17:02:34 #startmeeting service_chaining 17:02:34 Meeting started Thu Jun 18 17:02:34 2015 UTC and is due to finish in 60 minutes. The chair is cathy__. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:02:36 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:02:39 The meeting name has been set to 'service_chaining' 17:02:49 hi everyone 17:03:05 hi 17:03:35 yello 17:03:36 hi 17:03:45 o/ 17:03:59 o/ 17:04:49 here is today's agenda: 1. Update on repository creation 17:05:01 o/ 17:05:14 we have created a repository for the service chain development at the openstack repository 17:05:29 any question on that? 17:06:15 Agenda 2: Finalize the SFC Feature project scope: functional module breakdown and ownership as well as the types of service functions that will be chained in this feature development. 17:06:37 cathy__: I move the existing spec proposal over to the new repo 17:06:57 in last IRC meeting, we go through the list of functional modules and sign-up of each piece 17:06:58 all sfc-related tasks should be redirected to the new project setup 17:07:31 I think we still need to make one final push on the api proposal before we can go ahead with the functional decomposition of the project 17:07:33 armax: OK, let's take a look at the spec in the new repo after the meeting. 17:07:48 armax: can you point to the new repo 17:07:52 wtith no consensus I am not sure what we are coding for 17:08:09 LouisF: I sent an email yesterday night for details 17:08:25 #link github.com/openstack/openstack/networking-sfc 17:08:40 oops 17:08:41 #link https://github.com/openstack/networking-sfc 17:09:00 LouisF: I loaded your API proposal here: https://review.openstack.org/#/c/192933/ 17:09:05 the older spec with the reviews is at https://review.openstack.org/#/c/177946/ 17:09:19 armax: thanks 17:09:22 forget the older spec 17:09:23 The API discussion is Agenda #3 3. Service chain API and spec discussion and try to finalize the service chain API design so that we can start implementing it. 17:09:24 it’s abandoned 17:09:46 you’re still the author of https://review.openstack.org/#/c/192933/ 17:09:53 armax: the review history is there 17:10:09 armax: ok 17:10:11 LouisF: the history is mainly bollocks since the patch has been in a bad rebase for so long 17:10:17 but sure 17:10:26 the patch is there, no one is ever going to wipe it out of gerrit 17:10:30 Hi Cathy 17:10:39 I think it will be good to keep the original spec which has a lot of people's comments. 17:11:00 mohankumar__: hi 17:12:24 Let's also put a link to the original spec so that people knows what comments/input have been suggested before and how the spec has evolves based on those comments. 17:12:39 has evolved 17:13:28 cathy__: done 17:13:55 I mean a link on the new spec to the original spec so that people can view all the input/comments that have been made before 17:14:09 cathy__: yes 17:14:30 armax: you are quick, so you are done with putting the link to the original spec? Thanks! 17:14:42 cathy__: yes, it’s in the commit message 17:15:55 BTW, Swami sent me a message saying he can not attend today's meeting since he has an event or something. Also Vikram is on vacation. 17:17:07 armax: how would you suggest to go with the final push on the API proposal? 17:17:42 I agree that we really need to get this finalized so that we move forward with the design and implementation. 17:17:53 I was going to review it once again now, I would imagine that we need to turn it around quickly as soon as comments come in 17:18:10 armax: sure. 17:18:40 armax: Louis and I will turn this around as quick as possible now that the repo is created 17:19:14 Since people have given a lot of comments before on the original API spec, we are not seeing much comment coming in now 17:19:19 since the repo provisioning took longer than expected, I’d say let’s give us a few more days to make sure we catch as much feedback as we can get 17:19:47 we had an initial assessment of who was going to work on what…technically speaking code can start being pushed for review now 17:20:16 but it’s mostly likely going to be through drastic revisions if the shape of the proposal changes as well 17:20:38 armax: agree. Let's set it to next Wednesday. What do you think? 