Wednesday, 2014-02-19

*** dshurtle has quit IRC00:00
SpamapSmordred: right, and I'm with lifeless.. I want "something that resolves and is meaningless to what the server does"00:00
mordredsure. that's a great goal00:00
mordredbut since nova isn't plubmed to dns, I'd like it to not subvert my attempts to set it to meaningful things00:00
SpamapSBecause otherwise you need DNS as a service and something to guarantee unique hostnames globally.. allow reservations.. etc. etc..00:00
SpamapSI seem to recall there is a non-default setting one can do to have it apply a pattern and ignore the 'name' nova boot was given00:01
*** rsachary- has joined #tripleo00:01
mordredso, actually, doesn't cloud-init set the hostname from the metadata?00:02
lifelessit does00:02
mordredone would think, then, that both of us could be satisfied based on some sort of selectable choice00:02
SpamapSright so I am digging into this right now because my review brain is jet lagged.. but I think there's a setting to make it not do that.00:02
SpamapSto make nova not set the hostname in metadata to "name"00:03
mordredSpamapS: it's certainly not a consumer-visible option that I am aware of - I believe there might be a deployer option00:03
mordredSpamapS: honest to god though, I'm not trolling, when you're not jet lagged and when lifeless is not busy, I REALLY honestly want to understand the thinking00:04
*** ftcjeff has quit IRC00:04
mordredbecase, as a consumer of cloud resources  quite regularly, it's exactly the opposite of what I want00:04
mordredso I want to understand what I might be missing00:04
SpamapSmordred: right, I"m thinking deployer option00:05
lifelessmordred: so as a DC operator, we want things like 'hostname maps to physical location'00:05
*** CaptTofu has joined #tripleo00:05
lifelessmordred: or to 'CMDB record'00:05
mordredbecause it's possible that if I reframed my thinking on my problem space, perhasp something will become easier for me00:05
lifelessmordred: for troubletickets communications etc.00:05
SpamapSmordred: and the deployer option dhcp_domain is already a little bit WTF..00:05
mordredlifeless: yah - but couldn't you set that as the name you pass to nova boot?00:05
mordred(asking for real, trying to understand space)00:05
lifelessmordred: logical service -> host that something is running on is a separate layer in DNS etc for /this reason/00:05
lifelessmordred: no, you can't, because you don't know the node the thing will deploy to.00:06
lifelessmordred: nova knows.00:06
mordredok. I think we're talking about two different things, and I'm sorry for that00:06
lifelessmordred: likewise, when you ring rackspace, and you've got 40 nodes called 'jenkins.o.o', which one will they try to fix? Ah - the UUID will tell them.00:06
* mordred was going to disengage before SpamapS kept talking :)00:06
lifelessSRV vs CNAME vs A.00:07
lifelessI'm all for switching DNS around to rotate boxes in and out, but thats got nothing to do with what the identity of the box is.00:07
lifelessphysical or virtual00:07
lifelessmordred: I know, I blame SpamapS too00:07
mordredif you set the fqdn of the host as it sees itself to something related to random or uuid or something other than its logical name00:07
mordredpuppet will break completely00:08
mordredas in, it's 100% not possible to use it00:08
lifelessmordred: puppet supports an override for this (for this reason)00:08
lifelessmordred: which for your use case as described, I would feed in in the nova metadata00:08
mordredbut I did. I said "nova boot"00:08
lifelessmordred: and then you said 'and that affects hostname' and lost00:09
mordredok. look00:09
lifelessI would pass that name in - yes, and use it for puppet - yes, have it set hostname - no.00:09
mordredI'm saying "hi, I run a large cloud based application, and the behavior causes me pain"00:09
lifelessI don't get what I want, but thats what I want.00:09
mordredand, just like the hp  pulic cloud people, you're telling me that I'm just flat wrong00:09
