Thursday, 2015-12-10

*** gfidente has quit IRC00:02
*** panda has quit IRC00:03
*** panda has joined #tripleo00:04
*** mcornea has quit IRC00:06
*** chlong has quit IRC00:09
openstackgerritDan Sneddon proposed openstack/tripleo-heat-templates: Add IPv6 Support to Isolated Networks
*** dmacpher has quit IRC00:42
*** olaph has quit IRC00:56
*** eil397 has quit IRC01:02
*** chlong has joined #tripleo01:05
*** yuanying_ has joined #tripleo01:09
*** yuanying_ has quit IRC01:09
*** yuanying has quit IRC01:11
*** chlong has quit IRC01:23
*** yuanying has joined #tripleo01:23
openstackgerritDan Sneddon proposed openstack/tripleo-heat-templates: Add IPv6 Support to Isolated Networks
*** sthillma has quit IRC01:41
*** apetrich has quit IRC01:47
*** sthillma has joined #tripleo01:48
*** apetrich has joined #tripleo01:48
*** nijaba has quit IRC01:55
*** nijaba has joined #tripleo01:56
*** nijaba has quit IRC01:56
*** nijaba has joined #tripleo01:56
*** rcernin has quit IRC02:01
*** panda has quit IRC02:01
*** rhallisey has joined #tripleo02:09
*** davidlenwell has quit IRC02:10
*** dmacpher has joined #tripleo02:10
*** davidlenwell has joined #tripleo02:13
*** ChanServ sets mode: +v davidlenwell02:13
*** rhallisey has quit IRC02:13
openstackgerritSteve Baker proposed openstack/tripleo-heat-templates: Set the name property for all deployment resources
*** alop has quit IRC02:29
*** shivrao has quit IRC02:32
*** links has joined #tripleo02:33
*** cwolferh has quit IRC02:42
*** chlong has joined #tripleo02:44
*** yuanying has quit IRC03:19
*** yuanying has joined #tripleo03:37
*** anande has joined #tripleo03:58
openstackgerritTzu-Mainn Chen proposed openstack/tripleo-specs: Spec for TripleO overcloud deployment REST API
*** yuanying has quit IRC04:02
*** davidlenwell has quit IRC04:06
*** yuanying has joined #tripleo04:06
*** davidlenwell has joined #tripleo04:09
*** ChanServ sets mode: +v davidlenwell04:09
*** sthillma has quit IRC04:09
*** yamahata has joined #tripleo04:10
*** cwolferh has joined #tripleo04:47
*** masco has joined #tripleo04:56
*** shivrao has joined #tripleo05:08
*** shivrao has quit IRC05:22
*** sthillma has joined #tripleo05:27
*** sthillma_ has joined #tripleo05:27
*** sthillma has quit IRC05:31
*** sthillma_ is now known as sthillma05:31
*** tzumainn has quit IRC05:35
*** sthillma has quit IRC05:44
*** greghayn1 has joined #tripleo05:57
*** greghaynes has quit IRC05:58
*** greghayn1 is now known as greghaynes05:58
*** greghaynes has quit IRC05:59
*** greghaynes has joined #tripleo06:00
*** links has quit IRC06:01
*** links has joined #tripleo06:01
*** greghaynes has quit IRC06:05
*** greghaynes has joined #tripleo06:06
*** chlong has quit IRC06:28
*** jaosorior has joined #tripleo06:29
*** links has quit IRC06:49
*** chlong has joined #tripleo06:49
*** chlong has quit IRC07:04
*** links has joined #tripleo07:05
*** ukalifon has joined #tripleo07:10
*** ukalifon has quit IRC07:10
*** rcernin has joined #tripleo07:11
*** dmacpher has quit IRC07:16
*** rasca has joined #tripleo07:18
*** ukalifon has joined #tripleo07:19
*** chlong has joined #tripleo07:20
*** bvandenh has joined #tripleo07:21
*** rasca has quit IRC07:24
*** rasca has joined #tripleo07:26
*** tserong has joined #tripleo07:26
*** tserong has joined #tripleo07:26
*** jprovazn has joined #tripleo07:43
*** sthillma has joined #tripleo07:48
*** tremble has joined #tripleo07:56
*** tremble has joined #tripleo07:56
*** mkovacik has joined #tripleo07:58
*** chlong has quit IRC08:05
*** bhunter has quit IRC08:05
*** liverpooler has joined #tripleo08:07
*** liverpooler has quit IRC08:10
jaosoriortremble: Hey dude, got any idea if apache (used for horizon) is currently managed by pacemaker?08:10
*** liverpooler has joined #tripleo08:10
tremblejaosorior: I thought it was...  There were some discussions around apache taking too long to restart08:11
*** bhunter has joined #tripleo08:11
*** aufi has joined #tripleo08:11
*** yamahata has quit IRC08:14
*** regebro has quit IRC08:14
*** links has quit IRC08:14
*** regebro has joined #tripleo08:15
*** devvesa has joined #tripleo08:19
*** athomas has quit IRC08:22
*** mbound_ has joined #tripleo08:23
*** athomas has joined #tripleo08:23
*** sthillma has quit IRC08:26
*** sthillma has joined #tripleo08:26
*** ifarkas has joined #tripleo08:39
*** ifarkas has joined #tripleo08:40
*** shardy has joined #tripleo08:43
*** bvandenh has quit IRC08:43
*** pblaho has joined #tripleo08:44
*** Marga_ has quit IRC08:44
*** links has joined #tripleo08:52
*** nico_auv has joined #tripleo09:00
*** sthillma has quit IRC09:02
*** Marga_ has joined #tripleo09:03
*** Marga_ has quit IRC09:03
*** Marga_ has joined #tripleo09:03
*** sthillma has joined #tripleo09:04
*** Marga_ has quit IRC09:04
*** Marga_ has joined #tripleo09:05
*** gfidente has joined #tripleo09:10
*** gfidente has quit IRC09:10
*** gfidente has joined #tripleo09:10
*** yuanying has quit IRC09:10
*** gfidente has quit IRC09:16
*** gfidente has joined #tripleo09:28
*** gfidente has quit IRC09:28
*** gfidente has joined #tripleo09:28
*** paramite has joined #tripleo09:30
*** paramite is now known as paramite|afk09:31
*** openstackgerrit has quit IRC09:32
*** openstackgerrit has joined #tripleo09:32
*** sthillma has quit IRC09:33
*** sthillma has joined #tripleo09:33
jaosoriorgfidente: Do you remember why this was commented out?
