| *** jiaopengju has joined #openstack-karbor | 00:30 | |
| *** yamamoto_ has quit IRC | 00:37 | |
| *** yamamoto has joined #openstack-karbor | 00:39 | |
| *** liujiong has joined #openstack-karbor | 00:51 | |
| *** saggi has joined #openstack-karbor | 01:27 | |
| openstackgerrit | chenying proposed openstack/karbor master: Spec: Implement policies in code https://review.openstack.org/492143 | 01:44 |
|---|---|---|
| openstackgerrit | Pengju Jiao proposed openstack/karbor master: Fix log info error https://review.openstack.org/492343 | 02:04 |
| *** edisonxiang has joined #openstack-karbor | 02:44 | |
| *** yamamoto has quit IRC | 03:36 | |
| *** yamamoto has joined #openstack-karbor | 03:47 | |
| *** yamamoto has quit IRC | 03:48 | |
| *** yamamoto has joined #openstack-karbor | 03:49 | |
| *** gouthamr has joined #openstack-karbor | 04:31 | |
| *** gouthamr has quit IRC | 05:16 | |
| *** liujiong has quit IRC | 05:44 | |
| *** liujiong has joined #openstack-karbor | 05:45 | |
| *** yamamoto has quit IRC | 06:01 | |
| *** yamamoto has joined #openstack-karbor | 06:05 | |
| openstackgerrit | Merged openstack/karbor master: Ignore plan parameters for unloaded plugins https://review.openstack.org/492054 | 06:06 |
| openstackgerrit | Merged openstack/karbor master: Imported Translations from Zanata https://review.openstack.org/491999 | 06:29 |
| *** saggi has quit IRC | 06:33 | |
| *** saggi has joined #openstack-karbor | 06:39 | |
| openstackgerrit | Yuval Brik proposed openstack/karbor-dashboard master: Plan: send only parameters for protected resources https://review.openstack.org/492393 | 06:39 |
| jiaopengju | ping yuval | 07:22 |
| yuval | hey jiaopengju | 07:22 |
| jiaopengju | yuval chenying: I am doing the integration with freezer and found some issues. Freezer can backup two tpye of openstack resources: volume and nova instance(contains image, network or system volume). I try to use freezer plugin support volume resource and server resource, but failed. | 07:26 |
| yuval | jiaopengju: what kind of failure? | 07:27 |
| chenying | Just now I and jiaopengju discuss the freezer plugin, he want to support volume and server in one plugin. | 07:27 |
| jiaopengju | yuval chenying: It is because that karbor will split server resource into ‘image, volume, network’ | 07:28 |
| jiaopengju | yuval: but freezer supports nova instance backup, which already contains image, volume and network | 07:29 |
| yuval | jiaopengju: "nova instance backup" - freezer feature or nova feature? | 07:29 |
| jiaopengju | yuval: freezer feature | 07:29 |
| yuval | jiaopengju: this is exactly why we then protection plugins have 3 kinds of "contexts" | 07:30 |
| jiaopengju | yuval: I think the root of this issue is that the resource graph is stable for all providers | 07:31 |
| yuval | jiaopengju: the first one is the ProtectionPlugin instance itself, which is created in the context of the whole operation | 07:31 |
| yuval | jiaopengju: the second one is the Operation instance which is created in the context of the operation for each resource | 07:31 |
| yuval | jiaopengju: the third one is the ProtectionPlugin class variables, which persist across the lifetime of the service, but that's less important | 07:32 |
| yuval | jiaopengju: what I'm saying is: consider an operation spanning multiple resource types, like freezer's nova instance backup | 07:32 |
| yuval | jiaopengju: Karbor will create a Protection Plugin for volume and image, which can just gather the volume and image information or even do nothing | 07:33 |
| yuval | jiaopengju: then the protection plugin for server can initiate the freezer operation for the nova server, backuping the image and volume as well | 07:34 |
| yuval | jiaopengju: what do you mean "is stable for all providers"? | 07:37 |
| jiaopengju | yuval: can not be changed | 07:38 |
| jiaopengju | yuval: you mean that I can use ‘do nothing ’ when creating Protection Plugin? | 07:39 |
| yuval | jiaopengju: a protection plugin can be "registered" for multiple resource (using the 'get_supported_resources_types' method) | 07:40 |
| jiaopengju | yuval: Yes, I have used this | 07:40 |
