| xuhaiwei | Qiming, morning | 00:44 |
|---|---|---|
| Qiming | morning | 00:44 |
| xuhaiwei | sorry for missed yesterday's meeting, just checked the log | 00:45 |
| Qiming | okay, pretty interesting a discussion | 00:45 |
| xuhaiwei | one question is about the summit proposal | 00:45 |
| Qiming | yes | 00:45 |
| xuhaiwei | there is no proposal of Senlin introduction? | 00:45 |
| Qiming | we continued the discussion on #senlin again after the meeting | 00:45 |
| xuhaiwei | I saw it too | 00:46 |
| Qiming | working on it | 00:46 |
| Qiming | ethan has proposed one | 00:47 |
| Qiming | I'm revising it into a generic introduction | 00:47 |
| Qiming | we are thinking of a second one | 00:47 |
| xuhaiwei | so up to now, how many proposals are decided to be made to summit? Two? Senlin introduction and Container Clustering? | 00:47 |
| xuhaiwei | Is there anything I can help? | 00:48 |
| Qiming | the second one would be about auto-scaling | 00:48 |
| Qiming | I'm not sure we can make container clustering work by the summit | 00:49 |
| Qiming | we are working on the auto-scaling scenario | 00:50 |
| xuhaiwei | In fact yuanying and me are also thinking of 'Container as a service' topic, but it will be discussed first in Magnum team | 00:50 |
| xuhaiwei | there maybe some relationship | 00:51 |
| Qiming | we need work on that to make it happen | 00:52 |
| Qiming | as you know, we are being challenged by some early users that there are quite some bugs | 00:53 |
| Qiming | people want to compare senlin's stability to heat resourcegroup | 00:53 |
| Qiming | it is unfair, but it is understandable | 00:53 |
| Qiming | those will be the top priority | 00:53 |
| Qiming | then comes the fixes to known issues | 00:54 |
| xuhaiwei | so currently we should try our best to fix bugs? | 00:54 |
| Qiming | then comes typical scenarios | 00:54 |
| Qiming | then comes new features | 00:54 |
| Qiming | however, sometimes we cannot wait | 00:54 |
| Qiming | if we wait too long, we are losing the time window | 00:54 |
| Qiming | all teams/communities/projects are moving fast | 00:54 |
| xuhaiwei | ye | 00:55 |
| Qiming | we may have to work on both | 00:55 |
| xuhaiwei | s | 00:55 |
| Qiming | make the best of resources available | 00:55 |
| Qiming | make senlin usable and valuable | 00:55 |
| Qiming | just got a request to backport senlin to liberty | 00:56 |
| xuhaiwei | the next big topic should be done in m3 is NODE CREATE/DELETE action fix? | 00:56 |
| Qiming | and desired_capacity | 00:56 |
| Qiming | and an end-to-end auto-scaling solution | 00:57 |
| Qiming | it would be highly desireable to have a POC of container clusters | 00:57 |
| xuhaiwei | ok, got it | 00:57 |
| Qiming | when I'm revisiting the journey we have gone so far, I'm a little bit concerned that we may be a little bit conservative | 00:58 |
| xuhaiwei | what do you mean? | 00:59 |
| *** dixiaoli has joined #senlin | 00:59 | |
| Qiming | but it is always a tradeoff, 1) getting the existing features stabilized 2) add new features to demonstrate our capabilities | 00:59 |
| Qiming | there is nothing wrong to stabilize the existing features | 01:00 |
| Qiming | but we should also encourage experimentations of new features | 01:00 |
| xuhaiwei | there is a desire from our team which is baremetal autoscaling by Senlin | 01:02 |
| Qiming | oh, thanks god | 01:03 |
| Qiming | that would be so sexy | 01:03 |
| xuhaiwei | sometime we got ideas but are not so active to make it happen :) | 01:04 |
| Qiming | yep | 01:05 |
| Qiming | I was experimenting some container support recently | 01:05 |
| Qiming | I was using coreos as the guest OS | 01:05 |
| Qiming | there is no problem to create and manage a cluster of nova servers running CoreOS | 01:06 |
| Qiming | to support container clustering, we need to have senlin probe into those VMs to manage docker instances | 01:06 |
| Qiming | it is an interesting scenario | 01:06 |
| xuhaiwei | you don't want to use magnum? | 01:07 |
| Qiming | having senlin sit above magnum? | 01:07 |
| Qiming | I was expecting magnum to use senlin | 01:08 |
| xuhaiwei | use Magnum to do the preparement job | 01:08 |
