13:59:23 <uli-k_> #startmeeting Octopus weekly meeting
13:59:23 <collabot> Meeting started Mon Oct  5 13:59:23 2015 UTC.  The chair is uli-k_. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:59:23 <collabot> Useful Commands: #action #agreed #help #info #idea #link #topic.
13:59:23 <collabot> The meeting name has been set to 'octopus_weekly_meeting'
13:59:52 <uli-k_> #topic roll call
14:00:12 <trozet> #info Tim Rozet
14:00:16 <uli-k_> Hello everybody! Type #info if you want to participate in octopus meeting
14:00:24 <[1]JonasB> #info Jonas Bjurel
14:00:47 <fdegir> #info Fatih Degirmenci
14:01:31 <uli-k_> My chinese colleagues will probably be not here, since it is a holiday in China.
14:02:13 <uli-k_> So looks like we are a smaller group today.
14:02:25 <uli-k_> #topic Agenda review
14:02:58 <uli-k_> I just saw I forgot to update the date ....
14:03:37 <uli-k_> #info besides standard topics, I just put there "Starting work on B Jira topics"
14:03:56 <uli-k_> Should we add the lab configuration question Fatih had in the email?
14:04:22 <trozet> yes
14:04:22 <fdegir> yes
14:04:54 <fdegir> it makes your action item more important :)
14:05:01 <uli-k_> #info Adding lab configuration to the agenda.
14:05:07 <uli-k_> Yes.
14:05:27 <uli-k_> And sorry, I didn't press the lab owners yet. :(
14:05:53 <uli-k_> anything else for agenda?
14:06:15 <uli-k_> Then let's go for Action Items.
14:06:24 <uli-k_> #topic action item review
14:06:49 <uli-k_> #info first AI: uli-k trozet ChrisPriceAB: clarify the need of doing Brahmaputra release on LF Lab
14:07:16 <uli-k_> No progress....
14:07:29 <trozet> uli-k: I personally think if the lab meets pharos requirements it shouldn't matter if its LF or not
14:07:31 <fdegir> can this go to the top of the agenda for tomorrow's TSC?
14:07:37 <fdegir> it's been there for more than a month
14:07:50 <fdegir> +1 trozet
14:08:07 <uli-k_> I can try persuade ChrisPriceAB
14:08:35 <uli-k_> I'll put the action again here.....
14:08:57 <uli-k_> #action uli-k trozet ChrisPriceAB: clarify the need of doing Brahmaputra release on LF Lab - put high priority on TSC agenda
14:09:40 <uli-k_> #info next: uli-k talk to Huawei lab guys to support Pharos
14:09:58 <uli-k_> Need to wait for them coming back from holiday now.
14:10:06 <uli-k_> #action uli-k talk to Huawei lab guys to support Pharos
14:10:26 <uli-k_> #info next: uli-k to trigger other lab reps according to community lab overview wiki
14:10:28 <fdegir> uli-k_: have a comment regarding that
14:10:45 <fdegir> as I put into the wiki, we have lots of huawei slaves there
14:10:48 <fdegir> which is good
14:10:55 <fdegir> but connection is not stable
14:11:15 <fdegir> can you push lab guys to increase the bandwidth, check the network, etc. to make sure they're up at all times
14:11:28 <fdegir> or else, it will fail on connectivity tests which I'm about to write
14:11:37 <uli-k_> OK. Point taken.
14:11:44 <fdegir> checking how much a certain slave is up and utilized
14:11:48 <fdegir> thx
14:12:04 <uli-k_> #action uli-k to talk to Huawei lab guys to increase the bandwidth, check the network, etc. to make sure they're up at all times
14:12:29 <uli-k_> As already said, my action to other lab reps is also still open. Sorry.
14:12:39 <fdegir> np
14:12:43 <uli-k_> #action uli-k to trigger other lab reps according to community lab overview wiki
14:13:02 <uli-k_> But the next one I can close :D
14:13:12 <uli-k_> #info next AI: uli-k to remove 3 committers from INFO file and inform TSC
14:13:22 <uli-k_> #info done
14:13:57 <uli-k_> I reached chengappa via slack. He wanted to call in today. Chengappa, are you here?
14:14:52 <uli-k_> For another HP guy whom I was not able to contact, and for Palani, I start TSC process for committer removal.
14:15:09 <uli-k_> After that we are down to 8 committers.
14:15:32 <uli-k_> So I will go on talking to people to come back or resign.
14:15:58 <uli-k_> Which was the last AI on the list.
14:16:11 <uli-k_> Shall we go for the lab issue first?
14:16:31 <fdegir> yes
14:16:51 <uli-k_> #topic Reconfiguration of LF POD 1
14:16:57 <fdegir> the reason I sent mail is the mail morgan_orange sent last week
14:17:04 <fdegir> regarding no of needed PODs
14:17:27 <fdegir> also looking at our sorry resource state
14:17:53 <fdegir> made me ask myself if we should keep the POD since LF PODs are the only ones that usable
14:18:24 <fdegir> and we have that discussion going on
14:18:35 <fdegir> #info The discussion is related to reconfiguration of LF POD1
14:18:57 <fdegir> #info Should we reconfigure LF POD1 and convert all servers to standalone servers
14:19:02 <uli-k_> That was your option 1, right ? Use some of the servers in POD 1 for virtual deploy.