| yuval | jiaopengju: you can create a freezer plugin which will be registered for 'image', 'volume', 'server' | 07:41 |
| yuval | jiaopengju: it will do nothin for image and volume, and for server, it will protect/restore all three | 07:41 |
| chenying | jiaopengju mean that now the resource dependency in karbor only consider the default usecase. For example, in karbor the volume mount to a server all the sub reousrce of this servce. But in freezer's plugin usecase, he want to backup service meatadata and bootable volume when taskflow pass a service id to this plugin. | 07:43 |
| jiaopengju | yuval: when doing server backup( the server contains system volume), it seems karbor will send two request to freezer, one for server, one for system volume | 07:43 |
| chenying | he think only the no bootable volume are the sub resource of the server in his usecase. | 07:43 |
| yuval | jiaopengju: I don't get it. You said freezer has "nova instance backup" which takes care of server+image+volume | 07:44 |
| jiaopengju | yuval: yes. But the task flow was generated by karbor | 07:45 |
| yuval | jiaopengju: if so, then for a volume and an image, do nothing. For a server, protect all three | 07:45 |
| yuval | jiaopengju: then register your plugin to volume and image and do a no-op | 07:45 |
| jiaopengju | yuval: As you say, how can I do volume backup in the same plugin? | 07:46 |
| jiaopengju | yuval: I want to support both freezer server and freezer volume in one plugin | 07:46 |
| chenying | jiaopengju: In this usecase, you need devlep anonther plugin only backup volume. | 07:46 |
| yuval | chenying: no need, just see if the volume is attached | 07:47 |
| yuval | chenying: if it is attached - leave it to the server call to backup. If not attached, call freezer to back it up | 07:47 |
| yuval | chenying: we can actually try to see if the attached server is in the plan or not, but that's an edge case | 07:48 |
| chenying | Or as yuval said, you also can check the attached or not attached in one plugin itself. | 07:49 |
| jiaopengju | yuval chenying: freezer server backup, only backup the system volme | 07:50 |
| jiaopengju | yuval chenying: so if we have a nova instance, which has a system volume and two other data volume(also attached) ? | 07:51 |
| yuval | jiaopengju: call freezer only for the server (and if I understand what you say correctly, freezer will take care of the volumes) | 07:52 |
| jiaopengju | yuval: freezer only care of system volume when doing server backup | 07:53 |
| jiaopengju | yuval: not data volumes | 07:53 |
| yuval | jiaopengju: then for volumes, use freezer to backup data volumes only, and in server backup, call freezer. On restore, restore them all and attach | 07:53 |
| yuval | jiaopengju: pretty sure you can query cinder to understand if the volume is attached and whether a server is using it to boot from (aka system volume) | 07:54 |
| jiaopengju | yuval: get it , I understand what you mena | 07:54 |
| yuval | jiaopengju: :) | 07:54 |
| jiaopengju | yuval: thank you ! | 07:54 |
| yuval | jiaopengju: no, thank you! :D | 07:55 |
| yuval | jiaopengju: feel free to consult more if you want | 07:55 |
| jiaopengju | yuval: I would like to do that :) | 07:56 |
| yuval | chenying: any idea why our gate keep failing with timeouts? | 08:00 |
| chenying | yuval I only know the fullstack test about volume restore, the error is in cinder-backup service, these are some exceptions about rpc when restore a volume. I ask the guy from cinder team to ses this error several times, he also can not find the reason. | 08:07 |
| yuval | chenying: :\ | 08:08 |
| yuval | chenying: frustrating | 08:08 |
| chenying | yuval cinder project jenkins don't have a test for volume restore. | 08:08 |
| chenying | yuval: I aslo think the fullstack of karbor spend too much time to run all the tests. | 08:12 |
| yuval | chenying: do you suggest we remove them? | 08:12 |
| chenying | yuval nearly one hour and half. | 08:12 |
| yuval | chenying: only the devstack takes half an hour | 08:13 |
| chenying | yuval: don't remove. For example, there are several fullstack tests about volume restore, only need one test to run actual volume restore, other tests can use the provider of noop.conf. | 08:15 |