| Qiming | well, we can do that if it makes sense | 01:08 |
| Qiming | personally, I'm very very concerned of integrating mesos, marathon, kubernetes into openstack | 01:09 |
| Qiming | you will have python, Java, go, scalar, C programs at different layers | 01:10 |
| Qiming | it would be very difficult (if possible) to do integration tests | 01:10 |
| xuhaiwei | yes, that's the problem | 01:11 |
| Qiming | here is what the SUR team has done: https://github.com/tengqm/SUR/tree/SURmagnum/magnum/sur/SURspec | 01:11 |
| Qiming | the students were hard coding everything, which is bad | 01:12 |
| Qiming | https://github.com/tengqm/SUR/blob/SURmagnum/magnum/sur/cluster_function.py | 01:12 |
| xuhaiwei | you want container to run on baremetal or vm? | 01:13 |
| Qiming | however, the POC did show us it is possible | 01:13 |
| Qiming | ideally, they should run on baremetal | 01:13 |
| Qiming | however, I don't think Linux kernel is ready for that | 01:14 |
| Qiming | at its hard core, container is about building isolated runtime environments based on kernel support | 01:16 |
| Qiming | I'm not that sure the isolation provided by Linux kernel is complete | 01:17 |
| xuhaiwei | yes | 01:17 |
| Qiming | that is why running containers inside VMs is still the common practice | 01:17 |
| Qiming | but looking into near future, we can expect the gaps will be filled pretty quickly | 01:18 |
| Qiming | so one of the option to support a flexible deployment/management of container clusters could be using docker-py | 01:18 |
| *** pratikmallya has quit IRC | 01:18 | |
| xuhaiwei | yes | 01:19 |
| xuhaiwei | docker is doing container on baremetal, right? | 01:19 |
| Qiming | it doesn't care | 01:19 |
| Qiming | in docker community, there is a project named docker-machine | 01:20 |
| Qiming | which is an attempt to abstract the underlying infrastructure | 01:20 |
| xuhaiwei | as I discussed with yuanying, he wants to create a container compute service and use docker to start container directly | 01:21 |
| xuhaiwei | on baremetal | 01:22 |
| Qiming | that is promising | 01:22 |
| *** Liuqing has joined #senlin | 01:34 | |
| *** Yanyanhu has joined #senlin | 01:49 | |
| *** Qiming has quit IRC | 01:53 | |
| *** lixinhui_ has joined #senlin | 02:01 | |
| *** elynn has joined #senlin | 02:04 | |
| *** elynn_ has joined #senlin | 02:09 | |
| *** elynn has quit IRC | 02:09 | |
| *** Qiming has joined #senlin | 02:22 | |
| openstackgerrit | Merged openstack/senlin: Fix a bug about cluster-node-add/del https://review.openstack.org/271984 | 03:12 |
| *** yuanying has quit IRC | 03:21 | |
| *** yuanying has joined #senlin | 03:23 | |
| *** yuanying has quit IRC | 03:28 | |
| *** yuanying has joined #senlin | 03:33 | |
| *** yuanying has quit IRC | 03:40 | |
| *** yuanying has joined #senlin | 03:40 | |
| *** Qiming has quit IRC | 03:54 | |
| *** yuanying has quit IRC | 03:56 | |
| *** yuanying has joined #senlin | 03:57 | |
| *** yuanying_ has joined #senlin | 03:58 | |
| *** yuanying has quit IRC | 04:01 | |
| *** Qiming has joined #senlin | 04:14 | |
| *** elynn_ has quit IRC | 04:18 | |
| openstackgerrit | Merged openstack/python-senlinclient: Change 'status_code' to 'http_status' to parse SDK's exception https://review.openstack.org/272360 | 04:26 |
| *** elynn_ has joined #senlin | 04:51 | |
| *** elynn_ has quit IRC | 04:55 | |
| *** elynn_ has joined #senlin | 04:56 | |
| elynn_ | Hi Qiming , Are you working on change the cluster_delete returned location to action id? if not , I can help | 05:01 |
| Qiming | yes | 05:01 |
| Qiming | I'm working on that right now | 05:01 |
| elynn_ | Then I will wait for your patch :) | 05:01 |
| openstackgerrit | Merged openstack/senlin: Updated from global requirements https://review.openstack.org/272830 | 05:02 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Make cluster-delete return action id https://review.openstack.org/272904 | 05:16 |
| *** elynn__ has joined #senlin | 05:24 | |
| *** elynn_ has quit IRC | 05:26 | |
| *** elynn_ has joined #senlin | 05:29 | |
| *** elynn__ has quit IRC | 05:31 | |
| *** Ayush has joined #senlin | 05:36 | |
