 15:14 | |
| rosmaita | jokke_: that's basically correct | 15:16 |
| rosmaita | if multihash is not populated, then we do md5 (by default) | 15:17 |
| rosmaita | if multihash is populated but the algo implementation is not available client-side we fail | 15:17 |
| jokke_ | ohh | 15:17 |
| jokke_ | I misunderstood that | 15:18 |
| rosmaita | the option on the patch allows you to use md5 in that case where the multihash failure is due to missing implementation | 15:18 |
| jokke_ | so the option is there just to fallback if the algo is not available at the client end? | 15:18 |
| rosmaita | yes | 15:18 |
| jokke_ | ok ... pheew | 15:18 |
| rosmaita | :) | 15:18 |
| jokke_ | you almost caused a heart attack for me :P | 15:18 |
| imacdonn | jokke_: I *really* need some sort of decision on the add vs. replace thing .. I have developers on my back, wanting to know what the solution is - I've been telling them for months to wait until I get it settled with the glance team | 15:19 |
| imacdonn | internal developers of software that consumes the glance API(s), that is | 15:19 |
| imacdonn | since we can't seem to decide if replace is supposed to work on an empty set or not, I propose reverting the previous fix, and using add only | 15:20 |
| imacdonn | I'll update my smuggling spec to pertain only to add | 15:20 |
| jokke_ | imacdonn: I hear ya, like said when I discovered that the add changes the status you can rely that behavior being there. So that will just work for your use case | 15:24 |
| imacdonn | jokke_: Yes, I'm fine with that (I said so in my ML post) ... but I need the decision to be made so I can finalise the spec for the smuggling thing | 15:25 |
| jokke_ | if you _really_ want to use the replace instead we're working on that but if you look the meeting logs I just explained there why this is taking time | 15:25 |
| imacdonn | if replace is going to work (set status to active), the smuggling thing should work there too ... if not, it should pertain only to add | 15:25 |
| imacdonn | I still don't really understand why it's considered a change of API behaviour, as opposed to a bug fix | 15:26 |
| imacdonn | could you describe a scenario where someone would use "replace" on an image in "queued" status, and NOT need the image to transition to "active" ? | 15:26 |
| jokke_ | we want to make the right thing for you but same time we have responsibility to make the right thing for everyone else as well without ending into huge war over our api again | 15:26 |
| imacdonn | I would be willing to bet a moderate amount that no one has ever tried using replace on locations with an image in queued status | 15:27 |
| imacdonn | (other than me) | 15:27 |
| jokke_ | imacdonn: say someone has more than one backend and they want to make sure the image really is available before making it active | 15:27 |
| imacdonn | so how would they make it active in that case ? | 15:28 |
| jokke_ | they might be updating the locations as they go with replace and only add when all the backends are populated | 15:28 |
| jokke_ | or say all -1 | 15:28 |
| jokke_ | so you can still do add after replace | 15:28 |
| jokke_ | and it will still activate the image | 15:28 |
| imacdonn | huh? | 15:28 |
| imacdonn | yeah, it will still activate, but ... that's way more perverted than my use of replace | 15:29 |
| jokke_ | yup | 15:29 |
| jokke_ | specially as where we stood at the time all those backends would have needed to be different types (apart from some production clouds I know of who have already their own implementations to have multiple ceph clusers for example) | 15:30 |
| jokke_ | but my point is that if there is workflows relying that replace not activating the image we want to know about them before breaking it ;) | 15:32 |
| imacdonn | OK, so if that's the case, I propose: 1) revert the change in master 2) abandon the backports 3) update my spec to pertain to add only | 15:33 |
| imacdonn | this is a major problem for me ... I can't move forward with Rocky deployment/upgrade until it's figured out | 15:34 |
| imacdonn | I brought it up many months ago, and I was told to wait, because nothing could be done until the policy stuff got implemented | 15:35 |
| imacdonn | I could tell the developers to use "add", and implement checksum smuggling, in anticipation of the spec getting approved, but then you guys might turn around and reject that too | 15:36 |
| jokke_ | yeah and that work did not get done :( | 15:36 |
| jokke_ | imacdonn: based on the PTG discussion that checksum work is going to happen | 15:37 |
| jokke_ | did you look the etherpad for the PTG session? | 15:37 |
| imacdonn | I want an approved spec ... otherwise you might decide that is has to be done some different way (again) | 15:38 |
| jokke_ | https://etherpad.openstack.org/p/glance-stein-checksum-hash-at-patch | 15:38 |
| imacdonn | guess what "op" is used in that etherpad :P | 15:38 |
| imacdonn | I can't keep jerking my colleagues around on this | 15:39 |
| jokke_ | yes so only thing we want to change there is where in that json blob the data sits in. Also we need to obviously make sure that is there for the add | 15:39 |
