| *** jamielennox|away is now known as jamielennox | 00:05 | |
| *** gouthamr has quit IRC | 00:24 | |
| *** gouthamr has joined #openstack-shade | 00:32 | |
| *** gouthamr has quit IRC | 05:07 | |
| *** yfried has joined #openstack-shade | 06:18 | |
| *** abregman has joined #openstack-shade | 07:09 | |
| *** abregman has quit IRC | 08:24 | |
| *** abregman has joined #openstack-shade | 08:29 | |
| *** yolanda has quit IRC | 09:02 | |
| *** yolanda has joined #openstack-shade | 09:07 | |
| *** yolanda has quit IRC | 09:15 | |
| *** abregman has quit IRC | 09:18 | |
| *** abregman has joined #openstack-shade | 09:19 | |
| *** yolanda has joined #openstack-shade | 09:24 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade: Update coding document to mention direct REST calls https://review.openstack.org/424014 | 09:26 |
|---|---|---|
| openstackgerrit | Monty Taylor proposed openstack-infra/shade: Copy in needed template processing utils from heatclient https://review.openstack.org/424016 | 09:27 |
| frickler | mordred: https://review.openstack.org/#/c/423595/3/shade/openstackcloud.py would be giving me the same issues that I had with image_endpoint_override earlier, would it make sense to add the possibility to explicitly override the network endpoint the same way? | 09:55 |
| frickler | strangely though, even though I have the wrong URL in endpoint_override, it doesn't seem to get used, listing networks is still working for me | 09:55 |
| frickler | ah, I see you are semi-reverting that one anyway in https://review.openstack.org/#/c/423604/2 , so the total result should be fine again. | 09:58 |
| mordred | frickler: oh - thanks for pointing it out - should definitely honor the endpoint_override if it's set if we decide to go the discovery route | 10:10 |
| mordred | (best I can tell, actually doing the discovery call seems to just be a waste of a REST round trip on every cloud out there :) ) | 10:10 |
| yfried | mordred: ping re shade cache | 10:11 |
| mordred | yfried: whassup? | 10:11 |
| yfried | mordred: ansible os_server_facts != nova list | 10:12 |
| yfried | mordred: I'm guessing this is due to shade cache | 10:12 |
| yfried | mordred: can I flush it somehow? | 10:12 |
| mordred | yfried: you have a persistent cache configured? | 10:13 |
| yfried | mordred: IDK | 10:13 |
| mordred | if you don't have one explicitly configured in clouds.yaml, then there should be no persistent cache | 10:13 |
| mordred | like, I hve this in mine: | 10:14 |
| mordred | cache: | 10:14 |
| mordred | max_age: 3600 | 10:14 |
| mordred | class: dogpile.cache.dbm | 10:14 |
| mordred | arguments: | 10:14 |
| mordred | filename: /home/mordred/.cache/openstack/shade.dbm | 10:14 |
| mordred | but if you don't - something else must be wrong and we should track down the bug I think | 10:14 |
| yfried | mordred: http://paste.openstack.org/show/596009/ os_server_facts | 10:15 |
| yfried | mordred: | 10:17 |
| yfried | http://paste.openstack.org/show/596010/ | 10:17 |
| yfried | mordred: note the diff between networks in json and shell | 10:17 |
| yfried | mordred: this might be better @ #ansible-devel not sure if this is shade or ansible issue | 10:18 |
| mordred | yfried: all the logic is in the shade part - but I don't see the difference, - in your shell output, it shows three: yfried-trial1-management=192.168.1.13; yfried-trial1-data=172.16.1.13; yfried-trial1-external=172.31.0.13 | 10:19 |
| mordred | each with one address | 10:19 |
| yfried | mordred: in json, you see a floating ip | 10:20 |
| yfried | yfried-trial1-management | 10:20 |
| yfried | mordred: on ^ | 10:20 |
| mordred | AH | 10:20 |
| yfried | and so, the server's public_v4 is set | 10:20 |