| openstackgerrit | Merged openstack/senlin: Make cluster-delete return action id https://review.openstack.org/272904 | 05:57 |
| *** Yanyan has joined #senlin | 05:58 | |
| *** Yanyanhu has quit IRC | 05:59 | |
| *** Qiming has quit IRC | 06:45 | |
| *** Qiming has joined #senlin | 06:46 | |
| *** Qiming has quit IRC | 06:54 | |
| zigo | elynn_: xuhaiwei: qiming: There's no tag for senlin-dashboard, could someone make one? | 06:58 |
| zigo | Otherwise, I can't package it ... | 06:58 |
| elynn_ | I'm not sure that senlin-dashboard is ready for use now... | 07:00 |
| elynn_ | Liuqing, Could you help with zigo ? | 07:00 |
| Liuqing | zigo, the tag is about instance or anything else? | 07:02 |
| zigo | Liuqing: I just need a git tag in the repo, so that I can package that for Mitaka b2. | 07:02 |
| elynn_ | Add git tags to senlin-dashboard | 07:02 |
| zigo | (I don't use tarball, I generate them out of git tags) | 07:02 |
| Liuqing | ok, got that | 07:03 |
| *** wei1 has joined #senlin | 07:05 | |
| elynn_ | Thanks Liuqing :) | 07:07 |
| *** wei1 has quit IRC | 07:11 | |
| *** Qiming has joined #senlin | 07:19 | |
| *** elynn_ has quit IRC | 07:21 | |
| *** elynn_ has joined #senlin | 07:21 | |
| Qiming | Liuqing, I'm producing a 0.1.0 tag for senlin-dashboard | 07:21 |
| Qiming | okay, it is there | 07:23 |
| Liuqing | thanks Qiming, | 07:23 |
| *** Qiming has quit IRC | 07:32 | |
| *** dixiaoli has quit IRC | 07:46 | |
| *** Qiming has joined #senlin | 07:56 | |
| xuhaiwei | hi Qiming, if generally comparing Senlin with Heat auto-scaling, what is the merit point of Senlin? | 07:57 |
| xuhaiwei | Senlin has policy and can make auto-scaling more convenient and more powerful? | 07:59 |
| Qiming | yes, more flexible | 08:00 |
| xuhaiwei | Senlin also has health management function is also a point i think | 08:01 |
| *** dixiaoli has joined #senlin | 08:03 | |
| *** Qiming has quit IRC | 08:05 | |
| *** Yanyan has quit IRC | 08:07 | |
| *** Yanyanhu has joined #senlin | 08:07 | |
| *** Qiming has joined #senlin | 08:07 | |
| Qiming | xuhaiwei, what's the question for? | 08:09 |
| xuhaiwei | the question is Senlin is better than Heat auto-scaling in where? | 08:10 |
| Qiming | what's your observation? | 08:10 |
| xuhaiwei | my idea about it? | 08:11 |
| xuhaiwei | I have thought of some merits, but afraid it's not enough | 08:11 |
| xuhaiwei | I am often asked by someone about this question | 08:12 |
| Qiming | okay, I have a list of comparison, :) | 08:13 |
| Qiming | but I'm not sharing that with you | 08:13 |
| xuhaiwei | :( | 08:13 |
| Qiming | it has to come from your own belief | 08:13 |
| *** lixinhui_ has quit IRC | 08:13 | |
| xuhaiwei | can I make a reference of it? | 08:14 |
| Qiming | we can exchange our thoughts, but I'm not sharing mine with you first | 08:15 |
| Qiming | :) | 08:15 |
| Qiming | when you are deploying a cluster for auto-scaling and operating it | 08:16 |
| Qiming | you will see the real difference | 08:16 |
| Qiming | especially when you are trying to operate such a cluster | 08:16 |
| xuhaiwei | as far as I can see, due to the policies in senlin we can do auto-scaling more flexible than heat if we only think about auto-scaling | 08:17 |
| Qiming | okay, that is a good point | 08:17 |
| xuhaiwei | of course, senlin has its own API, DB and engine, that must be much more powerful | 08:18 |
| xuhaiwei | that is from effectiveness point | 08:19 |
| xuhaiwei | the people who asks this question usually think from function point | 08:20 |
| Qiming | APIs are providing functions, right? | 08:21 |
| xuhaiwei | hm, yes | 08:22 |
| Qiming | DBs and engine doesn't matter that much | 08:22 |
| Qiming | you are basically comparing a heat resource type to a service | 08:23 |
| xuhaiwei | actually no | 08:23 |
| *** Qiming has quit IRC | 08:27 | |
| *** Qiming has joined #senlin | 08:27 | |
| Qiming | no? | 08:28 |
| xuhaiwei | yes, creating a heat resource is just doing a task in heat engine? | 08:29 |
| xuhaiwei | not understand your question well | 08:29 |
| Qiming | in heat, you are managing template and stacks, not resources | 08:30 |