jaosoriorgfidente: or how does httpd interact with pacemaker?09:37
gfidentejaosorior, so it is started/stopped by pcmk09:40
jaosoriorgfidente: I didn't really understand where in the manifest it's set to be managed as a pacemaker service :/09:41
gfidentejaosorior, the reason for commenting out service_manage is in bug 1247547 in openstack-tripleo-heat-templates "overcloud: HA: httpd service should not be started on boot by systemd." [High,Post] - Assigned to gfidente09:42
*** links has quit IRC09:42
gfidentejaosorior, and it's added to pcmk here
jaosorioraah, I understand now09:44
jaosoriorgfidente: thanks man09:44
*** mcornea has joined #tripleo09:48
gfidentejaosorior, I think we wanted to get away without importing ::apache at all but wasn't possible09:50
gfidente(without importing specifically)09:50
*** athomas has quit IRC09:51
*** jcoufal has joined #tripleo09:51
jaosoriorgfidente: I'm running into those issues, cause running keystone in wsgi ends up iporting ::apache09:51
*** paramite|afk is now known as paramite09:51
*** sthillma has quit IRC09:54
*** links has joined #tripleo09:55
*** athomas has joined #tripleo09:56
*** jistr has joined #tripleo09:59
*** links has quit IRC10:04
*** mkovacik has quit IRC10:12
*** links has joined #tripleo10:17
*** electrofelix has joined #tripleo10:19
*** akrivoka has joined #tripleo10:27
*** dtantsur|afk is now known as dtantsur10:30
*** paramite is now known as paramite|afk10:33
openstackgerritSteven Hardy proposed openstack/tripleo-heat-templates: Fix wrong keypair parameter description
openstackgerritMartin AndrĂ© proposed openstack/tripleo-specs: Spec for validation API
*** tosky has joined #tripleo11:06
dtantsurshardy, hi! do you have release rights? I would really appreciate a DIB release ASAP, if it's possible11:13
openstackgerritDougal Matthews proposed openstack/tripleo-common: TripleO API v1
openstackgerritDougal Matthews proposed openstack/tripleo-common: Create basic API layout
shardydtantsur: Hi, it looks like I do, but I've never done a TripleO release before11:16
shardymarios: Hey, are you familiar with the process?11:16
shardyI see some docs in the wiki but I'm not sure if they're current11:17
mariosshardy: hey, sec, reading back11:17
mariosshardy: yeah i haven't done a release in a long time i expect the wiki won't include that new stuff like tripleo-common11:18
mariosdtantsur: sure gimme few mins and i'll try do that for you11:19
shardythanks marios!11:19
*** mkovacik has joined #tripleo11:21
*** lucas-dinner is now known as lucasagomes11:28
mariosdtantsur: do you know of any reason not to just bump patch here? current is 1.4.0, prevsiou two are 1.3.0 and 1.2.011:32
dtantsurmarios, well, changes look like features.. but I don't mind11:32
mariosshardy: don't know if that is a dib convention... at least from my notes from the last release I did we always bumped patch, unless there was/is a known backwards compat thing11:32
mariosgreghaynes: there ^^^ about to cut dib release for dtantsur ... yo udid the last one @ 1.4.011:33
*** anande has quit IRC11:34
openstackgerritAna Krivokapic proposed openstack/tripleo-specs: Add spec for TripleO UI
*** Marga_ has quit IRC11:38
*** Marga_ has joined #tripleo11:38
mariosdtantsur: yeah so i see things like "Add dynamic-login element" as well as bugixes/docs etc so ... also going with convention gonna go with 1.5.011:38
mariosgreghaynes: ^^^^11:39
*** apetrich has quit IRC11:41
*** paramite|afk is now known as paramite11:46
*** anande has joined #tripleo11:48
*** paramite is now known as paramite|afk11:49
mariosdtantsur: so, sorry for delay, has been so long since release i have lost my pgp key i was using for this. i am just waiting for the new one to be available (may take 10 mins)11:49
*** paramite|afk is now known as paramite11:50
openstackgerritLucas Alvares Gomes proposed openstack/instack-undercloud: Set Nova's ram_allocation_ratio configuration option to "1.0" By default
gfidenteshardy, on the hostname/ip customization12:04
*** bvandenh has joined #tripleo12:04
mariosdtantsur: so is in progress now12:04
gfidenteshardy, I had also another question about merging together the pattern logic we currently have with anything provided from a map similar to ControllersIP12:04
gfidenteshardy, does that sound valid?12:04
mariosdtantsur: e70138a12:04
mariosdtantsur: will make its way to pypi and once that finishes12:05
mariosdtantsur: np sorry for delay. i only ever used that pgp signature for signing git tags and it didn't survive last os migrate it seems. delay was for the new key to showup in the keyserver. anyway, good excuse to have that updated :)12:06
mariosdtantsur: seems there12:08
shardygfidente: Yeah it could be, but I was keeping the hostname/IP/placement thing separate as folks may not want to use all of them12:08
dtantsurmarios, yeah, I see it, thanks a lot!12:08
gfidenteshardy, separate yes, in a different map12:08
*** paramite is now known as paramite|afk12:08
shardygfidente: athomas made the good observation that if we could make the ctlplance IP's predictable via Ironic/DHCP, we could use a DNS lookup to determine the hostname12:08
shardye.g assuming there is a local DNS that can resolve the provisioning network IP allocation12:09
gfidenteshardy, do a reverse lookup in heat to determine the name: param?12:09
shardygfidente: Yeah, well not in heat, just on the node12:09
gfidentebut the IP is a heat param12:09
shardyit doesn't look like cloud-init support that, but it'd be easy enough in a script12:09
shardygfidente: not for the provisioning network12:09
shardygfidente: your patch only fixes the IP's for the overlay networks, doesn't it?12:10
gfidenteshardy, the existing patch only works for overlay networks yes12:10
gfidenteshardy, but I was looking at it from user perspective12:10
gfidenteuser will probably want to paste the ctlplane IPs in a param as well12:10
shardylucasagomes: Is it possible to make Ironic deploy with predictable DHCP assignments per-node, vs just using a pool of IP's12:12
shardyI saw the external DHCP providers patches, but I wondered if you can just configure the undercloud DHCP with mapping of MAC's->IP's12:12
shardyI've done that before with hand-rolled DHCP/PXE setups12:12
*** paramite|afk is now known as paramite12:13
shardygfidente: I'm not sure it matters that much to operators where it's configured provided placement and IP assignement is predictable12:14
shardyI guess it'd be nice if it was all heat parameters, but it's probably not essential for the provisioning network12:14
lucasagomesshardy, hi there, I think it may be possible by using neutron + nova12:14
lucasagomesif you create a port in neutron with that IP and give it to nova when booting the instance12:14
lucasagomeslike --nic port_id=<id> (or something like that)12:15
shardylucasagomes: Ah, OK thanks!12:15
shardygfidente: So that is more like the fixed_ip's approach you suggested12:15
lucasagomesshardy, I didn't try that, but I don't see why it wouldn't be possible12:15
lucasagomeslemme know if it works please12:16
*** liverpooler has quit IRC12:16
shardylucasagomes: Ok, thanks - sounds like we need to do some testing12:16
*** liverpooler has joined #tripleo12:16
athomasThanks, lucasagomes. That's a great help.12:16
lucasagomesathomas, np12:17
athomasWe'll get someone to try that out. If it works, and we could support setting the hostname from the DNS entry for the IP, we could solve a bunch of problems.12:18
hewbroccavery cool12:18
gfidenteshardy, ack I'll trying the fixed_ips12:22
gfidenteif you don't get there first :P12:23
*** weshay_xchat has joined #tripleo12:25
openstackgerritDougal Matthews proposed openstack/tripleo-common: TripleO API v1
openstackgerritDougal Matthews proposed openstack/tripleo-common: Create basic API layout
*** thrash|g0ne is now known as thrash12:35
*** weshay_xchat is now known as weshay12:39
*** anande has quit IRC12:40
openstackgerritMarios Andreou proposed openstack/tripleo-heat-templates: Fixup default open_files_limit for galera
mariosmandre: ^^^ review please :)12:43
EmilienMgood morning12:43
*** apetrich has joined #tripleo12:45
*** mbound_ has quit IRC12:45
*** jaosorior has quit IRC12:47
*** openstackgerrit has quit IRC12:47
*** jaosorior has joined #tripleo12:47
*** openstackgerrit has joined #tripleo12:48
*** devvesa has quit IRC12:48
openstackgerritMarios Andreou proposed openstack/tripleo-heat-templates: Fixup default open_files_limit for galera
*** panda has joined #tripleo12:49
*** crinkle has quit IRC12:57
*** zaneb has quit IRC12:57
*** zaneb has joined #tripleo12:57
*** crinkle has joined #tripleo12:58
*** mgould has joined #tripleo13:05
*** masco has quit IRC13:08
jistrshardy: hi, should we branch stable/liberty for puppet-tripleo?13:14
gfidentejistr, yeah I was looking into that but don't have write permissions13:15
gfidenteslagle, ^^ >13:15
jistrOPM folks asked me about it last week too13:15
jistri think we talked about it on the last tripleo mtg13:15
jistrthat puppet-tripleo branching would be in upstream tripleo team scope rather than upstream puppet team scope13:16
*** mgrohar has joined #tripleo13:17
openstackgerritGiulio Fidente proposed openstack/puppet-tripleo: Allows customization of the HAProxy default timeouts
*** dprince has joined #tripleo13:21
openstackgerritDougal Matthews proposed openstack/tripleo-common: TripleO API v1
