Friday, 2016-05-20

*** r-mibu has quit IRC00:20
*** r-mibu has joined #tripleo00:25
*** mbound has joined #tripleo00:30
*** rbrady has quit IRC00:32
*** mbound has quit IRC00:35
*** khrystoph has joined #tripleo00:50
*** khrystoph has quit IRC00:54
*** dixiaoli has joined #tripleo01:04
*** TSCHAK has quit IRC01:07
larsksayoung: I don't know why you are going through all the rigamarole setting up an OVS bridge on your guest.  It doesn't make any sense from what I understand of your use case.01:18
larsksIn particular, this part:
*** fzdarsky_ has joined #tripleo01:24
*** penick has quit IRC01:24
*** fzdarsky|afk has quit IRC01:28
*** tzumainn has quit IRC01:49
ayounglarsks, does network isolation mean exactly the opposite of what it seems to mean?01:50
larsksayoung: probably not?  It means you get multiple overcloud networks, rather than having everything running over the ctlplane network.01:51
ayounglarsks, is that all the vlans?  Is network isolation the default?01:52
larsksIt is not all vlans, and it is the default using the quickstart (because it is set to true in playbooks/centosci/minimal.yml).01:53
larsksI don't know if it's the default to tripleo in general, although I am sure it's recommended.01:53
larsksBut I'm not sure it matters if your goal is to get a host connected to the overcloud and external networks.01:54
larsksThat's going to look the same regardless of the network isolation setting.01:55
ayounglarsks, just making sure I know the lay of the land.  So the step that you have calling virsh to launch the VM seems to be comparable to the call here:
ayoungthe xml from the previous step shows the two bridges01:56
larsksRight.  The virt-install command I use will result in the same network configuration.01:57
larskseth0 connected to brext, eth1 connected to brovc.01:57
ayounglarsks, but there was nothing inside the VM until I did all the ovs rigamarole01:57
larsksYour vm had an eth0 and an eth1.01:58
larsksThat's all you need.01:58
ayoungI tried ip addr add dev eth101:58
larsksThat's exactly what I did, and it works just fine.  eth1  is connected to brovc, as is eth1 on the undercloud *and* eth0 on the overcloud hosts.01:58
ayoungOK..let me give that a go...the ovs stuff is neat and all, but I'd rather it be the simpler approach you show01:58
ayounglarsks, yeah, not working for me02:13
ayounglarsks, granted, I have not set up the undercloud yet, so maybe just doing02:13
ayoungsudo ip addr add dev eth102:13
ayoungis not sufficient02:14
ayounglet me undo that and setup the undercloud02:14
larsksThat doesn't make any sense, though, because if just setting up an ovs bridge makes things work, you obviously already have the necessary connectivity.02:16
larsksLet me know if you want me to take a look.02:16
larsksOr it you want to poke around my setup...02:17
*** rwsu has quit IRC02:17
larsksayoung: ^^^02:19
*** cwolferh has quit IRC02:22
*** dmacpher has quit IRC02:23
*** _prad_ has quit IRC02:23
ayounglarsks, ok,  I have the undercloud set up02:24
ayoungon the second VM (identity) I just did02:24
ayoungsudo ip addr add dev eth102:24
larsksDid you remember 'ip link set eth1 up' also?02:25
ayounglarsks, missed that02:26
larsksInterfaces are down by default unless something brings them up...02:26
ayounglarsks, yes that was that I was missing02:26
larsksAlso, note the bit at the bottom of about allocating a dhcp address from neutron.02:27
larsksI think that's probably the way to go.02:27
ayounglarsks, absolutely02:27
ayoungI was wondering about that earlier02:27
ayoungso...if that is so simple, why is the undercloud install using the ovs stuff?02:28
larsksneutron uses the ovs bridges to link things like dhcp services, neutron routers, etc that run in their own network namespace to the appropriate networks.02:28
ayoungOK...I'll update.  The ovs logging of the commands was still an interesting find02:30
*** dmacpher has joined #tripleo02:31
*** sthillma has quit IRC02:43
*** thrash is now known as thrash|g0ne02:44
*** dmacpher_ has joined #tripleo02:44
*** dmacpher has quit IRC02:46
*** sthillma has joined #tripleo02:47
*** sthillma has quit IRC03:13
*** cwolferh has joined #tripleo03:23
*** greghaynes has quit IRC03:31
*** bkero has quit IRC03:31
*** cwolferh has quit IRC03:32
*** ramishra has quit IRC03:35
*** ramishra has joined #tripleo03:36
*** ramishra has quit IRC03:38
*** ramishra has joined #tripleo03:39
*** ramishra has quit IRC03:39
*** cwolferh has joined #tripleo03:39
*** cwolferh has quit IRC03:44
*** ramishra has joined #tripleo03:44
*** cwolferh has joined #tripleo03:46
*** masco has joined #tripleo03:49
*** veteran has joined #tripleo03:50
*** veteran has quit IRC03:50
*** cwolferh has quit IRC03:55
*** mburned has quit IRC04:04
*** bvandenh has joined #tripleo04:05
*** mburned has joined #tripleo04:05
*** links has joined #tripleo04:13
*** bvandenh has quit IRC04:13
*** ccamacho1 has joined #tripleo04:15
*** ccamacho has quit IRC04:15
*** dmacpher_ is now known as dmacpher-afk04:17
*** TSCHAK has joined #tripleo04:18
*** khrystoph has joined #tripleo04:20
*** yamahata has joined #tripleo04:33
openstackgerritIan Wienand proposed openstack/diskimage-builder: Add cloud-initramfs-growroot for Precise
openstackgerritIan Wienand proposed openstack/diskimage-builder: Add release to pkg-map
*** jcoufal has joined #tripleo04:39
openstackgerritGiulio Fidente proposed openstack/tripleo-heat-templates: Pass MysqlVirtualIP via EndpointMap
*** rlandy has quit IRC05:01
openstackgerritMerged openstack/tripleo-heat-templates: Filter for local nodes in check_resource function
*** jaosorior has joined #tripleo05:02
*** Nakato has quit IRC05:07
*** paramite has joined #tripleo05:28
*** khrystoph has quit IRC05:35
*** rcernin has joined #tripleo05:47
*** aufi has joined #tripleo05:47
*** apetrich has joined #tripleo05:55
*** greghaynes has joined #tripleo06:02
*** khrystoph has joined #tripleo06:14
*** khrystoph has quit IRC06:14
openstackgerritAndreas Florath proposed openstack/diskimage-builder: [WIP] Add: lvm element
*** tremble has joined #tripleo06:21
*** milan has joined #tripleo06:22
*** _milan_ has quit IRC06:22
*** david-lyle_ has joined #tripleo06:26
*** SlickNik has quit IRC06:26
*** SlickNik has joined #tripleo06:26
*** Tristitia has quit IRC06:27
*** jrist has quit IRC06:27
*** fpan has quit IRC06:27
*** fpan has joined #tripleo06:28
*** david-lyle has quit IRC06:28
*** shanky has quit IRC06:28
*** shanky has joined #tripleo06:29
*** jrist has joined #tripleo06:29
*** jrist has joined #tripleo06:29
*** Tristitia has joined #tripleo06:30
*** Nakato has joined #tripleo06:31
*** shardy has joined #tripleo06:35
openstackgerritJiri Tomasek proposed openstack/tripleo-ui: Initial zaqar websockets
*** jprovazn has joined #tripleo06:42
*** gfidente has joined #tripleo06:43
*** jokke_ has quit IRC06:45
*** jokke_ has joined #tripleo06:45
openstackgerritMerged openstack/instack-undercloud: Move 02-undercloud-stack-heat-metadata to python module
openstackgerritMerged openstack/instack-undercloud: Don't use os-net-config element
openstackgerritMerged openstack/instack-undercloud: Consolidate undercloud* elements
openstackgerritMerged openstack/instack-undercloud: Remove truncate nova config workaround
*** spredzy has quit IRC06:56
*** spredzy has joined #tripleo07:00
*** jaosorior has quit IRC07:00
*** jaosorior has joined #tripleo07:01
*** jtomasek has quit IRC07:01
*** florianf has joined #tripleo07:06
*** spredzy has quit IRC07:08
*** devvesa has joined #tripleo07:08
*** devvesa has quit IRC07:09
*** mcornea has joined #tripleo07:09
*** devvesa has joined #tripleo07:09
openstackgerritMarios Andreou proposed openstack/tripleo-heat-templates: Adds memcached as a composable service
*** mikelk has joined #tripleo07:25
*** jpena|off is now known as jpena07:37
*** abehl has joined #tripleo07:38
jaosoriorshardy: how does the "name" parameter in the OS::Nova::Server resource get used?07:40
*** yamahata has quit IRC07:41
*** zoli_gone-proxy is now known as zoliXXL07:42
*** ohamada has joined #tripleo07:42
shardyjaosorior: it gets passed to nova to set the name of the server07:45
shardywhich in turn influences things like the hostname set by cloud-init07:45
jaosoriorseems to me like we should have been appending the domain there O_O07:46
jaosorior(slow as hell yes, but I FINALLY have a dev environment...also had problems with my hardware)07:46
*** ebarrera has joined #tripleo07:47
*** dtantsur|afk is now known as dtantsur07:47
*** mbound has joined #tripleo07:48
jaosoriorshardy: So it seems to me like we should be adding the CloudDomain here too;a=blob;f=puppet/controller.yaml;h=d5dcc7d8f3036b4f3b34a34ce81ca9df299e3ede;hb=HEAD#l70007:49
