Friday, 2015-04-24

*** sdake has quit IRC00:03
*** ukalifon has quit IRC00:20
*** noslzzp has joined #tripleo00:22
*** Marga_ has quit IRC00:23
*** morazi has quit IRC00:32
*** eghobo has quit IRC00:34
*** Marga_ has joined #tripleo00:45
*** Marga_ has quit IRC00:46
*** spzala has joined #tripleo00:48
*** Marga_ has joined #tripleo00:53
*** Marga_ has quit IRC00:54
*** michchap has quit IRC00:55
openstackgerritgreghaynes proposed openstack/diskimage-builder: Make backwards compatible qcow2s by default
*** alop has quit IRC01:03
*** alop has joined #tripleo01:06
*** spzala has quit IRC01:09
*** alop has quit IRC01:18
*** ChuckC has quit IRC01:21
*** sdake has joined #tripleo01:22
*** noslzzp_ has joined #tripleo01:25
*** sdake_ has quit IRC01:26
*** michchap has joined #tripleo01:26
*** wfoster has quit IRC01:26
*** wfoster has joined #tripleo01:29
*** noslzzp has quit IRC01:29
*** wfoster has quit IRC01:34
*** wfoster has joined #tripleo01:36
*** ChuckC has joined #tripleo01:49
*** sdake_ has joined #tripleo01:50
*** shardy_ has joined #tripleo01:50
*** shardy has quit IRC01:51
*** sdake has quit IRC01:53
*** ChuckC has quit IRC01:53
*** ChuckC has joined #tripleo01:54
*** shardy_ has quit IRC01:55
*** shardy has joined #tripleo01:55
*** kbyrne has quit IRC02:05
*** noslzzp_ has quit IRC02:08
*** michchap_ has joined #tripleo02:08
*** kbyrne has joined #tripleo02:10
*** michchap has quit IRC02:12
*** eghobo has joined #tripleo02:16
openstackgerritDan Prince proposed openstack-infra/tripleo-ci: puppet: use a single image for CI
*** noslzzp has joined #tripleo02:47
*** wuhg has quit IRC02:49
*** eghobo has quit IRC02:54
*** mikal has left #tripleo03:00
*** untriaged-bot has joined #tripleo03:00
untriaged-botNo untriaged bugs so far! \o/03:00
*** untriaged-bot has quit IRC03:00
*** sdake has joined #tripleo03:00
*** sdake_ has quit IRC03:04
*** sdake_ has joined #tripleo03:04
*** sdake has quit IRC03:08
*** tzumainn has quit IRC03:12
*** BadCub is now known as BadCub_Hiding03:15
*** yamahata has joined #tripleo03:23
openstackgerritSteve Baker proposed openstack/diskimage-builder: Mount with -o nouuid for XFS base images
*** eghobo has joined #tripleo04:00
*** eghobo_ has joined #tripleo04:06
*** eghobo has quit IRC04:10
*** subscope_ has joined #tripleo04:30
*** Marga_ has joined #tripleo04:41
openstackgerritMerged openstack/diskimage-builder: Update cloud-init-datasources README
*** wuhg has joined #tripleo04:52
greghaynesthat is an odd fail04:55
*** tserong has quit IRC05:02
*** eghobo_ has quit IRC05:02
*** Marga_ has quit IRC05:10
*** Marga_ has joined #tripleo05:11
*** vinkman has quit IRC05:17
*** bvandenh has quit IRC05:25
*** vinkman has joined #tripleo05:26
*** Marga_ has quit IRC05:36
*** Marga_ has joined #tripleo05:36
*** Marga_ has quit IRC05:36
*** Marga_ has joined #tripleo05:37
*** bvandenh has joined #tripleo05:40
*** Marga_ has quit IRC05:42
*** bvandenh has quit IRC05:56
*** masco has joined #tripleo05:58
*** david-ly_ has joined #tripleo06:01
*** aufi has joined #tripleo06:02
*** david-lyle has quit IRC06:03
*** subscope_ has quit IRC06:11
*** dasm|afk is now known as dasm06:11
*** lan has joined #tripleo06:13
*** ukalifon has joined #tripleo06:28
*** jprovazn has joined #tripleo06:29
*** yog__ has joined #tripleo06:30
*** tserong has joined #tripleo06:31
*** tserong has joined #tripleo06:31
*** chlong has quit IRC06:39
*** ukalifon has quit IRC06:45
jprovaznshardy, morning06:50
*** jtomasek has joined #tripleo06:53
stevebakerworking quintupleo! \o/07:01
*** dtantsur|afk is now known as dtantsur07:02
*** gfidente has joined #tripleo07:17
*** sdake_ has quit IRC07:20
*** Goneri has joined #tripleo07:24
*** yog__ has quit IRC07:26
*** ukalifon1 has joined #tripleo07:28
*** wuhg has quit IRC07:29
*** ifarkas has joined #tripleo07:29
shardyjprovazn: hi!07:32
*** ukalifon1 has quit IRC07:33
shardystevebaker: Hey, if you're still around, did dprince catch up with you about Deployment outputs?07:35
shardystevebaker: we've been puzzling over ways to get the IP address os-net-config generates from the provided index out of the node and exposed to the operator07:36
shardyI can think of a few possible approaches, but wanted to get your take on it when you have the time07:36
shardy(I can take it to the list if this is too late for you)07:36
jprovaznshardy, Hi, I'm playing with your patch but I'm quite sure I'll not have anything reviewable before afternoon :/. when doing general poll w/o restricting search to a resource subset, then traversing through all nested events is better than fetching resources and then check events for them. On the other side this approach is better when a user restricts checking to a resource subset. So maybe combination of both approaches might be ideal07:37
shardyjprovazn: Ok, no worries - feel free to push what you have to gerrit and WIP it if you'd like feedback07:38
