| *** jiaopengju has joined #openstack-karbor | 00:40 | |
| *** zhurong has joined #openstack-karbor | 00:49 | |
| *** liujiong has joined #openstack-karbor | 01:42 | |
| openstackgerrit | chenying proposed openstack/karbor master: Spec: Add Quotas to Karbor https://review.openstack.org/496579 | 02:50 |
|---|---|---|
| *** gouthamr has quit IRC | 03:49 | |
| *** zhurong has quit IRC | 04:29 | |
| *** zhurong has joined #openstack-karbor | 04:48 | |
| openstackgerrit | chenying proposed openstack/karbor master: Implement Karbor REST API microversions https://review.openstack.org/499465 | 06:15 |
| *** chenying has quit IRC | 06:45 | |
| *** chenying has joined #openstack-karbor | 07:09 | |
| jiaopengju | hi, I have a question about status_poll in karbor. I found that we use ‘_get_volume_status’, ‘get_image_status’ etc to poll the resource status with the specify client(never renew it). In some user cases, the client token may be expired, so the status poll will fail. Should we see this as a bug or just modify the auth expire time in keystone? | 07:11 |
| openstackgerrit | Merged openstack/karbor master: Fix restore security group error https://review.openstack.org/498795 | 07:12 |
| openstackgerrit | Merged openstack/karbor master: Add project_id in neutron operations https://review.openstack.org/498794 | 07:12 |
| yuval | jiaopengju: have you encountered that? | 07:13 |
| jiaopengju | yuval: yes, I’m doing integration with freezer. When I doing backup a cinder volume with using freezer plugin, it cost a long time, so the freezer client expired and throws 401 error | 07:14 |
| jiaopengju | yuval: so I must renew it every time | 07:15 |
| jiaopengju | yuval: I think freezer client is same as glance client, cinder client etc | 07:16 |
| yuval | we once spoke about renewing tokens or getting tokens with very long timeout | 07:17 |
| yuval | jiaopengju: let me check, one minute | 07:17 |
| *** zhonghua has quit IRC | 07:21 | |
| *** zhonghua has joined #openstack-karbor | 07:22 | |
| chenying | hi yuval jiaopengju I have submitted the patch about API microversions for karbor. welcome to review them. : ) | 07:26 |
| yuval | chenying: already have | 07:26 |
| chenying | I implement API microversions for karbor refer to the microversions implementation of manila and cinder projects. | 07:27 |
| jiaopengju | chenying: I will review it later | 07:27 |
| chenying | hi yuval I have noticed your comment about API microversions. Why I introduce microversions to karbor? -- Now the karbor has been deployed to huawei public cloud. So thera are also multiple cases like nova cinder projects have ,need microversions to resolve. REST API backwards incompatible. Older clients/API with new cloud. | 07:59 |
| chenying | yuval: IMO, introducing microversions doesn't mean encourage developers to break the API with microversion. | 08:01 |
| *** chenying has quit IRC | 08:08 | |
| *** chenying has joined #openstack-karbor | 08:14 | |
| chenying | yuval: I agree with your partial view about microversions. It will bring complexity to existed API change. We need try to limit and avoid these changes. | 08:14 |
| yuval | chenying: I don't understand you first sentence (Now the karbor has been...) | 08:15 |
| *** openstackgerrit has quit IRC | 08:17 | |
| chenying | yuval: Now the pike version of karbor have been deployed to huawei public cloud. In the future, the karbor will be upgraded to new version of karbor. | 08:22 |
| chenying | yuval: Top services and UI interfacing with karbor have the usecase of REST API backwards incompatible. | 08:25 |
| chenying | yuval: As I know, some dr service also use karbor service as the control plane service. So the version upgrade of karbor service also bring about the usecase of API backwards incompatible. | 08:30 |