| xuhaiwei | ok | 08:30 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Make node-delete return an action ID https://review.openstack.org/272947 | 08:54 |
| *** bdrich has joined #senlin | 09:02 | |
| *** bdrich has quit IRC | 09:22 | |
| *** Liuqing has quit IRC | 09:33 | |
| *** Yanyanhu has quit IRC | 09:40 | |
| openstackgerrit | xu-haiwei proposed openstack/senlin: Cleanse set_status method used in cluster_action https://review.openstack.org/272962 | 09:44 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Improve size parameter checking function https://review.openstack.org/272963 | 09:47 |
| Qiming | xuhaiwei, I have some disagreements regarding the changes to set_status patch | 09:50 |
| Qiming | We are supposed to distinguish action status and cluster status instead of mixing them together | 09:51 |
| Qiming | any patches in alignment with that are welcomed | 09:51 |
| Qiming | for example, when we do a CLUSTER_CHECK | 09:53 |
| Qiming | the check operation itself may fail, but we should not change the cluster status after cluster check is completed | 09:53 |
| *** lixinhui_ has joined #senlin | 10:12 | |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Fix cluster_create function https://review.openstack.org/272985 | 10:27 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Fix cluster_create function https://review.openstack.org/272985 | 10:31 |
| *** lixinhui_ has quit IRC | 10:37 | |
| *** lixinhui_ has joined #senlin | 10:39 | |
| *** elynn_ has quit IRC | 10:56 | |
| *** elynn_ has joined #senlin | 10:56 | |
| *** elynn_ has quit IRC | 10:58 | |
| *** dixiaoli_ has joined #senlin | 11:17 | |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Fix cluster-resize parameter checking https://review.openstack.org/272998 | 11:19 |
| *** dixiaoli_ has quit IRC | 11:25 | |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Remove NodeNotOrphan exception type https://review.openstack.org/273016 | 12:23 |
| *** dixiaoli_ has joined #senlin | 12:37 | |
| *** elynn_ has joined #senlin | 13:01 | |
| cschulz_ | elynn: Has anything changed in how the senlin client in heat builds webhook receivers? | 13:03 |
| *** jun__ has quit IRC | 13:04 | |
| cschulz_ | We had been having success with creating receivers until yesterday when the request started failing because of a missing host attribute in the configuration. | 13:04 |
| cschulz_ | Do we need to add a ip address to a configuration file? | 13:04 |
| Qiming | cschulz_, refer to: http://git.openstack.org/cgit/openstack/senlin/tree/senlin/api/common/wsgi.py#n97 | 13:08 |
| Qiming | the host and the port option in the webhook section in the senlin.conf file will be used as the defaults for webhooks | 13:09 |
| Qiming | I'm not aware of any changes to these options | 13:10 |
| Qiming | if things used to work and then fail now, it could be some situation we didn't capture, possibly a bug? | 13:10 |
| cschulz_ | Qiming, Thanks. Actually discovered that it works on one of my hosts an not on another. Will need to get the configurations synchronized. | 13:26 |
| *** elynn_ has quit IRC | 13:26 | |
| *** elynn has joined #senlin | 13:39 | |
| *** elynn has quit IRC | 13:59 | |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Add size checking for cluster scaling https://review.openstack.org/273074 | 14:31 |
| openstackgerrit | Merged openstack/python-senlinclient: Updated from global requirements https://review.openstack.org/271681 | 14:40 |
| *** Qiming has quit IRC | 15:18 | |
| *** dixiaoli_ has quit IRC | 15:31 | |
| *** lixinhui_ has quit IRC | 15:33 | |
| *** pratikmallya has joined #senlin | 15:49 | |
| *** bdrich has joined #senlin | 16:19 | |
| *** bdrich has quit IRC | 17:01 | |
| *** bdrich has joined #senlin | 17:35 | |
| *** pratikmallya has quit IRC | 18:34 | |
| *** bdrich has quit IRC | 19:33 | |
| *** pratikmallya has joined #senlin | 19:35 | |
| *** bdrich has joined #senlin | 20:46 | |
| *** yuanying_ has quit IRC | 23:02 | |
| *** yuanying has joined #senlin | 23:12 | |
| *** pratikmallya has quit IRC | 23:13 | |
| *** bdrich has quit IRC | 23:22 | |
| *** Qiming has joined #senlin | 23:56 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!