| *** mhen_ is now known as mhen | 01:14 | |
| mhen | Hi! I am reading that DEFAULT.api_paste_config is a deprecated option as per https://docs.openstack.org/oslo.service/2025.1/configuration/#wsgi | 09:05 |
|---|---|---|
| mhen | However, the audit middleware of Keystone requires the paste ini to be edited: https://docs.openstack.org/keystonemiddleware/latest/audit.html#enabling-audit-middleware | 09:06 |
| mhen | Is there any successor to this option once it is removed? | 09:06 |
| hberaud[m] | o/ | 09:06 |
| hberaud[m] | Well, the api_past_config is deprecated because of the removal of Eventlet from OpenStack | 09:07 |
| hberaud[m] | oslo.service rely on the WSGI module of Eventlet | 09:07 |
| hberaud[m] | Keystone also have to be migrated off of Eventlet | 09:08 |
| hberaud[m] | So, normally both projects will be aligned once the migration will be done | 09:08 |
| hberaud[m] | In all the case oslo.service will not remove these deprecated things until services are not fully migrated | 09:09 |
| mhen | Alright, thank oyu hberaud[m] | 09:09 |
| mhen | *thank you | 09:09 |
| hberaud[m] | Keystone will have to find a replacement for the WSGI features of oslo.service, so it will surely expose its own config options | 09:09 |
| hberaud[m] | You are welcome, hopefully these details help you, you can grab more details about the status of the migration here https://removal.eventlet.org/, there https://www.youtube.com/watch?v=RIc-Tut95YM and also there https://www.linkedin.com/groups/13183090/ | 09:11 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: docs: Remove use of sphinxcontrib-apidoc https://review.opendev.org/c/openstack/pbr/+/949170 | 09:38 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: tox: Stop calling 'setup.py' https://review.opendev.org/c/openstack/pbr/+/949188 | 09:38 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Do not use the onerror parameter in shutil.rmtree() https://review.opendev.org/c/openstack/pbr/+/924803 | 09:40 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: docs: Remove use of sphinxcontrib-apidoc https://review.opendev.org/c/openstack/pbr/+/949170 | 10:30 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: tox: Stop calling 'setup.py' https://review.opendev.org/c/openstack/pbr/+/949188 | 10:30 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: docs: Update usage guide and README https://review.opendev.org/c/openstack/pbr/+/949051 | 10:30 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Deprecate '[entry_points]' section https://review.opendev.org/c/openstack/pbr/+/949052 | 10:30 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Deprecate '[files]' section https://review.opendev.org/c/openstack/pbr/+/949053 | 10:30 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Deprecate various '[metadata]' options https://review.opendev.org/c/openstack/pbr/+/949054 | 10:30 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Deprecate various '[backwards_compat]' options https://review.opendev.org/c/openstack/pbr/+/949055 | 10:30 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: WIP: Remove need for setup.cfg if pyproject.toml present https://review.opendev.org/c/openstack/pbr/+/949049 | 10:30 |
| opendevreview | Merged openstack/pbr master: Modernize tests to use EXT_SUFFIX, fix PyPy https://review.opendev.org/c/openstack/pbr/+/940773 | 12:47 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Add black https://review.opendev.org/c/openstack/pbr/+/954040 | 13:49 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Tweak loading distribution version https://review.opendev.org/c/openstack/pbr/+/954041 | 13:49 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Use importlib.metadata where possible https://review.opendev.org/c/openstack/pbr/+/954042 | 13:49 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Use packaging where possible https://review.opendev.org/c/openstack/pbr/+/954043 | 13:49 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Cleanup removal of 'build_sphinx' command doc https://review.opendev.org/c/openstack/pbr/+/954044 | 13:49 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Remove 'test' distutils command override https://review.opendev.org/c/openstack/pbr/+/954045 | 13:49 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Move