| *** anastzhyr has quit IRC | 00:07 | |
| *** slaweq has joined #openstack-meeting-4 | 00:11 | |
| *** slaweq has quit IRC | 00:16 | |
| *** igordc has quit IRC | 00:28 | |
| *** jokke_ has quit IRC | 00:43 | |
| *** igordc has joined #openstack-meeting-4 | 00:59 | |
| *** ricolin has joined #openstack-meeting-4 | 01:14 | |
| *** ricolin has quit IRC | 01:17 | |
| *** ricolin_ is now known as ricolin | 01:18 | |
| *** Liang__ has joined #openstack-meeting-4 | 01:40 | |
| *** tinwood has quit IRC | 02:10 | |
| *** slaweq has joined #openstack-meeting-4 | 02:11 | |
| *** tinwood has joined #openstack-meeting-4 | 02:12 | |
| *** slaweq has quit IRC | 02:16 | |
| *** ktibi has joined #openstack-meeting-4 | 02:53 | |
| *** ktibi has quit IRC | 02:57 | |
| *** psachin has joined #openstack-meeting-4 | 03:35 | |
| *** igordc has quit IRC | 04:08 | |
| *** slaweq has joined #openstack-meeting-4 | 04:11 | |
| *** slaweq has quit IRC | 04:16 | |
| *** ricolin has quit IRC | 04:42 | |
| *** links has joined #openstack-meeting-4 | 05:00 | |
| *** slaweq has joined #openstack-meeting-4 | 06:11 | |
| *** slaweq has quit IRC | 06:16 | |
| *** Liang__ has quit IRC | 06:30 | |
| *** slaweq has joined #openstack-meeting-4 | 07:05 | |
| *** eharney has quit IRC | 07:08 | |
| *** eharney has joined #openstack-meeting-4 | 07:21 | |
| *** ralonsoh has joined #openstack-meeting-4 | 07:31 | |
| *** gcheresh_ has joined #openstack-meeting-4 | 07:46 | |
| *** gcheresh_ has quit IRC | 08:09 | |
| *** gcheresh_ has joined #openstack-meeting-4 | 08:13 | |
| *** gcheresh_ has quit IRC | 08:47 | |
| *** gcheresh_ has joined #openstack-meeting-4 | 08:50 | |
| *** e0ne has joined #openstack-meeting-4 | 09:07 | |
| *** gcheresh_ has quit IRC | 09:15 | |
| *** gcheresh_ has joined #openstack-meeting-4 | 09:22 | |
| *** gcheresh_ has quit IRC | 09:36 | |
| *** gcheresh_ has joined #openstack-meeting-4 | 09:41 | |
| *** e0ne has quit IRC | 09:54 | |
| *** yebinama_ has quit IRC | 09:57 | |
| *** salmankhan has joined #openstack-meeting-4 | 10:09 | |
| *** dviroel has joined #openstack-meeting-4 | 10:36 | |
| *** pcaruana has quit IRC | 10:46 | |
| *** e0ne has joined #openstack-meeting-4 | 11:07 | |
| *** pcaruana has joined #openstack-meeting-4 | 11:48 | |
| *** pcaruana has quit IRC | 12:21 | |
| *** jokke_ has joined #openstack-meeting-4 | 13:24 | |
| *** enriquetaso has joined #openstack-meeting-4 | 13:27 | |
| *** rishabhhpe has joined #openstack-meeting-4 | 13:45 | |
| *** rosmaita has joined #openstack-meeting-4 | 13:49 | |
| *** yebinama_ has joined #openstack-meeting-4 | 13:58 | |
| *** pcaruana has joined #openstack-meeting-4 | 14:03 | |
| *** alistarle has joined #openstack-meeting-4 | 14:07 | |
| *** alistarle has quit IRC | 14:08 | |
| jokke_ | startmeeting glance | 14:08 |
|---|---|---|
| jokke_ | #startmeeting glance | 14:08 |
| openstack | Meeting started Thu Jan 30 14:08:25 2020 UTC and is due to finish in 60 minutes. The chair is jokke_. Information about MeetBot at http://wiki.debian.org/MeetBot. | 14:08 |
| openstack | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 14:08 |
| *** openstack changes topic to " (Meeting topic: glance)" | 14:08 | |
| openstack | The meeting name has been set to 'glance' | 14:08 |
| jokke_ | #topic roll-call | 14:08 |
| *** openstack changes topic to "roll-call (Meeting topic: glance)" | 14:08 | |
| jokke_ | o/ | 14:08 |
| rosmaita | o/ | 14:08 |
| yebinama_ | o/ | 14:08 |
| rosmaita | looks like a short agenda today, just open discussion | 14:09 |
| rosmaita | i am behind on my reviewing, is there anything in particular that i should look at first? | 14:09 |
| jokke_ | #topic open discussion | 14:10 |
| *** openstack changes topic to "open discussion (Meeting topic: glance)" | 14:10 | |
| jokke_ | ok, yebinama_ looking into the hashing | 14:11 |
| jokke_ | sorry for noticing it this late | 14:11 |
| yebinama_ | No problem | 14:11 |
| yebinama_ | Did you see my update? | 14:12 |
| rosmaita | link to patch? | 14:12 |
| yebinama_ | review.opendev.org/#/c/667132 | 14:13 |
| rosmaita | ty | 14:13 |