mordredI'm saying that there should be a way for me to ask nova to pretty please set the damned hostname00:10
mordredbecause I might have a different worldview00:10
lifelessok, sorry. I'm trying to tell you that a) I've seen that pain and b) I've seen the pain the middle house everyone other than AWS does and c) I solved that pain in a way that scales, doesn't introduce support problems00:10
lifelessand that I like00:10
lifelessmordred: you can of course to whatever-you-want to hostname in your instance00:10
mordredgreat. I'm honestly thrilled there is a better way to scale the information system00:11
mordredwhat I'm saying is that I'd like to be able to ask nova to set the hostname to what I requested, rather than what I requested + novalocal - and I'd like to not then just have to follow up the boot with "ssh host ; hostname $hostname" because that's lame00:11
mordredbut, we shodl have disengaged from this a while ago00:11
lifelessSpamapS: the setting you're looking for is #dhcp_domain=novalocal00:11
lifelesswhich is a deployer setting, one off, affects every node using that API server00:12
lifelessso essentially its per-region00:12
lifelessmordred: user metadata scripts can do that setting for you during first boot00:13
lifelessmordred: I think my point is that wearing my ops hat, the *very last* thing I ever want is two shells to look like the same machine00:14
lifelessmordred: because sudo poweroff on the wrong box is bad00:14
lifelessmordred: so I have a visceral OMG WHY WOULD YOU DO THAT reaction00:15
lifelessquite apart from anything else00:15
*** NearlyFunctional has quit IRC00:16
mordredI hear that - I really do00:19
mordredI guess the thing that irks me is that there are a bunch of simple cases that this behavior makes much harder than they need to be00:20
mordredsure, if you have 100 nodes that share a role, awesome00:20
mordreda lot of the time, I'm spinning up one server for one thing, and the fact that I have to then shell in and set the hostname even though I knew what I wanted it to be first00:21
mordredis irksome00:21
mordredif I wanted to do more complex hostname management, I'd do that myself at deployment time00:21
mordredand user metadata scripts are not helpful, because there is almost nothing I've been able to use them for anyway00:21
mordredagain - if the problem was more complex, I'm sure they'd be a great answer00:22
*** matsuhashi has joined #tripleo00:22
*** cadenzajon_ has quit IRC00:23
*** matsuhashi has quit IRC00:32
*** matsuhas_ has joined #tripleo00:34
*** cd-undercloud has joined #tripleo00:50
cd-undercloud************** overcloud complete status=0 ************00:50
*** cd-undercloud has quit IRC00:50
lifelessnow to try live migration [again]00:53
openstackgerritlifeless proposed a change to openstack/diskimage-builder: Permit using arbitrary PyPI urls.
lifelessderekh: you're in ireland right?01:13
derekhlifeless: yup01:13
lifelessderekh: not galway by chance?01:13
derekhlifeless: about 90 miles north of galway, sligo01:14
lifelessSpamapS is in galway this week01:14
derekhin the HP office? I meet a few of the openstack people at a meetup a few months back, don't remember any of the names though01:15
lifelessgreghaynes: StevenK: btw DIB_DISTRIBUTION_MIRROR doesn't update the security.u.c reference01:17
lifelessderekh: yeah01:17
lifelessderekh: giving them the old tripleo induction01:17
greghaynesah, right01:17
lifelesshmm, ramdisk doesn't use COMMON_ELEMENTS01:22
lifelessit probably should01:22
StevenKlifeless: Yes, it doesn't.01:22
*** mestery has quit IRC01:24
*** derekh has quit IRC01:24
openstackgerritlifeless proposed a change to openstack/tripleo-incubator: Use DIB_COMMON_ELEMENTS in ramdisk creation.
openstackgerritlifeless proposed a change to openstack/tripleo-incubator: Point at apt-sources element in variables.
openstackgerritlifeless proposed a change to openstack/tripleo-incubator: Use DIB_COMMON_ELEMENTS in ramdisk creation.