fixtures to their own module https://review.opendev.org/c/openstack/pbr/+/954046 | 13:49 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Remove support for custom compilers https://review.opendev.org/c/openstack/pbr/+/954047 | 13:49 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Move distutils commands to _compat https://review.opendev.org/c/openstack/pbr/+/954048 | 13:49 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Move functional tests to their own directory https://review.opendev.org/c/openstack/pbr/+/954049 | 13:49 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: WIP: Deprecate tests_require option https://review.opendev.org/c/openstack/pbr/+/954050 | 13:49 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: tox: Stop calling 'setup.py' https://review.opendev.org/c/openstack/pbr/+/949188 | 14:32 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: docs: Update usage guide and README https://review.opendev.org/c/openstack/pbr/+/949051 | 14:34 |
| opendevreview | Merged openstack/pbr master: docs: Remove use of sphinxcontrib-apidoc https://review.opendev.org/c/openstack/pbr/+/949170 | 14:36 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: docs: Update usage guide and README https://review.opendev.org/c/openstack/pbr/+/949051 | 14:37 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Deprecate '[entry_points]' section https://review.opendev.org/c/openstack/pbr/+/949052 | 14:37 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Deprecate '[files]' section https://review.opendev.org/c/openstack/pbr/+/949053 | 14:37 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Deprecate various '[metadata]' options https://review.opendev.org/c/openstack/pbr/+/949054 | 14:37 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Deprecate various '[backwards_compat]' options https://review.opendev.org/c/openstack/pbr/+/949055 | 14:37 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Add black https://review.opendev.org/c/openstack/pbr/+/954040 | 14:40 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Tweak loading distribution version https://review.opendev.org/c/openstack/pbr/+/954041 | 14:40 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Use importlib.metadata where possible https://review.opendev.org/c/openstack/pbr/+/954042 | 14:40 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Use packaging where possible https://review.opendev.org/c/openstack/pbr/+/954043 | 14:40 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Cleanup removal of 'build_sphinx' command doc https://review.opendev.org/c/openstack/pbr/+/954044 | 14:40 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Remove 'test' distutils command override https://review.opendev.org/c/openstack/pbr/+/954045 | 14:40 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Move fixtures to their own module https://review.opendev.org/c/openstack/pbr/+/954046 | 14:40 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Remove support for custom compilers https://review.opendev.org/c/openstack/pbr/+/954047 | 14:40 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Move distutils commands to _compat https://review.opendev.org/c/openstack/pbr/+/954048 | 14:40 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Move functional tests to their own directory https://review.opendev.org/c/openstack/pbr/+/954049 | 14:40 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: docs: Update usage guide and README https://review.opendev.org/c/openstack/pbr/+/949051 | 14:40 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Deprecate '[entry_points]' section https://review.opendev.org/c/openstack/pbr/+/949052 | 14:40 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Deprecate '[files]' section https://review.opendev.org/c/openstack/pbr/+/949053 | 14:40 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Deprecate various '[metadata]' options https://review.opendev.org/c/openstack/pbr/+/949054 | 14:40 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Deprecate various '[backwards_compat]' options https://review.opendev.org/c/openstack/pbr/+/949055 | 14:40 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Deprecate 'tests_require' option https://review.opendev.org/c/openstack/pbr/+/954050 | 14:40 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: tox: Stop calling 'setup.py' https://review.opendev.org/c/openstack/pbr/+/949188 | 14:40 |