shardyjprovazn: I'm making progress on fixing event filtering in the heat API07:38
shardyobviously we may still need the client workarounds if I can't get those into kilo07:39
jprovaznshardy, cool, I will send when I make it consumable07:40
shardyjprovazn: Ok, sounds good, thanks for looking into it! :)07:40
*** rdopiera has joined #tripleo07:41
*** jistr has joined #tripleo07:44
*** lucasagomes has joined #tripleo07:54
gfidentejistr, on your local dev env07:57
*** jcoufal has joined #tripleo07:57
gfidentejistr, did you had to hack the images to fix the puppet-concat issue?07:58
jistrgfidente: i commented out the load balancer for now (acutally i commented out everything except pacemaker to be able to focus on fixing the pacemaker hostname/IP issue)07:58
jistrbtw it requires patches to 3 repos :D07:58
gfidenteyeah I think I lost track of this indeed07:59
jistri'm not sure if it's enough priority to have it fixed *now* but since i already have the necessary bits ready, i'm going to submit them today to gerrit at least, so that it doesn't get lost07:59
gfidentejistr, but then it would work with hostnames?08:00
*** dtantsur is now known as dtantsur|brb08:00
jistryeah it would. some of that stuff might be quite invasive so i'm not sure if the CI will pass on first touch for all jobs and what will people think about the changes08:00
jistranother alternative is to patch puppet-pacemaker to somehow make it work with IPs. The word is that it would be possible probably, but there's no documentation on how to do it with Pacemaker, since it's not really a recommended setup ATM.08:01
gfidentethe loopback mapping to local hostname was actually due to cloud-init ?08:02
jistrgfidente: yes, you were right, thanks for that hint :)08:02
gfidenteI remember vaguely of a setting file we build from one of the elements08:02
gfidentebut not where it is08:03
gfidenteso I couldn't really point at anything more specific08:03
gfidente < ?08:03
jistryeah it's in the base element, in DIB itself, actually. And i think it shouldn't be in the base element, because not all images would want that setting i think. One of the 3 patches will actually address this.08:03
gfidentejistr, a while ago, I was discussing with shardy of how we could pass cloud-init params from heat08:04
gfidenteat that time, change seemed too invasive <08:05
gfidentebut more recently something along those lines came up again >08:06
jistrgfidente: well if we don't do something like this, we won't have pacemaker with hostnames :) tbh i think that should be fixed in DIB rather08:07
gfidentejistr, as in DIB should not enforce settings08:07
jistrgfidente: yes. i think setting the FQDN to localhost is wrong in more cases than it's right, so i don't think DIB should do that by default. If someone really wants that in their image, they can always enable this with their own image element.08:08
gfidentejistr, totally ack from me08:08
gfidenteso hopefully this lands soon08:08
gfidenteonly needs +A it seems?08:08
jistri'll look at that in a bit08:08
*** jistr is now known as jistr|mtg08:09
*** yog__ has joined #tripleo08:12
*** mcornea has joined #tripleo08:13
gfidentejistr|mtg, btw change for the concat pin is here:
*** mmagr has joined #tripleo08:16
*** zhiwei has joined #tripleo08:20
openstackgerritGiulio Fidente proposed openstack/tripleo-heat-templates: Map NovaEnableRbdBackend to ephemeral_storage from nova::compute::rbd
*** mmagr is now known as mmagr|afk08:34
*** derekh has joined #tripleo08:35
*** weshay has joined #tripleo08:40
*** mmagr|afk is now known as mmagr08:44
*** mmagr has quit IRC08:49
openstackgerritOpenStack Proposal Bot proposed openstack/diskimage-builder: Updated from global requirements
*** mmagr has joined #tripleo09:04
*** cody-somerville has joined #tripleo09:07
shardyjistr|mtg, gfidente: re - there's one comment re a moved link in a comment09:08
shardyI'd suggest I fix that in a follow-up patch unless there's other rework required09:08
* shardy doesn't want to lose 3 +2's for a comment09:08
gfidenteindeed :)09:09
shardyTripleO review(ers) are weird, why would you +2 something 3 times but not +A it?09:09
shardyEvidently a workflow difference between TripleO and Heat, where 2x+2 is the norm09:10
derekhshardy: maybe it was waiting on ci ?09:10
derekhshardy: just fix the link, I'll A+ it straigt away since its a trivial change09:10
shardyderekh: it passed CI 3 days ago09:10
shardyderekh: Ok, will do, thanks!09:10
*** rdopiera has quit IRC09:12
*** rdopiera has joined #tripleo09:20
*** pelix has joined #tripleo09:20
openstackgerritSteven Hardy proposed openstack/tripleo-heat-templates: Enable passing optional first-boot user-data
openstackgerritSteven Hardy proposed openstack/tripleo-heat-templates: Example of post-deploy hook doing rhel registration