| imacdonn | you've seen this, right? http://logs.openstack.org/48/597648/7/check/openstack-tox-docs/4ae1ed8/html/specs/stein/approved/glance/spec-lite-locations-with-validation-data.html | 15:40 |
| jokke_ | so I'll go through the spec today (I just need to get something to eat first) and we can get the checksum speck moving from there | 15:40 |
| imacdonn | so I'm proposing to remove references to "replace", and change the example to "add" | 15:40 |
| imacdonn | OK, I feared that you were going to disappear into the wilderness for another week! | 15:41 |
| jokke_ | so if you want to make the change there, (op add and move it to verification_data or what ever it is called in the etherpad instead of in the location metadata (which is actual thing in database) I'll review it and get the rest doing the same | 15:42 |
| imacdonn | that change is already made - see above link | 15:42 |
| jokke_ | imacdonn: nope ... have burned my holidays ... I'm stuck here until Berlin at least :D | 15:43 |
| imacdonn | I made it like the same day (of PTG) | 15:43 |
| imacdonn | OK, good - we lock you in your office now ;) | 15:43 |
| jokke_ | this week I have been just fighting the aftermath ofliving 3 different timezones in like 4 weeks | 15:44 |
| imacdonn | pondering this a bit more ... maybe it's actually OK to have the smuggling work with replace, even if the status doesn't get updated to active | 15:44 |
| jokke_ | so I'm sleeping ~12hr every two days and online at really weird times when I'm awake enough to work | 15:45 |
| jokke_ | imacdonn: yeah, I'd like to see it on both as long as the image is queued | 15:45 |
| imacdonn | yeah, that sounds like fun to recover from | 15:45 |
| imacdonn | OK, then I think the spec is ready for approval | 15:45 |
| *** jistr|call is now known as jistr | 15:46 | |
| jokke_ | rosmaita: smcginnis: would be great if you looked through that spec as well ^^ | 15:53 |
| *** gyee has joined #openstack-glance | 15:53 | |
| smcginnis | Will try to on the plane later. | 15:54 |
| rosmaita | ack | 15:54 |
| *** gyee has quit IRC | 15:54 | |
| jokke_ | ty | 15:54 |
| *** dpawlik has quit IRC | 15:54 | |
| *** gyee has joined #openstack-glance | 15:57 | |
| *** dpawlik has joined #openstack-glance | 15:58 | |
| imacdonn | BTW, implementation is here too: https://review.openstack.org/#/c/597368/ | 15:59 |
| *** dpawlik has quit IRC | 15:59 | |
| *** dpawlik has joined #openstack-glance | 15:59 | |
| imacdonn | and the glanceclient part... which I do have a question about: https://review.openstack.org/#/c/597368/ | 15:59 |
| imacdonn | er sorry, paste-o .... https://review.openstack.org/602794 | 16:00 |
| *** ratailor has quit IRC | 16:04 | |
| jokke_ | https://review.openstack.org/602794 | 16:14 |
| imacdonn | the question on that one: should the CLI arg be '--validation-data "{...}"', or should there be separate args like '--checksum', '--os-hash-algo', .... ? | 16:21 |
| imacdonn | .... and how are such things spec'ed ? | 16:22 |
| *** dpawlik has quit IRC | 16:26 | |
| jokke_ | so one thing we learned from the tasks api is that users really hate writing json | 16:26 |
| imacdonn | damn humans | 16:27 |
| jokke_ | so it would probably be sensible doing separate options | 16:27 |
| jokke_ | ikr | 16:27 |
| imacdonn | though in this case, they already have to know how to make json for metadata | 16:30 |
| *** dpawlik has joined #openstack-glance | 16:30 | |
| *** a-pugachev has quit IRC | 16:34 | |
| *** dpawlik has quit IRC | 16:58 | |
| *** dpawlik has joined #openstack-glance | 16:59 | |
| *** dpawlik has quit IRC | 17:36 | |
| *** dpawlik has joined #openstack-glance | 17:39 | |
| *** dpawlik has quit IRC | 17:39 | |
| *** dpawlik has joined #openstack-glance | 17:40 | |
| *** dpawlik has quit IRC | 17:40 | |
| *** mvkr has quit IRC | 17:42 | |
| *** goldenfri has joined #openstack-glance | 17:50 | |
| *** rosmaita has quit IRC | 18:02 | |
| *** dims_ is now known as dims | 18:05 | |
| *** rosmaita has joined #openstack-glance | 18:26 | |
| *** jistr has quit IRC | 18:47 | |
| *** jistr has joined #openstack-glance | 18:49 | |
| *** jistr has quit IRC | 19:08 | |
| *** jistr has joined #openstack-glance | 19:08 | |
| *** itlinux has quit IRC | 19:09 | |
| *** jistr has quit IRC | 19:23 | |
| *** jistr has joined #openstack-glance | 19:26 | |
| *** jistr has quit IRC | 19:28 | |
| *** jistr has joined #openstack-glance | 19:29 | |
| *** jistr has quit IRC | 19:34 | |
| *** jistr has joined #openstack-glance | 19:37 | |
| *** jistr has quit IRC | 19:39 | |
| *** jistr has joined #openstack-glance | 19:49 | |
| *** pcaruana has quit IRC | 19:54 | |
| *** dpawlik has joined #openstack-glance | 20:42 | |
| *** dpawlik has quit IRC | 20:46 | |
| *** rosmaita has left #openstack-glance | 22:00 | |
| *** mvkr has joined #openstack-glance | 22:13 | |
| *** rcernin has joined #openstack-glance | 22:29 | |
| *** dpawlik has joined #openstack-glance | 22:42 | |
| *** dpawlik has quit IRC | 22:47 | |
| *** dpawlik has joined #openstack-glance | 23:20 | |
| *** dpawlik has quit IRC | 23:25 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!