Earlier  
Posted Nick Remark
#openstack-sdks - 2019-09-16
11:35:41 gtema this is something we can't do in SDK as of now
11:35:54 gtema but this also leads to inability to do checksum check
11:36:22 mordred yeah. glance cli I think also supported image from stdin ... is this a thing that people really do?
11:36:31 gtema so I'm curios, what should we better do - deprecate stdin/flush to temp file/teach SDK to accept this
11:36:56 gtema no idea, but exactly this is used in the devstack, so even devstack setup already fails
11:37:38 mordred does devstack use this because it needs to stream something - or it just happens to do it?
11:38:01 gtema seems just do it this way. First download image and then pass it as stdin
11:38:02 gtema https://refstack.openstack.org/#/community_results
11:38:08 gtema sorry, wrong link
11:38:14 gtema https://github.com/openstack/devstack/blob/master/functions#L380
11:38:35 gtema sidenote: using github instead of opendev because of response performance
11:38:50 mordred I mean - we could teach sdk to understand it - but it would make some parts, like multi-part chunked upload in swift, much slower, since you can't seek on stdin
11:39:05 gtema exactly
11:39:18 gtema but deprecating this is already quite a "loud" change
11:39:31 gtema or we can flush stdin silently in OSC
11:39:36 mordred yeah. oh wow - it's actually zcatting stuff
11:39:38 gtema hehe, funny
11:39:58 gtema yeah, in archive mode - agree
11:40:02 gtema it make sense
11:40:07 mordred I mean - I guess that's an example of a use case where stdin makes sense
11:40:09 mordred yeah
11:41:12 gtema I think then really flushing stdin to temp file might be a relatively good idea, since you can use swift for upload
11:41:33 mordred it might be - although it would cause more disk to be used
11:41:46 gtema agree
11:41:55 mordred but yeah - multi-threaded upload to swift is, you know, important :)
11:42:01 gtema sure
11:42:09 gtema and also md5 check is
11:42:17 mordred yeah
11:42:59 gtema ok, then will implement this in OSC
11:43:22 gtema I don't think it make sense to be done in SDK
11:43:24 mordred maybe that's the best bet then - and then we update docs (and devstack) to use a direct filename argument when we can - because it has better performance
11:43:27 mordred yah
11:43:37 gtema ok, cool
11:43:39 gtema thks
11:44:01 gtema btw, project cleanup. We need to start working on it
11:44:24 gtema and solve caching
11:44:55 mordred yah
11:45:44 gtema I'm currently having own need for project cleanup
12:11:57 mordred gtema: got a sec for an easy one? https://review.opendev.org/#/c/601659 <-- I didn't actually write it, ignore gerrit. :)
12:12:14 gtema sure
12:12:50 gtema sone
12:13:14 gtema done I mean
12:13:38 mordred \o/
13:29:36 openstackgerrit Artem Goncharov proposed openstack/python-openstackclient master: WIP: Switch image to use SDK https://review.opendev.org/650374
13:52:14 openstackgerrit Artem Goncharov proposed openstack/python-openstackclient master: WIP: Switch image to use SDK https://review.opendev.org/650374
16:15:27 openstackgerrit Artem Goncharov proposed openstack/python-openstackclient master: WIP: Switch image to use SDK https://review.opendev.org/650374
16:57:03 openstackgerrit Artem Goncharov proposed openstack/python-openstackclient master: WIP: Switch image to use SDK https://review.opendev.org/650374
17:14:26 openstackgerrit Artem Goncharov proposed openstack/python-openstackclient master: WIP: Switch image to use SDK https://review.opendev.org/650374
18:41:25 openstackgerrit Artem Goncharov proposed openstack/openstacksdk master: Fix image create with tags https://review.opendev.org/682454
18:42:19 gtema_ mordred: ^^^. This came up in some of the devstack functional tests. We cast tags to string currently for the image
18:45:31 mordred gtema_: you left in a print - otherwise LGTM
18:46:29 gtema_ right
18:46:32 openstackgerrit Artem Goncharov proposed openstack/openstacksdk master: Fix image create with tags https://review.opendev.org/682454
18:46:47 gtema_ fixed
18:47:29 gtema_ I hoped to escape issues with followup SDK changes, since I need to raise SDK in lower-constraints as well (and would require release)
18:49:28 openstackgerrit Artem Goncharov proposed openstack/python-openstackclient master: WIP: Switch image to use SDK https://review.opendev.org/650374
19:03:23 gtema otherwise remaining tests in the OSC are green, so the change is on the good track
#openstack-sdks - 2019-09-17
00:09:17 openstackgerrit Merged openstack/keystoneauth master: Fix misspell word https://review.opendev.org/680600
07:04:56 gtema morning, was anyone able to find the reason for devstack-networking failures? It blames octavia (argument --bind_ip: Invalid IPAddress value:), but I can't find any recent suspicious change
07:23:28 gtema solved, the fix already landed
07:45:44 mordred gtema: oh yeah? good - I was just looking at that
07:45:54 mordred gtema: was just a devstack issue?