17:20:49 cathy__: btw as far as the management groups for the repo is concerned 17:21:09 armax: In our last meeting, I think we said one week time window. Now let';s extend to another week 17:21:12 you and I are currently member of the -core and -release them…but I would love for someone else to step in 17:21:48 to help the management and progress of the project 17:21:55 What do you mean by "someone else to step in"? Do you mean you have a representative? 17:22:03 as much as I willing to help I am stretched thin as is 17:22:03 armax: we discussed task assignments last week: http://eavesdrop.openstack.org/meetings/sfc_project/2015/sfc_project.2015-06-11-17.01.log.html 17:22:46 I can manage the progress of the project if you trust me. I have done project management before 17:22:57 anyone can review code, but only the members of the -core and -release team can merge code, create tags, etc 17:23:36 completed cli python neutron client changes .. not able to.submit for review getting remote rejection error . .. 17:23:37 we should share this responsibility amongst people who are going to work the most on the project 17:23:41 who’s intereted? 17:23:45 interested? 17:24:39 mohankumar__: great. Given the time constraint of this meeitng, we may need to work with you on the error off line. 17:25:07 Hi cathy__ 17:25:16 * armax thinks no-one is interested in the extra chores 17:25:22 interesting... 17:25:37 armax: quite some people have contacted me and expressed interest on contributing to this project. Also in our last meeting, we have people signed up on the tasks. 17:25:49 I have added their names to the core team 17:26:04 ramanjaneya_: hi 17:26:11 armax: add my name to core 17:26:11 * mestery wants to be a core too :) 17:26:24 * mestery lurks 17:26:24 ok, cathy__ if you have taken care of this, fine 17:26:30 Very ominous sounding I know 17:26:32 LouisF: I have added your name to the core 17:26:40 mestery: you’re core for sure since you’re the Neutron PTL 17:26:40 thanks 17:26:50 armax: danke 17:26:58 mestery: done 17:27:05 mestery: yes, sure you will be core:-) 17:27:12 cathy__: danke as well :) 17:27:27 mestery: you and armando will be core for sure 17:27:42 not that anyone needs to express an opinion right now, but happy to see more people involved to get this in the direction that suits everybody 17:28:11 reaching us out publicly is best 17:29:02 armax: agree. I sent out an email to the openstack-dev with all the names of developers who have signed up for the contribution. the contributors include people from multiple different companies as well as individual contributors. 17:29:09 ok, perhaps we want to move on from the bureocratic trivia 17:29:59 ok, let's come back to the API discussion 17:30:30 armax: OK with we wait until next Wednesday to finalize it? 17:30:43 sounds ok 17:31:04 cathy__: sure, let’s aim for that, ideally it’ll depend on the level of feeback being gathered 17:31:13 cathy__: I am personally going to review it by the end of the week 17:31:18 try to finilize it by next WED 17:31:28 armax: great. 17:32:21 let me summarize today's topic discussion 17:32:37 #topic repository has been created 17:33:00 nbouthors: saw you email on adding AppId to flow-filter 17:33:08 #topic API discussion and finalize by next Wednesday 17:33:36 nbouthors: Hi Nicolas 17:33:54 sdake: hi 17:34:15 LouisF: I am still on the fence if we we should tie NSH specifics to the abstract API, maybe I got it wrong 17:34:23 #topic project management and core members 17:34:30 but I’d rather avoid the pollution for now, if we can 17:34:53 cathy__: LouisF Hi 17:36:35 btw I’d like to understand the relationship between this effort and use case https://review.openstack.org/#/c/169201/ 17:36:51 armax: we do not tie the NSH specifics to API. But to support service chain, the vSwitch needs to know how to communicate with the service function VM about service treatment. The communication could be via NSH or VLAN or just raw packet without any chin identification 17:36:59 the person filed the use case and then disappeared 17:37:11 cathy__: ok 17:37:35 I mean packet without any chain identification 17:38:01 armax: it can be added as an item in the optional parameters 17:38:17 LouisF: ok I’ll ponder on this a bit 17:38:20 more 17:39:07 armax: ok 17:39:52 In the first phase, there will be no service function VM that supports