| mordred | so - we use neutronclient in shade to ask neutron what floating ips a server has | 10:20 |
| mordred | becuase the nova cache breaks sometimes | 10:21 |
| mordred | and can't be trusted | 10:21 |
| yfried | mordred: openstack --os-cloud prod floating ip show 03c16249-853f-40f2-8d86-33facf279b65 | 10:21 |
| yfried | | fixed_ip_address | 192.168.1.13 | | 10:21 |
| yfried | | floating_ip_address | 10.8.182.222 | | 10:21 |
| yfried | mordred: so the bug is here: ^ | 10:21 |
| yfried | mordred: there are several network clusters on this tenant with overlapping IPs | 10:22 |
| yfried | mordred: and the neutron query should be much more specific | 10:22 |
| yfried | openstack --os-cloud prod server show 7ff77f73-9264-4569-8ba1-25ba955ee25f | 10:23 |
| yfried | addresses | yfried--management=192.168.1.13, 10.8.182.222; yfried--data=172.16.1.13; yfried--external=172.31.0.13 | 10:23 |
| mordred | yfried: so shade is incorrectly showing the floating ip on the wrong fixed address? | 10:24 |
| yfried | mordred: it queries floating ip based on fixed_ip | 10:24 |
| yfried | mordred: but not based on fixed (sub)net | 10:24 |
| yfried | mordred: and/or server ports | 10:25 |
| mordred | gotcha. yah - that's a bug | 10:25 |
| mordred | it would be SO nice if nova returned port ids | 10:25 |
| yfried | mordred: you don't need that - neutron port list can filter on "device_id" | 10:27 |
| yfried | mordred: that's the nova id | 10:27 |
| yfried | of the server | 10:27 |
| yfried | openstack --os-cloud prod port list --server af742eb9-7637-4422-883d-17ec6002421e | 10:28 |
| yfried | mordred: this bug is not in shade 1.7.0 | 10:31 |
| mordred | 1.7.0 is likely before we started pulling fip info directly from neutron | 10:34 |
| mordred | but yeah - we need to do matching on port_id instead of the fixed_id | 10:35 |
| mordred | i'll work on a patch | 10:35 |
| mordred | unless you want to :) | 10:36 |
| mordred | shade.meta._get_suplemental_addresses is where the code for it is | 10:36 |
| *** cdent has joined #openstack-shade | 10:42 | |
| *** purplerbot has quit IRC | 10:46 | |
| *** purplerbot has joined #openstack-shade | 10:48 | |
| yfried | mordred: sure, np | 10:58 |
| yfried | mordred: but only if you can think of a way for me to use it now instead fo waiting for it to be published in shade via pip | 10:59 |
| yfried | mordred: because shade release cycles are too long for us | 10:59 |
| yfried | mordred: https://storyboard.openstack.org/#!/story/2000845 | 11:31 |
| yfried | mordred: could you please bump this priority? | 11:31 |
| mordred | yfried: I can make a shade release any time - like as soon as we have this fixed - but also I'll work on it right now (just got back with coffee, which is important) | 11:39 |
| yfried | mordred: ok. | 11:40 |
| yfried | mordred: I'm curretnly bumping our framework to use shade 1.9 | 11:40 |
| mordred | yfried: awesome. you know we're on 1.14.1 yeah? | 11:40 |
| yfried | mordred: I was using 1.13.2 until I found this bug | 11:41 |
| mordred | ah - I understand | 11:41 |
| yfried | mordred: can you think of a way to run https://github.com/openstack-infra/shade/blob/master/shade/meta.py#L308 | 11:44 |
| yfried | given a clouds.yml file | 11:45 |
| yfried | ? | 11:45 |
| yfried | mordred: I would like to fix this but I need a good/quick reproduction script to test the filter | 11:45 |
| mordred | yfried: yes - one sec | 11:48 |
| mordred | yfried: http://paste.openstack.org/show/596028/ | 11:50 |
| mordred | yfried: or this one: http://paste.openstack.org/show/596029/ | 11:51 |
| yfried | mordred: tnx | 11:51 |
| mordred | yfried: I believe unfortunately we're going to have to essentially do a join - the port list tells us which things are attached to the server, but we then need to fetch the fip info to get the actual fip address | 11:53 |