lifelessmordred: so01:27
lifelessValueError: I/O operation on closed file01:27
lifelesspip :/01:27
mordredyup. pip blows01:27
StevenKlifeless: I decided it was an acceptable trade-off given security updates will hit -updates too01:28
lifelessI'm guessing01:28
lifelessStevenK: you misunderstand01:28
lifelessStevenK: it still hits security01:28
lifelessStevenK: so in a behind firewall setup, it will fail01:28
StevenKlifeless: Let me see if I can write a quick patch01:30
*** mestery has joined #tripleo01:30
*** rsacharya has quit IRC01:31
StevenKHeh, it even works.01:31
*** spzala has joined #tripleo01:32
StevenKlifeless: Why apt-sources to just a set a mirror, in 73045?01:33
lifelessStevenK: well, several reasons. right now security isn't set :)01:35
openstackgerritSteve Kowalik proposed a change to openstack/diskimage-builder: Replace when setting mirror
lifelessholy crap this machine compiles fast01:36
*** nosnos has joined #tripleo01:36
lifelessabout 3 seconds to compile lxml01:36
greghayneshrm, it seems odd to do that for security.u.c01:37
greghaynesI assumed theres a reason sec.u.c is not arch.u.c01:37
StevenKI thought it was keep load off the primary mirrors for security updates, but the point is totally moot nowdays since they're both the same IPs01:39
lifelessisn't archive.u.c geodns for clouds or something?01:40
lifelesswe could - shock, horror, ask.01:40
* greghaynes gasp01:40
StevenKlifeless: That's crazy talk!01:41
* mordred hides his eyes at the shame01:52
*** ramishra has joined #tripleo01:53
*** weshay has quit IRC01:57
*** dkehn has quit IRC01:57
*** dkehn has joined #tripleo01:57
openstackgerritlifeless proposed a change to openstack/tripleo-image-elements: Restart libvirt-bin when reconfiguring.
*** lazy_prince is now known as killer_prince02:08
*** cwolferh has quit IRC02:12
*** spzala has quit IRC02:16
openstackgerritRyan Brady proposed a change to openstack/tripleo-image-elements: Adds NFS backend support for cinder-volume
lifelessthe default apparmor rules so don't fit openstack02:24
lifeless2014-02-19 00:49:37.964+0000: 2405: warning : virAuditSend:135 : Failed to send audit message virt=kvm vm="instance-00000001" uuid=fe7ed8d3-ddad-4e9f-ad9c-0b68c9622e1c vm-ctx=libvirt-fe7ed8d3-ddad-4e9f-ad9c-0b68c9622e1c img-ctx=libvirt-fe7ed8d3-ddad-4e9f-ad9c-0b68c9622e1c model=apparmor: Operation not permitted02:24
mikallifeless: so, python tutorial time...02:25
lifelessmikal: shoot02:26
mikallifeless: how do I do a singleton right in python? A metaclass?02:26
lifeless1+1= ?02:26
lifelessmikal: uhm, most people just do02:26
lifelessat global scope02:26
*** UtahDave has quit IRC02:26
lifelessmikal: there are patterns for doing fancy things, but they fail the simpleness test.02:26
mikalI started with just hiding the bits I had in a module02:26
ccrouch1(2:51:09 PM) lifeless: -> storage_config: {'swift': 100, 'cinder': 50} - 2/3 swift, 1/3 cinder02:26
ccrouch1from our conversation earlier02:26
ccrouch1i see how this support the case of swift and cinder being colocated, but how would you indicate you wanted cinder to use all available disks when its deployed on its own node and swift to use all available partitions when it is deployed on its own node?02:26
mikalBut decided that a singleton would be less ugly02:26
mikalI think you're saying plan A is more pythonic02:27
lifelessmikal: I think an object is fine02:27
lifelessmikal: for instance02:27
lifelessclass Bar: ....02:27
lifelessthe_one_true_bar = Bar()02:27
mikalOh I see02:27
mikalYeah, that would work02:27
lifelessmikal: then, if youneed to test the behaviour of the thing, you can make a new what etc02:27
lifelessmikal: but its not super fancy02:28
lifelessmikal: so its easy to understand02:28
mikalAnd avoids having more than one global variable02:28
*** newell has quit IRC02:28
lifelessccrouch1: look up one line higher ;)02:28
ccrouch1lifeless so you mean we'd have02:29
ccrouch1storage_config: {'swift': 100} and02:29
ccrouch1storage_config: {'cinder': 100}02:29
ccrouch1ah ok :-)02:29