| stephenfin | mhen: Are you sure you're talking about the same thing. The oslo.service variant of the option may be deprecated, but most services provide their own implementation. For example, here is nova's https://opendev.org/openstack/nova/src/branch/master/nova/conf/wsgi.py | 14:42 |
| stephenfin | hberaud[m]: ^ that is likely something to be addressed for projects who currently inherit the option from oslo.service. The option isn't strictly related to the eventlet. It can also be used under WSGI deployments as nova does | 14:44 |
| hberaud[m] | ack | 14:44 |
| stephenfin | IMO, we would benefit from an oslo.wsgi project (and perhaps and oslo.asgi project) to store things like this, along with e.g the openapi validation stuff | 14:44 |
| stephenfin | I started bootstrapping such a project but haven't had time to complete it | 14:45 |
| hberaud[m] | indeed, the adaptation of services depends on their rootage over oslo.service | 14:45 |
| mhen | I'm a bit confused by this. I was able to activate the middleware through '[DEFAULT] api_paste_config =' adjustment in Nova. If I put it into '[wsgi]' as well, which value will be used? | 14:45 |
| hberaud[m] | stephenfin: Since the begining of the eventlet removal intiative I proposed, several times, discussions around wsgi/asgi in OpenStack for the coming years, but for now I didn't get tons of feedback, let me know if you want to revive this topic | 14:47 |
| stephenfin | mhen: '[DEFAULT] api_paste_config' is the deprecated alias for '[wsgi] api_paste_config'. You can switch over https://docs.openstack.org/nova/latest/configuration/config.html#wsgi.api_paste_config | 14:49 |
| hberaud[m] | I think this topic is for now, more or less, out of the boundaries of the eventlet removal | 14:49 |
| hberaud[m] | but there is interconnection points | 14:49 |
| hberaud[m] | IMO it would deserve a side community goal | 14:49 |
| mhen | stephenfin: thank you. What if both are specified but carry different values? Is [wsgi] always winning? | 14:50 |
| hberaud[m] | to define and formalize what we want to achieve around wsgi/asgi | 14:50 |
| stephenfin | I agree it's mostly out of scope. The relationship is there because oslo.service also holds things like Router, Request, and Loader | 14:50 |
| hberaud[m] | exactly | 14:51 |
| stephenfin | and this error message means the whole thing is deprecated, rather than just the server https://github.com/openstack/oslo.service/blob/master/oslo_service/wsgi.py#L40-L50 | 14:51 |
| stephenfin | I think that's wrong actually | 14:51 |
| stephenfin | only Server needs to be deprecated, no? | 14:52 |
| stephenfin | mhen: That would be up to oslo.config and I'm not sure what the behaviour is | 14:52 |
| hberaud[m] | What do you mean by server? Just to be sure to be on the same page | 14:53 |
| stephenfin | this class https://github.com/openstack/oslo.service/blob/master/oslo_service/wsgi.py#L70 | 14:53 |
| stephenfin | we're deprecating that because it uses ServiceBase, which is an eventlet thing (right?) | 14:53 |
| stephenfin | but there's nothing eventlet'y in Router https://github.com/openstack/oslo.service/blob/master/oslo_service/wsgi.py#L261 | 14:54 |
| hberaud[m] | this class is deprecated because it rely on import eventlet.wsgi | 14:54 |
| stephenfin | Yes, i understand. But we haven't deprecated the class. We've deprecated the whole module | 14:54 |
| stephenfin | We should only be deprecating the class | 14:55 |
| stephenfin | IMO | 14:55 |
| stephenfin | at least until we have a new home for the wsgi stuff, which never should have been put here | 14:55 |
| hberaud[m] | indeed, I think you are right, the other class could remains valid (at first glance) | 14:55 |
| hberaud[m] | *classes | 14:56 |
| stephenfin | yes, fwiw, those are copy-pasted from Nova (at least fwict) and Nova still uses them just fine | 14:56 |
| hberaud[m] | good to know that | 14:56 |
| hberaud[m] | Then they could be moved to oslo.sgi | 14:57 |
| hberaud[m] | but, I also think that without the Server class, this module is a bit meaningless | 14:58 |
