Earlier  
Later
New Search
Posted Nick Remark
#openstack-sdks - 2018-04-20
13:48:10 pooja_jadhav ??
13:49:05 cdent I can talking with you about it, but it will mostly be you explaining what you're trying to do to me, and maybe as part of that things will be more clear
13:49:48 pooja_jadhav cdent: Actually, I am facing issue while using split logger parameter in nova-cinder interaction.
13:50:07 pooja_jadhav cdent: i will tell you what i am trying to do.
13:50:59 pooja_jadhav cdent: after line https://github.com/openstack/nova/blob/master/nova/volume/cinder.py#L86.. i am setting _SESSION._split_loggers = True just for time being testing.
13:51:45 pooja_jadhav so when request goes from nova to cinder ( i should request into see split logging) as passed split_logger = True
13:53:25 pooja_jadhav cdent: But the code in keystoneauth at line https://github.com/openstack/keystoneauth/blob/master/keystoneauth1/session.py#L613-L617 is making split_logger = False if logger is already passed. and because of this request-ids are not getting logged
13:53:25 cdent I think instead of doing _SESSION._split_loggers = True after that call you instead need to pass split_loggers=True in the call to load_session_from_conf_options
13:53:33 cdent does that make any difference
13:54:13 pooja_jadhav but this load_session_from_conf_options method dont have this parameter
13:55:00 pooja_jadhav how can I pass that split_logger to load_session_from_conf_options
13:56:27 cdent pooja_jadhav: from what I can tell load_session_from_conf_options takes kwargs:
13:57:25 cdent https://github.com/openstack/keystoneauth/blob/master/keystoneauth1/loading/session.py#L246
13:57:53 pooja_jadhav ok
14:00:13 pooja_jadhav but the issue is, logger is getting passed from some where
14:00:47 pooja_jadhav and that if condition in keystoneauth at line https://github.com/openstack/keystoneauth/blob/master/keystoneauth1/session.py#L613-L617 making split_logger to False
14:02:23 pooja_jadhav so it only logs (request, response and response body) except request-id
14:05:38 openstackgerrit Mário Santos proposed openstack-infra/shade master: Adds kwargs support when creating a Neutron subnet https://review.openstack.org/563134
14:09:25 cdent sorry pooja_jadhav got distracted for a moment, re-reading
14:09:36 pooja_jadhav no issue
14:10:26 cdent Based on what you are saying I think you've found a bug.
14:12:23 cdent if you can't find mordred, you might try speaking with cmurphy or lbragstad ?
14:12:51 cdent If you're able to come up with a minimal test case that demostrates the problem, that would probably be useful too
14:14:24 cmurphy kmalloc might also be able to help with that
14:14:35 cmurphy I can try but would have to refresh my memory a bit
14:16:33 pooja_jadhav ok
14:17:39 pooja_jadhav in keystonenauth, in def request()- if logger is passed then split_logger sets to False, and because of this request-ids are not getting logged
14:17:50 pooja_jadhav In my opion also, it a bug
14:18:47 pooja_jadhav My aim is to use split_logger functionality wherever keystoneauth session is getting used.
14:19:51 pooja_jadhav I already had discussion with kmalloc
14:20:50 cmurphy pooja_jadhav: I think if a logger is passed then that indicates you only want to use one logger, ie the one that you passed, and so splitting loggers wouldn't make sense
14:22:40 pooja_jadhav cmurphy: ohk, But I am able to see (3 parts of request in splits like(keystoneauth.session.request, keystoneauth.session.response, keystoneauth.session.body)
14:23:16 pooja_jadhav cmurphy: Only the part I am not able to see is keystoneauth.session.request-id
14:26:18 cmurphy pooja_jadhav: so you think something around here is going wrong? https://github.com/openstack/keystoneauth/blob/master/keystoneauth1/session.py#L772-L775
14:26:52 pooja_jadhav cmurphy: yes
14:27:35 pooja_jadhav cmurphy: Split_logger is getting as False and because of that request-is is not get logged.
14:28:42 dtroyer_zz adriant: it has been the plan for OSC to use the SDK once it has a 1.0 release since the SDK was started. We relented on the 1.0 for Network in order to not add the neutronclient dependency and other reasons, and it has been painful as the SDK has made at least two changes we have been required to work around and handle multiple SDK versions for.