openstackgerritDougal Matthews proposed openstack/tripleo-common: Create basic API layout
shardyjistr: Hey, yes we should, I've been meaning to cut it but wanted to confirm the SHA, is latest master looking OK to branch from?13:23
*** trown|outttypeww is now known as trown13:24
shardytrown: ^^ discussing creating stable/liberty branch for puppet-tripleo and looking to confirm the best SHA to branch from13:25
trownshardy: I would guess latest puppet-tripleo is in the latest OPM built in delorean?13:26
trownthat is still a bit magic to me13:26
*** jprovazn has quit IRC13:26
shardytrown: Yeah I'm pretty sure stable CI is using lastest master, and last time I checked that was working OK13:26
shardyI just wanted to ensure you didn't have a downstream branch already we needed to align with13:26
jistrstable CI is using master of puppet-tripleo?13:26
*** jayg|g0n3 is now known as jayg13:27
shardyjistr: it must be, there are no other branches13:27
jistrwouldn't it be using master of OPM instead?13:27
*** mkovacik_ has joined #tripleo13:27
shardyhmm, let me check13:27
jistri'm not 100% sure, but i think the OPM rpm is built from here
trownshardy: for RDO I pretty much only care about delorean, and it promoted a repo yesterday so we should be good as long as puppet-tripleo is represented there13:27
slaglegfidente: what do you need write permissions for?13:28
*** tzumainn has joined #tripleo13:28
*** rbrady has joined #tripleo13:28
*** mkovacik has quit IRC13:29
shardyslagle: the tripleo-release group to cut a puppet-tripleo branch13:29
*** pradk has joined #tripleo13:29
shardyI'm on it now tho13:29
slagleoh ok13:29
shardycan add gfidente anyway :)13:29
gfidenteack thanks, useful next time13:29
shardyjistr: Yeah sorry you're right, I got confused by what goes into o-p-m13:30
jistrshadower: actually OPM even has a stable/liberty branch here
* jistr pinged social to join this channel13:31
shardyOk, but that will build from puppet-tripleo master if there's no stable branch right?13:31
shardyis there a delorean build log we can check to confirm the sha for the latest package?13:31
trownshardy: looking13:32
jaygjistr: I was just asking about this on #rdo yesterday too, it seems to me nobody really knows how opm gets there....13:32
jaygwe just built for downstream, so we have a new build with lots of fixes, probably another coming today or tomorrow13:32
jistrah ok13:32
jistri recall social pinged me a few days ago re "where's the stable/liberty of puppet-tripleo, should we just use master?" and i think he wanted that ref for the stable/liberty branch of OPM13:33
gfidenteI think this is the hash which OPM liberty uses of tripleo13:34
trownhmm... not sure how to see what actually goes into OPM... is the latest build in liberty delorean13:34
gfidenteand that is latest from master yes13:35
gfidenteso I think we're good branching from current master13:35
EmilienMshould I abandon ?13:35
shardyOk, thanks guys, I'll branch now from 74b589497c9efe400403f2d446aa6335abdb89d513:35
trowngfidente: shardy I do not think that commit has been built in delorean OPM13:36
jistrshardy: ack. I'm not sure if we're actually using the OPM from stable/liberty, but we probably should be, so +1 on branching puppet-tripleo from the ref which is in OPM stable/liberty.13:36
*** nico_auv has quit IRC13:38
jistrtrown: ok that's a good point too :) maybe we should find it somewhere first, like where stable/liberty OPM gets built, if anywhere13:38
trownhmm... I have no idea how OPM is being built actually... looks like from master-patches branch13:39
* shardy holds off branching until we figure it out13:39
jaygtrown: I am pretty sure the 7.0.2 tag we just built from for downstream yesterday is on stable/liberty13:39
EmilienMgfidente, jistr: I would like your thought on this old patch: -- should I abandon it? Look at the BZ in the patch topic13:39
trown is the latest commit in delorean liberty13:39
trownwhich is HEAD of master-patches branch on github13:40
gfidentetrown, eheh and in there the tripleo hash is different13:40
jaygsocial is on a meetingm, I'll remind him to join when done to help answer questions - but we are a little unclear on how it gets to delorean/rdo as well, think there has been some shuffling of who does what13:40
jayg is what you want for liberty13:41
jistri asked this on #rdo-puppet
jaosoriorany +A for this?
trownjayg: that would make sense... however that is not what is currently building in delorean13:43
tzumainndprince, hiya!  I addressed your comments for the API spec, let me know if you think there are still -2 worthy issues with it13:43
jaygtrown: ok, so if we can fix that setup, you should be good13:43
jaygnote I have no idea how to do so13:44
jistrtrown: so we're using master-patches branch of OPM in stable/liberty CI TripleO runs? perhaps we should branch puppet-tripleo from that then13:44
trownjayg: :) I will check rdoinfo13:44
dprincetzumainn: thanks, I'll have a look13:44
trownjistr: I would be +1 to that from RDO perspective, since that has passed RDO CI13:44
jaygjust to make it perfectly clear, whatever you guys recommend for puppet-tripleo hash to be in liberty we can switch to, if the hash in stable/liberty is incorrect, just let us know13:45
jistrtrown: right... and it also passed the TripleO CI13:45
*** derekh has joined #tripleo13:45
gfidentetrown, jistr I see but I'd follow jayg13:45
gfidenteif the hash in stable/liberty is wrong, let's update it13:46
gfidenteand possibly build OPM from stable/liberty for tripleo ci13:46
jistryeah, +1 let's start from what we already test13:46
jaygwe are going to be working on a few updates to go into another liberty build anyway, so now is the time to coordinate this13:46
jistrbecause updating a hash in OPM is easier than removing commits from a stable/liberty of puppet-tripleo via reverts or even force pushes13:47
gfidentejistr, ack13:47
jaygand if someone can tell me if we need to do something to help get delorean squared away later, we can do that as well13:47
openstackgerritMerged openstack/instack-undercloud: Remove Ceilometer Alarm from the undercloud
gfidentederekh, do you know where we configure in delorean which branch of OPM to use for the tripleo stable/liberty CI?13:47
jaygright, it would be great it opm can have a stable branch in puppet-tripleo to track13:48
trowngfidente: the issue is
gfidentetrown, that is the 'master' branch of rdoinfo13:51
gfidentetrown, don't we need a liberty branch of rdoinfo to point at the liberty branch of opm?13:51
trowngfidente: ya same branch is used for all deloreans, but with a different delorean config13:51
trowngfidente: we do if we have source-branch override in rdoinfo... OPM is the only project doing that13:52
gfidentetrown, ok can you change the source-branch in rdoinfo?13:52
gfidenteI will try to update opm stable/liberty to use same hash that master-patches is using13:53
trowngfidente: it is unfortunately not so simple... we do not use multiple branches of rdoinfo in delorean, so we would also have to change the delorean server to do that13:54
jayggfidente: are you meaning only for puppet-tripleo?13:54
derekhgfidente: that would be in rdoinfo, one sec13:54
gfidentejayg, only puppet-tripleo yes13:54
jayggfidente: k13:54
trownit would be much better if OPM could use master branch for master and remove that source-branch override from rdoinfo13:55
jaygtrown: I am not familiar with this, but the top of the file you linked says each package can have its own 'releases' section13:55
trownthen stable/liberty would just work13:55
* shardy is confused13:55
derekhgfidente: you probably need to ask apevec to be sure but I think it might be here
trownjayg: I am not too sure about the stuff at the top of that file :(13:56
shardyIt may just be easier to update TripleO CI to build the modules from source, for stable/liberty, like it does for master13:56
trownjayg: lots of legacy there13:56
jaygah, ok13:56
shardythat's on my todo list but not got to it yet13:56
beaglesmarios, thanks for your feedback on the QoS review!13:56
*** social has joined #tripleo13:57
jistrshardy: hmm but we should have it for dev setups too... this should be fixed on RDO delorean side, not on tripleo side13:57
socialjang: yere?13:57
beaglesmarios, I need a bit of a sounding board though for my next set update, do you have a second?13:57
jaygsocial: hello!13:57
*** openstackstatus has quit IRC13:57
*** links has quit IRC13:57
mariosbeagles: np, thanks for looking13:58
shardyjistr: Yeah, agreed, I'm just saying from the TripleO perspective from-source may be better13:58
jistrsocial: hi, just to update you on the latest findings, seems like liberty delorean is building from master-patches branch instead of stable/liberty, probably because of this
*** mgrohar has quit IRC13:58
jaygtrown | it would be much better if OPM could use master branch for master and remove that source-branch override from rdoinfo13:58
jistr^ yeah was just pasting that13:58
*** dmacpher has joined #tripleo13:58
socialjistr: that should be changed already by apevec13:58
jaygbut I am not yet quite sure how these branches are organized to say if that makes any sense for us13:58
*** jprovazn has joined #tripleo13:59
beaglesmarios, I find myself weighing two options - add the qos as default to N-AgentExtensions, etc and remove the NeutronEnableQoS flag or keep the flag and remove the N-AgentExtensions N-ExtensionDriver parameters ...13:59