openstackgerritSteven Hardy proposed openstack/tripleo-heat-templates: Add hooks for extra post-deployment config
openstackgerritMerged openstack/tripleo-heat-templates: Enable passing optional first-boot user-data
*** athomas has joined #tripleo09:25
gfidentethanks guys :)09:27
*** morazi has joined #tripleo09:31
*** jistr|mtg is now known as jistr09:32
*** jerryz has joined #tripleo09:33
*** sdake has joined #tripleo09:37
openstackgerritGiulio Fidente proposed openstack/tripleo-incubator: Announce deprecation of --no-mergepy option
*** sdake_ has joined #tripleo09:39
*** sdake has quit IRC09:43
*** jprovazn has quit IRC09:45
*** Marga_ has joined #tripleo09:48
*** Marga_ has quit IRC09:50
*** Marga_ has joined #tripleo09:50
*** therve has joined #tripleo09:58
therveJust encountered the issue fixed by
*** mmagr is now known as mmagr|afk09:59
*** mmagr|afk is now known as mmagr09:59
thervePatch seems good to go, but has been stalling, can someone have a look?09:59
*** mmagr is now known as mmagr|afk09:59
thervejdob, You put a -1 a while ago on that one10:00
openstackgerritGiulio Fidente proposed openstack/tripleo-heat-templates: Map NovaEnableRbdBackend to ephemeral_storage from nova::compute::rbd
gfidentederekh, can you check and see if we should merge or abandon?10:06
*** pdb has quit IRC10:06
*** Marga_ has quit IRC10:06
*** pdb has joined #tripleo10:06
*** mmagr|afk is now known as mmagr10:07
derekhgfidente: will do10:07
*** mmagr is now known as mmagr|afk10:09
gfidentederekh, thanks, I am fine abandoning as well, it's just that it was stalling there10:14
derekhgfidente: ack10:14
openstackgerritGiulio Fidente proposed openstack/tripleo-incubator: Announce deprecation of --no-mergepy option
*** trown|outttypeww is now known as trown10:27
*** dtantsur|brb is now known as dtantsur10:33
*** yuanying has quit IRC10:39
*** sdake_ has quit IRC10:53
*** dguerri is now known as _dguerri10:59
*** _dguerri is now known as dguerri11:00
jistrtherve: hi, thanks for bringing this up. I think the concern jdob mentioned is valid, i commented there with additional explanation.11:03
jistrtherve: in case the author doesn't update the patch, i think someone else could do it. Alternatively we could add a dependent patch on top (to keep authorship clear) which would address the issue, and then we could land both11:05
* jistr might get a peek at that today but not 100% sure11:07
*** zhiwei has quit IRC11:08
openstackgerritGiulio Fidente proposed openstack/tripleo-puppet-elements: Revert "Temporarily remove dashboard"
gfidentejistr, ^^ seems to work for me now on fc2111:10
gfidentejistr, derekh I'm using delorean /current though11:11
jistrgfidente: +2'd and waiting for CI11:12
gfidenteso hopefully we can finally land
*** mmagr|afk is now known as mmagr11:13
jistrgfidente: re delorean current -- that might be an issue... it could be solved by bumping the delorean hash in TOCI to something newer (but also working :) ) and make the t-p-e patch depend on it11:14
gfidentejistr, yeah I should probably update that, set horizon revert as depend ...11:14
openstackgerritGiulio Fidente proposed openstack-infra/tripleo-ci: Bump up delorean to allow for openstack-dashboard installation
openstackgerritGiulio Fidente proposed openstack/tripleo-puppet-elements: Revert "Temporarily remove dashboard"
gfidentewith many thanks from horizon11:19
*** dguerri is now known as _dguerri11:20
gfidentederekh, do you remember if usage of 'rhel7' in pkg-map is safe?11:21
derekhgfidente: afraid I dunno11:23
gfidenteit looks like there is something for it in DIB as DISTRO_NAME11:23
gfidentewill try11:23
*** tzumainn has joined #tripleo11:24
openstackgerritJiri Stransky proposed openstack/tripleo-image-elements: Downcase /etc/hosts entries
gfidentejistr, chimney sweep11:28
gfidenteyou only take the hard ones though11:29
jistrheh... it looked like an easy one when i picked it up11:29
jistrtwo more patches incoming, hopefully :)11:30
gfidenteoh for the glory, ImportError: No module named pathlib from httpd11:31
gfidenteafter horizon is added ...11:31
*** yamahata has quit IRC11:31
gfidentelooks like a dep missing11:31
*** _dguerri is now known as dguerri11:37
*** rhallisey has joined #tripleo11:41
*** Marga_ has joined #tripleo11:41
*** trown is now known as trown|afk11:45
thervejistr, Thanks, will look11:45
*** lucasagomes is now known as lucas-hungry11:45
*** dprince has joined #tripleo11:49
therveIndeed the test is useless11:49
openstackgerritGonéri Le Bouder proposed openstack/tripleo-incubator: add sbin in $PATH
*** funzo has joined #tripleo11:51
openstackgerritThomas Herve proposed openstack/os-collect-config: Use non-versioned auth_url for keystone
thervejistr, ^^^ should be better11:52
*** bvandenh has joined #tripleo11:54
openstackgerritDan Prince proposed openstack/tripleo-incubator: Add diskimage-builder/bin to PATH.