| yuval | chenying: I agree | 08:15 |
| chenying | yuval: The test about test_scheduled_operations_create_and_scheduled aslo take too much time. The tests about checkpoint are tested already, this test aslo can use noop provider. | 08:18 |
| chenying | yuval We only need test the scheduler in this one, we don't need actually test volume backup and restore. | 08:19 |
| *** yamamoto has quit IRC | 08:27 | |
| *** yamamoto has joined #openstack-karbor | 08:28 | |
| *** yamamoto has quit IRC | 08:31 | |
| *** yamamoto has joined #openstack-karbor | 08:32 | |
| yuval | chenying: have you tested karbor lately? On a deployment? | 08:47 |
| chenying | yuval: I only have one deployment. all the services are enable. | 08:49 |
| chenying | yuval I do all the test and development in this depolyment. | 08:50 |
| yuval | chenying: are all services up to date? I'm getting some weird things, want to know if you found them also | 08:50 |
| chenying | yuval only karbor service is up to date. Other serivces are installed at Jul 11. | 08:52 |
| openstackgerrit | chenying proposed openstack/karbor master: Add Protectable Plugins Development Guide doc to Karbor https://review.openstack.org/492442 | 09:06 |
| openstackgerrit | chenying proposed openstack/karbor master: Add the extra_info parameter when checkpoint being created manually https://review.openstack.org/492447 | 09:27 |
| chenying | yuval: If you get some weird things about your environment, it's better to instatll a new devstack environment. I often do like this. :D At least the devstack environment in karbor jinkens can expose most of failures about karbor. | 09:48 |
| openstackgerrit | chenying proposed openstack/karbor master: [WIP]Optimize the execution time of karbor fullstack tests https://review.openstack.org/492457 | 09:51 |
| *** jiaopengju has quit IRC | 10:14 | |
| *** yamamoto has quit IRC | 10:21 | |
| *** liujiong has quit IRC | 10:22 | |
| *** yamamoto has joined #openstack-karbor | 10:25 | |
| *** yamamoto has quit IRC | 10:28 | |
| *** yamamoto has joined #openstack-karbor | 10:28 | |
| *** jiaopengju_ has joined #openstack-karbor | 10:31 | |
| *** yamamoto has quit IRC | 10:41 | |
| yuval | chenying: I have a fresh devstack | 10:49 |
| *** yamamoto has joined #openstack-karbor | 11:09 | |
| *** yamamoto_ has joined #openstack-karbor | 11:20 | |
| *** yamamoto has quit IRC | 11:24 | |
| *** NatanBro has joined #openstack-karbor | 11:54 | |
| *** jiaopengju__ has joined #openstack-karbor | 12:36 | |
| *** jiaopengju_ has quit IRC | 12:39 | |
| openstackgerrit | Yuval Brik proposed openstack/karbor master: Support swift auth domain name https://review.openstack.org/492523 | 12:51 |
| *** gouthamr has joined #openstack-karbor | 14:00 | |
| *** jiaopengju__ has quit IRC | 14:44 | |
| *** yamamoto_ has quit IRC | 15:08 | |
| *** yamamoto has joined #openstack-karbor | 15:10 | |
| *** yamamoto has quit IRC | 15:11 | |
| *** saggi has quit IRC | 15:31 | |
| *** saggi has joined #openstack-karbor | 15:36 | |
| *** saggi has quit IRC | 15:39 | |
| *** saggi- has joined #openstack-karbor | 15:39 | |
| *** saggi- is now known as saggi | 15:39 | |
| *** yamamoto has joined #openstack-karbor | 16:12 | |
| *** saggi- has joined #openstack-karbor | 16:16 | |
| *** yamamoto has quit IRC | 16:17 | |
| *** saggi has quit IRC | 16:19 | |
| *** saggi- is now known as saggi | 16:19 | |
| *** NatanBro has quit IRC | 16:30 | |
| *** gouthamr_ has joined #openstack-karbor | 17:51 | |
| *** gouthamr has quit IRC | 17:51 | |
| *** openstackgerrit has quit IRC | 19:03 | |
| *** gouthamr_ has quit IRC | 20:50 | |
| *** gouthamr has joined #openstack-karbor | 21:23 | |
| *** yamamoto has joined #openstack-karbor | 22:51 | |
| *** lihi has quit IRC | 23:10 | |
| *** dimak has quit IRC | 23:10 | |
| *** dimak has joined #openstack-karbor | 23:11 | |
| *** lihi has joined #openstack-karbor | 23:11 | |
| *** gouthamr has quit IRC | 23:12 | |
| *** yamamoto has quit IRC | 23:33 | |
| *** yamamoto has joined #openstack-karbor | 23:40 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!