socialjistr: because he patches build-opm.sh13:59
socialah I don't see any patch14:00
beaglesmarios, my thinking is this - if I keep the string lists and default them to QoS then I open the door for the unexpected if a deployer removes qos from the list, but leaves the NeutronEnableQoS as true...14:00
mariosbeagles: yeah i don't know. if you see it as something that will be/remain 'qos' for a while (e.g. cos new feature) then maybe just add the NeutronEnableQoS flag.14:00
beaglesmarios, if I want to keep the flag, I'm probably better to remove the N-ExtensionDriver and N-AgentLists14:00
openstackgerritDerek Higgins proposed openstack-infra/tripleo-ci: Switch to F22 jobs
jaygsocial: where is this
mariosbeagles: yeah so my primary concern was user facing interface with the params. but in general wrt sanity checking, that is aproblem, and not one we always can/do deal with14:01
jistrsocial: hmm sorry i don't know what's, but let me just follow up on what trown said as it sounded right to me, from what i can tell as a non-delorean person :) IIUC if we'd rename 'master-patches' to 'master', and 'master' to something else (in OPM repo), and remove that source-branch line from rdoinfo, then things could just work without further patching14:01
jistrtrown: ^ sounds correct?14:02
derekhTripleo jobs are now all F22 can somebody review ^^ to keep the status page looking at the correct thing................... >>
beaglesmarios, yeah.. wow. I forget how much your brain can twist when you have to think about human-facing-interfaces14:02
socialI don't see change done by apvec14:02
beaglesmarios, okay thanks, I'll keep it simple!14:02
mariosbeagles: yeah, i think we can make it configurable (the drivers/extensions) if/when that is necessary/requested i think14:03
*** julim has quit IRC14:03
beaglesmarios, right14:03
*** julim has joined #tripleo14:03
jistrsocial, trown: sounds to me like there's some convention on branch names, 'master' going to delorean master, 'stable/liberty' going to delorean liberty etc. so perhaps it would be easiest if OPM followed that convention14:04
* jayg has to walk the dogs, back in about 20 minutes14:04
trownjistr: that would be my vote14:04
socialjistr: we do follow that, opm in rhos8 is from stable/liberty liberty-patches14:05
*** rlandy has joined #tripleo14:05
openstackgerritDerek Higgins proposed openstack-infra/tripleo-ci: [NO MERGY] Test a update in trunk repository version
openstackgerritRyan Brady proposed openstack/tripleo-common: Ensures plan exists
*** julim has quit IRC14:08
*** morazi has joined #tripleo14:08
jistrsocial: hehe now i'm getting confused :) mostly about what's the difference between stable/liberty vs. liberty-patches, and other RDO packages also have these two branches14:09
socialI'll try to explain but I'm in meeting14:09
jistrsure thing, let's pause this for a bit14:10
*** paramite is now known as paramite|afk14:13
*** paramite|afk is now known as paramite14:15
socialjistr: 15:18 < apevec> social, on hold until testday(s) are over, please do not rebase master-patches this week!14:24
socialso yeah I'm going to bump opm master for you guys14:24
jistrlet's discuss this in person to make sure we're on the same page what the issue is :)14:25
EmilienMslagle: hey, you asked me to come back if no reviews -- could you help on ?14:27
*** devvesa has joined #tripleo14:28
*** jcoufal has quit IRC14:34
*** jcoufal has joined #tripleo14:34
jistralright this is way more complicated than i thought :D14:35
*** lucasagomes is now known as lucas-hungry14:35
gfidenteand if we wanted to take it as a black box14:35
gfidentewhat is that we need to know?14:35
jistr* the other projects should be following what OPM does for mitaka, so changing conventions on OPM at this point doesn't make sense14:36
socialjistr: trown: so opm, we have stable/liberty for first time only because RDO moved out of fedora. It would be f24 and f24-patches instead14:37
*** rhallisey has joined #tripleo14:37
*** rpothier has joined #tripleo14:37
socialjistr: trown: yet the branch without -patches is mere copy of upstream while -patches contains our patchset and is used by rdopkg for package building14:37
openstackgerritMiles Gould proposed openstack/python-tripleoclient: [WIP] Support new ironic "enroll" state
jistrand the change to make liberty delorean build liberty OPM is ready i think, but held off for now due to upcoming test day14:38
socialjistr: trown: It's easy now to hack it temporarly in delorean as opm in mitaka+ will change a lot switching to opm-ci which means it'll get bit more automated and will contain stable stable-tag and stable-patches branches14:38
socialwhere stable-tag will be the one to be used and tracked14:38
openstackgerritRyan Brady proposed openstack/tripleo-common: Ensures plan exists
openstackgerritBrent Eagles proposed openstack/tripleo-heat-templates: Enable setting Neutron QoS feature
*** liverpooler has quit IRC14:41
*** lblanchard has joined #tripleo14:44
*** nico_auv has joined #tripleo14:44
*** paramite is now known as paramite|afk14:45
gfidentejistr, so it'll be something like liberty-patches but it's not there yet?14:47
gfidentesocial ^^14:48
jistrso the gist of what we just discussed in person with social and paramite is: we're testing everything with puppet-tripleo ref which is in master-patches branch of OPM. liberty-patches of OPM is ahead of master-patches wrt puppet-tripleo ref, as it got branched just a few days ago from puppet-tripleo master, because there was no stable/liberty to branch from. What we want to do is un-bump puppet-tripleo in liberty-patches branch of OPM to the14:48
jistrversion we're testing now, so that when liberty delorean is fixed to build OPM from liberty-patches and not master-patches, we'd keep using the same version of puppet-tripleo14:48
jistrsounds ok?14:48
gfidentethat is great14:49
gfidenteso we need to change the hash in liberty-patches to match master-patches14:49
rhalliseyshardy, ping.  I need some wisdom14:50
gfidenteand we need to branch puppet-tripleo from the hash in master-patches14:50
jistrgfidente: yea14:50
gfidentejistr tks14:50
trownjistr: gfidente, makes sense to me. thanks!14:50
gfidenteshardy, so the branch hash is ae595ce7731f2b286d4ffd280ed735b61d86b63c14:50
jistrso we'd branch puppet-tripleo stable/liberty from ae595ce7731f2b286d4ffd280ed735b61d86b63c14:50
shardygfidente: thanks, done!14:53
derekhTripleo jobs are now all F22 can somebody review status page update to keep  it looking at the correct thing
shardyderekh: done14:54
*** julim has joined #tripleo14:54
*** Goneri has quit IRC14:54
derekhshardy: gfidente thanks, going to approve as CI does nothing on this file14:55
openstackgerritGiulio Fidente proposed openstack/puppet-tripleo: Set defaultbranch in .gitreview to stable/liberty
*** pradk has quit IRC14:57
*** pradk_ has joined #tripleo14:57
d0ugaltripleo_common has a bunch of exceptions in tripleo_common.core.exception. Does it make sense for the API to use these? Or should the API catch and re-raise an exception for each that controls the message and status code.14:57
d0ugalrbrady: ^ thoughts?14:57
*** pradk_ is now known as pradk14:58
rbradyd0ugal: If it uses the exceptions directly, we'd need to change them to add a status code and treat them more like HTTP exceptions, right?15:00
d0ugalrbrady: Yeah15:01
d0ugalrbrady: I guess that is a bit of an abuse of the exceptions15:01
rbradyd0ugal: If the CLI is going to use the API anyway, then I don't see an issue with changing the exceptions and just using them15:01
d0ugalrbrady: I added status code's to a couple here to test things out:,cm15:01
rbradyd0ugal: If the CLI or any other code will use the library directly, then there might be an issue with it15:01
d0ugalso only status code are needed15:01
d0ugalThe CLI currently uses bits, but not these exceptions and I understand we are migrating everything to using the API15:02
d0ugalso I think that is fine.15:02
rbradyd0ugal: okay...let's add the attributes needed.  seems like a small patch atm15:03
d0ugalrbrady: ok, thanks. I guess we can always split it out if it feels weird after doing a few more15:03
rbradyd0ugal: ack15:04
openstackgerritSteven Hardy proposed openstack/tripleo-heat-templates: Remove not needed completion-signal
*** mgould has quit IRC15:10
openstackgerritMarios Andreou proposed openstack/tripleo-heat-templates: Sample environment with old ServiceNetMap value
gfidentemarios, slagle ^^ you know that using internal_api other than being the default before is also the default for the external lb15:12
gfidentemarios, slagle and the only reason why we switched that to ctlplane was os-cloud-config ...15:12
openstackgerritMerged openstack-infra/tripleo-ci: Switch to F22 jobs
gfidentemarios, slagle so I think internal_api was actually a better default, except we had to cope with os-cloud-config15:13
slaglegfidente: i've argued for that and ctlplane was still chosen15:13
slagleimo, it doesnt even need a specific example15:13
slagleit's an internal api15:13
gfidenteand can we make internal_api the default for 8 though?15:13
slaglehmm, i wonder what network it should go on???? :)15:14
devvesadprince: can you ask you to do some reviews for the midonet integration with TripleO?15:18
devvesaThey are small15:18