*** Marga_ has quit IRC11:57
openstackgerritDan Prince proposed openstack-infra/tripleo-ci: puppet: use a single image for CI
*** lblanchard has joined #tripleo11:58
*** jerryz has quit IRC12:00
jistrtherve: indeed, thanks :)12:00
*** EmilienM has quit IRC12:01
*** EmilienM has joined #tripleo12:01
*** dasm is now known as dasm|afk12:10
*** subscope has quit IRC12:11
openstackgerritJiri Stransky proposed openstack/diskimage-builder: Do not manage host entries by default
*** jayg|g0n3 is now known as jayg12:19
*** sseago has left #tripleo12:24
*** subscope has joined #tripleo12:24
dprincederekh: as we are adding some puppet jobs should we increase our ratio of Fedora nodes for now?12:25
openstackgerritJiri Stransky proposed openstack/tripleo-heat-templates: Switch FQDN/hostname entries in Heat hosts
jistrgfidente: ^ marked the 2 latest patches as WIP as i think they're a little invasive, still verifying if i can make it happen without fiddling with FQDN that much12:28
*** athomas has quit IRC12:28
openstackgerritMarek Aufart proposed openstack/python-tuskarclient: Add /v2 to TUSKAR_URL in README example
*** zhiwei has joined #tripleo12:31
*** Goneri has quit IRC12:43
jistrhmm ok so the *only* reason i fiddle with FQDN is that `puppet apply` requires `hostname -f` and `facter fqdn` to be the same. They are the same if everything is configured correctly, but in PoC environments they are not. I think a better option than forcing the machine's FQDN to be something via /etc/hosts would be to do FACTER_fqdn=`hostname -f` when running puppet apply -- it won't break anything for properly configured systems and it will12:46
jistrfix PoC environments.12:46
*** lucas-hungry is now known as lucasagomes12:46
* jistr never understood why `facter fqdn` doesn't simply take the `hostname -f` value anyway...12:46
* jistr proceeds to try and patch the puppet config element in heat-templates12:47
*** rlandy has joined #tripleo12:53
*** bvandenh has quit IRC12:53
* gfidente still prefers IPs12:58
derekhdprince: yup, we probably should12:59
*** athomas has joined #tripleo13:01
*** funzo has quit IRC13:01
dprincejistr: I've used FACTOR_fqdn before myself and it works fine...13:04
jistrgfidente: yeah but we'd be pioneers in using IPs with Pacemaker it seems, and i'm not sure that's the best thing to pioneer :) i'd wait till it becomes more "usual" way of running pacemaker13:04
jistrdprince: thanks13:05
gfidentejistr, indeed I upvoted the change about cloud-init, I think that is cleaning up things regardless of what we end up doing with corosync13:06
*** radez_g0n3 is now known as radez13:07
gfidentethe issue with the horizon packaging seems not trivial instead, we actually miss the python-pathlib in fedora13:07
gfidente bug 1215150 in Package Review "Review Request: python-pathlib - Object-oriented filesystem paths" [Medium,New] - Assigned to nobody13:07
gfidenteso I thought it would be just an update in the spec, but it isn't13:07
jistrgfidente: re change about cloud-init -- yeah we'd still need that one, but we wouldn't need the tripleo-heat-templates patch (which hardcodes FQDNs) if patching heat-config-puppet element goes fine13:08
gfidentedespite this passing
gfidenteI think there isn't much benefit in bumping up the delorean hash13:08
gfidentebecause the horizon install would still be broken13:09
jistrgfidente: ack13:09
*** masco has quit IRC13:10
*** morazi has quit IRC13:13
*** weshay has quit IRC13:13
gfidentejistr, on
gfidenteshouldn't we just remove the .novalocal domain part?13:15
jistrgfidente: maybe, but just by itself it won't do the trick, because then `hostname -f` reports just <hostname>, but hostname -f reports <hostname>.openstacklocal, so that breaks puppet too13:18
gfidenteyeah so openstacklocal is taken from resolv.conf13:19
jistrthe command reporting with .openstacklocal is `facter fqdn`13:19
zhiweiandreaf: Hi, are you Martin André?13:35
zhiweimandre: Hi13:38
gfidentejistr, so how about using systemd: in pacemaker for the services which don't require a resource agent?13:53
gfidentejistr, can you see any reason for not doing it?13:54
jistrgfidente: i'd be +1 on doing that, i think the arch doc has that
gfidentejistr, cool so we can just let puppet do the configuration AND start the service, then we add the pacemaker resource13:55
gfidentelike haproxy13:55
gfidentemakes sense?13:56
gfidenteseems easy to control with the enable_pacemaker boolean without messing too much with the manifests13:56
jistrgfidente: hmm i'm not sure about that. jayg maybe you would know? ^ is it ok to start and/or enable the service the usual way and then add it to pacemaker, or does pacemaker have to start it?13:58
jaygwell, that is fine for certain resources, but in quickstack we always ran the setup first on whatever node had the vip, which was at least in some cases because we didn't want all the nodes trying to run db_sync commands and potentially conflicting14:00
jayghaving one run first makes sure there are not collisions14:00
gfidentejayg, yeah for the db_sync we have same problem14:01
*** dguerri is now known as _dguerri14:01
jaygthere may be a few special cases, but needing one node set up first has consistently been an issue,14:02
jaygeven the github ref arch docs show it this way (granted, they don't use puppet, but they are avoiding similar issues)14:03
*** mcornea has quit IRC14:03
*** mcornea has joined #tripleo14:04
*** _dguerri is now known as dguerri14:05
*** ChuckC_ has joined #tripleo14:06
*** BadCub_Hiding is now known as BadCub14:06
*** ChuckC has quit IRC14:06
*** jp_at_hp1 has joined #tripleo14:09
jistrjayg: yeah db_sync is an issue as well, but i think gfidente meant if it's ok to do "systemctl enable <service>" and/or "systemctl start <service>" prior to doing "pcs resource create"... I vaguely recall that having the service enabled in systemd caused some issues when the nodes were restarted? (*very* vaguely actually, hopefully i'm not making this up :D)14:10
*** jtriley has joined #tripleo14:10
jaygyeah, you dont want systed trying to manage it, that'll mess things up for sure14:11
jaygrrr, *systemd*14:12