| stephenfin | as I said, this code should never have been put into oslo.service: it's unrelated to the core mission of that library. However, it is there now | 14:59 |
| stephenfin | until we find a new home, I don't believe we can or should remove it | 14:59 |
| stephenfin | for example: you're breaking neutron https://opendev.org/openstack/neutron/src/branch/master/neutron/api/wsgi.py#L94 | 15:00 |
| stephenfin | and manila https://opendev.org/openstack/manila/src/branch/master/manila/api/openstack/__init__.py#L142 | 15:01 |
| stephenfin | and cinder https://opendev.org/openstack/cinder/src/branch/master/cinder/api/openstack/__init__.py#L104 | 15:01 |
| stephenfin | you get the idea :) | 15:01 |
| hberaud[m] | I agree. Good point | 15:01 |
| stephenfin | I can submit a fix if you want | 15:01 |
| hberaud[m] | Yes please, if you do not mind | 15:02 |
| opendevreview | Stephen Finucane proposed openstack/oslo.service master: Don't deprecate the entire oslo_service.wsgi module https://review.opendev.org/c/openstack/oslo.service/+/954055 | 15:07 |
| stephenfin | hberaud[m]: damani[m]: ^ | 15:07 |
| stephenfin | IMO that needs to go back to stable/2025.1 also | 15:08 |
| stephenfin | I'm shocked no one raised this before now, btw. That could be a symptom of a larger issues wrt warnings | 15:08 |
| opendevreview | Stephen Finucane proposed openstack/oslo.service stable/2025.1: Don't deprecate the entire oslo_service.wsgi module https://review.opendev.org/c/openstack/oslo.service/+/954056 | 15:09 |
| hberaud[m] | ack | 15:17 |
| hberaud[m] | I wonder if the remaining classes should not be moved to the common module of the backend | 15:35 |
| frickler | the good thing about deprecation warnings is that there are so many of them that it is easy to get trained to ignore them, yes :-/ | 15:35 |
| stephenfin | hberaud[m]: wdym? | 15:35 |
| hberaud[m] | the backend notion of oslo.service introduced a common module to store all the things shared by all the backends, so if we decide to keep the remaining classes of the WSGI module, then I think we would have to store them into the common module to avoid duplicating the code in both backends. | 15:37 |
| hberaud[m] | else, if I'm right, the threading backend won't seen them | 15:38 |
| stephenfin | why would the backend need to see these? | 15:38 |
| hberaud[m] | https://opendev.org/openstack/oslo.service/src/branch/master/oslo_service/backend/common | 15:39 |
| hberaud[m] | to expose them to users | 15:39 |
| stephenfin | sorry, I'm missing something | 15:39 |
| hberaud[m] | but I may be wrong | 15:39 |
| stephenfin | as user currently does 'from oslo_service import wsgi; wsgi.Request(...)'. What's changing? | 15:39 |
| stephenfin | s/as/a/ | 15:40 |
| stephenfin | or rather what will need to change? | 15:40 |
| hberaud[m] | My memory is surely wrong, so take it with with tweezers with my following assertions. For now the threading backend expected that the wsgi module would disappear. there is a mapping for each backend (https://opendev.org/openstack/oslo.service/src/branch/master/oslo_service/backend/eventlet/__init__.py, https://opendev.org/openstack/oslo.service/src/branch/master/oslo_service/backend/eventlet/__init__.py). So I'm not sure that if a | 15:43 |
| hberaud[m] | service switching the backend will still be able to import the wsgi module of oslo.service | 15:43 |
| hberaud[m] | s/My memory is surely wrong, so take it with with tweezers with my following assertions./My memory is surely wrong, so take my following assertions with tweezers/ | 15:44 |
| damani[m] | stephenfin, thanks a lot for the patch | 15:45 |
| hberaud[m] | damani: correct me if I'm wrong | 15:45 |
| hberaud[m] | (about the common and the import) | 15:46 |
| damani[m] | so the server must be deprecated yes | 15:48 |
| damani[m] | the rest can be useful and can be probably move the the common yes | 15:48 |
| hberaud[m] | ack, but is the move to common optional or mandatory? from a user perspective who imports the wsgi module with the threading backend enabled? | 15:50 |
| hberaud[m] | damani: ^ | 15:51 |
| opendevreview | Merged openstack/oslo.service master: Fix inconsistent interface of ThreadGroup https://review.opendev.org/c/openstack/oslo.service/+/953032 | 16:22 |
| *** haleyb is now known as haleyb|out | 21:52 | |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!