openstackgerritLukas Bezdicka proposed openstack/puppet-tripleo: Set tunnel timeout for nova_novncproxy
openstackgerritLukas Bezdicka proposed openstack/puppet-tripleo: MidoNet services manifests
openstackgerritLukas Bezdicka proposed openstack/puppet-tripleo: Fix unit tests failing against Puppet 4.3.0
dprincedevvesa: I will try in a bit, link them?15:22
*** trown is now known as trown|brb15:23
dprincegfidente: once we use Puppet to configure keystone we can forget about os-cloud-config15:23
dprincegfidente: but that patch was reverted15:24
dprincegfidente: forget about os-cloud-config for keystone that is15:24
gfidentedprince, yeah was just speculating to set back the default to internal_api for mitaka15:24
dprincegfidente: please don't, until we get the Puppet keystone configuration landed15:24
gfidentedprince, sure, only after we get that15:24
gfidenteand IF we get that15:25
gfidenteactually, what is the status of that patch? anybody working on it?15:25
*** mgould has joined #tripleo15:25
dprincegfidente: I think bnemec posted it15:25
dprincegfidente: but is is blocked on "updates testing" or something15:25
gfidenteyeah I don't know the details of that though15:26
*** trown|brb is now known as trown15:27
shardydprince: It was proven to break things on update/upgrade, which is why it was reverted15:29
openstackgerritMerged openstack/python-tripleoclient: Increase test coverage for utils and plugin
dprinceshardy: yep, I approved the revert :)15:31
dprinceshardy: I'm just very keen to have it back because it solves some problems. Namely keystone v3 and gets us of os-cloud-config15:31
*** paramite|afk is now known as paramite15:32
shardydprince: Cool - IMHO nothing should be blocked on automated updates testing, we just can't merge anything risky without someone testing the process manually and proving the world doesn't break15:32
shardyobviously automating it is preferable in the long term15:32
dprinceshardy: sure, I spoke to slagle with regards to this yesterday as well. manual is fine for now15:33
dprinceshardy: anyways, it is blocked until someone does this I guess15:34
*** Goneri has joined #tripleo15:44
*** lucas-hungry is now known as lucasagomes15:51
*** paramite is now known as paramite|afk15:52
*** paramite|afk is now known as paramite15:55
openstackgerritDougal Matthews proposed openstack/tripleo-common: TripleO API v1
openstackgerritMarios Andreou proposed openstack/tripleo-heat-templates: Set default start/stop timeout for pcmk services to 95s
openstackgerritMarios Andreou proposed openstack/tripleo-heat-templates: Bump further the stop/start timeout for pcmk/systemd services
EmilienMif you have some time, please help me to have reviewed16:00
openstackgerritMarios Andreou proposed openstack/tripleo-heat-templates: Fixup default open_files_limit for galera
openstackgerritRonelle Landy proposed openstack/tripleo-common: Merge "Fix node group name when deleting a node"
openstackgerritMerged openstack/puppet-tripleo: loadbalancer: add Aodh API support
*** dtantsur is now known as dtantsur|brb16:08
openstackgerritMarios Andreou proposed openstack/tripleo-heat-templates: honor the rabbit user and password provided
*** apetrich has quit IRC16:14
*** pblaho has quit IRC16:14
openstackgerritDougal Matthews proposed openstack/tripleo-common: TripleO API v1
openstackgerritDougal Matthews proposed openstack/tripleo-common: Create basic API layout
openstackgerritJohn Trowbridge proposed openstack/tripleo-common: [stable/liberty]Fix node group name when deleting a node
*** dprince has quit IRC16:19
*** aufi has quit IRC16:19
*** tremble has quit IRC16:20
*** rohitpagedar__ has joined #tripleo16:20
*** dprince has joined #tripleo16:21
jpeelerd0ugal: is there anybody that can be poked to get merged?16:21
d0ugaljpeeler: We need CI to pass and we need another code to +216:23
jpeelerd0ugal: it won't pass CI until the delorean packaging is fixed though, which can't be merged before, right?16:23
d0ugaloh yeah16:23
d0ugalsorry, I forgot it was *this* one16:23
openstackgerritLennart Regebro proposed openstack/python-tripleoclient: More tests for overcloud deploy
d0ugalAny cores around with a few mins to take a look?16:24
openstackgerritDougal Matthews proposed openstack/tripleo-common: TripleO API v1
openstackgerritDougal Matthews proposed openstack/tripleo-common: Create basic API layout
jpeelerdprince: can you give a quick look at (my review is 3 dependencies deep currently)16:26
*** Marga_ has quit IRC16:27
openstackgerritDerek Higgins proposed openstack/tripleo-common: Don't install git-hg
derekhshardy: stable jobs are failing because of the switch to F22, fix is there ^^16:32
shardyderekh: kk, thanks16:41
dprincejpeeler: yes, in a bit. I've got to run out now... I've got it up now though16:41
*** dprince has quit IRC16:41
shardyderekh: +2, feel free to approve when it passes CI if I don't check back first16:42
*** rcernin has quit IRC16:42
openstackgerritEmilien Macchi proposed openstack/tripleo-heat-templates: Manage keystone initialization directly in t-h-t manifests
openstackgerritEmilien Macchi proposed openstack/tripleo-heat-templates: WIP - Deploy Aodh services, replacing Ceilometer Alarm
rhalliseyderekh, ya that's why the containers job was failing :(16:45
rhalliseyIt's run on f2216:45
derekhrhallisey: for liberty yes probably, it should be ok for master as that patch was already in.16:49
*** paramite is now known as paramite|afk16:50
*** paramite|afk is now known as paramite16:52
*** david-lyle has quit IRC16:53
*** dtantsur|brb is now known as dtantsur17:01
*** dprince has joined #tripleo17:02
*** ibravo has joined #tripleo17:04
*** yamahata has joined #tripleo17:04
*** jprovazn has quit IRC17:05
openstackgerritJiri Stransky proposed openstack/tripleo-heat-templates: Pacemaker maintenance mode for the duration of Puppet run on update
openstackgerritDerek Higgins proposed openstack/instack-undercloud: Change undercloud replicat count to 1
openstackgerritDerek Higgins proposed openstack-infra/tripleo-ci: [NO MERGY] Test a update in trunk repository version
*** ayoung has joined #tripleo17:15
ayoungjaosorior, so, on,cm  do we have a catch 22?  We need to affect change in both puppet and triple-heat at the same time, or can we do this all in the heat template?17:17
openstackgerritBen Swartzlander proposed openstack/tripleo-heat-templates: Enable Manila integration
*** dprince has quit IRC17:22
*** alop has joined #tripleo17:24
openstackgerritBen Swartzlander proposed openstack/tripleo-heat-templates: Enable Manila integration
openstackgerritBen Swartzlander proposed openstack/tripleo-heat-templates: Add NetApp integration to Manila
openstackgerritBen Swartzlander proposed openstack/tripleo-heat-templates: Add Gluster integration to Manila
rhalliseyshardy, every so often my deployment hangs here:
openstackgerritGiulio Fidente proposed openstack/tripleo-heat-templates: WIP: Allow predictable IPs for Controllers on the ctlplane
openstackgerritgreghaynes proposed openstack/diskimage-builder: Allow grub2 to build with opensuse
rhalliseyany idea why?  Heat is constantly signaling, but never passing a create complete when a `heat stack-list` shows a create-complete17:30
shardyrhallisey: I'm confused, if heat stack-list says CREATE_COMPLETE, it is complete?17:32
rhalliseyshardy, ya17:32
rhalliseyexcept it never signals back17:32
*** richm has joined #tripleo17:33
rhalliseyit just hangs there17:33
ayoungrichm, cool, you are here17:33
shardyOh, you mean the event polling in tripleoclient?17:33
shardyrhallisey: If heat stack-list show CREATE_COMPLETE, tripleoclient should complete and exit17:33
*** mkovacik_ has quit IRC17:33
shardyI suspect it's more a client issue, but it's hard to say what17:34
rhalliseyshardy, ok will look there. Thanks17:34
ayoungrichm, trying to get a handle on what the puppet manifest will do if we change the heat template.  jaosorior has a comment, but I am not clear on what it means17:34
shardyrhallisey: heat doesn't signal anything on completion, tripleoclient is polling the heat event API17:34
shardymaybe that is what's getting stuck17:34
rhalliseyshardy, oh that's what I see in the logs then17:34
shardyrhallisey: Yeah, if you run tripleoclient with --debug it'll show each call to heat17:34
shardyhopefully you'll see where it's getting hung up then17:35
ayoungrichm, "That is because httpd will be managed by pacemaker instead." is that as significant a statement as I think it is, or does that just mean that heat is trying to do too much, and we can remove things17:35
*** Marga_ has joined #tripleo17:35
rhalliseyshardy, cool thx17:35
richmI don't know how pacemaker works - is it ok to restart httpd "out from under" pacemaker?17:36
*** dmacpher is now known as dmacpher-afk17:36
*** athomas has quit IRC17:37
gfidenterichm, nope it isn't17:37
*** ukalifon has quit IRC17:37
*** openstackstatus has joined #tripleo17:38
*** ChanServ sets mode: +v openstackstatus17:38
gfidenterichm, it might work on occasion, but it might trigger bad events in others17:38
richmThen I'm not sure how puppet is supposed to work in this case - you modify a config that requires a service restart to take effect - what service is puppet supposed to know to notify?17:38
richm~> Service['some service provided by caller']17:39
richmtell puppet that service management is outside of puppet, then17:39
richm~> Service['/dev/null']17:40