shardyjaosorior: hrm, that's certainly worth trying - we'll end up with some pretty verbose names in nova list, but cloudinit would then be able to split the domain off07:49
*** electrofelix has joined #tripleo07:49
shardygfidente: ^^ did you try that approach?07:50
jaosoriorshardy: Just one question, that name be the hostname linked to the internal-api network?07:50
*** electrofelix has quit IRC07:50
jaosoriorah, nevermind07:51
jaosoriorwe do use that name to create the hosts file07:51
jaosoriorso this is wrong:;a=blob;f=puppet/controller.yaml;h=d5dcc7d8f3036b4f3b34a34ce81ca9df299e3ede;hb=HEAD#l163907:52
shardyjaosorior: Hmm, yeah you may be right - give it a try :)07:52
shardybe sure to reboot one of the deployed nodes and check cloud-init doesn't do bad things on reboot to update the name07:53
shardy(well, it will update it, we need to check it's still correct)07:53
jaosoriorEmilienM: Hey dude07:53
shardyHey EmilienM, good morning07:53
jaosoriorshardy: will take it into account07:54
jaosorioronly problem is that it changes the scheme we used to have for /etc/hosts07:54
jaosoriorso now we have to figure something else out07:54
openstackgerritCarlos Camacho proposed openstack/tripleo-heat-templates: Composable roles within services - NTP
openstackgerritCarlos Camacho proposed openstack/tripleo-heat-templates: Composable roles within services - MongoDB
shardyjaosorior: pls update with your findings when tested07:57
openstackLaunchpad bug 1581472 in tripleo "CloudDomain doesn't correctly set hostname" [High,Triaged] - Assigned to Giulio Fidente (gfidente)07:57
jaosoriorwill do07:57
*** ifarkas has joined #tripleo08:07
mandreshardy: there is a bit of resistance against having the ansible validations in instack-undercloud08:09
mandrereason being it creates unnecessary coupling with the undercloud08:10
mandreslagle suggested tripleo-common could be a better place, since they would live along with the mistral validation workflow08:10
mandredprince argued it should be a new client library08:11
*** chem`` has joined #tripleo08:12
*** spredzy has joined #tripleo08:12
shardymandre: Hi, so I agree instack-undercloud probably isn't the ideal location08:13
shardytripleo-common is probably OK, as tripleoclient already depends on it, and as you say the mistral workflows will most likely reference the validations08:14
*** chem` has quit IRC08:14
shardytripleo-common is kind of getting into a weird state because we're implementing all the custom actions there, so we may eventually need some new repos, but if you'd rather not create one right now, I think tripleo-common is OK08:15
*** jistr has joined #tripleo08:15
mandreshardy: ok, so you're not opposed to create a new repo for tripleo-validations?08:16
shardymandre: I'm not, or alternatively we could add them to tripleo-common and plan to break out all of the API stuff into another repo08:16
mandretripleo-common is definitely better than instack-undercloud, still my preference would go to a separate repo08:16
shardyas, really, tripleo-common should only contain the pieces shared between the client and the backend of the API08:17
shardymandre: I'm fine with a new repo if you think that will be cleaner08:17
mandreshardy: it will be more work for me, but yeah, I think it's cleaner :)08:17
shadowerhow would we go about getting a new repo? What's the proocess?08:28
openstackgerritMerged openstack/tripleo-docs: Document profile and node placement redundancy
shardyshadower: propose a patch to project-config and openstack/governance, e.g see this for tripleo-quickstart08:31
shardypls add me as a reviewer to both08:31
shadowershardy: thanks!08:31
openstackgerritCarlos Camacho proposed openstack/tripleo-docs: Composable roles within services Tutorial
*** lucas-afk is now known as lucasagomes08:32
*** fzdarsky_ has quit IRC08:32
*** fzdarsky_ has joined #tripleo08:32
openstackgerritEmilien Macchi proposed openstack/instack-undercloud: Cleanup hieradata to reduce Puppet warnings
*** fzdarsky_ is now known as fzdarsky08:33
*** spredzy has quit IRC08:36
openstackgerritPanFengyun proposed openstack/python-tripleoclient: fix the output message about 'driver_info/deploy_kernel'
gfidentejaosorior, shardy ack, no I didn't08:44
*** derekh has joined #tripleo08:44
jaosoriorgfidente: No worries, I'm looking into it08:44
gfidentejaosorior, it's for the hosts entry I think we could just use the new str_split and get the first item08:48
gfidentejaosorior, set to PRIMARYHOST08:48
jaosoriorI was thinking of appending the domain to the name that's passed to the OS::Nova::server resource08:49
jaosoriorand then changing the scheme (again) for the /etc/hosts08:49
jaosoriorthe /etc/hosts file is not problematic, we know that's being set to whatever we put in that template. But the issue is the domain being overwritten by cloud-init. So my hope is that setting the hostname with the proper domain via nova will do the trick08:50
gfidentewhy would the domain get overwritten?08:50
gfidenteI was looking after the ::server name customization08:51
jaosoriorcause cloud-init uses localdomain as default08:51
gfidentejaosorior, no man I am much past that08:51
jaosoriorI see08:51
jaosoriorso what's up?08:51
gfidentethe /etc/hosts would get fqdn as short name and fqd.domain twice08:52
gfidenteso I was thinking we could do str_split by .08:52
gfidentethere to get the shortname08:52
gfidenteI'm +2 the ::server name change08:52
*** rasca has quit IRC08:52
jaosoriorwell, if the deployer uses a . in the template (which is customizable) the str_split becomes problematic08:53
jaosoriorthat's why I was just appending the domain there08:53
gfidenteno hostname is always the leftmost08:53
*** rasca has joined #tripleo08:53
rascarieccomi scusa ma si รจ piantato tutto08:53
gfidentejaosorior, let's talk on an actual submission which is easier08:54
jaosorioryes, but what if the deployer decides to put in the demplate something such as '' and then sets the domain to something (lets say ''08:54
gfidentethey can't use . in hostname08:54
jaosoriorthey can't?08:55
jaosoriorwould be a bad idea (I agree)08:55
jaosoriorbut they could, in theory08:55
gfidenteI think it's not compliant08:56
gfidente. has special meaning08:56
shardyWe could use a constraint on the parameter to prevent it08:57
jaosoriorif that's the case then we're good08:59
openstackgerritMerged openstack/tripleo-heat-templates: Tighten the access rules for galera
*** hewbrocca is now known as hewbrocca-afk09:00
gfidenteguys believe it or not, I have a change that only passes upgrades job :)09:00
*** hewbrocca-afk is now known as hewbrocca09:00
jaosoriorI was just thinking of re-constructing the hostname the same way it's passed to the server's name09:00
openstackgerritCarlos Camacho proposed openstack/tripleo-heat-templates: Test do not merge
jaosoriorgfidente: pass it09:00
gfidentenah probably the non-upgrade failures are unrelated, was joking about it09:01
*** saneax_AFK is now known as saneax09:01
openstackgerritQasim Sarfraz proposed openstack/puppet-tripleo: add plumgrid neutron profile
*** dmk0202 has joined #tripleo09:19
*** sambetts|afk is now known as sambetts09:24
openstackgerritMerged openstack/tripleo-common: Updated from global requirements
openstackgerritMerged openstack/tripleo-heat-templates: Updates ControlPlaneSubnetCidr to be a string
*** spredzy has joined #tripleo09:31
openstackgerritSteven Hardy proposed openstack/tripleo-heat-templates: Fix Neutron enable_*_agent* Hiera parameters
openstackgerritSteven Hardy proposed openstack/tripleo-heat-templates: Fix Neutron enable_*_agent* Hiera parameters
*** jtomasek_ has joined #tripleo09:34
openstackgerritMerged openstack-infra/tripleo-ci: Update 60-setup-testenvs to match what we use
openstackgerritEmilien Macchi proposed openstack/instack-undercloud: Cleanup hieradata to reduce Puppet warnings
openstackgerritMerged openstack/tripleo-heat-templates: Use a different ceph key for admin/client user
*** spredzy has quit IRC09:50
*** ccamacho1 is now known as ccamacho09:51
*** spredzy has joined #tripleo09:57
openstackgerritMerged openstack/puppet-tripleo: Adds the base and pacemaker profile for the memcached service
openstackgerritEmilien Macchi proposed openstack/instack-undercloud: Deploy and Upgrade Ironic to run in mod_wsgi
*** hdsdman_ has quit IRC10:03
bandinimarios: ROTFL @ enterprise. I am totally getting t-shirts when this lands10:04
openstackgerritAthlan-Guyot sofer proposed openstack/puppet-pacemaker: WIP: integrate PCS provider in the merge.