| jokke_ | https://review.opendev.org/#/c/667132/21/glance/location.py | 14:13 |
| jokke_ | yebinama_: I saw your comment and that you had update, didn't have my head with me glancing it through to follow what was going on there :P | 14:14 |
| jokke_ | I need to look through it properly | 14:14 |
| yebinama_ | Ok :) | 14:14 |
| yebinama_ | I didn't do the revert part | 14:15 |
| yebinama_ | I had a question regarding it | 14:15 |
| jokke_ | yeah, that might get hairy | 14:15 |
| jokke_ | go for it rosmaita might be able to help as well | 14:15 |
| yebinama_ | Today, if an import fails we don't revert to queued state? | 14:16 |
| yebinama_ | Or I didn't see the code for this | 14:16 |
| jokke_ | IIRC it depends, I think web-download reverts to queued glance-direct might go to uploading or something like that as we leave the data in staging | 14:17 |
| rosmaita | i think that's correct | 14:17 |
| jokke_ | but neither should leave it active with locations in there | 14:17 |
| *** eharney has quit IRC | 14:18 | |
| yebinama_ | Ok so I don't have to change that part | 14:18 |
| yebinama_ | Just make sure in case of revert to delete os_hash ans other related | 14:19 |
| jokke_ | no, what I meant and might have frased it wrong was that when we rever to the point that we have no locations in the image, we should not have hashes there either | 14:19 |
| jokke_ | yep! | 14:20 |
| jokke_ | and this is the part that might get hairy, rosmaita do you remember if we have some protections at the db lever for those properties? | 14:21 |
| rosmaita | you mean whether a user can modify them? | 14:21 |
| rosmaita | they're protected however we protect 'checksum' | 14:22 |
| jokke_ | I mean wether we can actually internally remove them once they have been set | 14:22 |
| rosmaita | that's a good question | 14:22 |
| jokke_ | as I can't remember which point of the code path those protections kicks in | 14:22 |
| yebinama_ | I think we can do it as with the code I wrote today, they are changed each time an import is done | 14:23 |
| rosmaita | i think they get filtered at the api layer | 14:23 |
| jokke_ | yebinama_: that sounds great | 14:23 |
| rosmaita | i don't think we have any hard core protections in the DB | 14:23 |
| yebinama_ | And that's what you asked to change | 14:23 |
| jokke_ | yebinama_: yeah, true that's where my initial concern came from so they are not protected on this level ;) | 14:24 |
| jokke_ | I really need to take my head with me next time I sit back to the computer :D | 14:25 |
| yebinama_ | :) | 14:25 |
| rosmaita | sounds like we are OK thne | 14:25 |
| jokke_ | yebinama_: so it's clear for you now what I was looking for? | 14:25 |
| rosmaita | *then | 14:25 |
| yebinama_ | My only concern is where is the revert code you talked about | 14:26 |
| yebinama_ | There is no revert in image import flow | 14:26 |
| jokke_ | and rosmaita does that make sense to you? | 14:26 |
| rosmaita | you mean the part about if there's no image data, then os_hash_* and checksum should not be set? | 14:27 |
| yebinama_ | The part where we go back from active to queued | 14:28 |
| jokke_ | rosmaita: I mean the 2 cases, first we should not set the checksums every time we upload image, but rather validate against the first one if we do multi store import and if it fails we should revert the checksums and algo out of the image metadata as well when we clean up the locations | 14:28 |
| rosmaita | i definitely agree with that | 14:29 |
| jokke_ | yebinama_: so each task in taskflow has it's rever function (if it's defined) | 14:29 |
| jokke_ | revert | 14:29 |
| jokke_ | taskflow calls those on failure (this happens each time there comes exception through the flow instead of sucessful return) | 14:30 |
| yebinama_ | Yep I added one for ImportToStore | 14:30 |
| jokke_ | so basically what we want to do is, in the code that deletes the locations upon failure, add a check that looks if we delete last location and clean out the chekcsum and _os_hash_* | 14:31 |
| yebinama_ | But you told that today if an import fails, glance set the image status back to queued | 14:31 |