*** ChuckC_ has quit IRC14:13
jistrjayg: thanks  /cc gfidente14:15
gfidentejayg, jistr so that means we can use the systemd script, but it should not be enabled14:16
gfidenteam I reading it right?14:16
jtrileyis this the right place to ask questions about kolla?14:16
rhalliseyjtriley, #kolla14:16
jtrileyshould update their README :D14:17
jtrileyi'll send a pR14:17
*** jtriley has left #tripleo14:17
jayggfidente: there are a handful of services (like galera, rabbit, for example) that must be ocf, others are systemd start scripts, but managed by pacemaker14:17
*** mmagr is now known as mmagr|afk14:17
gfidentejayg, yeah we are using ocf for those, see
gfidentejayg, got it, thanks14:17
jistrgfidente: i think so. The takeaway as i see it is that we should use the systemd resource provider in pacemaker, but we'll probably need to pass manage_service => false to the openstack service puppet classes when using pacemaker.14:17
gfidentejistr, yeah was going to double check that14:18
openstackgerritGiulio Fidente proposed openstack/tripleo-heat-templates: Move Ceph OSD data path configuration into static ceph.yaml file
*** MasterPiece has joined #tripleo14:21
*** ChuckC_ has joined #tripleo14:28
jistrhmm it seems that the FACTER_fqdn override turned the puppet run to a noop... it succeeds but doesn't do anything14:30
* jistr investigates14:30
jayggfidente jistr: fwiw, we set manage service (for most services anyway) to true for the first pass, as many of the services need to start up for configuration purposes, but then on the second puppet run, we set it to false14:33
openstackgerritMerged openstack/tripleo-image-elements: SELinux keepalived read denials
openstackgerritMerged openstack/os-cloud-config: Updated from global requirements
openstackgerritMerged openstack/os-cloud-config: Set kernel and ramdisk ID for ironic nodes
*** Marga_ has joined #tripleo14:50
*** mmagr|afk is now known as mmagr14:53
*** julim has joined #tripleo14:54
*** untriaged-bot has joined #tripleo15:00
untriaged-botUntriaged bugs so far:15:00
openstackLaunchpad bug 1447943 in python-tuskarclient "Missing /v2 path of TUSKAR_URL example in README" [Undecided,In progress] - Assigned to Marek Aufart (maufart)15:00
*** untriaged-bot has quit IRC15:00
jistrLOL i just forgot we flipped EnablePacmaker to false15:04
*** ifarkas has quit IRC15:04
jistrthe FACTER_fqdn bit actually does seem to work15:04
* jistr loves the smell of good news on friday afternoon15:05
*** mcornea has quit IRC15:11
dprincethis shaves around (over) 10 minutes off of the puppet job
dprinceby building only 1 image instead of 215:14
gfidentedprince, custom_env was as well one of the smallest but greatest change I got to use :)15:15
openstackgerritDerek Higgins proposed openstack-infra/tripleo-ci: Abort failed cherry-pick and reverts
dprincegfidente: awesome!15:16
dprincegfidente: so yeah shardy wants to make it support custom_env's (multiple)15:16
dprincegfidente: which we can do eventually as well...15:16
*** david-ly_ is now known as david-lyle15:22
*** lucasagomes_ has joined #tripleo15:26
*** mcornea has joined #tripleo15:26
*** lucasagomes has quit IRC15:28
*** lucasagomes_ is now known as lucasagomes15:28
*** cody-somerville has quit IRC15:29
jistr[17:28:41] <openstackgerrit> Jiri Stransky proposed openstack/heat-templates: Ensure Puppet can run even on PoC net setups
*** cody-somerville has joined #tripleo15:40
openstackgerritJiri Stransky proposed openstack/diskimage-builder: Do not manage host entries by default
*** Marga_ has quit IRC15:45
*** stendulker has joined #tripleo15:48
*** cody-somerville has quit IRC15:48
*** alop has joined #tripleo15:51
*** sdake has joined #tripleo15:51
gfidentejistr, I actually prefer
gfidentewhat do you mean by hardwiring?15:52
gfidenteactually, regardless of the other change you also have for FACTER_fqdn15:52
gfidentewhat are the downs of ?15:52
*** dtantsur is now known as dtantsur|afk15:53
*** Marga_ has joined #tripleo15:55
openstackgerritPino Toscano proposed openstack/diskimage-builder: Cleanup /tmp in the guest
jistrgfidente: i think if we do that ^^ then regardless how you set up your DNS, then running `hostname -f` would always print "<something>.novalocal" rather than "<something>". Which for setups where DNS is actually configured correctly wouldn't be very desirable. I'm only about 90% sure about that though. I think the configs for determining FQDN locally can vary, but i think /etc/hosts is really high on the prio list by default, i15:57
jistrthink it can override the setup in the network (just for queries local on that machine).15:57
*** yog__ has quit IRC15:57
jistrs/local on that machine/done locally on that machine/15:57
*** pdb has quit IRC15:59
gfidentejistr, so I read 'search' and 'domain' in resolv conf have different meaning16:00
gfidentewould use a name without domainpart in /etc/hostname and /etc/hosts fix it?16:01
gfidenteI am not sure where the .novalocal comes from in the /etc/hostname file16:01
gfidenteI suppose that is cloud-init again?16:02
*** bvandenh has joined #tripleo16:02
jistrgfidente: it's from the heat templates + os-*-config, though i'm not sure what is its purpose16:03
gfidenteit's the .novalocal I am unsure16:03
gfidenteneutron seems to pass openstacklocal via dhcp16:04
gfidentesigh :(16:04
*** aufi has quit IRC16:04
*** eghobo has joined #tripleo16:04
jistr.novalocal is the hard-wired one generated from cloud-config, so maybe the aim from t-h-t was to match that, but i'm still not sure of the purpose16:04
jistri think we should be able to remove it, but if the facter_fqdn patch passes, we won't need to, as it shouldn't have any further effect on our ability to run corosync16:06
*** rdopiera has quit IRC16:06
gfidentejistr, yeah I was trying to figure if/how we could make that work better16:07
gfidentefacter_fqdn might as well merge16:07
openstackgerritDan Prince proposed openstack/tripleo-heat-templates: Add support for Glance RBD backend
*** dprince has quit IRC16:13
stendulkerjistr: Hi16:14
stendulkerjistr: Can you please have a look at this review when you find time.