gfidenterichm, we used to pass something like manage_service false to the puppet module so that it won't trigger restart when notified17:41
*** dprince has joined #tripleo17:41
*** weshay has quit IRC17:41
richmgfidente: then how does pacemaker know that it needs to restart httpd if some config changes, if it is not getting those puppet notifies?17:42
gfidenterichm, currently the sad story is that we can't do this on a per-service basis and tell pcmk to restart the services17:44
richmgfidente: so do you just tell pacemaker to restart httpd unconditionally?17:45
*** masco has joined #tripleo17:45
richmevery time puppet runs?17:45
gfidenterichm, yes unfortunately that is the effect17:45
richmgfidente: ayoung: Then this is a bug (or rfe) for puppet-keystone - need to be able to disable all ~> and notifies for Service['httpd'] and/or Service['keystone']17:46
ayoungFcuking Hell.17:46
ayoungCan we please not make things impossible?17:46
*** weshay_xchat has joined #tripleo17:46
ayoungPretty please?17:47
*** weshay_xchat is now known as weshay17:47
*** paramite has quit IRC17:48
richmI don't know if it is possible to work around this - maybe there is some puppet hack whereby you can define/alias/redirect notifies to Service['httpd'] to a no-op17:48
ayoungrichm, so let me get this straight.  I took us a couple years to finally get HTTPD support into puppet-keystone, and, now that we have it, we can't use it because of Pacemaker.  I assume this change is a small change?17:48
richmwill need a puppet guru for that17:48
ayoungPuppet needs to die17:48
openstackgerritEmilien Macchi proposed openstack/tripleo-heat-templates: WIP - Deploy Aodh services, replacing Ceilometer Alarm
gfidenteI think if you pass service_manage false to ::apache the -> won't trigger a restart17:49
* ayoung gonna go rewrite everything in ansible.17:49
*** trown is now known as trown|lunch17:49
richmayoung: a) right - can't use it because pacemaker b) it is a small change c) puppet cannot die17:49
ayoungI know, you shoot the puppet, all you do is injure the puppeteer17:49
richmgfidente: I don't think that will suffice -
richmgfidente: afaict, keystone::wsgi::apache needs to be fixed17:50
EmilienMrichm: for what?17:50
pino|work(don't kill puppet, think about the jistr involved)17:50
richmEmilienM: Then this is a bug (or rfe) for puppet-keystone - need to be able to disable all ~> and notifies for Service['httpd'] and/or Service['keystone']17:50
EmilienMrichm: oh I can imagine that's because of pacemaker?17:51
richmEmilienM: correct17:51
jistri think what we really need is a faux service provider for services managed by pacemaker17:51
EmilienMI agree, we need to fix it17:51
jistrwhich wouldn't react on restarts17:51
richmjrist: +117:51
EmilienMjistr: yeah17:51
ayoungcould we tell puppet somehow that Keystone == pacemaker?17:52
richmayoung: not exactly17:52
ayoungSo the puppet modules wouldn't know what was implementing it17:52
ayoungjust defere to a later definition of Keystone?17:52
richmayoung: gfidente says that you cannot notify pacemaker via a puppet notify17:52
gfidentejistr, faux service provider ...17:52
richmayoung: what jrist is proposing is better17:52
jistrrichm: btw jrist vs jistr :))17:53
jristnothing I'm not used to17:53
ayoungjistr, I learned the diff between you two the hard way17:53
jistryeah me too :)17:53
jistrayoung: everybody does it seems :)17:53
ayoungjistr should be read as Jester17:53
richmjistr: I think you meant gfidente17:53
ayoungjrist should be read as Jay Wrist17:54
jristfor the record I'm older than jistr17:54
jristso I think I came first?17:54
ayoungFor the record I'm older than all of you excpet maybe for richm, and even that is close17:55
jristright but we're not comparing your name to mine :)17:55
richmjistr: sorry, yes, I meant jistr17:55
ayoungjrist, when I first stared on openstack, everyone thought I as Andrew Young.  But his irc was "sleepsonthefloor"17:55
devvesadprince: please remember reviews.17:56
gfidenteso if the keystone module was notifying class[::apache:service] instead of service[keystone] it should work?17:56
devvesaMy managers are pushing me to have the changes merged by tuesday...17:57
ayounggfidente, I thought that was what I was doing, but apparently not17:57
richmgfidente: Does notifying class[::apache:service] restart httpd?  If not, what exactly does it do?17:57
*** devvesa has quit IRC17:57
jistrrichm, ayoung: btw we don't want to necessarily notify pacemaker from puppet anyway, as some config changes can mean a two-way dependency between service restarts. They need to be restarted at the same time, or things break.17:57
*** athomas has joined #tripleo17:58
gfidente richm it'd let the class decide, based on the service_manage thing17:58
ayoungjistr, agreed.  And this is all initial configuration stuff.  We should not start/stop/restart 1000000 times17:58
gfidenteand pcmk is pretty good in maintaining that relationship amongst the services17:59
gfidentelet's give it what has to be given17:59
ayoungThis is why I really don't like the all-in-one approach.  While Keystone should be running in apache, it really should be able to vary separately from the resot of openstack.  Should be in its own container17:59
ayoungwould make upgrades much easier17:59
ayoungand allow people to run later keystones than the rest of openstack...17:59
*** ifarkas has quit IRC18:00
gfidenteayoung, so we have also other issues with restarting httpd18:00
gfidenteit's going to affect other services, poentially more and more while the apis all move to wsgi18:00
ayounggfidente, what is in Apache besides Keystone and Horizon?18:00
gfidenteand I think we wanted to explore the possibility of running the wsgis in some lightweight server which we would spawn in multiple copies anf configurations, depending on service18:01
jistrayoung: currently nothing i think, but everybody plans to move there, at least that's what the rumor is :)18:01
ayoungCan HTTPD be told to re-read config without restarting?18:02
ayoungSeems to me that is the heart of the problem18:02
gfidenteI think running an isolated server also works better with containers18:03
ayoungaccording to this yes
gfidentebut on the other hand, -HUP is a good interim solution18:03
gfidenteayoung, for now, I would just try noticing ::apache::service18:04
ayounggfidente, here?
gfidentejistr, ^^ ?18:05
richmwill that work both with and without pacemaker?18:05
*** shardy has quit IRC18:05
jistri don't think this is guaranteed to work with pacemaker18:05
jistrwhat are we trying to achieve anyway? i don't think i noticed it in the conversation, sorry18:06
gfidenterichm, yes in non-pacemaker service_manage is true18:06
ayoungjistr, run Keystone in HTTPD instead of eventlet18:06
richmis someone going to file a bug against puppet-keystone?18:06
ayoung  We are coing up on 4 years, so I figured it was time18:07
ayoungrichm, I can do that18:07
richmayoung: ok - please add me to cc or just send me a link18:07
ayoungrichm, can you take a tuick look at  and se if it is actually covered under one of the exsitng?18:08
ayoung"when WSGI enabled: domain service is unavailable during first Puppet run"  almost seems to  be it18:08
*** dtantsur is now known as dtantsur|afk18:08
jistrgfidente: i don't know what notifying ::apache::service does, not *that* much of a puppeteer. It might do the service restart.18:08
jistrbut i don't think any of this is going to play well with pacemaker18:09
jistrwe can just prevent it from freaking out if we put it into maintenance mode, which we do for stack-updates18:09
gfidentejistr, I think I've seen it working for rabbitmq module18:11
richmayoung: no, that bug is the "set default domain bootstrap problem"18:12
richmayoung: looks like this problem isn't covered by an existing bug18:12
ayoungrichm, the whole "we don't know the id until after we create it" cycle?18:12
richmayoung: yes18:13
openstackLaunchpad bug 1524927 in puppet-keystone "Puppet keystone cannot restart HTTPD managed by Pacemaker" [Undecided,New]18:13
ayoungI used jaosorior 's comment to populate18:13
richmayoung: looks good - thanks!18:13
ayoungrichm, I have a git checkout of the tripleo heat templates that I was running with.  I'll roll back to master there, and then, I want to try with that change to the puppe modules18:15
ayoungI assume I can clone them to the undercloud stack users dir and run overcloud deploy pointing at them somehow?18:15
openstackgerritRyan Hallisey proposed openstack/tripleo-heat-templates: Convert JSON generations from bash to python
richmayoung: If you're talking to me, I have no idea18:17
ayoungrichm, you were not testing them via tripleo deploys, I take it then18:18
richmayoung: I've never even touched tripleo - any puppet-keystone testing I did was with spec or packstack18:20
*** dsneddon has quit IRC18:20
ayoungrichm, OK...I'll figure this out.  I know jistr has made something like this work18:21
patrickeastHi, maybe a simple question... I'm trying to add in a new package to my nodes so I've added in a definition like package {  'device-mapper-multipath': ensure => installed}" to the compute.pp manifest thats in tripleo-heat-templates. I'm seeing in the log that puppet is picking up the package
patrickeastany reason why the package then would not actually be installed?18:21
dprincejpeeler, trown|lunch: this looks fine, but have we added the actual RDO package update yet?