| *** Liang__ has joined #openstack-meeting-4 | 14:32 | |
| yebinama_ | I couldn't find that revert in the code | 14:32 |
| *** psachin has quit IRC | 14:32 | |
| *** links has quit IRC | 14:33 | |
| jokke_ | yebinama_: web_download.py L:132 does | 14:34 |
| *** pcaruana has quit IRC | 14:34 | |
| jokke_ | so like said, we do it all the way to queued on the case of web-download method as we do not have data in staging | 14:34 |
| yebinama_ | Oh ok, not the image import. I missed that point. | 14:35 |
| yebinama_ | So I don't have to deal with it, thanks for the clarification. | 14:36 |
| jokke_ | sorry, you don't need to deal with what? | 14:37 |
| yebinama_ | Reverting the state in image import workflow | 14:37 |
| jokke_ | ah, no we don't need to change into which status we are reverting the image to | 14:37 |
| yebinama_ | Sorry for the confusion :) | 14:38 |
| jokke_ | np, it's good that we're on the same page. I love taskflow for what it enables us to do but It can be confusing | 14:39 |
| yebinama_ | Yes it' a great feature. | 14:40 |
| jokke_ | if you ever want to get few more grey hair, look our first try with it on the tasks api (the code lives directly under async_/flows/) | 14:41 |
| jokke_ | It was nice try, but the user experience was from the horror movies | 14:41 |
| yebinama_ | I'll give it a look :) | 14:42 |
| jokke_ | ok, we've spent 40min on this (which is good) ... anything else? | 14:42 |
| yebinama_ | It's good for me. | 14:43 |
| jokke_ | I'm good too, rosmaita? | 14:44 |
| rosmaita | sounds good to me (given that i haven't looked at the code yet) | 14:44 |
| rosmaita | i have something else | 14:44 |
| jokke_ | rosmaita: more like timecheck for the meeting | 14:45 |
| jokke_ | ah, exactly | 14:45 |
| *** gcheresh_ has quit IRC | 14:45 | |
| rosmaita | i wonder if we should deprecate 'checksum' this cycle | 14:45 |
| rosmaita | i think the md5 reliance is becoming a problem | 14:45 |
| rosmaita | both for us calculating it, but also for people confirming it | 14:45 |
| rosmaita | i think the multihash has been in place for a few cycles | 14:46 |
| rosmaita | hopefully people are starting to use that instead | 14:46 |
| rosmaita | but, i admit that it's going to be a problem for legacy images | 14:46 |
| jokke_ | I'm even less worried it being there, but lots of environments are removing the support to actually calculate them | 14:46 |
| rosmaita | yeah, so i guess we can leave it there, but stop computing it in Victoria | 14:47 |
| rosmaita | that way we don't change the image-show response | 14:47 |
| jokke_ | need to see what abhishek thinks, but yeah I agree, we could potentially stop adding it, or at least write checks if md5 is not supported in the env, leaving it out | 14:48 |
| rosmaita | also, i'm not sure how glanceclient would handle md5 being unavailable | 14:48 |
| *** gcheresh_ has joined #openstack-meeting-4 | 14:48 | |
| rosmaita | it uses multihash if its available | 14:48 |
| rosmaita | but i think it will always try to compute md5sum | 14:49 |
| jokke_ | the nice part of md5 is that it's super light to calculate (that's why it's still used in so many places) | 14:49 |
| rosmaita | although there is a really weird case in that code | 14:49 |
| jokke_ | it might calculate it but it definitely doesn't expect it to be there | 14:50 |
| jokke_ | as all the ceph snapshots comes without any checksums and they still work | 14:50 |
| rosmaita | yeah, a null checksum is ok, i think the problem case will be checksum is there, no multihash, it will try to compute the checksum, get an exception, and then the image can't be downloaded | 14:51 |
| jokke_ | yeah, we need to look into the client code if we're finally dropping this for sure | 14:52 |
| jokke_ | ok, 7min left, something else? | 14:53 |
| *** andrebeltrami has joined #openstack-meeting-4 | 14:54 | |
| yebinama_ | Just remembered something | 14:54 |
| rosmaita | ok, i'll put it on the agenda for next week and we can see what abhishek thinks | 14:54 |
| jokke_ | rosmaita: gr8 | 14:54 |