*** sshnaidm has quit IRC10:06
*** milan has quit IRC10:06
gfidenteshardy, hey thanks for checking the cinder rols10:06
*** mbound has quit IRC10:06
openstackgerritSteven Hardy proposed openstack/tripleo-heat-templates: Replace NodeConfigIdentifiers with DeployIdentifier
gfidenteso I can add the rabbit params to scheduler and volume10:07
gfidentebut do you think it's better to split this into multiple submissions or keep it into a single one?10:07
shardygfidente: Personally I would have done three patches instead of one, but you don't have to break it up if it's too much effort10:07
shardythe main reason for the -1 was the rabbit thing10:08
gfidentethough splitting might help testing them too10:08
gfidenteand we have cinder/backup in a separate submission anyway10:08
gfidentenot the puppet manifests though10:09
*** stendulker has joined #tripleo10:10
EmilienMgfidente: i've noticed some patches in puppet-tripleo do some hiera lookup on pacemaker manifest, we should rather feed the hieradata in THT/service/pacemaker/role yaml file, right?10:14
gfidenteEmilienM, yes I think that'd be nicer10:14
EmilienMit's cosmetic though10:15
openstackgerritMichele Baldessari proposed openstack/tripleo-specs: Add initial next generation HA architecture spec
gfidenteI kept the hiera calls where the parameter is to be shared with other stuff10:15
EmilienMbut will make puppet-tripleo code much cool10:15
gfidentebut other than that it's better to turn those into class params yes10:15
EmilienMgfidente: I've done some reviews all the morning - do you have patches that need review in particular?10:16
gfidentesure cinder roles!10:16
EmilienMI'm reviewing your cinder bit10:16
gfidenteyeah it's 800 lines :(10:16
EmilienMI'm already on it10:16
openstackgerritMerged openstack/tripleo-heat-templates: Add ExtraConfig example that always runs on update
*** rasca has quit IRC10:16
gfidenteI am not sure if there is a better mechanism to make an array with the list of enabled backends10:16
gfidentegiven the backend names can be customized via hiera as well10:16
EmilienMlet me review that10:16
gfidenteso I had to write some conditionals in volume.pp10:17
gfidentewhich aren't very cool10:17
gfidenteif you've better ideas -1 ! :)10:17
*** rasca has joined #tripleo10:17
*** fragatina has quit IRC10:20
*** fragatina has joined #tripleo10:20
*** rhallisey has joined #tripleo10:21
EmilienMI'll -210:22
EmilienMeverything from you :P10:22
*** fragatin_ has joined #tripleo10:24
*** fragatina has quit IRC10:24
gfidentetks :)10:26
EmilienMok I found a limitation in your code10:28
EmilienMit's not a blocker I would say but it's a limitation we need to document in the volume.pp I guess and work on it later10:28
*** fragatin_ has quit IRC10:28
EmilienMyour current code won't allow (now) to deploy multiple same backends10:28
EmilienMie: multi backends with 2 rbd backends10:28
EmilienMsame for dell, etc10:28
EmilienMbut I won't -1 for that, it's ok to iterate later.10:29
EmilienMwe had the same issue in spinalstack, let me show you how we resolved it10:29
EmilienMactually it's a whole different design I think10:30
EmilienMwhich is ~ your volume.pp10:30
EmilienMand see
EmilienMwhich is an example of driver10:31
EmilienMwe use Define, so we can deploy the same backend multiple times10:31
EmilienMgfidente: ^10:31
openstackgerritDerek Higgins proposed openstack-infra/tripleo-ci: Allow the mirror server to be publicly accessible
derekhtrown|outtypewww: ^^10:34
gfidenteEmilienM, so cinder_backends there has to be a dict with all key/values for all backends right?10:35
*** thrash|g0ne is now known as thrash10:35
*** fragatina has joined #tripleo10:36
gfidentebecause presently to do that in tripeo we just use ceph::conf10:37
EmilienMgfidente: we define cinder backends in a hash10:38
EmilienMI can provide an example10:38
EmilienMa sec10:38
gfidenteand you can add with that any backend of any type, including any of the types we also deploy from tripleo10:38
EmilienMlet me a sec10:39
EmilienMso here's an example of backend hash:
EmilienMthis is ONE parameter volume_cinder_backends that contains ALL backends config10:39
EmilienMevery key is a backend type (rbd, iscsi) etc and has its own hash for every backend name (rbd_fast, rbd_slow)10:40
*** fragatina has quit IRC10:40
gfidenteright so user would have to mess with that to deploy their own custom backend10:40
EmilienMthat way, we allowed to enable 1) multi backends 2) multi backends with same driver10:40
gfidenteright that's possible with existing manifests too10:41
EmilienMin THT, we can default to single backend name per driver in the templates10:41
EmilienMbut at least we have the interface to allow multiple instances of the ssame driver10:41
gfidenteagain, we can do multiple instances of same driver with those manifests too10:41
EmilienMerr sorry10:41
EmilienMyou need to transform the class in a define10:42
gfidenteyes we can using ceph::conf10:42
gfidenteand cinder_user_enabled_backends10:42
gfidenteand I can tell you instead that if a user wanted to deploy ceph with tripleo and plug into cinder their own dellsc I don't think they can10:42
gfidentewhile with ceph::conf and cinder_user_enabled_backends they can10:42
EmilienMI don't understand your ceph::conf mention10:43
EmilienMI'm going to take my lunch break now, maybe we can talk about this afternoon10:43
gfidente bug 1312000 in openstack-puppet-modules "[RFE] support for cinder multiple heterogeneous backends for a single deployment" [High,Closed: errata] - Assigned to emacchi10:43
EmilienMdo not use cinder::config10:44
gfidenteso people won't even have to hack our manifests to configure their backends10:44
EmilienMno, please don't do that10:44
gfidentewe only need to write manifests in puppet-tripleo for the backends which tripleo manages10:44
EmilienMwe have the interface in puppet-cinder to do all of this already10:44
EmilienMdon't use cinder::config10:44
EmilienMok I'm stepping down a bit, ttyl10:45
gfidenteoh it's not that *I* do it10:45
gfidenteit's that people has cinder::config and can use to add their backends10:45
gfidenteso I don't see the point of trying to wrap puppet-cinder backends in puppet-tripleo10:45
gfidentewe'd only add limitations10:45
gfidentewith ::config they can configure whatever they want10:45
gfidenteand we only write manifests for the backends which tripleo itself deploys10:46
openstackgerritMerged openstack/tripleo-heat-templates: Fix Neutron enable_*_agent* Hiera parameters
openstackgerritGiulio Fidente proposed openstack/tripleo-heat-templates: Switch Cinder Api/Scheduler/Volume to composable roles
*** milan has joined #tripleo10:51
*** zoliXXL is now known as zoli|wfh10:54
jaosoriorshardy, gfidente: Do you guys know what in the nodes post deployment (or even the loadbalancer deployment) is trying to do getaddrinfo ?10:54
jaosoriorI'm getting failures there when adding the domain10:54
jaosoriorError: Could not run: Could not retrieve facts for overcloud-controller-1: getaddrinfo: Name or service not known10:54
*** spredzy has quit IRC10:55
jaosoriorso it seems it's trying to access the old hostname, not taking into account the domain10:56
gfidentejaosorior, looks like puppet trying to make up some facts10:56
gfidentejaosorior, I think it split by . and is trying to resolve it10:56
gfidentedo you have a mapping for the shortname in /etc/hosts ?10:56
openstackgerritMerged openstack/tripleo-ui: Disable links when deploy is in progress
jaosoriorI didn't, first wanted to try without it11:01
