14:30:09 <debrascott> #startmeeting Brahmaputra Daily Standup
14:30:09 <collabot> Meeting started Fri Feb  5 14:30:09 2016 UTC.  The chair is debrascott. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:30:09 <collabot> Useful Commands: #action #agreed #help #info #idea #link #topic.
14:30:09 <collabot> The meeting name has been set to 'brahmaputra_daily_standup'
14:30:29 <debrascott> #link agenda https://wiki.opnfv.org/releases/brahmaputra/minutes
14:30:39 <narindergupta> #info Narinder Gupta
14:30:46 <debrascott> #topic roll call
14:30:48 <joekidder> #info Joe Kidder
14:30:55 <bin_> #info Bin Hu
14:30:57 <anac1> #info Ana Cunha
14:31:09 <narindergupta> #info Narinder Gupta
14:32:39 <debrascott> #topic Test status
14:33:00 <debrascott> morgan_orange: functest progress today?
14:33:10 <ashyoung> #info Ashlee Young
14:33:14 <ChrisPriceAB> #info Chris Price
14:33:17 <morgan_orange> #info Morgan Richomme
14:33:31 <morgan_orange> did not check the nightly build, so a bit late for the update
14:33:57 <morgan_orange> vIMS should work on joid at the end of the day
14:34:10 <trevor_intel> #info Trevor Cooper
14:34:25 <morgan_orange> #info tempest info sent to apex (where it is critical to reach the 90%)
14:34:47 <morgan_orange> #info functest doc updated
14:34:54 <debrascott> morgan_orange: vIMS is good news
14:35:19 <morgan_orange> basically joid on Orange POD2 (not a CI POD) would probably meet the release criteria
14:35:37 <morgan_orange> Intel POD5 and 6 will be modified soon, so hopefully we may reach the same results on the target POD
14:36:08 <morgan_orange> that is all for me today
14:36:14 <debrascott> morgan_orange: is it difference in networking equipment?
14:36:42 <morgan_orange> surely, but I think they are facing the same issues as on Intel POD2
14:36:53 <morgan_orange> as far as I understand operations are plan to migrate the PODs
14:37:20 <trevor_intel> Morgan: POD 2 was migrated a week ago
14:37:26 <debrascott> morgan_orange: interesting. Something that will need to be documented.
14:37:45 <morgan_orange> morgan_orange: what? that hardware has an influence on results???
14:37:50 <trevor_intel> THere is only one known issue POD 5
14:38:13 <morgan_orange> trevor_intel I was thinking to POD5 as it is the target CI POD for joid/stable
14:39:01 <trevor_intel> Morgan: What are the issues?
14:39:16 <trevor_intel> I don't think we are aware of any
14:39:24 <morgan_orange> trevor_intel narindergupta wrote a Jira on that, I do not have the details
14:40:06 <narindergupta> trevor_intel morgan_orange deployment was it is working but when we run test then results were not same.
14:40:11 <debrascott> morgan_orange: yes, that diifferent hardware will change results. I guess I would have thought we were more abstracted away from hardware than that
14:40:42 <trevor_intel> The issue with POD 2 was an Apex limitation, not a broken POD
14:40:45 <narindergupta> it was mentioned to me that it could be network issues in pod and there was a plan to change the switch environment
14:41:41 <narindergupta> morgan_orange: in that case can we debug the functest different behavior. even pod5 and pod6 behaving differently
14:42:12 <trevor_intel> We want to migrate those PODs to a new switch
14:42:44 <trevor_intel> But need some down time to do so
14:43:13 <narindergupta> trevor_intel no issue let me know whenever you are ready
14:44:22 <debrascott> lets come back to pod management in a few. We need to add topic to the agenda: how will we manage migrating all pods to stable as noted in ChrisPriceAB email
14:44:36 <morgan_orange> narindergupta: trevor_intel so shall we investigate on POD5? or on the most favorable env? on POD5 looking at tempest the success rate is 14%, it is 92% on Orange POD (and 54% on POD6 as far as I remember)
14:45:25 <morgan_orange> narindergupta: trevor_intel it is probably a network config (more than hardware issue)
14:45:45 <narindergupta> morgan_orange: thats what i am thinking.
14:46:20 <narindergupta> morgan_orange: i am open for any options. What do you think trevor_intel
14:46:23 <narindergupta> ?
14:48:18 <trevor_intel> How to investigate?
14:48:28 <trevor_intel> We need a plan
14:48:44 <debrascott> morgan_orange: do we have a standard (ie recommended settings) for setting up network configs as part of setup for functest? Is that possible to do?