14:29:01 cmurphy pooja_jadhav: you said you're passing in a logger and that's why it's getting set to false, can you not pass in a logger?
14:30:16 dtroyer_zz The api.object layer in OSC was largely lifted from swiftclient since swiftclient was still only available in the swift repo at the time. I put it into the form I was proposing for a low-level SDK that was subsequently not merged. and it has sat there ever since waiting for the SDK 1.0 as explained above
14:31:45 dtroyer_zz so yes, once we have an SDK 1.0 folks can go nuts converting commands to use it, with adequate compatibility testing in place
14:32:22 pooja_jadhav pooja_jadhav: in cinder client they are getting logger, I have comment out that code, i am able to see request-ids but, it will solve only for cinder client.
14:32:36 pooja_jadhav cmurphy: in cinder client they are getting logger, I have comment out that code, i am able to see request-ids but, it will solve only for cinder client.
14:33:32 pooja_jadhav cmurphy: at line https://github.com/openstack/python-cinderclient/blob/master/cinderclient/v3/client.py#L115-L116
14:34:12 pooja_jadhav cmuphy: IMO, we should fix at central place so that, i will solve for all other services issue
14:34:41 pooja_jadhav cmuphy: But still Not clear how to fix at keystoneauth.
14:35:50 cmurphy pooja_jadhav: I don't really agree, I think any client that want to opt into using the split loggers should be explicit about it by setting split_loggers to true and stopping passing in its own logger
14:37:01 pooja_jadhav cmurphy: Ohk, I will try to fix that in cinder client then,
14:37:29 cmurphy ok :)
14:37:57 pooja_jadhav cmurphy, cdent : Thanks for your inputs :)
15:28:42 openstackgerrit Mário Santos proposed openstack-infra/shade master: Adds kwargs support when creating a Neutron subnet https://review.openstack.org/563134
16:21:08 openstackgerrit Stephen Finucane proposed openstack/os-api-ref master: Raise warning if top-level value is not a dict https://review.openstack.org/563178
16:21:09 openstackgerrit Stephen Finucane proposed openstack/os-api-ref master: Use 'sphinx.util.logging' https://review.openstack.org/563179
16:33:12 melwitt dtroyer_zz: hi, we're trying to resolve a problem where we couldn't update the constraint for python-novaclient to 10.1.0 because of removed functionality that OSC couldn't leverage in queens. because of this, users that install queens are unable to use the volume multi-attach feature with the older novaclient (old review https://review.openstack.org/#/c/538070),
16:34:10 melwitt we've proposed a backport of the adjustments from rocky to queens in an effort to be able to update the queens constraint to 10.1.0 https://review.openstack.org/#/c/562840
16:34:34 melwitt your review would be appreciated
18:50:48 openstackgerrit melissaml proposed openstack/keystoneauth master: Trivial: Update pypi url to new url https://review.openstack.org/563227
20:26:46 dtroyer_zz melwitt: +2 I'd like amotoki to stamp it too since this is adding functionality to a stable release. We can't properly version this when we release it…
20:30:56 melwitt understood. thanks dtroyer_zz
20:31:18 openstackgerrit melissaml proposed openstack/openstacksdk master: Trivial: Update pypi url to new url https://review.openstack.org/563287
22:16:41 kmalloc Ok
#openstack-sdks - 2018-04-21
10:37:24 openstackgerrit melissaml proposed openstack-infra/shade master: Trivial: Update pypi url to new url https://review.openstack.org/563412
18:52:59 fungi not sure how many in here follow the general openstack@l.o.o ml, but in case you missed it there's been a unified sdk vs client libs thread going there since wednesdayish: http://lists.openstack.org/pipermail/openstack/2018-April/046375.html
18:54:34 fungi might be nice to double-check for fud/misinformation as well as popular feature parity gaps
#openstack-sdks - 2018-04-23
04:51:19 openstackgerrit Merged openstack/python-openstackclient master: Fix functional job failed https://review.openstack.org/562345
06:55:11 openstackgerrit Theodoros Tsioutsias proposed openstack/python-openstackclient master: Add project info to server list from all projects https://review.openstack.org/562300
06:55:33 openstackgerrit Theodoros Tsioutsias proposed openstack/python-openstackclient master: Add force parameter for quota set https://review.openstack.org/551811