| mordred | also - when did server fixed ips stop having neutron ports associated with them? am I just dreaming that that used to be a thing? | 11:53 |
| yfried | mordred: I don't think there ever where neutron ports there | 11:53 |
| mordred | weird. I must just be insane then :) | 11:57 |
| mordred | yfried: how does this look: http://paste.openstack.org/show/596031/ | 11:57 |
| mordred | gah. bad patch -one sec | 12:00 |
| mordred | http://paste.openstack.org/show/596033/ | 12:01 |
| yfried | mordred: why do you need to keep the fixed_ip check as well? | 12:06 |
| yfried | if fip['fixed_ip_address'] in fixed_ip_mapping: | 12:08 |
| yfried | mordred: ^ this line can be removed | 12:08 |
| yfried | mordred: you already verify port list, which should be identical to fixed_ip_mappings | 12:09 |
| yfried | mordred: my test works with you patch and with my suggestion | 12:10 |
| yfried | mordred: could you please submit this for review? | 12:11 |
| mordred | yfried: yup - one sec - I put in one more thing since I sent it to you based on local testing | 12:17 |
| mordred | and then I need to fix a unit test :) | 12:17 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade: Use port list to find missing floating ips https://review.openstack.org/424087 | 12:32 |
| mordred | yfried: ^^ there you go (I changed your suggestion a little - in the VERY off chance something is broken there, I'd like to be able to return a good error message. I'm 99% sure it'll never happen | 12:33 |
| mordred | Shrews: ^^ when you awaken, a review of that patch would be much appreciated so that we can potentially cut a release for yfried (although I'd also love to get that swiftservice patch landed too) | 12:34 |
| yfried | mordred: assuming Shrews approves, what's the eta of a pypi release? | 12:34 |
| mordred | no reason we can't get one done today | 12:35 |
| mordred | although I'm about to be AFK for about 5 hours, so it won't be sooner than 5 hours | 12:35 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade: Use port list to find missing floating ips https://review.openstack.org/424087 | 12:38 |
| mordred | added a reno note | 12:38 |
| mordred | Shrews: also, in testing this, I see "The cloud returned multiple addresses, and none of them seem to work. That might be what you wanted, but we have no clue what's going on, so we just picked one at random" more than I think we should | 12:40 |
| mordred | Shrews: I think we should investigate that - but nodepool is still working, so it doesn't seem to be an actual problem affecting folks | 12:40 |
| mordred | rcarrillocruz: if you happen to be around and have a few minutes ^^ https://review.openstack.org/424087 and https://review.openstack.org/#/c/415739 could use some love | 12:41 |
| mordred | and with that - I will afk - back in a few | 12:41 |
| *** _sigmavirus24 is now known as sigmavirus | 13:02 | |
| *** sigmavirus has joined #openstack-shade | 13:02 | |
| *** yfried has quit IRC | 13:10 | |
| *** gouthamr has joined #openstack-shade | 13:57 | |
| Shrews | mordred: i agree that message appears disturbingly too often. also, that patch failed the neutron test which might possibly be related since you touch the list_servers() path (thru meta) | 14:34 |
| Shrews | mordred: also, the bug referenced by jordan in 415739 seems to have interesting comments | 14:35 |
| Shrews | not sure how jordan was aware of that bug, but perhaps we should reference it somewhere? comments or commit msg | 14:37 |
| Shrews | at the "workaround" seems appropriate | 14:39 |
| *** cdent has quit IRC | 14:40 | |
| morgan | o/ | 15:06 |
| morgan | mornin | 15:06 |