stendulkerjistr: thank you.16:16
openstackgerritGiulio Fidente proposed openstack/tripleo-puppet-elements: Update pkg-map to reflect Ceph packages split on RHEL
*** MasterPiece has quit IRC16:21
*** eghobo_ has joined #tripleo16:22
*** bvandenh has quit IRC16:23
*** mcornea has quit IRC16:24
*** eghobo has quit IRC16:25
*** adrianopetrich has quit IRC16:29
*** jcoufal has quit IRC16:31
*** lblanchard has quit IRC16:32
*** lblanchard has joined #tripleo16:33
*** jcoufal has joined #tripleo16:33
*** jcoufal has quit IRC16:36
*** jistr has quit IRC16:36
openstackgerritBen Nemec proposed openstack/diskimage-builder: Add functional smoke test for disk-image-create
openstackgerritMerged openstack/tripleo-image-elements: Add openstack-keystone to pkg-map
*** lucasagomes is now known as lucas-beer16:41
*** vinkman1 has joined #tripleo16:44
*** vinkman1 has left #tripleo16:44
stendulkerbnemec: Hi16:46
stendulkerbnemec: Can you please have a look at this review when you find time.
stendulkerbnemec: Thank you.16:47
*** lucas-beer has quit IRC16:47
*** derekh has quit IRC16:48
greghaynesbnemec: thanks on
greghaynesSpamapS: can I get a review ^16:53
greghaynesor anyone else16:53
*** sdake has quit IRC16:53
bnemecI am well-motivated since I think it's blocking my other dib patches. :-)16:54
greghaynesyea, sorry about that16:54
greghaynesbut, there will be tests soon!16:54
*** bnemec changes topic to "Deploying OpenStack Using OpenStack | Summit proposed topics: |"16:54
*** bnemec is now known as beekneemech16:54
* beekneemech is a fan of more tests16:54
SpamapSgreghaynes: done16:55
greghaynesdid you see ?16:55
greghaynesSpamapS: ty16:55
openstackgerritgreghaynes proposed openstack/diskimage-builder: Initial element tests
*** pblaho has quit IRC16:57
SpamapSgreghaynes: been dancing around that one. I'll review hard on it soon. :)16:57
greghaynesuh oh16:58
greghaynesits nasty, but I couldnt find a better way16:58
*** stendulker has quit IRC16:59
*** lucas-beer has joined #tripleo16:59
*** eghobo_ has quit IRC17:07
*** mmagr has quit IRC17:08
*** eghobo has joined #tripleo17:10
*** sdake has joined #tripleo17:12
*** jp_at_hp1 has quit IRC17:15
*** dprince has joined #tripleo17:18
*** harmw has joined #tripleo17:21
harmwanyone here using Kolla?17:21
gfidentemuch ♥ everyone17:21
*** gfidente has quit IRC17:22
*** MasterPiece has joined #tripleo17:23
rhalliseyharmw, ya #kolla17:23
*** hewbrocca has quit IRC17:24
openstackgerritgreghaynes proposed openstack/diskimage-builder: Fail functests if no qemu-img is found
harmwoh ok, didn't know there ws a dedicated chnnel17:24
*** dguerri is now known as _dguerri17:24
*** harmw has left #tripleo17:25
*** _dguerri is now known as dguerri17:25
SpamapSgreghaynes: 2015-04-24 16:38:38.136 | + echo 'Warning: No qemu-img binary found, cowardly refusing to run tests.'17:26
SpamapS2015-04-24 16:38:38.136 | Warning: No qemu-img binary found, cowardly refusing to run tests.17:26
greghaynesyes, see parent patch17:26
SpamapSoh hah he's on it17:26
*** penick has joined #tripleo17:26
greghayneswe need to get
SpamapSgreghaynes: interesting17:30
*** eghobo has quit IRC17:38
*** Goneri has joined #tripleo17:40
*** prad_ has joined #tripleo17:44
*** prad_ is now known as prad17:44
*** cody-somerville has joined #tripleo17:47
*** cody-somerville has joined #tripleo17:47
*** Goneri has quit IRC17:52
*** Marga_ has quit IRC17:56
*** rhallisey has quit IRC18:00
*** rhallisey has joined #tripleo18:00
*** vinkman has quit IRC18:07
*** dguerri is now known as _dguerri18:09
*** vinkman has joined #tripleo18:13
*** vinkman has left #tripleo18:13
jdobdprince: ping18:18
dprincejdob: yo18:19
jdobhey, gfidente mentioned that pacemaker is disabled by default in the upstreadm THT templates, but I can't find the parameter to enable it. do you know what he was talking about?18:19
dprincejdob: this?