08:12:26 openstackgerrit Merged openstack/python-openstackclient master: Network: Add tag support for floating ip https://review.openstack.org/547316
08:41:05 openstackgerrit Jens Harbott (frickler) proposed openstack/python-openstackclient master: Add dns-domain support to Network object https://review.openstack.org/516701
08:41:05 openstackgerrit Jens Harbott (frickler) proposed openstack/python-openstackclient master: Add DNS support to floating IP commands https://review.openstack.org/558824
09:37:36 openstackgerrit Jens Harbott (frickler) proposed openstack/python-openstackclient master: Add DNS support to floating IP commands https://review.openstack.org/558824
09:49:35 openstackgerrit Merged openstack/python-openstackclient master: Fix limits show command without Nova and Cinder https://review.openstack.org/558315
10:40:06 frickler mordred: should we make bifrost checks non-voting for shade until someone fixes their pip10 issues? https://review.openstack.org/563134 http://logs.openstack.org/34/563134/2/check/bifrost-integration-tinyipa/edf7b16/job-output.txt.gz#_2018-04-23_08_02_49_967207
10:46:52 adriant fungi: I kind of responded to that thread, although I probably come across a bit ranty at the standalone clients :P
10:47:12 adriant mordred: apologies for maybe sounding a bit ranty :P
10:49:11 adriant and I wish I had more time to contribute to the SDK :(
10:49:16 adriant soon, maybe...
11:12:03 cdent I enjoyed that response adriant. Didn't sound ranty to me. More frustrated.
12:56:59 frickler amotoki: dtroyer_zz: dhellmann: osc-cores: these patches would like to get reviewed before I need to start rebasing again ;) https://review.openstack.org/516701 https://review.openstack.org/558824 https://review.openstack.org/559742 https://review.openstack.org/562300
12:57:41 rabel hi there. when i go to https://bugs.launchpad.net/python-openstackclient/+bugs there is a message "python-openstackclient must be configured in order for Launchpad to forward bugs to the project's developers. " and i seem not to be able to report a bug
13:01:15 frickler rabel: yes, the bugs have been migrated to storyboard
13:01:38 frickler rabel: https://storyboard.openstack.org/#!/project/975
13:01:52 rabel frickler: thanks. so i have to click "add story" to report a bug?
13:02:28 frickler rabel: correct
13:02:35 rabel frickler: kthx
13:24:34 elmiko cdent: ack about the single day
13:24:47 cdent ack ack
13:34:51 elmiko cdent: now you're making me think about the aliens from the movie _Mars Attacks_ XD
13:35:06 cdent you're welcome
13:45:17 dtroyer_zz frickler: I'm done travelling for a while, hope to get caught up Real Soon Now
13:55:51 amotoki frickler: ack. sorry for late. I have been busy due to the beginning of FY here.
14:26:00 openstackgerrit Mário Santos proposed openstack-infra/shade master: Adds kwargs support when creating a Neutron subnet https://review.openstack.org/563134
14:49:31 openstackgerrit Merged openstack/api-wg master: Remove use of Python builtin name for variables https://review.openstack.org/560994
20:19:48 openstackgerrit Ilya Margolin proposed openstack/openstacksdk master: Fix openstack-inventory https://review.openstack.org/563754
20:40:51 openstackgerrit Ilya Margolin proposed openstack/openstacksdk master: Fix openstack-inventory https://review.openstack.org/563754
#openstack-sdks - 2018-04-24
02:33:20 adriant mordred: not sure if bug or something silly I'm doing... but list(conn.image.images()) against v2 throws me a not found error because it is trying to call "/v2/v2/images"
02:33:49 adriant also, I didn't realise this at first, but because of how my interpreter piggybacks on the OSC if you call it with --debug... it work!
02:33:55 adriant which is weird, but so very useful
02:34:24 adriant you get a shit load of debug!
02:53:52 adriant also GAH!!! WHY GLANCE! The API limits to 25 items by default... so the fucking glance client uses the pagination to get around that and 'list all' which means what you think is one API call without a limit set is like 20
02:54:50 adriant with the sdk you have to explicitly set a crazy high limit to get around this, which is fine, but a failing of the API (which can't be changed)

Earlier   Later