*** spredzy has joined #tripleo11:02
gfidenteso I think it was unable to resolve the shortname11:02
jaosoriorit was11:03
jaosoriorsooo I gotta do that11:03
jaosoriorwell, I'll finish it on monday11:04
openstackgerritMerged openstack/tripleo-heat-templates: Adds memcached as a composable service
jaosoriorhave a good weekend guys!11:04
*** jaosorior has quit IRC11:04
openstackgerritGiulio Fidente proposed openstack/python-tripleoclient: Generate a password for the Mysql 'clustercheck' user
openstackgerritGiulio Fidente proposed openstack/python-tripleoclient: Generate a password for the Mysql 'clustercheck' user
*** veteran has joined #tripleo11:14
mariosshardy: gfidente thanks gents for landing the memcached11:16
gfidentemarios++ :)11:16
shardymarios: np :)11:23
ccamachoHey guys, testing this patch for compatibility with mitaka ( the nonha job is failing with this message ( ConnectFailure: Unable to establish connection to Do you think it can be related to overhead as the ha job have passed?11:23
ansiwenso, I'm running on CentOS 7, but first, the libselinux-python package was missing on the target machine (--install-deps only installs it on the source machine)11:24
ansiwenand now I have this error:11:25
ansiwenfatal: [localhost]: FAILED! => {"changed": false, "failed": true, "gid": 1001, "11:25
ansiwengroup": "stack", "mode": "0664", "msg": "unlinking failed: [Errno 13] Permission denied: '/var/cache/tripleo-quickstart/images/_undercloud.qcow2' ", "owner": "stack", "path": "/var/cache/tripleo-quickstart/images/_undercloud.qcow2", "secontext": "unconfined_u:object_r:var_t:s0", "size": 2145906454, "state": "file", "uid": 1001}11:25
*** veteran has quit IRC11:25
ccamachohey ansiwen, what are you trying  to do? Deploying tripleo? Which is you HW config? Are you trying to deploy it virtualized?11:28
ansiwenccamacho: deploy tripleO with the script on a baremetal machine with CentOS 711:29
ccamachoI'm just about going to have lunch, but if you need it I can help you to deploy it.11:30
shardygfidente: Hey, did you have a patch to make the per-service bind IPs generic?11:32
gfidenteshardy, yeah11:32
ansiwenccamacho: sure, thanks, but I though the idea of is, that you don't need help ;-)11:32
shardyI need a way to reference the proxy/management network for swift11:32
ccamachoansiwen +111:33
gfidenteshardy, oh we can't use it yet though because we need to update hiera in the images to >= 211:33
*** ccamacho is now known as ccamacho|lunch11:33
gfidenteshardy, patches are and
*** athomas has joined #tripleo11:42
shardygfidente: Nice, thanks - what's the status of the hiera upgrade?11:45
*** weshay has joined #tripleo11:46
gfidenteshardy, so given this is distro-dependent, we were tracking this with a BZ bug 1337335 in distribution "Hiera 2.x packaging" [Urgent,New] - Assigned to lars11:47
shardygfidente: Hmm, thanks, the comments from apevec on that aren't very encouraging11:49
shardyis there any interim workaround we can use?11:49
gfidenteI think we could write down each ip and ip/subnet in a different hiera key11:49
*** morazi has joined #tripleo11:51
shardygfidente: Yeah, so then we can just do e.g hiera tenant_network instead of the hash lookup11:51
gfidenteshardy, yes but we'd need that for all three maps, ip, subnet and ip_uri11:51
gfidenteso we could do it but meh11:51
*** mbound has joined #tripleo11:51
gfidenteif we can get the hash lookup to work it'll turn useful :)11:52
shardyI agree, but we need to understand the blockers mentioned in the bug and decide if we should just go with the interim solution11:52
shardyotherwise several services aren't really composable11:53
*** julim has joined #tripleo12:02
*** paramite is now known as paramite|afk12:02
*** lucasagomes is now known as lucas-hungry12:02
*** jtomasek_ has quit IRC12:05
*** jtomasek_ has joined #tripleo12:05
*** jayg|g0n3 is now known as jayg12:10
openstackgerritImre Farkas proposed openstack/python-tripleoclient: Add Ironic password generation during deployment
openstackgerritImre Farkas proposed openstack/tripleo-puppet-elements: Add Ironic packages to overcloud controller
*** dprince has joined #tripleo12:14
openstackgerritMarios Andreou proposed openstack/puppet-tripleo: Adds the puppet-tripleo manifests for manila
*** jtomasek_ has quit IRC12:16
*** ayoung has quit IRC12:16
*** jtomasek has joined #tripleo12:16
openstackgerritBrad P. Crochet proposed openstack/puppet-tripleo: Change default CloudFormation ssl port to 13005
openstackgerritBrad P. Crochet proposed openstack/os-cloud-config: Add Heat CloudFormation service for endpoint creation
*** paramite|afk is now known as paramite12:19
*** zoli|wfh is now known as zoli_gone-proxy12:25
*** trown|outtypewww is now known as trown12:25
*** masco has quit IRC12:28
openstackgerritMerged openstack/tripleo-quickstart: Update undercloud default variable location
trownadarazs: mind taking a look at it is pretty simple12:30
adarazstrown: looking12:30
adarazstrown: actually the clear thing would be to do it with awk and add 1 to the variable ;)12:33
adarazstrown: but I think we can get away with this.12:33
adarazsthough I thought I can get away with 1->2 :)12:33
trownadarazs: ya, I hit some issues with puppet modules when testing out the gating role on master12:34
adarazstrown: aaah! I think I hit the same issue just now, I was looking into the conflicts :D12:34
trownadarazs: it was trying to install openstack-puppet-modules from deps repo12:34
adarazsI was confused why it was doing it12:35
trownwhich is pre-split, so a ton of conflict with the split packages12:35
adarazsyou spared me some headache. :)12:35
*** zoli_gone-proxy is now known as zoliXXL12:35
adarazsall right, I keep working on the gating feature requests in that case ;)12:36
trownthat is really a bit of a packaging bug... the post split should have an epoch bump, but that is what yum plugin priorities is for... hiding packaging bugs12:36
adarazswould be so nice if ansible would support with_items on blocks. -_- I'll mock around with includes.12:36
trownadarazs: ya, I will respond shortly on those 2 issues. I have a hacked up version for the packaging one locally12:37
adarazstrown: great, because I don't have anything for that yet :)12:37
trownperfect! I like what you laid out for the other one12:38
*** rbrady has joined #tripleo12:39
*** rlandy has joined #tripleo12:45
*** _prad_ has joined #tripleo12:46
*** marcusvrn_ has joined #tripleo12:47
openstackgerritBrad P. Crochet proposed openstack/puppet-tripleo: Add Sahara profiles
*** jpena is now known as jpena|lunch12:48
*** julim has quit IRC12:49
openstackgerritBrad P. Crochet proposed openstack/tripleo-heat-templates: Composable Sahara services
openstackgerritMerged openstack/tripleo-quickstart: Make sure delorean.repo takes precedence over delorean-deps.repo
openstackgerritwes hayutin proposed openstack/tripleo-quickstart: localhost is a reserved term by ansible, use instead
*** julim has joined #tripleo12:57
*** tzumainn has joined #tripleo12:57
*** lucas-hungry is now known as lucasagomes13:00
pabelangerderekh: dprince: slagle: I need to learn more information about tripleo-cloud-test-rh1, is there any diagrams for the current network setup?  I am seeing some UDP issue on the AFS mirror and think there is an issue with the NAT. I'd likely need to get some tcpdumps from the gateway that is doing NAT13:01
*** paramite is now known as paramite|afk13:02
*** zoliXXL is now known as zoli|lunch13:03
openstackgerritCarlos Camacho proposed openstack/tripleo-heat-templates: Composable roles within services - MongoDB