dprincejdob: the way to enable that top level (because it is a nested stack) is to use default_parameters in the resource registry18:21
jdobgah, I'm an idiot. I git pulled the wrong repo and wasnt seeing it locally18:21
jdobi'm good now, thanks dprince18:21
dprincejdob: cool, If it were me I'd enable it like this:
jdobany reason to do it that way instead of setting it like we do other config values?18:23
*** Marga_ has joined #tripleo18:26
openstackgerritDan Prince proposed openstack/tripleo-heat-templates: Add support for Glance RBD backend
*** cody-somerville has quit IRC18:40
*** jtomasek has quit IRC18:46
*** aufi has joined #tripleo18:50
*** spzala has joined #tripleo19:03
*** masco has joined #tripleo19:19
*** trown|afk is now known as trown|outttypeww19:24
*** _dguerri is now known as dguerri19:30
*** prad has quit IRC19:39
openstackgerritMerged openstack/diskimage-builder: Add functional smoke test for disk-image-create
greghaynes\O/ ^19:45
*** jdob has quit IRC20:03
mordredgreghaynes: my dib-nodepool dib patches all failed things - but I think I remember you saying somehtin gwas broke? should I just recheck?20:07
greghaynesthat merge right there ^ just unbroke things20:07
greghaynesso yes, recheck time20:07
greghaynesmordred: I could also use an infra +A to make our tests actually test things
*** aufi has quit IRC20:08
mordredI betcha I can do that20:09
openstackgerritMonty Taylor proposed openstack/diskimage-builder: Split the debootstrap functions into an element
mordredgreghaynes: so ... remember that debian kernel package name thing?20:14
mordredgreghaynes: it turns out solving it is much harder20:15
mordredgreghaynes: because ARM20:15
greghaynesdid you not see my patch?20:15
mordredgreghaynes: there is not a direct one ot one mapping of arch to kernel name20:15
mordredgreghaynes: I did - it's sadly not enough20:15
greghaynesyou just have to list them out20:15
mordredyou can't tell from arch20:15
mordredarm7 could be any one of 3 different kernel packages20:15
mordredand you just apparently kinda sortof have to know20:15
greghaynesThat sounds like were misuing arch20:16
greghaynesarch should be arm7hf, etc20:16
mordredI think it's actually a debian thing20:16
mordreddavidlenwell has a thing running on arm20:16
greghaynesnah, they are different architectures20:16
mordredand apparently there is an installer utility that figures it out or something20:16
greghayneshf is hardfloating, but theres other places where theres bigger differences20:16
greghayneslike armtiny, etc20:16
mordred13:21:45 davidlenwell | linux-image-3.2.0-4-mx5 - Linux 3.2 for Freescale i.MX51/5320:17
mordred13:21:47 davidlenwell | linux-image-3.2.0-4-omap - Linux 3.2 for TI OMAP3+20:17
mordred13:21:49 davidlenwell | linux-image-3.2.0-4-vexpress - Linux 3.2 for ARM Ltd. Versatile Express20:17
mordredthose are teh 3 kernel packages available on his system20:17
mordredand at this point my brain boggles20:17
greghaynesyea, the way arm works there are drastic changes between those versions20:17
greghaynesso you cant make a common binary20:18
greghaynesso everyone treats them as separate architectures20:18
mordredsure - but apparently all three of those show up as arm7 to debian because bonghits20:18
greghayneswait, what20:18
greghayneshow does that work20:19
mordredI don't know20:19
mordredI keep mentioning davidlenwell hoping he'll respond20:19
mordredbut he also may not like me20:19
greghaynesso debian has the per-arch images20:22
greghaynesI dont see an arm7 image20:22
greghaynesIMO arch for us should be something like OMAP4/etc. Theres a ton of varians and its a PITA to enumerate so as theres a use case I think wed just add them20:24
greghaynesand in debian we just linux-image-omap arch: omap4, repeat for omap320:24
greghayneslooks like now we just have armv and armhf20:25
greghayneswhich is definitely not gonna work20:25
greghaynesactually, how does this work at all for people on arm20:27
greghaynesoh, I bet folks overide the cloud image url to point at a specific arch20:28
*** jayg is now known as jayg|g0n320:28
mordredgreghaynes: so - honestly - I'm happy to solve this in any direction - but I have 0 arm things, so I don't really know how to make good decisions20:29
mordredgreghaynes: let me know if you think what I put in the debootstrap element is non-vomit-inducing20:29
greghaynesyes, for now I would say dont worry about it - I think the solution I mentioned should work for how we need to do this long term20:30
greghaynesjust say "this element doesnt work for arm as is"20:30
greghaynesIll review that patch after I nom nom20:30
greghaynesmordred: do we have folks who are using our stuff for arm that I could ask?20:31
*** penick has quit IRC20:31
mordredgreghaynes: no clue20:32
mordredgreghaynes: I mean, let's be honest - I don't think 32-bit is worth supporting20:32
greghaynesthe real motivation we had for supporing 32bit was tripleo ci was 32bit, but that is no longer true20:33
greghaynesthe arm stuff is potentially useful though20:33
greghaynesmaybe ill send out an email asking if anyone uses it or knows someone who does20:34
*** masco has quit IRC20:38
*** penick has joined #tripleo20:45
mordredgreghaynes, SpamapS: 166641 is updated after a merge conflict - it still has 2x+2 on it, just needs a +A trigger20:50
*** pelix has quit IRC20:52
greghayneswhoa what is that bird icon doing on there20:53
greghayneslooks like it didnt pass tripleo CI20:53
*** spzala has quit IRC20:56
*** dguerri is now known as _dguerri20:57
*** untriaged-bot has joined #tripleo21:00
untriaged-botUntriaged bugs so far:21:00
openstackLaunchpad bug 1447943 in python-tuskarclient "Missing /v2 path of TUSKAR_URL example in README" [Undecided,In progress] - Assigned to Marek Aufart (maufart)21:00
*** untriaged-bot has quit IRC21:00
*** sdake_ has joined #tripleo21:04
*** lblanchard has quit IRC21:05
*** sdake has quit IRC21:08
*** rhallisey has quit IRC21:12
*** Goneri has joined #tripleo21:14
*** sdake has joined #tripleo21:19
*** rhallisey has joined #tripleo21:22
*** sdake_ has quit IRC21:23
*** Marga_ has quit IRC21:23
openstackgerritMerged openstack/diskimage-builder: Make troubleshoot work with dracut ramdisks
davidlenwellyou guys remember me? how sweet ;)21:28
greghaynesso yea, we have this issue where we want to install the right kernel package for a given architecture21:29
greghaynesthis is a non straightforward feat on arm21:29
davidlenwellI'm learning that21:30
davidlenwellI got it in my head that I wanted to use this embedded linux switch as a bootstrap node21:30
greghaynesoh wow21:30
davidlenwellI got ironic bifrost working on it21:30
greghaynesso have you actually build arm images using dib?21:30
greghayneser, built21:31
davidlenwellno.. im not booting arm hardware with this.. its bootstrapping intel atom machines21:31
davidlenwellI have a bunch of these supermicro 8 core servers with ipmi21:31
davidlenwellthey are super low power..21:31
lifelessdavidlenwell: the thing is itself arm though?21:32
davidlenwellso I wanted an even lower power bootstrap node ..21:32
davidlenwelllifeless: yes21:32
davidlenwellits this ..