| chenying | yuval: So what's your view about these usecases. | 08:40 |
| *** chenying has quit IRC | 08:50 | |
| *** zhonghua has quit IRC | 08:50 | |
| *** chenying has joined #openstack-karbor | 08:51 | |
| *** zhonghua has joined #openstack-karbor | 08:51 | |
| *** liujiong has quit IRC | 08:53 | |
| chenying | ping yuval | 08:57 |
| yuval | chenying: ok, I understand, but why microversions and not bump the major version? | 09:01 |
| jiaopengju | yuval: ‘we once spoke about renewing tokens or getting tokens with very long timeout’ ——— are there any results? | 09:04 |
| chenying | yuval: I am not sure that bumping the major version is a must. It is a big change for Karbor API. | 09:06 |
| chenying | yuval So what's your oppions about it? | 09:06 |
| *** jiaopengju has quit IRC | 10:15 | |
| yuval | chenying: which changes to the API are you talking about? Breaking changes (change, remove existing API) or adding new API routes? | 10:35 |
| chenying | yuval: which changes to the API are you talking about? ---- you mean that the api change about bumping the major version? | 10:59 |
| yuval | chenying: no. I'm asking why bumping major version is not enough, and why we need microversions | 11:00 |
| chenying | yuval: I mean that possible breaking changes (change, remove existing API) about karbor API. | 11:02 |
| yuval | chenying: that's not what microversions are for | 11:02 |
| yuval | chenying: that's major versions | 11:02 |
| chenying | yuval: how to solve the the backwards compatible of breaking changes. | 11:04 |
| yuval | chenying: support old major versions | 11:04 |
| chenying | yuval: Does it means that if there are some breaking changes about API, we need bump a new major version? The possible breaking changes about API go with every development cycle. | 11:10 |
| *** openstackgerrit has joined #openstack-karbor | 11:28 | |
| openstackgerrit | chenying proposed openstack/karbor master: Add kubernetes client to Karbor https://review.openstack.org/498779 | 11:28 |
| chenying | yuval so your oppion? | 11:35 |
| *** jiaopengju has joined #openstack-karbor | 11:41 | |
| yuval | chenying: yes, we need to bump major if we break api | 12:03 |
| chenying | yuval: I see. I will think it carefully. | 12:05 |
| chenying | yuval: I have another question about show_resource method of pod protectable. | 12:06 |
| chenying | yuval Although we have the uid of pod resource from the k8s cluster. But the k8s only expose a get pod API with the parameter pod name. | 12:07 |
| chenying | yuval: def read_namespaced_pod(self, name, namespace, **kwargs): | 12:08 |
| chenying | yuval: But the parameter of show_resource method is resource_id. | 12:09 |
| chenying | yuval: I am thinking about fill the Resource id with the pod name? | 12:10 |
| chenying | yuval: Do you have any idea about it? | 12:11 |
| *** jiaopengju has quit IRC | 13:07 | |
| *** jiaopengju has joined #openstack-karbor | 13:11 | |
| *** zhurong has quit IRC | 13:14 | |
| *** gouthamr has joined #openstack-karbor | 13:18 | |
| *** jiaopengju has quit IRC | 13:40 | |
| *** jiaopengju has joined #openstack-karbor | 13:53 | |
| openstackgerrit | Pengju Jiao proposed openstack/karbor master: [WIP]Add freezer protection plugin for karbor https://review.openstack.org/499685 | 15:32 |
| *** jiaopengju has quit IRC | 15:35 | |
| *** zhonghua2 has joined #openstack-karbor | 18:00 | |
| *** zhonghua has quit IRC | 18:02 | |
| *** gouthamr has quit IRC | 22:08 | |
| *** dimak has quit IRC | 23:28 | |
| *** dimak has joined #openstack-karbor | 23:29 | |
| *** yamamoto_ has joined #openstack-karbor | 23:31 | |
| *** yamamoto_ has quit IRC | 23:31 | |
| *** yamamot__ has joined #openstack-karbor | 23:31 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!