14:19:26 <fdegir> that is option 2
14:19:52 <fdegir> #info Above option is the original plan we decided some time ago
14:20:18 <fdegir> #info First option is not to convert LF POD1 and keep it as it is; to be used for work on metal
14:20:47 <fdegir> it seems we have slight majority regarding following the original plan
14:20:57 <fdegir> #info It seems we have slight majority regarding following the original plan
14:21:07 <uli-k_> Question to option 2: Can we run virtual deployment on all these servers then?
14:21:42 <fdegir> uli-k_: 1 (or maybe 2) will be build server
14:21:51 <fdegir> the rest will probably be virt deployment
14:22:02 <fdegir> depending on what other projects out there, needing resources
14:22:20 <fdegir> kvm is an example
14:23:13 <fdegir> here is how many servers we will have if we convert the POD
14:23:29 <fdegir> 1 ericsson + 5 intel + 6 LF = 12
14:23:37 <fdegir> 2 or 3 of them will be build servers
14:23:47 <uli-k_> For me, option 2 looks still good then.
14:23:51 <fdegir> majority of the rest will be virt deploys
14:24:09 <fdegir> it looks good for the timebeing
14:24:21 <fdegir> but for work on metal in coming months
14:24:25 <fdegir> it is a bit depressing
14:24:42 <uli-k_> Right. We need the community labs.
14:24:43 <trozet> i think we need to have more servers to launch virt deploy so i go with option 2
14:24:53 <trozet> we also need at least 2 pods to run baremetal deploy
14:24:55 <fdegir> uli-k_: yes
14:24:57 <trozet> but we can use intel lab or something else
14:25:02 <uli-k_> I think community labs should do bare metal
14:25:16 <fdegir> trozet: we need to get lab(s) connected to CI
14:25:21 <fdegir> if it happens, we should be fine
14:25:36 <trozet> fdegir: sure
14:25:42 <fdegir> but the situation is not encouraging
14:25:44 <fdegir> https://wiki.opnfv.org/pharos/hw_resource_allocation
14:26:14 <fdegir> trozet: maybe we can connect the POD (Intel POD2) you use to jenkins
14:26:18 <uli-k_> We have to make people aware of the situation. Great to have your wiki for that.
14:26:20 <fdegir> and you continue using it
14:26:37 <fdegir> and you can still manage your work yourself
14:26:47 <fdegir> enable/disable apex depending on what you need
14:27:01 <fdegir> then we offload POD2
14:27:06 <fdegir> LF POD2 I mean
14:27:26 <fdegir> and fuel will probably move to ericsson-lab
14:27:27 <trozet> fdegir: we can connect intel pod 2 to jenkins for baremetal deploys
14:27:41 <trozet> fdegir: why dont we do that and break down pod1 as our solution?
14:27:41 <fdegir> so apex go to intel pod2, fuel goes to ericsson lab
14:27:54 <fdegir> trozet: if this works for you
14:27:59 <fdegir> I'm more than happy to do that
14:28:00 <trozet> except for that last part
14:28:08 <fdegir> last part?
14:28:10 <trozet> apex goes to pod 2, fuel goes to ericsson lab?
14:28:18 <fdegir> apex -> intel pod2
14:28:23 <fdegir> fuel -> ericsson lab
14:28:35 <fdegir> so we utilize the community labs
14:28:43 <trozet> why do that though?  cant we just keep the baremetal pods as a pool
14:28:51 <trozet> and deploy on all of them
14:29:07 <trozet> and does that mean ericsson lab will provide a baremetal pod?
14:29:09 <fdegir> that might cause this "lab reconfig"
14:29:26 <fdegir> yes, we have plans to have baremetal pod
14:29:36 <trozet> ok so thats 3 baremetal pods
14:29:42 <fdegir> trozet: the best is what you say
14:29:49 <fdegir> see all metal pods as pool of pods
14:29:53 <trozet> can we pool them together?
14:29:56 <fdegir> and let jobs use whatever available
14:30:19 <fdegir> as long as we don't get into this LF networking/reconfig issues
14:30:23 <fdegir> that should work
14:30:38 <trozet> ok
14:30:41 <fdegir> but
14:30:47 <fdegir> can we start with intel pod2 first
14:30:54 <fdegir> and make it visible on jenkins
14:31:05 <trozet> fdegir: I think the networking/reconfig issues will be solved when the UCS patch that frankbrockners mentioned last week comes that no longer requires a Vlan 0 tag
14:31:23 <trozet> fdegir: yeah
14:31:26 <fdegir> trozet: can't wait to see it
14:31:32 <fdegir> then we should info this
14:31:37 <frankbrockners> :-) - me too
14:31:43 <fdegir> #info the networking/reconfig issues will be solved when the UCS patch that frankbrockners mentioned last week comes that no longer requires a Vlan 0 tag
14:31:59 <fdegir> #info Intel POD2 will be connected to jenkins
14:32:17 <trozet> fdegir: you want me to work with aricg- to connect intel pod2?