NSH, so the vSwitch needs to act as Proxy. But in the future, there will be quite some service VM that supports the new servic chain header since it is in IETF standard and it enables more rich service chain functionality such as metadata and with chain ID as the index for forwarding table, it greatly improves the performance of the vSwitch 17:41:30 armax: regarding use cases spec - that addresses sfc in a general way, this effort is to provide a means for supporitng those use cases 17:41:57 So our API needs to covers service VMs that do not support new chain header and VMs that support it, and the vSwitch needs to be told which format to use for communicating with the SF VMs. That is why we need this context parameter in the API to pass down to the vSwitch 17:42:11 LouisF: I understand, but the use spec case references your spec, but the author seems to have stopped addressing the patch 17:42:56 armax: Let me provide some information about the use case spec 17:43:36 cathy__: no need to do it here 17:43:58 I was only curious whether we should be taking over that one too 17:44:35 Based on my communication and comments with the author, the use case spec just provides the use case and our spec provides the design/code to realize the use case. So it is a good match. 17:45:15 cathy__: agree…but they’d need to be in synch 17:45:29 and the patch has a ton of -1’s and no feedback 17:46:52 I think the use case spec can be updated one more time to incorporate all the comments that have been given. The use case is in good sync with work that have been done in IETF and ONF L4~L7 work group. Our spec is also in good sync with IETF and ONF L4~L7 work group spec. 17:47:28 So ours is a good implementation of the use case described in the use case spec. 17:47:48 cathy__: want to change the meeting topic? 17:48:15 I think people put -1 because the latest comments are not incorporated in the use case spec. 17:48:32 I don’t disgree, I am just wondering who does what 17:48:40 anyway 17:48:43 moving on 17:48:58 s3wong: we are now on the topic of 4. Deep dive into technical questions of SFC 17:49:28 if no more questions. we can move on 17:49:49 s3wong: Any topic do you have in mind and want to discuss? 17:50:46 cathy__: not really, I merely pointed out that the meeting topic (reflected by #topic) is still "project management and core members) 17:50:49 topic 4. Tentative timeline for the feature development 17:51:06 s3wong: thanks! 17:51:14 s3wong: I did not know that 17:51:27 #topic Tentative timeline for the feature development 17:51:45 cathy__: there :-) 17:51:59 armax: any suggestion on the timeline? 17:52:58 un…I am never good at swag estimates 17:53:07 s3wong: btw, any interest to join development of some pieces? each piece has people signed up already, but it is always good to have more back-up and reviewers. 17:53:23 what are the dates needed for liberty? 17:53:48 cathy__: sure... I can help out on either the API/DB or OVS driver... whichever one needs more contributors 17:53:54 but I would expect having L1 over next week, we really only have L2 to get the bulk of the functionality in and L3 to refine, stabilize and testing 17:54:14 armax: very aggressive 17:54:23 armax: btw, s3wong sure. I will update the ownership sign up and add your names there. 17:54:26 s3wong: very aggressive is an understatement! 17:54:29 s3wong: thanks! 17:54:37 cathy__: sure. Thanks! 17:55:11 s3wong: wildy and preposterously aggressive might be more approriate 17:55:33 armax: agreed. Consider we only just got our repo set up... :-) 17:55:39 armax: a man for understatement .... 17:56:06 if the foundations lied down are solid, I am not worried if we ran out of time 17:56:11 Let's get the feedback from each piece owner about their timeline of completion in next IRC meeting 17:57:03 cathy__: sure 17:58:02 armax: btw, Andreas said the Neutron team has adopted this project, so there is no need for election of a PTL. Can you confirm this? This will save my effort of wondering how we should position this project. 17:58:48 cathy__: yes, correct 17:58:58 armax: thanks. 17:59:16 time is up. Let's stop here and meet next week. Thanks everyone! 17:59:25 bye 17:59:30 bye everyone 17:59:33 thanks 17:59:41 bye 17:59:44 bye 17:59:53 Bye cathy .. thanks :-) 18:00:01 #endmeeting