07:46:05 gtema https://review.opendev.org/#/c/682327/
07:48:30 mordred oh. hah
07:48:34 mordred yeah
07:49:50 gtema let's see whether it helps :)
07:50:48 gtema I didn't know the image is so deep in OSC. The change is now similar huge to the what it was in SDK :D
07:50:59 gtema it would be fun to wait for review
08:02:19 mordred gtema: image is such a fun topic
08:02:37 gtema yeah
11:20:55 openstackgerrit Vishakha Agarwal proposed openstack/keystoneauth master: Generate pdf documentation https://review.opendev.org/682272
11:34:20 gtema what the heck is going on, that still every change fails. Argh....
11:40:21 mordred gtema: there was a mirror issue for a second - although I think that's fixed now
11:40:48 gtema really?
11:41:20 gtema ah, yeah. lets see whether 3rd or 5th round of rechecks helps
13:11:13 openstackgerrit Merged openstack/openstacksdk master: Properly convert baremetal fields to server-side values https://review.opendev.org/680649
13:19:01 dtantsur mordred: we may need to release openstacksdk with stable/train, otherwise https://review.opendev.org/#/c/682412/ will happen
13:19:44 gtema we need to get image tags patch in, so that I finish OSC change
13:19:58 gtema s/I/I can/
13:40:43 gtema oh god, my change is again going to fail. What a wonderful day full of sporadic failures everywhere
13:53:06 dtantsur \o/
13:53:16 dtantsur it's been like that in the ironic world for weeks, I know your pain
13:53:34 dtantsur so, we need to release ASAP or at least communicate the plans to the release team
13:53:36 dtantsur mordred: ^^
13:54:17 gtema dtantsur: I need https://review.opendev.org/#/c/682454/ in the nearest release, so if you mind ...
13:54:30 gtema even tests are again failing (unrelated)
13:56:03 dtantsur gtema: please see if you want to fix my comment https://review.opendev.org/#/c/682454/
13:57:16 gtema well, I can quickly do this, if you both agree to put it into release ASAP, otherwise followup
13:59:31 dtantsur I can approve it, I think. it's a trivial modification to the already approved patch
13:59:38 dtantsur I mean, approve the updated version
14:00:01 gtema ok, gimme a sec. The current test run is anyway failing
14:02:12 openstackgerrit Artem Goncharov proposed openstack/openstacksdk master: Fix image create with tags https://review.opendev.org/682454
14:03:10 gtema done
14:03:12 dtantsur +A
14:03:32 gtema thks
14:06:09 mordred dtantsur, gtema: let me go ping the release team
14:06:16 gtema cool
14:14:09 mordred dtantsur, gtema: seems like bugfixes -so this might be an 0.35.1 - yeah?
14:14:23 gtema yupp, think so
14:52:43 openstackgerrit Merged openstack/openstacksdk master: baremetal-introspection: fix passing a Node to start_introspection https://review.opendev.org/681229

Earlier   Later