14:32:20 <fdegir> #info Ericsson will also have a metal POD
14:32:24 <fdegir> trozet: please do
14:32:33 <fdegir> #info LF POD2 will also be there
14:32:40 <trozet> #action trozet to connect intel pod 2 to OPNFV CI
14:32:47 <fdegir> #info This makes total 3 metal PODs; 1 OPNFV + 2 community
14:33:11 <trozet> just fyi, we are currently using LF POD1 jumphost to do our apex build, virt deploy
14:33:19 <trozet> so if we coudl leave that server as is
14:33:28 <fdegir> #info The feedback from Octopus is to continue with original LF POD1 and break it down, converting to standalone servers
14:33:29 <trozet> then split the rest out into individual broadcast domains
14:33:49 <fdegir> trozet: can you please make intel machines usable as well?
14:33:58 <fdegir> then we see all slaves as pool of resources
14:34:36 <fdegir> #info LF POD1 jumpstart should not be touched until further notice
14:35:23 <fdegir> trozet: pick one from https://etherpad.opnfv.org/p/cihw
14:35:24 <trozet> fdegir: what do you mean make intel machines usable?
14:35:40 <fdegir> trozet: we added 5 servers from another intel POD
14:35:45 <trozet> oh those servers
14:35:48 <fdegir> yes
14:35:58 <trozet> fdegir: OK
14:36:05 <fdegir> ok
14:36:27 <fdegir> I think we are good to move on if no other comments
14:36:40 <fdegir> I'll reply to the latest mail and point to minutes
14:37:17 <fdegir> did uli-k just drop?
14:41:43 <trozet> i guess so fdegir
14:41:48 <trozet> you drive the rest of ht meeting?
14:41:55 <fdegir> I'm not a chair
14:42:07 <fdegir> #topic B-Release Work
14:42:09 <trozet> oh :)
14:42:15 <trozet> heh
14:42:15 <fdegir> let's try :)
14:42:25 <trozet> the bot doesn't like you :)
14:42:33 <fdegir> :)
14:43:02 <fdegir> anyway may I ask a question about apex then?
14:43:28 <fdegir> can more than one virt deploy be run for apex on same server?
14:43:33 <fdegir> or it is only one?
14:43:44 <trozet> hmm good question
14:43:59 <trozet> we could make a modification so that more than 1 could run
14:44:20 <fdegir> then
14:44:25 <fdegir> we can spawn more deploys
14:44:39 <fdegir> how much time it takes for a deploy to complete?
14:44:43 <trozet> right now we cannot, but all it would need is change of bridges and domain names in libvirt
14:45:08 <trozet> fdegir: cannot comment on that yet as we don't have the deploy fully working.  build takes 45 min and builds all the ISOs
14:45:26 <fdegir> ok
14:45:39 <fdegir> this will be important when we start assigning roles to servers
14:45:56 <fdegir> will ask same questions for fuel as well
14:50:08 <uli-k_> Hi
14:50:20 <fdegir> hi
14:50:32 <uli-k_> Sorry, my connection broke.
14:51:00 <fdegir> np
14:51:14 <fdegir> we couldn't end the meeting even though we tried hard :)
14:51:15 <uli-k_> could you continue?
14:51:36 <fdegir> tried to switch to B-release work
14:51:43 <uli-k_> Ok, so i end the meeting now.
14:52:03 <fdegir> was just joking
14:52:15 <fdegir> we can perhaps switch the topic to B-release work and then we end
14:52:26 <uli-k_> :D
14:52:46 <uli-k_> #Topic B release
14:53:31 <uli-k_> #chair fdegir trozet
14:53:31 <collabot> Current chairs: fdegir trozet uli-k_
14:54:32 <fdegir> I think we're moving towards what we planned
14:54:36 <fdegir> a bit slow perhaps
14:54:53 <fdegir> but we have fuel, apex, and compass jobs there
14:55:06 <uli-k_> Yes. And the Pharos dependency bothers me.
14:55:28 <fdegir> uli-k_: yes, all the stuff I write is in order to make this obvious
14:55:42 <fdegir> trying to find workarounds
14:55:50 <uli-k_> I fear people dont know what it means.
14:56:21 <uli-k_> We cannot just kick Pharos out of B
14:56:25 <fdegir> yes
14:56:36 <fdegir> and we cannot just say "here is our lab"
14:56:39 <fdegir> it must work
14:57:20 <uli-k_> Right. I try to bring this to the release meeting...
14:57:41 <fdegir> I plan to create this script to visualize uptime of the slaves
14:57:54 <uli-k_> And now we have to give the channel to genesis
14:57:56 <fdegir> and then say "sorry you do not fulfil CI reqs"
14:58:02 <fdegir> you're out of jenkins
14:58:09 <fdegir> to get some attention
14:58:30 <fdegir> ok
14:58:33 <uli-k_> Yes.
14:58:45 <uli-k_> Ok thanks all!
14:59:01 <uli-k_> #Endmeeting