*** lblanchard has joined #tripleo13:07
openstackgerritCarlos Camacho proposed openstack/tripleo-heat-templates: Composable roles within services - NTP
openstackgerritCarlos Camacho proposed openstack/tripleo-docs: Composable roles within services Tutorial
openstackgerritCarlos Camacho proposed openstack/puppet-tripleo: Add gnocchi profiles
openstackgerritwes hayutin proposed openstack/tripleo-quickstart: update the readme to include a full end to end deployment --tags all
pabelangerderekh: derekh: slagle: If under the control of tripleo, or is that from the upstream provider?13:11
pabelangerI assume is nova13:11
mandreshardy: submitted and to create tripleo-validations repo13:12
mandreI hope everything is in order13:12
weshaytrown, would you prefer the note about root for localhost in the README or
trownweshay: maybe we could add a section to "Deploying" in the README stating that it can be run directly on a virthost (and here's how), but it requires root.13:16
weshaytrown, sure.. I think adding it to would be ok13:17
trownweshay: I just thought of another wrinkle there though... since we are not using ansible local connection I think it requires having root's pubkey in root's authorized_keys right?13:17
trownoh nice, ya it could go in that doc patch13:18
weshaytrown, I'm running a --tags all w/ root on localhost now, I'll find out13:19
trownweshay: ya just tested without having the root user pubkey in authorized keys, and it doesn't work :(13:21
weshaytrown, oh.. I see what you mean.. well that is the same requirement as running remotely I think.. just a nuance13:21
weshayI'll add it to the README13:21
trownweshay: I think we would need that on the USB key too (which makes the wrapper script more useful since I dont think I want to add that logic to
weshaysomething like.. ssh root@ should work w/o passwd13:22
weshaytrown, k13:22
trownand the wrapper script would need to make sure root has an ssh key, and generate it if not, then copy the pubkey to authorized keys (so it just works)13:23
weshaytrown, hrm.. and if the box does not have a ssh key created for root?13:23
trownit would have to have logic to `ssh-keygen`13:23
*** paramite|afk is now known as paramite13:25
trownlarsks: do you have any thoughts on using ssh to localhost vs. ansible local connection?
*** akshai has joined #tripleo13:31
*** ayoung has joined #tripleo13:33
openstackgerritLeif Madsen proposed openstack/tripleo-docs: Composable roles within services Tutorial
*** rodrigods has quit IRC13:34
*** rodrigods has joined #tripleo13:34
*** pabelanger has quit IRC13:39
*** pabelanger has joined #tripleo13:39
*** eggmaster has joined #tripleo13:40
*** rasca has quit IRC13:46
*** links has quit IRC13:46
larskstrown: That change smells a little hacky.  Before merging, I'd like to understand the problem better and see if we can't solve it differently.13:47
*** dtrainor has quit IRC13:50
*** dtrainor has joined #tripleo13:50
openstackgerritTzu-Mainn Chen proposed openstack/tripleo-common: Add TripleO Mistral documentation
*** rasca has joined #tripleo13:52
trownlarsks: ya, currently running against localhost does not work, and that change does make it work, albeit with a kind of large tradeoff13:54
larsksYeah, I understood that from the comments.13:54
larsksI just wonder if there isn't an alternative solution.13:54
derekhpabelanger: is the public IP of the bastion host, its on the tripleo Rack but the upstream team don't control that host13:54
larsksWe were previously able to target localhost successfully...13:54
larsks+1 on tests using localhost once this is fixed :)13:55
larsksWe'll add that to the test matrix I was wishing for yesterday...13:55
derekhpabelanger: I've suspected for some time now that we've been loosing UDP packets, this was the reason I think we were seeing unreliable DNS requests13:55
pradkslagle, I want to access a db string thats constructed in overcloud_controller_pacemaker.yaml from the aodh standalone manifest.. any way to do that without duplicating the logic again?13:55
trownlarsks: well for the "quick" tests more was always the intention, I think have some kind of "quick-teardown" test also makes sense13:56
pabelangerderekh: who is the owner of that system? Is is a linux box or some hardware?13:56
derekhpabelanger: its happening outside of the Rack though, as I can send DNS requests out from the bastion, and fail to get a response back intermittently I forget how often but can find out again if needed13:57
slaglepradk: which string is it? $ceilometer_database_connection?13:57
pradkslagle, yep13:57
openstackgerritRyan Brady proposed openstack/instack-undercloud: Revert "mistral: deploy API in WSGI with apache"
pabelangerderekh: Any way we can get support on it?13:58
pradkslagle, which inturn gets set to mongo, which is what i want13:58
openstackgerritEmilien Macchi proposed openstack/instack-undercloud: Cleanup hieradata to reduce Puppet warnings
*** jpena|lunch is now known as jpena13:58
derekhpabelanger: its a linux box, I would have always considered the RH lab team (internal team) the owners of the box, zoli|lunch I think has root on it and may be able to do some debugging13:59
ccamacho|lunchhey gfidente, question: I'm deploying an overcloud using master and getting "CREATE_FAILED resources.Controller: Property error: resources[0].properties: Property MysqlClustercheckPassword not assigned" do you think it might be related to ?13:59
openstackLaunchpad bug 1581677 in tripleo "Galera access rules need to be tightened" [High,Fix released] - Assigned to Michele Baldessari (michele)13:59
pabelangerderekh: okay, great. That gets me started13:59
slaglepradk: we can try and set that value via hiera13:59
pabelangerzoli|lunch: ohai, do you mind catching up on backscroll when you get back14:00
*** ccamacho|lunch is now known as ccamacho14:00
slaglepradk: like is done for ceilometer_coordination_url in puppet/controller.yaml14:00
derekhzoli|lunch: basically we think udp packets are being lost somewhere between the bastion and the outside world, gimme a few minutes and I'll find the commands I used to reproduce in the past14:00
weshaytrown, my local test of localhost worked w/ the change e-e14:00
bandiniccamacho: you need a more recent python-tripleoclient14:00
trownend to end?14:00
ccamachobandini, woow thanks!14:01
bandiniccamacho: no prob ;)14:01
pradkslagle, the mongodb string construction has a bit of logic involved including setting up the mongo_node_ips_with_port and replset .. i'll see how i can refactor that to hiera14:02
slaglepradk: oh, due to the case statement. i see14:03
*** milan has quit IRC14:03
pradkslagle, yea14:03
slaglei wouldnt spend too much time trying to avoid duplicating it if you have to14:04
pradkalso needs this
*** rasca has quit IRC14:04
pradkyea i dont know a better way other than duplicating here.. composable roles would come handy here but alas this is mitaka14:05
*** rlandy has quit IRC14:05
openstackgerritJiri Stransky proposed openstack/tripleo-specs: Composable Upgrades
openstackgerritwes hayutin proposed openstack/tripleo-quickstart: update the readme
*** rbrady has quit IRC14:10
*** stendulker has quit IRC14:12
*** rbrady has joined #tripleo14:13
openstackgerritRyan Brady proposed openstack/instack-undercloud: Revert "mistral: deploy API in WSGI with apache"
openstackgerritMichele Baldessari proposed openstack/tripleo-heat-templates: Tighten the access rules for galera
*** rasca has joined #tripleo14:16
weshaytrown, ya14:18
socialjistr: when doing refactorings => when refactoring14:22