| morgan | mordred: you were up early today | 15:06 |
| * morgan glares at keystone things and hurries up so can get back to shade things | 15:07 | |
| *** yolanda has quit IRC | 15:36 | |
| *** yolanda has joined #openstack-shade | 15:36 | |
| *** cdent has joined #openstack-shade | 15:37 | |
| *** yolanda has quit IRC | 15:37 | |
| *** yolanda has joined #openstack-shade | 15:38 | |
| *** ioggstream has joined #openstack-shade | 15:38 | |
| *** flvzsch50 has joined #openstack-shade | 16:06 | |
| *** yolanda has quit IRC | 16:06 | |
| *** yolanda has joined #openstack-shade | 16:06 | |
| *** gouthamr has quit IRC | 16:12 | |
| *** gouthamr has joined #openstack-shade | 16:12 | |
| *** abregman has quit IRC | 17:08 | |
| mordred | Shrews: actually - we see that error in list_servers from time to time - I have not tracked down _why_ yet | 17:53 |
| mordred | Shrews: popping a recheck in there | 17:53 |
| Shrews | poo | 17:54 |
| mordred | Shrews: yah. it's lkely the next openstack race condition we'll have the joy of debugging | 17:59 |
| morgan | yay openstack race conditions | 18:01 |
| morgan | no... wait... what is the opposite of "yay" | 18:01 |
| morgan | yeah that. | 18:01 |
| mordred | Shrews: ooh. on the 415739 bug - also, jordanP knows everything | 18:01 |
| mordred | Shrews: mind if I account for that in a follow up patch? | 18:01 |
| Shrews | mordred: that's fine. i haven't actually properly reviewed yet anyway | 18:02 |
| Shrews | mordred: let me do that now since I'm burned out on tracking down this thread problem | 18:03 |
| mordred | Shrews: well - just as a warning- it's a patch that deals wit hthreads :) | 18:03 |
| Shrews | mordred: your patch i understand. it's MY patch that is confusing me | 18:04 |
| mordred | Shrews: that's almost never true | 18:05 |
| mordred | morgan: :) | 18:05 |
| Shrews | mordred: oh, you DO reference the bug. my bad | 18:06 |
| Shrews | in _adapter.py | 18:06 |
| mordred | yah I was just about to mention that | 18:07 |
| Shrews | gah. rebased | 18:07 |
| Shrews | that makes comparing patch sets mucho harder | 18:07 |
| * Shrews haz sad | 18:07 | |
| mordred | Shrews: you can go back one patchset | 18:08 |
| mordred | the latest patchset is only a rebase patch | 18:08 |
| mordred | the previous patchset should have the latest version of that code pre-rebase | 18:08 |
| *** ioggstream has quit IRC | 18:10 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade: Upload images to swift as application/octet-stream https://review.openstack.org/424274 | 18:14 |
| mordred | Shrews: there's the followup patch | 18:14 |
| Shrews | mordred: for 423661, how difficult would it be to upload something and download to a file to test that it's the same thing? | 18:26 |
| Shrews | mordred: in a functional test, that is | 18:26 |
| mordred | Shrews: not hard - we already do an upload/download in a different test | 18:29 |
| Shrews | mordred: k. i'd really like to see one for that | 18:30 |
| mordred | cool | 18:31 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade: Upload images to swift as application/octet-stream https://review.openstack.org/424274 | 18:46 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade: Add ability to stream object directly to file https://review.openstack.org/423661 | 18:46 |
| mordred | Shrews: added test to the download patch, also changed the octet-stream patch slightly further | 18:46 |
| Shrews | mordred: k. i just exited my thread coordination hell, so will look again momentarily | 19:13 |
| *** gouthamr has quit IRC | 19:20 | |
| *** cdent has quit IRC | 19:37 | |
| *** flvzsch50 has quit IRC | 22:13 | |
| mordred | Shrews: yay no more thread contention | 22:37 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!