09:00:48 <yuval> #startmeeting karbor
09:00:48 <openstack> Meeting started Tue Mar 14 09:00:48 2017 UTC and is due to finish in 60 minutes.  The chair is yuval. Information about MeetBot at http://wiki.debian.org/MeetBot.
09:00:49 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
09:00:52 <openstack> The meeting name has been set to 'karbor'
09:00:59 <chenying_> hi
09:00:59 <yuval> Hello, and welcome to Karbor's weekly meeting!
09:01:14 <zhangshuai> hi, all
09:01:55 <yuval> On the agenda today: finalizing the list of goals for Pike, including schedule
09:02:14 <yuval> Let's wait 2 more minutes for people to arrive
09:05:34 <yuval> chenying_: zhonghua, xiangxinyong, chenzeng, wu, chenhuayi are coming?
09:06:13 <chenying_> yuval: waiting  I will make a call.
09:06:48 <chenying_> I am not at the office.
09:08:57 <zengchen> sorry, i'm late
09:09:51 <xiang123456> hello
09:09:55 <zhonghua> hi
09:10:17 <yuval> what about chenhuayi? wu?
09:10:46 <zhonghua> he is sick
09:10:52 <zhonghua> AFAIK
09:10:52 <yuval> let's start
09:10:58 <yuval> #topic Pike
09:11:34 <yuval> See the agenda and the link to the purposed final goals in https://wiki.openstack.org/wiki/Meetings/Karbor
09:12:27 <yuval> The Google document in the agenda includes a list of goals and schedule
09:13:06 <yuval> Besides specs and features, we should put heavy emphasize on deployment documentation
09:13:20 <yuval> Specifically plugin developer guides and API reference
09:14:54 <yuval> Does anyone has a suggestion about a missing feature, or priority for a listed goal?
09:15:23 <yuval> *have
09:15:48 <yuval> If you have, please burst in to say, I'll go over the high priority goals
09:15:49 <zhonghua> yuval, should we discuss the list one by one?
09:15:59 <yuval> I'll go over the important ones only
09:16:06 <chenying_> Hi yuval suggest that write the agenda to openstack etherpad. Most of time the Google document is unavailable for us.
09:17:15 <yuval> Copied it here: https://etherpad.openstack.org/p/karbor-Pike-planning
09:17:34 <chenying_> Ok Thanks.
09:18:14 <yuval> 1. release-with-intermediary
09:18:30 <yuval> We have adopted release-with-intermediary release model
09:18:57 <yuval> see https://releases.openstack.org/reference/release_models.html
09:19:50 <yuval> this basically means that we have to release python-karborclient on time (over a month before the actual release)
09:20:09 <yuval> and we will have a feature freeze in karbor, about a month before the actual release
09:20:27 <yuval> so remembering new features in the last minute will not get inside Karbor. Plan accordingly
09:21:04 <yuval> questions?
09:21:15 <zhonghua> does that mean we should make a milestone plan
09:21:20 <zhonghua> ?
09:22:01 <yuval> We are not bound to milestones, but we should try to follow them as best we can. We must comply to the feature-freeze deadlines
09:22:44 <zhonghua> only FF?
09:23:02 <yuval> This means that we should put very high effort on specs in the following 1-2 months, then high effort on features, then high effort on bugs in the last 1-2 months
09:23:41 <yuval> zhonghua: see here https://releases.openstack.org/pike/schedule.html
09:24:04 <zhonghua> ok
09:24:14 <zhonghua> thanks
09:24:41 <chenying_> Bank indexing: about this goal, do we need intreoduce a new indexing component to karor in pike?
09:24:59 <yuval> chenying_: we will get to that in a minute
09:25:11 <yuval> I opened a milestone for pike in launchpad: https://launchpad.net/karbor/+milestone/pike-ms
09:25:28 <yuval> Please bookmark it. Specs and bugs for pike should be attached to it
09:25:46 <yuval> Important specs for Pike:
09:26:12 <yuval> We have the Tiramisu spec and implementation in Cinder, which is relevant to Karbor:
09:26:44 <yuval> https://review.openstack.org/229722
09:26:50 <yuval> https://review.openstack.org/352228
09:26:53 <yuval> https://review.openstack.org/352229
09:28:28 <yuval> 2. Bank indexing spec - important spec about interface for indexing and querying checkpoints. The interface should provide a solution for single-site and multi-site scenarios
09:28:50 <yuval> chenying_: to your question: add the interface and reference implementation
09:29:48 <chenying_> yuval  single-site usecase, we can database bank. The Bank indexing may is not needed.