davidlenwellrunning debian wheezy21:33
davidlenwellA20 ARM Cortex A7 Dual-Core21:33
greghaynesooo neat21:33
davidlenwellI've got each port on its own network.. one goes into my ipmi network..21:34
davidlenwellone does routing for the data network21:34
*** ChuckC_ is now known as ChuckC21:34
davidlenwellthe problem I've always had with how tripleo does things is the whole dedicated node thing.. so I saught the lowest profile spec for my boot strap node in my lab21:35
davidlenwelllast night I successfully put 64 bit ubuntu images onto my entire rack from this thing..21:35
*** radez is now known as radez_g0n321:36
davidlenwellthough I generated the image with dib from another computer..21:36
davidlenwelltried it on the arm machine but dual 1ghz was super painful21:36
greghaynesyea, dib building on that little atom would be.... slow21:36
davidlenwellyeah .. after the first try.. I switched to my core7 desktop21:36
greghayneswell, I have a bunch of beaglebones, maybe I should just try making some images for it using dib21:37
davidlenwelldib is super fast when you have a lot of memory21:37
greghaynesand fast i/o21:37
greghayneswhich sd is not21:37
davidlenwellthe banana pi r1 has sata21:37
greghaynesoh, nice21:38
davidlenwellI have another one im making a cold storage array with21:38
davidlenwellgpio to turn on and off relays that power the drives.. a 5 port multiplier21:38
greghayneshaha, do you even need relays21:39
greghaynesits not *that* much power21:39
greghaynesbut thats neat though21:39
davidlenwellyeah .. the voltage is to high to go direct from gpio21:39
greghaynestotally, just use a mosfet21:39
davidlenwellI might be over thinking it21:40
davidlenwellI noticed there are no retail cold storage devices.. thought it would be rad to have a router with a cold storage backup driven by a rest api21:40
davidlenwell(this is why I can't sleep)21:40
davidlenwellgetting ironic to run on wheezy was painful.. and not because of the architecture .. mostly because sysvinit21:42
*** alop has quit IRC21:43
greghaynesmordred: you has +A
openstackgerritDan Sneddon proposed openstack/tripleo-heat-templates: Separate the network configuration per flavor.
mordreddavidlenwell: how in the _world_ could sysvinit make anything _hard_?21:51
mordredgreghaynes: \o/21:51
davidlenwellbecause I've forgotten how to use it21:51
mordredgreghaynes: now we just have to figure out debootstrap21:51
openstackgerritgreghaynes proposed openstack/diskimage-builder: Initial element tests
mordreddavidlenwell: 0_o21:51
mordreddavidlenwell: so - you write a shell script that takes a start or a stop option ...21:52
greghaynesthat element testing just found an awesome bug I think21:52
greghaynesthis bodes well21:52
mordredgreghaynes: w00t21:52
davidlenwellyeah.. I did that for all the ironic services last night and then bifrost installed21:52
greghaynesI think building on a centos7 host is broken in general21:52
mordredthat's ok - I don't care about doing that :)21:53
davidlenwelldo you think its a bug that installing ironic on older systems doens't create those shell scripts?21:53
mordredgreghaynes: building on centos7 in general - or the debootstrap element?21:53
greghaynesmordred: building on centos7 at all21:53
davidlenwellor should I just pretend everyone stopped using old distros21:53
mordreddavidlenwell: I don't think we install any service scripts for anything - but I also like to pretend that nobody uses old distros21:54
mordredof course, I'd prefer to pretend that systemd and upstart didn't exist21:54
davidlenwelljust run wheezy on your systesm21:55
davidlenwelljessie moves to systemd21:57
openstackgerritgreghaynes proposed openstack/diskimage-builder: Support arch-specific package-installs
greghaynesSpamapS: mordred curious what youall think of
mordredgreghaynes: I think it's a great idea22:01
greghaynesmordred: I thought linux-image-generic doesnt work?22:02
greghaynesbecause there is no debian package called that22:04
mordredgreghaynes: oh - I should have deleted that22:07
greghaynesah, and youre taking care of it in debootstrap22:08
greghaynesseems fine to me22:08
greghaynes'it' being installing the right kernel22:08
*** sdake_ has joined #tripleo22:11
*** _dguerri is now known as dguerri22:12
*** Marga_ has joined #tripleo22:13
greghaynesok, left that and one other comment22:14
greghaynesotherwise LGTM22:14
*** sdake has quit IRC22:15
openstackgerritMerged openstack/diskimage-builder: Port centos-minimal to yum-minimal
*** dguerri is now known as _dguerri22:22
*** sdake has joined #tripleo22:23
*** rlandy has quit IRC22:25
*** yamahata has joined #tripleo22:25
*** sdake_ has quit IRC22:27
*** dprince has quit IRC22:28
*** sdake_ has joined #tripleo22:53
SpamapSgreghaynes: I love it.22:54
SpamapSgreghaynes: also we've now confirmed we can't build on precise, right ? ;-)22:54
*** thrash is now known as thrash|wknd22:54
*** sdake__ has joined #tripleo22:55
*** sdake has quit IRC22:56
*** sdake_ has quit IRC22:59
greghaynesthats awesome23:00
clarkbI mean I confirmed that months ago and was told not a bug :P23:04
clarkbprobably the easy thing is to drop precise support23:04
greghayneswhat is really interesting is that since I rebased that patch f21 now fails23:04
*** jerryz has joined #tripleo23:33
*** Goneri has quit IRC23:34
*** ChuckC has quit IRC23:36
openstackgerritDan Sneddon proposed openstack/tripleo-heat-templates: Separate the network configuration per flavor.
*** sdake__ has quit IRC23:52
*** ChuckC has joined #tripleo23:56
*** shardy_ has joined #tripleo23:59

Generated by 2.14.0 by Marius Gedminas - find it at!