dprincethrash: ^^18:22
jpeelerdprince: my understanding is the delorean change had to happen afterwards -
jistrayoung, richm: openstack overcloud deploy --templates <path-to-templates>18:23
ayoungjistr, for puppet modules18:23
ayoungjistr, I need to change
jistrohh, i've had that working some time ago via rsync but that was in devtest times, and i don't think i got it to work after moving to instack, for some reason18:24
*** trown|lunch is now known as trown18:24
jistri've heard someone saves them to swift and fetches them (instead of rsync), but i haven't seen the code/snippets to make this happen yet18:24
*** dsneddon has joined #tripleo18:25
dprincejpeeler, thrash, trown okay, I commented on I think we could probably handle this in the .spec file to get CI passing. I've done similar things in the tripleo-heat-templates packaging before...18:25
trowndprince: the packaging update is waiting on that change you just reviewed :P18:25
jistrayoung: also you can use virt-customize or virt-edit to edit the image, then `openstack overcloud image upload --update-existing`18:25
jistrbut that can get heavy handed if you do a lot of changes18:26
jistrsorry to not have a good workflow here ATM18:26
*** dsneddon has quit IRC18:26
dprincetrown: ideally I think we'd do it the other way around. In other words make the packaging change work with both the old and new upstream code.18:26
*** dsneddon has joined #tripleo18:26
dprincetrown: a bit odd in the packaging world, yes but I think it is fine. Plus we can always drop any extra code afterwards.18:26
jistrdprince: did you do deploy-time OPM updates via swift?18:26
trowndprince: I think for the /* idea we should get opinions on #rdo18:27
ayoungjistr, hmmmm18:27
*** dsneddon has quit IRC18:27
trowndprince: merging that stub in tripleo-common with failing CI was meant to be the workaround18:27
ayoungjistr, when building the overcload images, we use the puppet modules as they are installed on the hypervisor?18:28
*** dsneddon has joined #tripleo18:28
dprincetrown: right, I'd rather see CI passing first though18:28
trowndprince: the real solution here is to have packaging in the same gerrit as the code so we can do depends-on ...18:28
gfidenteayoung, I think that alternatively you can use something like virt-customize to edit the overcloud image, reupload in undercloud glance, and reattempt deployment18:28
dprincetrown: I think we can do this18:28
dprincetrown: exactly18:28
dprincetrown: that is the real solution18:28
dprincetrown: without it we have to do package munging18:29
dprincetrown: which I've been doing in t-h-t here:
jistrayoung: hmm i don't know for sure but i don't think so. I think it's going to ensure that it's up to date with the repo location and ref, usually master branch.18:29
trowndprince: but I am not super opinionated on which less ideal solution to go with18:29
trowndprince: I am not even the maintainer of tripleo-common so even less opinionated :)18:30
* jistr in conversations for longer than intended, it's 7:30 PM, probably should go home :))18:30
dprincetrown: given this only affects tripleo* packages I think it is fine to treat the packages this way for now18:30
ayoungjistr, this call right here
ayoung/home/stack/tripleo-common/scripts/ --overcloud-images18:30
dprincejistr: I need more context. Swift updates for puppet?18:30
ayoungjistr, where does that get executed18:31
jistrdprince: yeah update a puppet module (puppet-keystone for ayoung / richm use case) on deploy-time, without rebuilding the image18:31
jistrayoung: on undercloud18:31
dprincejistr: oh, yeah. That is all done18:31
dprincejistr: reviews are posted, just stalled out in the review process now18:32
thrashdprince: i did add a .spec review18:32
dprincethrash: right, I commented on it18:33
thrashdprince: thanks18:33
jistri see18:34
dprincejistr: the mechanism would now support generic deployment of artifacts18:34
dprincejistr: so either a tarball (my favorite) or an RPM if you prefer18:34
jistrtarball ftw for dev work18:34
dprincejistr: exactly18:35
jistri used to use this
dprincejistr: this is super useful to me for composable roles work BTW (moving things out of t-h-t --> puppet-tripleo)18:35
dprincejistr: upload new puppet-modules tarball (contains etc/puppet/modules/*)18:35
dprincejistr: then heat stack-update18:36
jistryeah the firstboot only worked for stack-create iirc18:36
jistrok, i think i'll explore the swift way, would be good to converge on a single way to do these things18:36
dprincejistr: yep, check it out and see what you think18:37
jistrayoung, richm ^^18:37
* jistr leaving for real now :)18:37
dprincejistr: see ya18:38
*** jistr has quit IRC18:38
*** nico_auv has quit IRC18:38
*** olap has quit IRC18:38
ayoungjistr, dprince we have swift in the undercloud, right ?18:39
* ayoung neophyte18:39
dprinceayoung: yes, and we currently use it for very little18:39
dprinceayoung: introspection data goes there and that is it really18:40
*** masco has quit IRC18:40
dprinceayoung: there are however patches posted to make Glance use it as a backend, which would potentially allow image data to go directly from Swift -> IPA (our deployment ramdisk) skipping the conductor conversion bit in Ironic18:41
jpeelerthrash: you leaning towards putting that wildcard in?18:44
*** shivrao_ has joined #tripleo18:44
thrashjpeeler: I'm currently debating that with myself. :)18:44
thrashor trying to figure out how the conditional stuff would work.18:44
jpeeleri say do it!18:44
thrashjpeeler: the wildcard?18:45
dprinceayoung: and then the new idea of using it to deploy tarballs/RPMs is now posted as well18:45
jpeelerthrash: easy for a packager to remove it and explicitly list files if something needs to not go in the RPM. but chances are that won't happen in my opinion18:45
thrashjpeeler: agreed.18:46
ayoungdprince, I get it, but I think it does not help me trying to do it from git. Just adds another place I need to stick them18:46
*** Marga_ has quit IRC18:46
*** Marga_ has joined #tripleo18:47
openstackgerritMerged openstack/tripleo-common: Ensures plan exists
ayoungdprince, It seems to me "A Swift Temp URL"  could be replaced with a git repo and a tag, hash, or branch.18:48
ayoungdprince, can I git push to Swift?  I'm only 1/3 joking18:49
dprinceayoung: yeah, so if we'd stuck with the initial interface I had then yes. Initially the Heat parameter was called PuppetModulesURL18:50
dprinceayoung: and the deployment script would know where to deploy said module, be it a git URL, or a swift temp URL tarball18:50
*** apetrich has joined #tripleo18:50
dprinceayoung: but now (per comments in the spec) the mechanism is "generic". We *can* deploy RPMs now though18:50
dprinceayoung: so it is a trade off18:50
dprinceayoung: the difficulty with the generic ArtifactsURL deployment is you don't know what they are, or where to install them if it is a git repo18:51
dprinceayoung: so the question here is really do we specialize our Heat parameters for Puppet modules? or keep them generic. Most want them generic I think18:52
jpeelerthrash: unforunately, it still fails CI. forgot that it doesn't solve the chicken egg problem18:54
ayoungdprince, I'm still wrpping my head around what the mechanism flow is here.  I can't quite figure out where jistr's Ansible playbook is generating the images.  I assume that the scriopt it calls to do that is then getting the puppet modules from somewhere, but I don't want to guess where18:54
openstackgerritDougal Matthews proposed openstack/tripleo-common: TripleO API v1
openstackgerritDougal Matthews proposed openstack/tripleo-common: Add environment capabilities
openstackgerritDougal Matthews proposed openstack/tripleo-common: Create basic API layout
*** shivrao_ has quit IRC18:54
jpeelerd0ugal: would you now be okay with approving ?18:56
ayoung openstack overcloud image build --all18:56
*** jcoufal has quit IRC18:58
*** dprince has quit IRC18:58
*** dprince has joined #tripleo18:59
thrashjpeeler: thought so... It's complaining that nothing in in %{_bindir}19:00
rhalliseyhas only ever run into an issue with the triplo client polling heat forever even after heat is 'CREATE_COMPLETE'?19:01
trownthrash: dprince, I think we should just merge the stub as originally planned... In the extremely unlikely chance it breaks CI we can revert19:02