14:48:46 <trevor_intel> We will support whatever you need
14:49:31 <narindergupta> trevor_intel question is should we move to new environment first then debug? debrascott from host machine prspective we have same setting in both labs
14:49:51 <morgan_orange> debrascott: ???? Functest is not testing a system not defining this system, the reference is the Pharos architecture
14:50:22 <morgan_orange> Functest is testing a system not defining the network env of this system
14:51:37 <morgan_orange> trevor_intel narindergupta we can also support but when do you plan the operation? it is maybe useless to investigate if the network config is modified in a few days
14:51:38 <trevor_intel> narindergupta: POD 5 issue with BMC was closed yesterday ... POD 5 and 6 are identical
14:52:03 <debrascott> morgan_orange: just wondering what the setup process is for functest if outcome is dependent on hardware & configurations; Difficult to get high % runs if things are different in the different pods. Not really comparing apples to apples that way between the different installers/scenarios
14:52:49 <trevor_intel> Its not to be modified just migrated to a new switch
14:52:50 <debrascott> morgan_orange: makes troubleshooting more difficult
14:53:43 <trevor_intel> Differnt installer have different hw dependencies/assumptions .
14:53:55 <morgan_orange> debrascott: sure, that is why I mentioned this difference more than a week ago...so far we can see it with joid because we have 3 pods. compass and apex have only 1. For fuel we have 2 pods and we get similar results
14:54:58 <trevor_intel> Morgan: You mean 1 POD ... no they have 2
14:55:08 <morgan_orange> connected to CI?
14:55:19 <trevor_intel> yes
14:55:19 <debrascott> OK let’s continue offline. only 6 minutes left, want to get to yardstick status
14:55:22 <trevor_intel> https://wiki.opnfv.org/pharos_rls_b_labs
14:55:36 <debrascott> anac1: how is yardstick doing today?
14:55:41 <anac1> #info Yardstick succesfull run ODL scenario on Compass with ODL Be (no workaround needed)
14:55:51 <anac1> #info many thanks to Compass for the teamwork
14:56:07 <anac1> #info will try Apex next
14:56:11 <debrascott> anac1: yeah!!!!
14:56:30 <debrascott> that’s great news!
14:56:44 <anac1> that's all from me
14:56:52 <debrascott> OK thank you
14:57:14 <debrascott> Next topic
14:57:52 <debrascott> #topic how do we manage getting all available resources to stable as per ChrisPriceAB email?
14:57:54 <trevor_intel> Narinder / Morgan - I have to run but will be back online in an hour ... lets figure a plan
14:57:54 <ashyoung> anac1: that's awesome re: Be and Compass
14:58:05 <morgan_orange> trevor_intel ok
14:58:10 <narindergupta> sure
14:58:11 <anac1> ashyoung: yes, great teamwork
14:58:33 <JonasB> Why do we want to have all resources on stable, that breaks the whole idea of stable!
14:58:52 <debrascott> Did you see Chris’s email?
14:59:10 <debrascott> idea is to get test running through scenarios
14:59:22 <debrascott> find which scenarios work
14:59:36 <debrascott> put those that don’t work on back of queue
14:59:48 <debrascott> gives each scenario a chance to at least find out if it works
15:00:16 <debrascott> then start back through the queue to troubleshoot and get others working
15:00:54 <ashyoung> anac1: It means that we are at least narrowing on a configuration for supporting more scenarios. So I see this as a huge news.
15:01:26 <ashyoung> anac1: And in C release, we'll be looking at Be or it's successor anyway. We won't care about Li
15:01:43 <JonasB> I cant find that mail from chris, but is the meaning to get all PODs on stable or not?
15:02:06 <anac1> ashyoung: agree, trying to fix old faults perhaps not so effective
15:02:26 <debrascott> JonasB: I will forward it. This is a temporary move to get to release
15:02:36 <ashyoung> anac1: Agreed
15:03:07 <JonasB> But that is what I dont like, then the idea of stable is lost.
15:03:59 <radez> debrascott: anac1: are we ditching Li as release reqs?
15:04:18 <debrascott> JonasB: check your inbox
15:04:47 <debrascott> JonasB: no troubleshooting or fixes to “stable”, just test
15:04:55 <anac1> radez: yardstick tested successfully on fuel and compass with ODL Be
15:05:12 <radez> yes, but does that mean we are changing the release reqs?
15:05:14 <narindergupta> but that is shift to genesis requirement
15:05:15 <JonasB> Got the mail, but do not agree!
15:05:19 <radez> and we should not be building with Li?
15:05:28 <radez> anac1: debrascott: ^
15:05:33 <debrascott> JonasB: ok, take it up on the thread
15:05:48 <debrascott> radez anac1: yes watching
15:06:01 <debrascott> anac1 will this be ODL only?
15:06:06 <radez> debrascott: what do you mean watching?
15:06:17 <JonasB> We have master, stable and the cherry-pick process for good reasons. Do we want to go away from that in the last minute?
15:06:25 <debrascott> radez: two conversations, I’m paying attention
15:06:52 <radez> debrascott:  we should not be building with Li or Be?
15:07:06 <radez> sry that didn't make sense
15:07:13 <radez> debrascott: should we be building with Li or Be?
15:07:23 <JonasB> The idea is to make fixes on master, have ci verify it on master before you cherry pick it to stable. If it cant be verified on master, no reason to have stable!
15:07:24 <ashyoung> radez: build with ONOS ;)
15:07:32 <radez> ashyoung: we are! :)
15:07:33 <narindergupta> radez: that a CR should go to genesisreq
15:07:59 <debrascott> radez, anac1 Li is preferred, but if one controller needs Be I’m not sure what impact that might have to other projects. Does it impact installers?