lifelessany node is only gonig to see one storage_config key :)02:29
ccrouch1ah thats important02:30
ccrouch1thats why i was confused about moving this "higher" up in the template02:30
lifelessmikal: right02:31
ccrouch1"higher" up to me implied outside of the config for the specific node02:31
mikallifeless: thanks02:31
lifelessmikal: de nada02:31
mordredmikal: you could also do def foo(): def bar(val=Bar()): return val ; return bar()02:32
mordredmikal: but don't02:32
mikalYeah, stackforge is full of scary complex stuff02:34
mikalWhich I'd rather avoid02:35
mordredmikal: just use mongo02:37
mikalmordred: I'm working on that02:38
mikalmongo is the answer to all problems02:38
mikalCause its web scale02:38
mordredit is02:42
mordredis a fun read02:48
mordredand I want people to note I'm posting something that is a positive article about postgres02:48
greghaynesThe global thing can just get hairy if your class is complex and you want to avoid failing at import time02:48
lifelessmordred: only because you couldn'02:48
lifelesst find the same for drizzle :P02:48
lifelessok, this is driving me batty02:49
mordredthere's mark callaghan picking on mongo some too02:49
mordredjust for good measure02:49
lifelesstarget rich environment02:50
mordredit's like shooting crabs in room that's completely covered in crabs with a gun that shoots sharks02:50
lifeless| fe7ed8d3-ddad-4e9f-ad9c-0b68c9622e1c | demo                | MIGRATING | migrating  | Running     | default-net=, |02:51
lifelessstill stuck02:51
*** killer_prince is now known as lazy_prince02:52
StevenKmordred: Re: Pasting that article, are you feeling quite alright? :-P02:54
* mordred throws a crab at StevenK and picks up his shark gun02:54
StevenKmordred: Interestingly, in it's-a-small-world, one of my colleagues at C now works for Mongo02:55
mordredStevenK: yeah - I've got some old MySQL colleagues over there too02:55
lifelessStevenK: who?02:56
lifelessoh, I know02:56
* lifeless headdesks02:56
mordredmongo the company is the new MySQL, Inc, tbh - same sales tactics and everything02:56
StevenKlifeless: It's her fault I have a Mongo shirt02:56
mordredI drink coffee from Mongo mugs several times a week02:57
StevenKI have a Simpsons mug I use for tea02:57
greghaynesincidentally, I just got a great gag gift idea02:57
StevenKgreghaynes: A shark gun?02:57
greghaynesovbiously :)02:57
lifelessgreghaynes: a mongoDB CDROM ?02:58
lifelessgreghaynes: (useless on two counts...)02:58
greghayneswow, cruel02:58
StevenKI would pay money to see how mordred loads a shark gun without losing limbs02:58
StevenKlifeless: mongodb binary distribution for gentoo on 5 floppy disks?02:59
StevenKCompiled with -O4 -funroll-all-loops of course02:59
StevenKWe can call it the SIGFPE edition03:00
lifeless| 89bc404b-ed75-4f19-9fee-74666920792f | live-migration-test | BUILD     | scheduling | NOSTATE     |03:00
lifelessclearly I've broken this cloud03:00
StevenKlifeless: I was about say "Progress!" but then I saw the UUID is different03:00
lifelessyeah, its been scheduling for 10m now03:01
*** julim has quit IRC03:02
*** lazy_prince has quit IRC03:05
lifelesstwo days in a row fighting stuck vms in nova.03:05
* lifeless shakes fist03:05
lifelessshould we be using keystone v3 ?03:06
lifeless2014-02-19 03:06:28.476 3969 WARNING keystone.openstack.common.versionutils [-] Deprecated: v2 API is deprecated as of Icehouse in favor of v3 API and may be removed in K.03:06
lifelessspewing into our logs03:06
lifelessplus /opt/stack/venvs/nova/local/lib/python2.7/site-packages/amqp/ VDeprecationWarning: The auto_delete flag for exchanges has been deprecated and will be removed03:07
lifelessfrom py-amqp v1.5.0.03:07
lifelessStevenK: you got enough fodder to keep busy atm ?03:07
*** killer_prince has joined #tripleo03:12
StevenKlifeless: I have the nova patch to finish polishing, if you want to toss another at me, sounds fine03:15
lifeless| OS-EXT-SRV-ATTR:hypervisor_hostname  | overcloud-novacompute1-3zccbhid4vk6.novalocal            |03:15
lifelessok, so its one 103:15