| yebinama_ | I check checksum and hash for each import | 14:54 |
| yebinama_ | But I also added a check for the size | 14:54 |
| yebinama_ | Is it needed or not? | 14:54 |
| jokke_ | not necessarily, but it's lightweight check and kind of makes sense there | 14:55 |
| jokke_ | I didn't mind it when I saw it | 14:55 |
| yebinama_ | Ok, I'll keep it then. Thanks. | 14:56 |
| jokke_ | if our size suddenly changes, something is definitely wrong | 14:56 |
| jokke_ | specially if the hashes matches | 14:56 |
| rosmaita | no kidding | 14:57 |
| jokke_ | ok, last call, going first | 14:57 |
| rosmaita | nothing from me, see you next week | 14:58 |
| yebinama_ | Bye. | 14:58 |
| jokke_ | ok, thanks all! | 14:58 |
| jokke_ | #endmeeting | 14:58 |
| *** openstack changes topic to "OpenStack Meetings || https://wiki.openstack.org/wiki/Meetings/" | 14:58 | |
| openstack | Meeting ended Thu Jan 30 14:58:27 2020 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 14:58 |
| openstack | Minutes: http://eavesdrop.openstack.org/meetings/glance/2020/glance.2020-01-30-14.08.html | 14:58 |
| openstack | Minutes (text): http://eavesdrop.openstack.org/meetings/glance/2020/glance.2020-01-30-14.08.txt | 14:58 |
| openstack | Log: http://eavesdrop.openstack.org/meetings/glance/2020/glance.2020-01-30-14.08.log.html | 14:58 |
| *** ricolin has joined #openstack-meeting-4 | 14:58 | |
| *** pcaruana has joined #openstack-meeting-4 | 15:02 | |
| *** salmankhan1 has joined #openstack-meeting-4 | 15:03 | |
| *** gcheresh_ has quit IRC | 15:04 | |
| *** salmankhan has quit IRC | 15:04 | |
| *** salmankhan1 is now known as salmankhan | 15:04 | |
| *** trident has quit IRC | 15:06 | |
| *** trident has joined #openstack-meeting-4 | 15:17 | |
| *** eharney has joined #openstack-meeting-4 | 15:18 | |
| *** bobmel has quit IRC | 15:21 | |
| *** bobmel has joined #openstack-meeting-4 | 15:25 | |
| *** bobmel has quit IRC | 15:28 | |
| *** psachin has joined #openstack-meeting-4 | 15:28 | |
| *** k_mouza has joined #openstack-meeting-4 | 15:31 | |
| *** salmankhan1 has joined #openstack-meeting-4 | 15:35 | |
| *** k_mouza has quit IRC | 15:36 | |
| *** salmankhan has quit IRC | 15:38 | |
| *** salmankhan1 is now known as salmankhan | 15:38 | |
| *** ricolin has quit IRC | 15:45 | |
| *** rishabhhpe has quit IRC | 15:46 | |
| *** eharney has quit IRC | 16:16 | |
| *** eharney has joined #openstack-meeting-4 | 16:16 | |
| *** rishabhhpe has joined #openstack-meeting-4 | 16:33 | |
| *** michael-beaver has quit IRC | 16:50 | |
| *** rishabhhpe has quit IRC | 17:01 | |
| *** yebinama_ has quit IRC | 17:08 | |
| *** evrardjp has joined #openstack-meeting-4 | 17:33 | |
| *** e0ne has quit IRC | 17:36 | |
| *** salmankhan has quit IRC | 17:46 | |
| *** bobmel has joined #openstack-meeting-4 | 17:56 | |
| *** igordc has joined #openstack-meeting-4 | 18:09 | |
| *** ralonsoh has quit IRC | 18:11 | |
| *** andrebeltrami has quit IRC | 18:13 | |
| *** gcheresh has joined #openstack-meeting-4 | 18:19 | |
| *** k_mouza has joined #openstack-meeting-4 | 18:34 | |
| *** k_mouza has quit IRC | 18:38 | |
| *** k_mouza_ has joined #openstack-meeting-4 | 18:39 | |
| *** bnemec has joined #openstack-meeting-4 | 18:41 | |
| *** gcheresh has quit IRC | 18:45 | |
| *** bobmel has quit IRC | 18:55 | |
| *** bobmel has joined #openstack-meeting-4 | 18:56 | |
| *** bobmel has quit IRC | 19:00 | |
| *** gcheresh has joined #openstack-meeting-4 | 19:07 | |
| *** bnemec has quit IRC | 19:24 | |
| *** e0ne has joined #openstack-meeting-4 | 19:27 | |
| *** gcheresh has quit IRC | 19:28 | |
| *** Liang__ has quit IRC | 19:29 | |
| *** e0ne has quit IRC | 19:33 | |
| *** hamzy_ has joined #openstack-meeting-4 | 20:11 | |
| *** hamzy has quit IRC | 20:14 | |
| *** psachin has quit IRC | 20:16 | |
| *** eharney has quit IRC | 21:15 | |
| *** enriquetaso has quit IRC | 21:33 | |
| *** slaweq has quit IRC | 21:43 | |
| *** slaweq has joined #openstack-meeting-4 | 21:45 | |
| *** slaweq has quit IRC | 21:55 | |
| *** dviroel has quit IRC | 23:26 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!