Earlier  
Later
New Search
Posted Nick Remark
#openstack-sdks - 2017-05-24
17:58:32 openstackgerrit Monty Taylor proposed openstack/api-wg master: Describe the publication of service-types-authority data https://review.openstack.org/462815
17:58:32 openstackgerrit Monty Taylor proposed openstack/api-wg master: Add document describing consuming version discovery https://review.openstack.org/459405
17:58:33 openstackgerrit Monty Taylor proposed openstack/api-wg master: Add guideline about consuming endpoints from catalog https://review.openstack.org/462814
19:17:03 openstackgerrit Merged openstack/python-openstackclient master: Use _get_token_resource in role assignment list https://review.openstack.org/464684
19:17:53 openstackgerrit Merged openstack/python-openstackclient master: JSON-ify image functional tests https://review.openstack.org/466926
19:29:04 cdent elmiko, edleafe : slim chance I might miss api-wg meeting tomorrow. definitely won't be able to be the newsletter guy; have an evening thing that I'm not sure when it starts
20:13:57 edleafe cdent: I should be able to handle things
20:14:05 cdent thanks edleafe
20:57:14 elmiko cdent: ack
#openstack-sdks - 2017-05-25
02:09:53 openstackgerrit Hong Hui Xiao proposed openstack/python-openstackclient master: Don't show hint about vlan transparent in network set https://review.openstack.org/466002
04:52:31 openstackgerrit Reedip proposed openstack/python-openstackclient master: Add default-quota to subnet pool commands https://review.openstack.org/437367
08:46:15 openstackgerrit Rui Chen proposed openstack/python-openstackclient master: Rework server migrate to support cold and live https://review.openstack.org/460059
10:43:56 openstackgerrit Sean Dague proposed openstack/keystoneauth master: Add global_request_id param to Adapter constructor https://review.openstack.org/467991
10:44:25 sdague jamielennox: if you happen to still be awake ^^^^ would be great to have eyes on
11:35:03 jamielennox sdague: that spec looks remarkably like what request id was supposed to be before it got all committeed
11:40:42 sdague jamielennox: yeh, well what is old is new again?
11:41:13 jamielennox it's nice to see some sanity being restored
11:41:29 sdague so, all that previous work predates me, but we seemed to get agreement at the summit, so I'm running at it as fast as I can while we have that
11:42:19 sdague having the 2 ids does end up being useful, so you can distinguish the fact that there are a lot of cross calls some times
11:42:53 sdague the oslo side changes are all landed and released, so now it's just plumbing clients and callers
11:44:12 jamielennox sdague: commented, code feels odd, i don't mind which way it's fixed
11:44:44 jamielennox sdague: it does feel weird to be setting this on the Adapter, adapters are created per client so you're not going to share that request_id across a lot of connections that way
11:45:51 jamielennox though with sessions lasting a long time there's not really a good spot to put a request level middleware in currently
11:47:38 sdague jamielennox: ok, I'm not sure I quite understand the concerns there
11:48:15 sdague because we don't want to be sharing request_id more broadly, the important thing is that it's relevant to the flow right now
11:49:06 jamielennox note the way i always hoped this would work is that auth_token middleware currently passes down an auth plugin. that auth plugin when used to contact another service does all the correct things like uses a service token and presents the right catalog and stuff
11:49:30 jamielennox and would be able to pass through a request-id that it received to the auth plugin and auto end up in other requests
11:49:48 jamielennox but it was super slow work getting people to adopt that and it didn't get very far
11:51:00 jamielennox sdague: yea, if it's useful there then that's fine. Re code, i just mean to store the variable in one place. I don't mind if it's a property around additional_headers or if you manually add it to headers later like we do for user_agent, just not stored twice
11:52:32 sdague ok
12:09:22 sdague jamielennox: for reference - https://review.openstack.org/#/c/467242/ is what the nova change looks like with some passing on to clients
12:10:06 sdague https://review.openstack.org/#/c/467417/ is it all glued together, you'll see some global_ids logged in cinder / neutron with that
12:10:10 jamielennox sdague: yea, you can get a fair way with additional_headers now, but i'm happy to make it a real thing
12:12:00 sdague jamielennox: yeh, I've been just direct plumbing into neutron/cinder/glance clients, but the neutron folks wanted it in keystonauth. I'm kind of fine either way, because it's pretty much just a single field to pop out on the constructor if it's totally at the top level
12:12:51 sdague even though we're setting global_request_id in constructors quite often in time, in code, it's typically consolidated to a single location in the caller
12:27:39 openstackgerrit Matthew Booth proposed openstack/os-client-config master: Don't pop from os.environ https://review.openstack.org/468020
12:47:13 openstackgerrit Sean Dague proposed openstack/keystoneauth master: Add global_request_id param to Adapter constructor https://review.openstack.org/467991
12:53:01 openstackgerrit Rui Chen proposed openstack/python-openstackclient master: Rework server migrate to support cold and live https://review.openstack.org/460059
13:31:55 openstackgerrit Monty Taylor proposed openstack/os-client-config master: Don't pop from os.environ https://review.openstack.org/468020
13:32:01 mordred Shrews, dtroyer: if you get time, I'd really like to get the top five patches from https://review.openstack.org/#/q/project:openstack/os-client-config+status:open+label:Workflow=0 landed and a new occ out - they're all small/easy
13:32:16 mordred Shrews, dtroyer: but there are a few behavior bugs that are biting people
13:38:10 Shrews mordred: k
13:42:07 Shrews mordred: is https://review.openstack.org/465195 still relevant with https://review.openstack.org/468020 ?