15:08:17 <narindergupta> yes it does
15:08:27 <ashyoung> The earlier news about Be success on Yardstick w/o workarounds is good news and helps us to narrow working configs
15:08:35 <narindergupta> and that decision should not be done here
15:08:38 <radez> It just seems confusing for everyone to be asking for Be but for genesis to req Li
15:08:42 <debrascott> radez: agree a CR is needed so proper discussion can be made
15:08:43 <ashyoung> That's how I look at it-- not a decision
15:08:57 <anac1> debrascott, radez: yardstick works on Be without workaround
15:08:58 <debrascott> it is a tsc level discussion
15:09:08 <ashyoung> I agree with the issue about confusion
15:09:21 <ashyoung> Should be a Genesis decision
15:09:25 <ashyoung> Not TSC
15:09:26 <radez> anac1: right, and we've sliped to after Be release right? and SFC wants Be
15:09:44 <radez> so maybe we should just move to Be?
15:09:45 <ashyoung> radez: correct
15:09:54 <anac1> radez: yes
15:10:06 <ashyoung> The issue when we set the requirements was Be was going to be unstable
15:10:13 <ashyoung> Li would be "more stable"
15:10:22 <ashyoung> Seems flip flopped vs. reality
15:10:24 <radez> anac1: did compass wholesale switch to Be or do they just have an alternate build for YS?
15:10:24 <anac1> and late for the release date ?
15:10:30 <debrascott> I think Genesis first but it also needs to go to TSC for radification, it is a big change for last minute
15:10:54 <narindergupta> debrascott: +1 to your view
15:10:58 <ashyoung> I disagree, I don't think TSC ever set which versions of which components
15:10:59 <anac1> radez: not sure, i'd guess alternate build
15:11:06 <ashyoung> That was a genesis decision
15:11:26 <ashyoung> But I'm all for bringing it to the TSC
15:11:28 <JonasB> ashyoung: +1
15:11:36 <ashyoung> It's moving us forward vs. keeping us stuck
15:11:56 <anac1> radez: following yesterday's discussion, we tested with compass earlier today on Be and it works
15:12:22 <debrascott> ashyoung: we discussed all upstream revisions at TSC at beginning of release
15:12:23 <ashyoung> I don't believe compass team really cares what version is used
15:12:41 <debrascott> genesis decided and TSC bought into the Li due to some dependencies
15:12:53 <radez> anac1: I think it's less of an issue of if it works and more an issue of what's required by the release
15:12:56 <ashyoung> The documented notes was that we would not hold up the release for Be
15:13:10 <anac1> radez: i can only answer for yardstick
15:13:12 <ashyoung> Seems like Be might just get us through
15:13:25 <ashyoung> So I'm for going to TSC for clarifying it as a team
15:13:26 <debrascott> ashyoung: agree
15:13:36 <ashyoung> We want to all be on the same page
15:13:49 <debrascott> ashyoung: +1
15:13:53 <JonasB> Why work on something we know is broken and which will not fix?
15:14:07 <ashyoung> JonasB: that's my very concern
15:14:22 <debrascott> JonasB: agree but need to see what impact it has on the release as a whole
15:14:28 <ashyoung> In C release, we will be moving well beyond Li anyway
15:14:46 <radez> ashyoung: in C we'll move past Be too probably
15:15:04 <ashyoung> The question I have is how many scenarios get supported if we move from Li to Be-- do we gain?
15:15:13 <ashyoung> radez: agreed
15:15:20 <debrascott> ashyoung:+1 exactly
15:15:28 <ashyoung> It's forward progress and I am concerned about further slips for an old version
15:15:42 <ashyoung> Sounds like ODL folks fixed bugs in Be
15:15:53 <ashyoung> That's the vibe I get on our email thread on this
15:16:08 <radez> yea we've been told by ODL folks that Be is much better than Li
15:16:12 <radez> across the board
15:16:13 <ashyoung> This is with George Zhao and others
15:16:24 <ashyoung> And it supports SFC
15:16:36 <ashyoung> But I'm fine having the only SFC support in B release :)
15:16:47 <ashyoung> That's a joke
15:16:54 <debrascott> My concern is that we’ll have  some scenarios that work only with Be and some with Li
15:17:06 <ashyoung> That's what we need to discuss on the tech list
15:17:11 <radez> ashyoung: and we're all enjoying the joke :)
15:17:11 <anac1> debrascott: which ones ?
15:17:12 <ashyoung> Shall I start the discussion?
15:17:32 <ashyoung> radez: :)
15:17:48 <debrascott> anac1: I don’t know, maybe not an issue but that’s why it should be discussed with the broader community
15:17:53 <ashyoung> anac1: I think we need to take it to the list to see the risks
15:18:01 <debrascott> ashyoung: sure
15:18:11 <ashyoung> Okay
15:18:26 <ashyoung> posting now
15:18:41 <debrascott> Ok way past time for meeting. going to close it out. feel free to hang out on this channel and discuss if needed
15:18:46 <debrascott> #endmeeting