*** akshai has quit IRC14:23
* jistr amending locally14:23
socialjistr: I still think we should do puppet snippets with exec enclosed into role classes and just generate puppet files that ansible will orchestrate >.>14:24
socialbut end result might be as ugly as what we have now so dunno14:25
pradkjistr, Hi, not sure if slagle spoke to you about the software config for aodh upgrades.. we were discussing if it should be a stand alone template or part of say major_upgrade_pacemaker.yaml14:27
pradkjistr, you have any thoughts how this should be integrated in the upgrade workflow?14:27
jistrsocial: yeah maybe... it could be the way that provides most options, but also people might have trouble getting into that, and consider it super-weird if they didn't spend a few hours or so to think about the whole issue and the reasoning behind such choice. Generally the service owners might be responsible for service upgrades, also consultants might want to do amendments etc. So we need to consider "developer UX" too. That said, i think14:29
jistrwhat you mention is still on the table.14:29
*** bvandenh has joined #tripleo14:29
openstackgerritMerged openstack/tripleo-ui: Ignore the .idea folder (JetBrains)
*** zoli|lunch is now known as zoli|wfh14:30
*** akshai has joined #tripleo14:31
socialjistr: I'm not that good with ansible so my todo list on weekend is to play with it more but as far as I know you don't have declarative way to create dependancies and so on?14:31
jistrsocial: you do, but *much* more limited than puppet. You can do that just for roles (think Puppet classes), not micro-manage individual tasks (think resources). Roles (and their tasks) in a "play" are executed in the order specified. And you can only say "role X requires role Y", which can achieve that role Y will be added into the play before role X, (or if it was already in the play, it will change its spot to come before role X, IIUC).14:36
*** rasca has quit IRC14:37
jistri haven't played much with dependencies in Ansible, so take that with a grain of salt ^^14:37
*** rasca has joined #tripleo14:38
socialjistr: roles is doable14:38
openstackgerritTzu-Mainn Chen proposed openstack/tripleo-common: Add TripleO Mistral documentation
openstackgerritMichele Baldessari proposed openstack/tripleo-heat-templates: Tighten the access rules for galera
jistrsocial: well.. yea, except to utilize those deps we'd have to have "a role per service per step" instead of just "a role per service"14:39
trownlarsks: weshay I created a bug for the localhost issue: I will try to look into it a bit more this afternoon14:39
openstackLaunchpad bug 1584084 in tripleo-quickstart "Running against localhost does not work" [Critical,Triaged]14:39
jistrsocial: where by steps i mean "stop service", "upgrade service", "start service"14:39
larskstrown: thanks.  I will try to look also.14:39
jistrsocial: it's also somewhat explained (hopefully :) ) in the spec14:40
socialyeah, still reading that14:40
trownlarsks: weshay if there is not an obvious "good" fix, wdyt of going with weshay's patch tagged with Partial-Bug to that bug, and then work on restoring ansible local connection to close that bug?14:40
larskstrown: sure, if there isn't an obvious fix.  I would just like to understand what is going on first.14:40
trownlarsks: context is we are trying to create a USB key with tripleo-quickstart and RDO image on it for conferences, so we need localhost to work (somehow)14:41
trownback a bit later... 2 week checkup14:42
*** trown is now known as trown|afk14:42
jistrsocial: also i think Ansible dependencies cannot solve "only affect ordering, do not pull in this role if it wasn't already included", which makes them unusable for upgrades. (Yea, Puppet collectors would solve that :) )14:43
*** ramishra has quit IRC14:45
EmilienMdtantsur: I updated for Ironic deployed with apache/wsgi and it's still failing in undercloud CI, while it works fine in puppet CI, can you look at logs when you got time?14:47
*** ramishra has joined #tripleo14:47
jistrsocial: actually, i'm going to etch this conversation into the spec too. I think we already slightly discussed this before but it's hard to remember all the details, so i don't want to keep rediscovering these facts :)14:47
dtantsurEmilienM, will try, sure14:49
*** jprovazn has quit IRC14:50
*** milan has joined #tripleo14:52
*** ohamada has quit IRC14:52
gfidenteccamacho, you need to update tripleoclient14:53
gfidenteyou should see a CLUSTERCHECK password in tripleo-overcloud-password14:54
ccamachogfidente thanks! :) Already updated and working14:54
openstackgerritImre Farkas proposed openstack/puppet-tripleo: Add base ironic profile
openstackgerritJiri Stransky proposed openstack/tripleo-specs: Composable Upgrades
thrashjistr: I'm pretty sure the dependencies are only done once.14:59
thrashjistr: if multiple roles (A and B) depend on role C, then role C will only be executed once.15:00
thrashC -> A -> B15:00
jistrthrash: yea unless specifically overriden to repeat them.15:00
thrashjistr: right...15:00
openstackgerritShilla Saebi proposed openstack/tripleo-docs: corrected typo in instropecting doc
jistrthrash: the main blocker we have with the Ansible deps i think is "just make sure it's in correct order in case it's already included, but don't pull this into the play if it hasn't been included in the play manually"15:01
*** hewbrocca is now known as hewbrocca-afk15:02
openstackgerritMerged openstack-infra/tripleo-ci: Revert "Test with latest master heat"
*** diazjf has joined #tripleo15:10
socialjistr: rereading the spec15:11
*** yamahata has joined #tripleo15:14
*** akshai has quit IRC15:15
*** ebarrera has quit IRC15:16
*** paramite has quit IRC15:20
gfidentebnemec, the cloud-init change
gfidentebut it's really only meant to see if fqdn is in /etc/hosts when the nova metadata doesn't append any15:20
gfidenteit's supposed to preserve domainname on reboot though15:20
*** akshai has joined #tripleo15:21
bnemecgfidente: That may help.  The problem I'm hitting is that on deployment my compute nodes come up with service hosts like 'overcloud-novacompute-0', then after a reboot they change to 'overcloud-novacompute-0.localdomain'15:22
bnemecIf you've booted any instances prior to the reboot that breaks networking to them.15:22
gfidentebnemec, so jaosorir is trying to append .$domain to ::server resources15:23
gfidentewhich should be good as well15:23
gfidentedon't think he has a submission for it yet though15:23
shardybnemec: another workaround is to set dhcp_domain in nova.conf so it matches CloudDomain15:24
shardythe problem is cloud-init defaults to "localdomain" if the hostname provided by nova doesn't contain any domain15:24
shardybnemec: another workaround is to disable the update_hostname module in the /etc/cloud/cloud.cfg in the image15:25
shardy is the related bug15:26
openstackLaunchpad bug 1581472 in tripleo "CloudDomain doesn't correctly set hostname" [High,Triaged] - Assigned to Giulio Fidente (gfidente)15:26
*** mbound has quit IRC15:26
*** aufi has quit IRC15:28
bnemecshardy: I'm fine with the localdomain name, but it isn't being picked up on the initial deployment.  There's no domain at all in the host ids.15:28
bnemecI think we should be doing but I haven't had a chance to make the puppet changes Emilien wanted.15:28
shardybnemec: yeah, that's the problem, then update_hostname in cloud-init defaults it to localdomain on reboot15:29
shardybnemec: if you were to set e.g localdomain in nova.conf and CloudDomain, it will be applied on deloyment and on reboot15:29
