05:30:47 #startmeeting tacker 05:30:48 Meeting started Wed Jan 4 05:30:47 2017 UTC and is due to finish in 60 minutes. The chair is sridhar_ram. Information about MeetBot at http://wiki.debian.org/MeetBot. 05:30:49 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 05:30:51 The meeting name has been set to 'tacker' 05:30:56 hi 05:30:57 hi 05:30:58 #topic Roll Call 05:31:07 me here 05:31:07 o/ 05:31:22 gongysh: KanagarajM: janki: hi there! 05:31:23 hi all 05:31:37 Hi sridhar_ram, all 05:31:43 hi 05:31:50 hi 05:32:17 Happy new year team! 05:32:23 hi 05:32:25 let's start .. 05:32:37 #chair gongysh tbh KanagarajM 05:32:38 Current chairs: KanagarajM gongysh sridhar_ram tbh 05:32:46 #topic Agenda 05:32:50 #link https://wiki.openstack.org/wiki/Meetings/Tacker#Meeting_Jan_4th.2C_2016 05:33:21 #topic Announcements 05:34:15 nothing specific from my side, except the usual reminder of Ocata deadline coming up fast 05:34:31 #topic Ocata Blueprint Status 05:34:42 #topic NSD spec 05:34:56 tbh: dkushawa: can you please provide an update ? 05:35:24 * sridhar_ram realizes the topic should be NSD implementation status 05:35:49 sridhar_ram, pushed most of the functionality, the remaining items in the todo list are UTs/FTs 05:36:26 tbh, do we have a guide for testing? 05:36:32 tbh: that's wonderful ! 05:36:44 I just recorded a quick video of nsd https://www.dropbox.com/s/isu0fs1ebutg9gh/nsd.mp4?dl=0(not sure that is correct format to share) 05:36:57 gongysh, ^^ 05:37:20 tbh, cool, we should have a rst in our doc dir, I think. 05:37:33 gongysh, yes 05:38:07 gongysh: agree... 05:39:03 let's take 3 mins to quickly look at tbh's video 05:42:03 tbh, the video is cool 05:42:07 folks - please share your comments 05:42:22 some optimizations are needed as it involves building dict and yaml. so I think it will be covered mostly in reviews :) 05:42:33 tbh: the demo is cool 05:43:02 tbh: do we need those import line (for VNF1 and VNF2) in nsd.yaml ? 05:43:18 sridhar_ram, gongysh thanks 05:43:39 sridhar_ram, yes as per TOSCA nfv profile syntax 05:44:13 tbh, nice demo ! 05:44:21 tbh: sounds good, glad we made VNFD name unique 05:44:22 tbh: good one 05:44:43 sridhar_ram, tbh , i think import part is not required for tacker defintions 05:44:48 sridhar_ram, absolutely, that helped a lot in this patch 05:45:03 tbh: all this using mistral behind the scenes ? 05:45:05 as its already taken care in tacker while invoking the HT 05:45:31 sridhar_ram, yes, mistral occupied most of the time 05:46:16 KanagarajM, we aren't invoking HT directly from nsd perspective 05:46:40 nsd -> mistral -> tacker -> VNF 05:46:40 tbh, I need a guide to test, do we have one somewhere? 05:46:43 tbh, oh ! yes. I remember 05:47:02 sorry, late 05:47:09 gongysh, I will push the rst by EOD 05:47:24 tbh, thanks, looking forward to it. 05:47:48 sridhar_ram, tbh, so now, we could make NDS and VNFD all follow the same process in terms of importing the tacker definitions. is it ok ? 05:48:14 KanagarajM: i think you've this HT bug https://bugs.launchpad.net/heat-translator/+bug/1652928 in mind 05:48:16 Launchpad bug 1652928 in Heat Translator "Fails to translate the derived polices" [High,New] - Assigned to Sahdev Zala (spzala) 05:48:26 sridhar_ram, yes :) 05:48:55 tbh: while we don't use HT, we still use tosca-parser for NSD and tosca imports still play a role there. correct ? 05:49:01 s3wong: hi there 05:49:24 sridhar_ram, right 05:49:38 KanagarajM, I didn't get you 05:50:00 KanagarajM: now, i don't understand how can we get to follow the same process .. they already follow the same pattern 05:50:42 KanagarajM: we remove tacker custom types from the ToscaTemplate and pass a "sanitized" tree to HT 05:50:54 sridhar_ram, in case of VNFD, tacker takes the resposibility of importing the tacker defined tosca defitions, while for NSD user should include in TOSCA 05:51:20 o/ 05:51:24 sorry I am late 05:51:49 KanagarajM: that where i was kind of questioning tbh .. why we need an import. But now I understand it make sense to have them 05:52:05 sridhar_ram, you mean, will include imports in TOSCA before passign to HT 05:52:21 KanagarajM: without those import the "VNF1" type few lines further down won't parse 05:52:48 sridhar_ram, you are right 05:53:05 o/ 05:53:12 sridhar_ram, and these VNF1 has to be defined in VNFD1.yaml 05:53:36 KanagarajM: for VNFD the imports are custom variations on top of TOSCA_nfv_definitions in tosca-parser, so those custom types needs to be stripped before sending to HT. HT won't know an tacker custom types 05:53:46 sridhar_ram, tbh, for VNFD, HT passes the parsing without explicitily menting the import statements in VNFD 05:54:06 tbh: hmm.. you mean there should be a VNFD1 in the tacker vnfd catalog? 