09:30:24 <yuval> chenying_: database can be used for the bank backend, but karbor currently doesn't have a index/query interface
09:30:25 <zhonghua> chenying_, +1
09:30:56 <yuval> chenying_: by adding the index/query interface, we can use the database inherent indexing in single-site scenario
09:31:18 <chenying_> yuval: IMO introduce a new project or component to karbor , it is too heavy and make it more complex for bank module.
09:31:38 <chenying_> yuval IMO introduce a new project or component about indexing to karbor
09:31:50 <yuval> chenying_: there must be misunderstanding, as this is not about introducing a new project
09:32:35 <chenying_> yuval  a new project or component like searchlight
09:32:50 <zhonghua> yuval, Iam not sure is it neccessary to add this interface as a standard interface
09:33:11 <yuval> chenying_: as said, there is a misunderstanding, nobody said anything about adding a new project or component
09:33:35 <yuval> chenying_: we are speaking about karbor requirement to query checkpoints: by date, by plan, by metadata, etc
09:34:07 <yuval> this can be used by using database indexing, in single-site scenarios where database is used as a bank
09:34:24 <yuval> by indexing into a sqlite database in the bank itself
09:34:34 <yuval> or an external component
09:35:12 <yuval> index/query will be part of the bank interface. we will provide a reference implmentation for sqlite indexing in the bank, and database indexing for database bank plugin
09:37:20 <chenying_> yuval I know,  an external component will be intrduced about swift bank. Do we really add extra interface about indexing. In the usecase of database, the default query interface also support indexing  by date, by plan, by metadata, etc
09:38:16 <yuval> chenying_: no, in the case of a object storage bank (e.g swift) we will provide an sqlite index which will reside in the bank itself. Anyone who will want to index the bank in an external component will have to write that for his bank plugin
09:38:48 <yuval> we are short on time, let's continue this specific discussion in karbor channel after meeting
09:38:56 <yuval> moving on to the next item
09:38:59 <chenying_> OK
09:39:08 <yuval> Multi-Tenant Isolation spec
09:39:43 <yuval> very important spec about how to expose checkpoints to users without exposing them to unauthorized users
09:39:59 <yuval> this is relevant to single and multi site scenarios
09:40:00 <zhonghua> yuval, +1
09:40:09 <yuval> this spec is very high priority
09:42:09 <yuval> last critical spec - removing heat
09:42:53 <yuval> Regarding documentation - zhonghua is working on getting us a technical writer to assist with documentation
09:43:03 <yuval> this is a critical part of the Pike release
09:43:42 <zhonghua> yuval, od you know is there any community team can help us
09:43:54 <zhonghua> do
09:44:29 <yuval> zhonghua: yes, I spoke with docs team in PTG. They will be able to assist in reviewing, but they will obviously not write the docs for us
09:44:52 <zhonghua> got it
09:45:27 <yuval> Questions? Suggestions?
09:45:34 <zhonghua> I will try to find the right guys
09:46:17 <yuval> zhonghua: great
09:47:05 <yuval> #topic Open Discussion
09:47:25 <yuval> zengchen: xiang12345: chenying_: zhangshuai: ?
09:48:18 <chenying_> About the  goal :Deployment
09:48:50 <yuval> ?
09:48:57 <yuval> anything specific?
09:49:04 <chenying_> Do you have any suggestion about the tools or project which karbor should support.
09:49:56 <yuval> chenying_: we currently have Kolla, I think we should extend to OpenStack Ansible
09:50:00 <chenying_> You know, There are many project or tools about openstack deployment.
09:50:16 <yuval> chenying_: RPM/DEB packages are a low hanging fruit IMO
09:51:14 <chenying_> yuval: Of course.
09:52:13 <yuval> anything else?
09:54:25 <yuval> Thanks for attending, everybody!
09:54:27 <yuval> #endmeeting