openstackgerritBob Callaway proposed openstack/tripleo-heat-templates: Enable Neutron LBaaS Integration
EmilienMbnemec: yes, your patch will work until someone will patch puppet-nova with the param15:30
shardybasically we broke this when we removed the domain from dhcp_domain to enable the CloudDomain parameter15:30
EmilienMusing nova::config should not be something we used to do15:30
*** mgarciam has joined #tripleo15:31
bnemecEmilienM: Yeah, I'm not disagreeing, although we're already screwed if someone does that because we're setting it this way on compute nodes.15:31
bnemecI just need to find some time to sit down and push all the changes needed to do this the right way.15:32
EmilienMlet me know if I can help on this one15:32
dtantsurEmilienM, I don't see anything insane about this patch.. my only guess is selinux, but I don't see audit errors either (maybe I don't know where to look). probably we have to try it outside of the gate to be able to hack on it15:35
EmilienMdtantsur: puppet gate has selinux enforced and no problem15:36
dtantsurI see15:36
EmilienMdtantsur: no worries, I'll dig a bit more on my side15:36
*** dciabrin has quit IRC15:37
*** dciabrin has joined #tripleo15:37
*** devvesa has quit IRC15:38
*** mikelk has quit IRC15:38
*** penick has joined #tripleo15:46
*** rwsu has joined #tripleo15:46
*** Nakato has quit IRC15:49
openstackgerritJason E. Rist proposed openstack/tripleo-ui: Refactor notification for better display
jristflorianf, honza, jtomasek ^15:50
*** rcernin has quit IRC15:51
*** tremble has quit IRC15:53
jristEmilienM: what do you think of this tattoo? (joke)15:54
EmilienMjrist: you made me laugh :)15:54
EmilienMjrist: I'll have a problem with my gf15:54
jristit was rbrady's idea15:55
openstackgerritMerged openstack-infra/tripleo-ci: Add timeout to get_host_info task
rbradyjrist: the tattoo is too small, shows he is only minimally committed15:57
jristhe's committed enough to have the tattoo!15:57
*** cwolferh has joined #tripleo15:58
*** athomas has quit IRC16:02
*** mgould has joined #tripleo16:10
gfidentemarios, jistr, slagle, bnemec I wonder making network isolation the default16:11
gfidenteand singlenetwork optional env file16:11
gfidentewouldn't alone cover a good chunk of overriding issues?16:11
*** diazjf has quit IRC16:12
bnemecgfidente: The thing is that it complicates the initial setup quite a bit.16:13
mgouldhi everyone16:14
mgouldI wrote a patch recently that added a new parameter to puppet-ironic:
mgouldwhat do I need to do to enable people deploying an overcloud to use it?16:14
shardymgould: You'll need to add the option to instack-undercloud, e.g:16:17
*** dmk0202 has quit IRC16:17
openstackgerritBen Nemec proposed openstack-infra/tripleo-ci: TEST: Delete the overcloud when finished
openstackgerritBen Nemec proposed openstack-infra/tripleo-ci: Don't try to get stack details if the stack doesn't exist
*** diazjf has joined #tripleo16:18
mgouldshardy: OK, thanks. And those values come from environment variables?16:19
*** fzdarsky is now known as fzdarsky|afk16:19
shardymgould: basically yes - if you want it to be configurable then you'll add an option here:16:20
shardyand then it's wired in via converting the option variable to an env variable that's ultimately consumed by puppet via that template16:21
shardyRather, you'll add an option to and the line referenced is where the conversion happens16:21
mgouldOK, thanks!16:22
mgouldI feel better about not being able to follow the dataflow backwards now16:22
mgouldI'd got as far as puppet-stack-config.yaml, but couldn't see how to get there from undercloud.conf16:23
mgoulddoes include all the steps I'll need?16:23
mgouldand to test it I guess I'll need to set a *_REF option and run
EmilienMdprince, gfidente: please look - it's a cleanup to avoid warnings in Puppet catalog on the undercloud . I'll do overcloud next week16:26
*** myoung has quit IRC16:26
bnemecmgould: You don't need to do as much as before.  The mapping into puppet-stack-config.yaml is more automatic now:
shardybnemec: heh, I was just about to say that :)16:27
*** myoung has joined #tripleo16:27
bnemecBasically just add the opt, wire it in to puppet-stack-config.yaml, and it should be good to go.16:27
bnemecshardy: :-)16:27
*** mcornea has quit IRC16:28
mgouldbnemec, shardy: thanks!16:28
shardyre testing, I'm not sure the _REF approach works, you can build a delorean package tho with --delorean-setup && --delorean-build openstack/instack-undercloud16:28
shardyI actually hit an issue with that tho, as it seemed like the locally installed package got replaced during the openstack undercloud install, so you may need to juggle repos to ensure the locally built one gets priority16:29
*** dc has quit IRC16:29
mgouldoh dear16:30
mgouldyeah, when I was testing the puppet-ironic patch, I found I had to specify refs for every single puppet-* module or mine got overwritten16:30
* bnemec just pushes the patch then downloads the rpm from
mgouldbnemec: by setting an env var? or by running rpm manually in the undercloud?16:32
shardybnemec: heh, I didn't know about that - thanks! :)16:32
bnemecmgould: Just installing it manually.  Those rpms are versioned 999 so they shouldn't get "upgraded".16:33
mgouldOK, thanks16:33
*** florianf has quit IRC16:34
mgouldso the process is: push patch; create undercloud; ssh to undercloud; rpm install; openstack overcloud deploy?16:34
bnemecshardy: Yeah, I find it handy.  I mean, I could build rpms manually, but why should I when Jenkins can do it for me? :-)16:34
bnemecmgould: If you're changing instack-undercloud, then you need to push patch, install generated rpm, install undercloud, etc.16:35
mgouldoh, of course16:35
mgouldis this documented somewhere? I'm not having much luck with Googling16:36
openstackgerritwes hayutin proposed openstack/tripleo-quickstart: A helper script to launch quickstart from a usbkey
*** myoung has quit IRC16:39
jistrpradk: hey, sorry i totally forgot to reply to your question from earlier16:39
bnemecmgould: I think it's tribal knowledge for package testing.  There's official docs for using a proposed puppet change though:
*** myoung has joined #tripleo16:39
jristEmilienM: relevant to sharing
mgouldbnemec: do those instructions still work? I tried them about a month ago and they didn't16:41
* mgould should submit a bug16:41
EmilienMjrist: cool16:42
bnemecmgould: I recall someone saying that they only work if you set all the puppet modules to install from source.16:42
jristEmilienM: and it's by puppet!16:42
EmilienMjrist: I know him lol16:42
*** rajinir has quit IRC16:42
jristEmilienM: haha :)16:42
*** myoung has quit IRC16:42
dtrainoranyone got a few minutes to talk about or link me to info on how elements are discovered/used and in which order using diskimage-builder?  The "Developing Elements" documentation talks about what they are, but I'm more interested in how they're called, how they interact with each other etc16:42
*** myoung has joined #tripleo16:42
mgouldbnemec: yeah, that was what I found16:42
jistrpradk, slagle: i think the AODH workflow could be either of the options you mentioned. Either a separate YAML that would be run between major-upgrade-pacemaker-init.yaml and major-upgrade-pacemaker.yaml, or it could be within major-upgrade-pacemaker.yaml itself. Right now i don't see any blocker for either of those approaches (not sure what were your points for/against the respective approaches, i may have not noticed this particular topic on16:42