*** shivrao has joined #tripleo19:03
*** tosky has quit IRC19:03
jpeelertrown: the stub actually has two +2s19:03
thrashtrown: I think leaving the %{_bindir}/* should be fine though19:05
thrashonce we have *one* script in there, it alleviates future chicken-egg problems.19:05
trownthrash: ya I do not have a problem with that19:05
jpeelerso who can approve the stub patch?19:06
thrashd0ugal: ^^^?19:06
thrashtheoretically I can +A it, but I hate doing that on my own patch.19:07
jpeeleri'm imagine d0ugal is done for the day if he's still in the UK19:07
jpeelerreally hoping we can get this sorted like... nowish19:07
*** ukalifon1 has joined #tripleo19:09
trownjpeeler: thrash it would probably be good for dprince to +A if he agrees... he was the one with objections to that approach19:10
thrashtrown: ack19:10
dprincetrown: link me the patch again?19:10
dprincetrown: commented19:12
*** bvandenh has quit IRC19:13
jpeelerthrash: ok i guess you're stuck figuring out the conditional.19:13
trowndprince: the * trick does not work though with 0 files19:14
trownit works with any number of files >019:14
*** Marga_ has quit IRC19:15
*** mgould has quit IRC19:15
*** ukalifon1 has quit IRC19:16
openstackgerritDan Sneddon proposed openstack/tripleo-heat-templates: Add IPv6 Support to Isolated Networks
*** gfidente has quit IRC19:21
*** Marga_ has joined #tripleo19:25
dprincetrown: can we put a conditional around it then?19:26
*** ukalifon has joined #tripleo19:26
thrashjpeeler: dprince trown have a possibility on the conditional19:26
thrashI don't have a working delorean setup locally, so I'll push it up and see what it does. :)19:26
thrashit rpmlint's fine. :)19:26
*** bvandenh has joined #tripleo19:27
*** shivrao has quit IRC19:32
*** shivrao_ has joined #tripleo19:32
*** olap has joined #tripleo19:32
thrashdprince: trown jpeeler passed19:33
jpeeleryes, can trown approve it?19:33
trownthrash: thanks I will +A19:33
jpeelerone step closer!19:33
thrashtrown: theoretically, we can drop the 'add only' commit.19:34
thrashaka collapse the two.19:34
thrashtrown: should I collapse them?19:34
trownthrash: ya now we should be able to... I would like to remove that conditional once this all resolved as well... that is a bit ugly19:34
thrashtrown: agreed, but i'd rather have the ugly be in the packaging than in the upstream repo. :)19:35
thrashlet me collapse them.19:35
jpeeleranother rebase for me then19:35
trownthrash: ya as a temporary hack to get past chicken egg problem I have no issue with it19:36
openstackgerritEmilien Macchi proposed openstack/instack-undercloud: Deploy Monitoring on the undercloud with Puppet
openstackgerritEmilien Macchi proposed openstack/instack-undercloud: Manage keystone related configuration via Puppet
openstackgerritBrad P. Crochet proposed openstack/tripleo-common: Use the new yaml syntax in CI
openstackgerritBrad P. Crochet proposed openstack/tripleo-common: Build image files from definitions in yaml
thrashjpeeler: base your change on this:
thrashnot the CI one.19:39
thrashtrown: jpeeler squashed and rebased.. Let's see what happens. :)19:39
openstackgerritgreghaynes proposed openstack/diskimage-builder: Add dib-python element
*** electrofelix has quit IRC19:39
jpeelerthrash: thanks, yeah hopefully all is well and these patches get merged soon too19:40
thrashjpeeler: hold off on your rebase.19:40
thrashjpeeler: got a requirement mismatch.19:40
*** ibravo has quit IRC19:42
thrashgonna ignore that for now.19:44
jpeelerwhy not just go ahead and bump them19:45
openstackgerritDerek Higgins proposed openstack/instack-undercloud: Change undercloud replica count to 1
openstackgerritBrad P. Crochet proposed openstack/tripleo-common: Use the new yaml syntax in CI
openstackgerritBrad P. Crochet proposed openstack/tripleo-common: Build image files from definitions in yaml
jpeeleri mean, the job will restart immediately. so you won't lose that much testing time19:47
jpeelerthrash: hopefully my involvement has not been annoying on all this19:48
thrashjpeeler: not at all. :)19:48
*** Marga_ has quit IRC19:50
*** rcernin has joined #tripleo20:00
openstackgerritgreghaynes proposed openstack/diskimage-builder: Run package-intalls with py3k if we must
greghaynesSpamapS: pino|work ^ coded up what we chatted about20:03
openstackgerritJeff Peeler proposed openstack/tripleo-common: Upload docker image files from yaml config file
thrashjpeeler: trown will check on status when I get home.20:09
jpeelerthrash: appreciate it - i have high hopes though20:09
*** shivrao_ has quit IRC20:12
openstackgerritgreghaynes proposed openstack/diskimage-builder: Add a new element hpdsa
*** Marga_ has joined #tripleo20:12
*** shivrao has joined #tripleo20:14
*** lucasagomes is now known as lucas-beer20:19
*** ibravo has joined #tripleo20:26
*** bvandenh has quit IRC20:28
*** sthillma has joined #tripleo20:29
*** athomas has quit IRC20:32
*** lucas-beer has quit IRC20:34
*** dprince has quit IRC20:37
*** lucasagomes has joined #tripleo20:40
*** ibravo has quit IRC20:40
*** mcornea has quit IRC20:42
*** sthillma has quit IRC20:44
*** Hazelesque_ has joined #tripleo20:47
*** Hazelesque has quit IRC20:47
*** jaosorior has quit IRC20:47
*** jaosorior has joined #tripleo20:47
*** cwolferh has quit IRC20:53
*** sthillma has joined #tripleo20:58
*** akrivoka has quit IRC21:04
*** rebrego has joined #tripleo21:05
*** regebro has quit IRC21:05
*** rebrego is now known as regebro21:05
openstackgerritgreghaynes proposed openstack/diskimage-builder: Run package-intalls with py3k if we must
trownderekh: I would assume all of the tripleo jobs would fail on the undercloud install issue you found if they were not all failing on 'No package git-hg available' during delorean setup21:06
trownderekh: I thought we fixed that a while ago...21:06
*** Hazelesque_ has quit IRC21:10
*** cwolferh has joined #tripleo21:10
*** ukalifon has quit IRC21:13
*** Hazelesque has joined #tripleo21:13
*** julim has quit IRC21:14
*** sthillma has quit IRC21:17
openstackgerritEmilien Macchi proposed openstack/tripleo-heat-templates: pacemaker: run neutron-server-start-wait-stop only at step 4
openstackgerritMerged openstack/diskimage-builder: Add proliant-tools element
openstackgerritEmilien Macchi proposed openstack/tripleo-heat-templates: pacemaker: run neutron-server-start-wait-stop only at step 4
openstackgerritMerged openstack/diskimage-builder: Follow up patch for 25d3ee547176528e86d42eb026c99a134dff9452
openstackgerritDan Sneddon proposed openstack/tripleo-heat-templates: Add IPv6 Support to Isolated Networks
*** rpothier has quit IRC21:37
openstackgerritMerged openstack/diskimage-builder: Add a new element hpdsa
*** jayg is now known as jayg|g0n321:57
*** mcornea has joined #tripleo22:01
*** alop has quit IRC22:03
openstackgerritRyan Brady proposed openstack/tripleo-common: Fixes enabled bug
*** lblanchard has quit IRC22:05
*** derekh has quit IRC22:07
*** shivrao has quit IRC22:09
*** shivrao has joined #tripleo22:10
*** shakamunyi has quit IRC22:12
*** penick has joined #tripleo22:12
*** dsneddon has quit IRC22:12
*** mcornea has quit IRC22:13
*** trown is now known as trown|outttypeww22:14
*** shakamunyi has joined #tripleo22:15
*** dsneddon has joined #tripleo22:16
*** dsneddon has quit IRC22:18
*** dsneddon has joined #tripleo22:18
*** david-lyle has joined #tripleo22:20
*** derekh has joined #tripleo22:24
*** yamahata has quit IRC22:24
*** yamahata has joined #tripleo22:24
*** Goneri has quit IRC22:32
*** thrash is now known as thrash|g0ne22:38
*** pradk has quit IRC22:40
openstackgerritMerged openstack/tripleo-common: Don't install git-hg
*** rcernin has quit IRC22:45
*** alop has joined #tripleo22:51
*** david-lyle has quit IRC22:55
openstackgerritRyan Hallisey proposed openstack/tripleo-heat-templates: Remove hack the pulls latest docker
*** openstackstatus has quit IRC23:01
*** rhallisey has quit IRC23:12
*** alop has quit IRC23:12
*** apetrich has quit IRC23:16
*** yuanying has joined #tripleo23:18
*** morazi has quit IRC23:26
*** derekh has quit IRC23:35
openstackgerritDerek Higgins proposed openstack/instack-undercloud: Change undercloud replica count to 1
*** weshay has quit IRC23:40
openstackgerritDerek Higgins proposed openstack-infra/tripleo-ci: [NO MERGY] Test a update in trunk repository version
*** sthillma has joined #tripleo23:49

Generated by 2.14.0 by Marius Gedminas - find it at!