05:54:09 sridhar_ram, KanagarajM like this http://paste.openstack.org/show/593839/ 05:54:35 sridhar_ram, yes 05:55:00 tbh: so, it is not really a "file" per se ? 05:55:08 tbh: we can take the rest in the code review 05:55:40 overall .. happy w/ the progress you and dkushwaha has made! 05:55:42 good job 05:55:47 i think we should not bother user to include the tacker defined types in VNFD or NSD, because tacker should take care of it 05:55:52 sridhar_ram, we thought of having VNFD name in NSD instead of VNFD path 05:56:30 sridhar_ram, ^^ Tacker takes the control on the TOSCA definitions supported by it 05:57:21 tbh, s3wong , sridhar_ram but does HT expecting to have import in place, 05:57:42 KanagarajM, I think we should provide users to extend the TOSCA definition in the future. 05:57:52 KanagarajM, HT when invokes TP, it expects imports 05:57:55 KanagarajM: we can consider automating all the VNF type "references" to automatically lookup in the tacker catalog .. that is more a convenience (reducing few lines in NSD) than anything else 05:59:01 gongysh, not sure ! 05:59:10 KanagarajM: anyways, leave that comment in the gerrit .. we can let tbh and dkushwaha evaluation it 05:59:15 sridhar_ram, yes +1 05:59:23 to me it is not a deal breaker 05:59:42 sridhar_ram, yes, i would like tbh, s3wong would take a look on that defect 05:59:50 let's get the NS client changes in ASAP 06:00:21 KanagarajM: sure 06:00:29 s3wong, thanks 06:00:50 anything else on NSD? 06:01:16 moving on.. 06:01:34 had few issues in horizon part 06:02:11 but keeping it in the last part for now 06:02:19 there was an issue on VNFD catalog page which gongysh just pushed a patch (thanks gongysh!) 06:02:34 tbh: sure, make sense.. 06:03:09 KanagarajM: since we are in the topic, do you want to discuss https://bugs.launchpad.net/heat-translator/+bug/1652928 here ? 06:03:10 Launchpad bug 1652928 in Heat Translator "Fails to translate the derived polices" [High,New] - Assigned to Sahdev Zala (spzala) 06:03:11 sridhar_ram, it bothered me a few days. 06:03:44 gongysh: i finally hit the same problem today after a restack 06:04:10 sridhar_ram, yes 06:04:22 #topic HT failure due to tacker imports 06:05:10 my understanding is there is a new release of HT (and TP?) pending release any moment and that is failing tacker sanity tests ? 06:05:10 s3wong, we are waiting for this bug to get dfixed sooner for scaling seletive vdu 06:05:18 selective vdu scaling * 06:05:57 s3wong, sahdev is looking into already. and kindly help further on it 06:06:39 KanagarajM: from the bug notes, this sounds like a HT regression. 06:06:42 KanagarajM, sridhar_ram: seems like bobh's area --- looking at his comment on the bug now 06:06:53 s3wong: agree 06:07:07 o/ 06:07:12 sorry got late 06:07:13 bobh: are you here? 06:07:13 s3wong, sure. 06:07:18 diga: hi ! 06:07:19 sridhar_ram, yes 06:07:49 KanagarajM: okay, spzala is probably not online now 06:07:49 sridhar_ram: probably a little late for bobh 06:08:15 sridhar_ram, seems ! 06:08:34 KanagarajM: what is the specific error you get? unknown node type? 06:09:10 sridhar_ram, no. It is not returning the scaling yaml (provider template to be generated by HT as part policy->targets) 06:10:09 tung_doan, do you run into this bug? 06:10:51 to produced the bug, you need to use the master copy of the HT 06:10:57 KanagarajM: Hmm... HT should honor any derived types and parse properly 06:11:06 sridhar_ram, +1 06:11:45 KanagarajM: it will be good if we can narrow down to a specific HT patchset that broke this 06:12:13 KanagarajM: in that case, seems like there has been changes on master HT which caused this prpblem 06:12:16 problem 06:12:43 KanagarajM: obviously something post HT 0.7.0 release.. anyways spzala is looking into this 06:12:50 let's give him sometime 06:12:52 sridhar_ram, its a new feature added in HT, but it broken the back-ward comaptibility 06:13:13 i'll follow up him tomorrow US time 06:13:19 sridhar_ram, sure. thanks 06:13:19 KanagarajM: yeah, seems like 06:13:44 KanagarajM: so the breakage