jistrthe ML i think).16:42
*** trown|afk is now known as trown16:43
mgouldotherwise the puppet-modules package over-writes your change16:43
dtrainorI'm asking because I'm interested in DIB_CLOUD_IMAGES and finding out which diskimage-builder element the default specification comes from, basically what tells diskimage-builder which diskimage-builder/elements/<path> is used to resolve this16:44
openstackgerritMiles Gould proposed openstack/instack-undercloud: Add option to enable introspection of UEFI nodes
mgould^^^ bnemec shardy should that cover it?16:46
jistrpradk, slagle: separate yaml provides a smaller piece of workflow that can be tested separately (it's also one more step for everyone doing the upgrade in production, but that's probably fine). Doing the upgrade within major-upgrade-pacemaker.yaml gives more options perhaps (running stuff before we do the full yum update, as well as after we do yum update but all services are still switched off, etc.).16:46
slaglejistr: i think the goal is to migrate to aodh and make sure aodh is working before the rest of the upgrade16:48
*** panda has quit IRC16:48
slagleso whichever way makes that possible16:48
*** panda has joined #tripleo16:48
slagleonce they migrate to aodh, the need to finnish the rest of the upgrade though, they shouldnt leave the cloud as liberty + aodh from mitaka16:48
slaglebut that doesnt mean it couldn't be a seperate step16:49
*** lazy_prince has quit IRC16:50
jistrpradk, slagle: yeah that sounds like a good thing to aim for. Also, migrating from a "separate yaml" solution to making it a part of major-upgrade-pacemaker.yaml should be much easier than vice versa. So maybe aiming for a separate YAML first is a good idea, and if we hit some blocker with that approach, we could consider moving to major-upgrade-pacemaker.yaml (either just partially or fully) as a plan B?16:50
openstackgerritJames Slagle proposed openstack/instack-undercloud: Add timestamps to puppet apply output
openstackgerritJames Slagle proposed openstack/instack-undercloud: Minimize element usage
openstackgerritBrent Eagles proposed openstack-infra/tripleo-ci: Exit with non-zero status on create and update errors
pradkjistr, no worries, thx for the reply. Sounds good to me as well.. i'm working on getting this in place, will run it by you and slagle for input16:54
*** lucasagomes is now known as lucas-afk16:55
jistrpradk: awesome, thanks :)16:57
*** pblaho has quit IRC16:59
*** egafford is now known as egafford|appt16:59
gfidentetty monday tripleo17:00
*** gfidente has quit IRC17:00
*** mbound has joined #tripleo17:01
*** derekh has quit IRC17:03
*** sambetts is now known as sambetts|afk17:05
*** jistr has quit IRC17:09
*** noslzzp has joined #tripleo17:12
*** jtomasek has quit IRC17:14
*** jpena is now known as jpena|off17:14
openstackgerritAthlan-Guyot sofer proposed openstack/puppet-pacemaker: WIP: integrate PCS provider in the merge.
*** ccamacho has quit IRC17:15
*** ccamacho has joined #tripleo17:16
*** abehl has quit IRC17:22
*** dtantsur is now known as dtantsur|afk17:22
*** david-lyle_ is now known as david-lyle17:22
*** bvandenh has quit IRC17:25
*** diazjf has quit IRC17:29
*** mgould is now known as mgould|afk17:41
*** diazjf has joined #tripleo17:44
*** akshai has quit IRC17:44
*** akshai has joined #tripleo17:47
*** sthillma has joined #tripleo17:53
*** chem`` has quit IRC17:57
*** diazjf has quit IRC18:01
*** jcoufal has quit IRC18:01
*** fragatina has joined #tripleo18:06
*** chem`` has joined #tripleo18:07
*** jcoufal has joined #tripleo18:07
weshaygbraad, re: we're trying to imply here that users do not have to use this script18:09
weshaynaming suggestions?18:09 ?18:09
*** fragatina has quit IRC18:10
weshaytrown, ^ any suggestions?18:11
*** ifarkas has quit IRC18:11
trownhmm... what about just RUNME18:11
trownmake it executable, call it a day :)18:11
openstackgerritwes hayutin proposed openstack/tripleo-quickstart: A helper script to launch quickstart from a usbkey
weshaytrown, done18:12
*** fragatina has joined #tripleo18:13
*** fragatina has quit IRC18:15
*** fragatina has joined #tripleo18:15
*** mikelk has joined #tripleo18:15
*** julim has quit IRC18:27
*** jcoufal has quit IRC18:29
*** zeroshft has joined #tripleo18:29
openstackgerritwes hayutin proposed openstack/tripleo-quickstart: move most things out from the playbooks dir, and use config for release info
*** saneax is now known as saneax_AFK18:32
*** zoli|wfh is now known as zoli|gone18:37
zoli|gonehave a good weekend18:37
*** _prad_ has quit IRC18:39
*** pradk has quit IRC18:41
*** zoli|gone is now known as zoli_gone-proxy18:46
*** diazjf has joined #tripleo18:46
*** yamahata has quit IRC18:48
*** diazjf has quit IRC18:49
*** dmk0202 has joined #tripleo18:50
*** rajinir has joined #tripleo18:51
openstackgerritwes hayutin proposed openstack/tripleo-quickstart: move most things out from the playbooks dir, and use config for release info
*** pradk has joined #tripleo19:19
*** julim has joined #tripleo19:20
*** yamahata has joined #tripleo19:25
*** ayoung has quit IRC19:29
*** mikelk has quit IRC19:34
*** florianf has joined #tripleo19:38
openstackgerritwes hayutin proposed openstack/tripleo-quickstart: A helper script to launch quickstart from a usbkey
*** fzdarsky|afk has quit IRC19:44
openstackgerritAndreas Florath proposed openstack/diskimage-builder: [WIP] Add: lvm element
openstackgerritMichele Baldessari proposed openstack/tripleo-heat-templates: Tighten the access rules for galera
*** eggmaster has quit IRC19:56
*** jdob has quit IRC19:58
*** jdob has joined #tripleo19:58
openstackgerritMichele Baldessari proposed openstack/tripleo-heat-templates: Tighten the access rules for galera
openstackgerritwes hayutin proposed openstack/tripleo-quickstart: move most things out from the playbooks dir, and use config for release info
bandinislagle: apologies there. was totally not aware that the changed-id had to stay the same across branches20:03
slaglebandini: np20:03
openstackgerritJohn Trowbridge proposed openstack-infra/tripleo-ci: Fix workaround to not install packages during dlrn build
*** florianf has quit IRC20:03
*** diazjf has joined #tripleo20:16
*** diazjf has quit IRC20:25
*** ayoung has joined #tripleo20:25
openstackgerritBen Nemec proposed openstack/tripleo-docs: Recommend stopping all services before major uc upgrade
*** yamahata has quit IRC20:34
*** yamahata has joined #tripleo20:35
*** apetrich has quit IRC20:43
*** mgarciam has quit IRC20:44
*** MaxPC has quit IRC20:45
*** panda has quit IRC20:48
*** panda has joined #tripleo20:48
*** apetrich has joined #tripleo20:48
*** dprince has quit IRC20:58
*** akshai has quit IRC21:06
*** zeroshft has quit IRC21:06
*** jayg is now known as jayg|g0n321:10
*** lucas-afk has quit IRC21:11
*** pradk has quit IRC21:11
*** jdob has quit IRC21:12
*** trown is now known as trown|outtypewww21:13
*** lucasagomes has joined #tripleo21:18
*** rcernin has joined #tripleo21:20
*** lblanchard has quit IRC21:34
*** dmk0202 has quit IRC21:57
*** morazi has quit IRC22:01
*** apetrich has quit IRC22:15
*** dtrainor has quit IRC22:24
*** chem`` has quit IRC22:50
*** chem`` has joined #tripleo22:51
*** rwsu has quit IRC22:56
*** krotscheck is now known as krotscheck_dcm22:57
*** marcusvrn_ has quit IRC23:13
*** dtrainor has joined #tripleo23:14
openstackgerritColleen Murphy proposed openstack/tripleo-heat-templates: Use the 'password' auth type for nova/neutron

Generated by 2.14.0 by Marius Gedminas - find it at!