13:42:54 mordred Shrews: the reason for writing it isn't - but it's still a good idea I think - the OpenStackConfig object is really intended to be a singleton
15:19:49 openstackgerrit Reedip proposed openstack/python-openstackclient master: Introduce quota unset command https://review.openstack.org/376311
15:34:29 openstackgerrit Reedip proposed openstack/python-openstackclient master: Add extra dhcp option to 'port create/set/unset' https://review.openstack.org/356263
15:36:57 openstackgerrit Monty Taylor proposed openstack/api-wg master: Add guidelines on Version Discovery https://review.openstack.org/459710
15:36:58 openstackgerrit Monty Taylor proposed openstack/api-wg master: Describe the publication of service-types-authority data https://review.openstack.org/462815
15:36:58 openstackgerrit Monty Taylor proposed openstack/api-wg master: Add document describing consuming version discovery https://review.openstack.org/459405
15:37:32 mordred cdent, edleafe, elmiko: I believe I've gotten all current reviews taken care of
15:37:43 cdent yay!
15:38:18 edleafe mordred: on my list
15:46:52 elmiko mordred: sweet, i'm rounding the corner on my last lap for the reviews you posted
15:47:59 mordred elmiko: woot!
15:48:10 elmiko also kudos on consuming-version-discovery.rst, this thing is a beast!
15:52:17 mordred elmiko: it really is, isn't it?
15:52:34 mordred elmiko: the scary part is that somehow I actually have all of that stored in my head
15:53:09 elmiko mordred: that is scary...
15:53:31 elmiko mordred: i really like your enumeration of the algorithms though, i think it will help readers understand this better
15:54:57 mordred elmiko: thanks! ... I certainly hope so
15:55:13 mordred the fun part is that next I need to write a spec for keystone about implementing this in keystoneauth :)
15:56:35 elmiko good luck sir
15:58:48 openstackgerrit Yuval Shalev proposed openstack/python-openstacksdk master: Added server console output method https://review.openstack.org/468078
17:32:48 elmiko edleafe: draft up, https://etherpad.openstack.org/p/api-wg-newsletter
17:32:57 elmiko i realize you may be out, i'll wait 15-20 then post it
18:08:38 openstackgerrit Dean Troyer proposed openstack/python-openstackclient master: Add flavor value options https://review.openstack.org/302795
19:30:17 openstackgerrit Merged openstack/python-openstackclient master: Convert volume functional tests into JSON format https://review.openstack.org/465208
19:31:34 openstackgerrit Monty Taylor proposed openstack/api-wg master: Add guidelines on Version Discovery https://review.openstack.org/459710
19:31:34 openstackgerrit Monty Taylor proposed openstack/api-wg master: Add support for historical service type aliases https://review.openstack.org/460654
19:31:35 openstackgerrit Monty Taylor proposed openstack/api-wg master: Describe the publication of service-types-authority data https://review.openstack.org/462815
19:31:35 openstackgerrit Monty Taylor proposed openstack/api-wg master: Add document describing consuming version discovery https://review.openstack.org/459405
19:31:36 openstackgerrit Monty Taylor proposed openstack/api-wg master: Add guideline about consuming endpoints from catalog https://review.openstack.org/462814
19:53:51 openstackgerrit Merged openstack/python-openstackclient master: Add functional test for volume service https://review.openstack.org/432282
20:20:10 dhellmann dtroyer : in ATL we talked about a daemon mode for cliff/OSC. My notes say "add an option to emit a null value after the formatter output". Is that really all we need there?
20:50:05 dtroyer dhellmann: that is the minimum needed to delimit return data for each command. I didn't want to invent a protocol for this so hopefully that is all we would need
20:50:24 dhellmann dtroyer : ok
20:50:49 dhellmann do we want a --daemon-mode that implies --interactive and also adds the null? or just a --null-delimited-output option?
20:52:08 dtroyer I was only thinking about the —null option as interactive is already implied when no command is given
20:52:17 dhellmann makes sense
20:52:28 dhellmann just --null then?
20:52:44 dhellmann not --null-delimited-output?
20:53:27 dtroyer hmmm, it may need to be more specific, to be honest I have been thinking about cribbing the xargs -0 option, not sure if we need to be able to specify the delimiter char
20:53:38 dhellmann oh, -0 works
20:53:56 dhellmann ok, the other thing we talked about was "command extensions"
20:54:00 dhellmann I'm looking at that right now
20:54:04 dtroyer the semantics are not identical to xargs though so maybe that connection is not quite as useful
20:54:17 dhellmann I can either load them from a separate namespace for each use (args, pre, & post hooks)
20:54:34 dhellmann or I can load them from one namespace and expect a class with a specific API for each of those cases
20:54:39 dhellmann any preference?
20:55:01 dhellmann I'll make --null-delimited-output and -0 aliases
20:55:26 dtroyer I've imagined it as a single class with essentially hook methods to override, but I don't have a huge preference yet
20:55:33 dtroyer the alias sounds good
20:55:55 dhellmann ok, I was leaning toward a class, too
20:56:04 dtroyer having a single namespace does feel cleaner
20:56:27 dhellmann I was going to make the namespace the application namespace + cmd_name.replace(' ', '_')
20:56:39 dhellmann then plugins can have their own names
20:56:56 dtroyer sounds good
20:56:58 dhellmann k
21:42:20 dtroyer dhellmann: if you're still around… with the major rev of cliff on the near horizon, do you want to wait for that release for this stuff to land ^^^^
22:55:10 openstackgerrit Merged openstack/os-client-config master: Add helper method to fetch service catalog https://review.openstack.org/464238

Earlier   Later