is between heat-translator and latest HT? 06:13:52 like i mentioned, any incomp issue w/ tosca-parser or HT is a pain, need to be on the top of it 06:14:11 s3wong: yes 06:14:33 gongysh: no.. seems like it's the new bug 06:14:51 s3wong, yes i guess so, as tacker didn't face the issue in neuton 06:15:20 alright, let's circle back after the resp from spzala 06:15:29 moving on.. 06:15:56 #topic Ocata Client Release 06:16:42 In preparation for Ocata Final Client release on Jan 23rd (see https://releases.openstack.org/ocata/schedule.html) 06:16:53 .. we need to get all client changes in ASAP 06:17:25 cores: please prioritize the tackerclient patchsets in your review queue 06:18:01 once NSD client merges i plan to release an "interim" tackerclient 06:18:26 sridhar_ram, got it 06:18:28 this will help to pass dsvm gate tests for the NSD server sider patchset 06:18:35 gongysh: thanks! 06:19:17 apart from NSD and VNFFGD param support, are there any other feature pending tackerclient changes for Ocata ? 06:19:33 sridhar_ram, inline template? 06:19:46 I have pushed the client patch 06:20:06 janki: yes, thanks for reminding.. 06:20:16 janki: do you've a link? 06:20:47 sridhar_ram, https://review.openstack.org/#/c/414602/ - client patch 06:20:54 janki: thanks 06:21:04 I have 1 query sort of too for this feature, sridhar_ram 06:21:10 fyi, here is the NSD client patchset .. https://review.openstack.org/412188 06:21:29 sridhar_ram, i just +2 06:21:43 sridhar_ram, waiting for main feature to land 06:21:53 KanagarajM: thanks! 06:22:16 janki: let's move to open discussions 06:22:24 anything else on tackerclient ? 06:22:35 KanagarajM, why wait for main feature to land. I think main feature is waiting for it. 06:22:52 gongysh: +1 06:23:03 gongysh, i mean NSD feature support in Tacker 06:23:27 sridhar_ram, gongysh am i missing something :( 06:23:56 KanagarajM, Kind of. the codes in client and server sometimes depend on each other. 06:24:02 KanagarajM: yeah, I thought we are releasing client first --- so assuming NSD landing in Ocata is going to happen, we can merge the client changes now... 06:24:03 KanagarajM, to pass dsvm gate tests for NSD patches in Tacker we need client patch to be merged 06:24:13 KanagarajM, server code needs it to have functional tests. 06:24:34 KanagarajM: typically client changes goes in first w/o server, we make a client release and server will use to pass dsvm gate 06:24:41 yeah, only functional test will be pending i agree 06:24:49 KanagarajM: exactly1 06:24:56 KanagarajM, yes. 06:25:02 #topic Open Discussion 06:25:08 janki: what is your question ? 06:25:14 sridhar_ram, KanagarajM , I think we can let client merge. 06:25:28 diga: thanks for taking fwd the api-framework using falcon! 06:25:36 gongysh: agree 06:25:39 sridhar_ram, I am thinking in inline template, while deleting vnf the corresponding vnfd should also be deleted 06:25:44 sridhar_ram: yep 06:25:53 gongysh, sridhar_ram i think if we +A, gerrit does not allow as there is a depends-on tag in commit message 06:25:57 this point is mentioned in the spec 06:26:07 janki: make sense to me 06:26:16 sridhar_ram: can we go ahead & create feature/branch for falcon-api-framework 06:26:28 * sridhar_ram notes 4 mins left 06:26:31 sridhar_ram, gongysh so we need to remove depends-on tag from commit message ! 06:26:36 KanagarajM, yes. 06:26:41 sridhar_ram: I will address your comments by today EOD after discussion with monasca team 06:26:44 sridhar_ram, should I code it even without being mentioned in spec? 06:26:51 diga: let's take up api-framework discussion next week 06:26:57 sridhar_ram: okay 06:27:00 KanagarajM, I will update the commit message 06:27:25 sridhar_ram, and also to delete vnfd if vnf is in error state. not sure how valid this would be 06:27:32 sridhar_ram, what happened to odl sfc? 06:27:45 tbh, sure. thanks. 06:28:35 janki: it is fine unless it is against what is mentioned in the spec.. 06:29:04 sridhar_ram, it is not against the spec. infact the point is not mentioned. so will add it 06:29:15 updated the commit message for client patch https://review.openstack.org/#/c/412188/ 06:29:18 gongysh: that is a bigger topic, another time 06:29:18 sridhar_ram, how about deleting vnfd if vnf is in error state? 06:29:44 janki: let's take it offline 06:29:55 times up folks 06:29:59 thanks for joining 06:30:04 bye 06:30:04 bye 06:30:07 #endmeeting