lifelessI live migrate03:15
lifelessroot@overcloud-novacompute0-pvxacl6m7vvt:~# virsh list --all03:16
lifeless Id    Name                           State03:16
lifeless 5     instance-00000003              running03:16
lifelessso its migrated03:16
lifelessbut nova doesn't realise it03:16
lifelessjog0: seen that before ?03:16
lifelessthe good news is my ping wasn't interrupted from the vm03:17
lifeless*that* is sexy03:17
greghaynesDo you lose open connectinos?03:17
lifelesswas the flags I used03:17
lifelessgreghaynes: no, I'm sshed in and all03:17
lifelessI'd turn off hypervisor 1 to test fidelity03:17
lifelessbut I don't want to lose whateve rstate would reproduce this bug03:18
lifeless -     instance-00000003              shut off03:18
lifeless^ thats on compute103:18
lifelessso it hasn't done the cleanup03:18
lifelessthere is something odd here for sure03:18
*** rsachary- has left #tripleo03:24
mordredlifeless: that IS sexy03:26
*** CaptTofu has quit IRC03:26
mordredlifeless: what's the storage backing?03:26
lifelessmordred: local disk03:27
mordredlifeless: so - you live migrated without losing conenctions without shared storage?03:28
lifelessmordred: mostly. nova doesn't think its' finished.03:28
mordredjust wondering why everyone keeps telling me "you'll need shared storage for that"03:28
lifelessmordred: It's possible the rabbit hole here includes network traffic being routed via compute103:29
*** matsuhas_ has quit IRC03:29
lifelessmordred: because they've been wrong for at least a year03:29
StevenKHow does it copy the storage?03:29
mordredI was pretty sure they were, but I don't like directly telling them so until I've seen you say something in IRC03:29
StevenKrsync the image?03:29
lifelessnova sets up the same base image to start with03:29
lifelessthen the libvirts talk to each other to copy the blocks across03:30
lifelessonce thats synced they keep it updating in realtime as they start copying memory across, then finally cpu state and a pivot03:30
lifelessat which point the libvirt call is meant to return and we should be active03:30
StevenKlifeless: And nova is supposed to kill the source instance and then forget about it?03:32
lifelessthe default flags are03:32
lifelessStevenK: no, libvirt kills it03:33
lifelessnova is mean to undefine it03:33
lifelessnote the _LIVE which is what tells libvirt to do the right thing03:33
StevenKlifeless: I note there is a VIR_MIGRATE_UNDEFINE_SOURCE missing03:34
lifelessI'm not sure why03:34
lifelessthis is why I've been reading libvirt sources today03:34
lifelessbut, its C time now03:34
StevenKSo maybe that's why it is hanging around03:34
lifelessStevenK: the thing is why *nova* hasn't realised03:35
lifelessthat concerns me03:35
StevenKlifeless: My point is perhaps nova is expecting libvirt to do so, and it hasn't done so due to no VIR_MIGRATE_UNDEFINE_SOURCE flag03:35
lifelessStevenK: I agree that that is a good point :)03:36
lifelessStevenK: as for more patches for you, I had something. Oh right, I think pushing on xtradb cluster is probably back top of the list - or joining me on this live migration arc03:37
StevenKgreghaynes: xtradb was you, wasn't it?03:38
lifelessand jprovazn03:41
*** ramishra has quit IRC03:47
*** lsmola2 has quit IRC03:51
*** ramishra has joined #tripleo03:51
*** ramishra has quit IRC04:03
*** ramishra has joined #tripleo04:03
*** ramishra has quit IRC04:04
*** rsachary- has joined #tripleo04:04
greghaynesYes, although progress is epicly slow for me due to being new :)04:05
*** rsachary- has quit IRC04:06
greghaynesStevenK: You wanting to poke at it a bit?04:06
*** rsacharya has joined #tripleo04:07
StevenKgreghaynes: When I get undistracted by this nova patch04:12
lifelessgreghaynes: would you like to talk about it?04:17
*** ramishra has joined #tripleo04:19
*** coolsvap has joined #tripleo04:25
greghaynesoh, I got some help from SpamapS earlier. Ive also been trying to soak up some knowlege on things like Heat04:28
greghaynesty though04:29
jog0lifeless: what version of libvirt are we using?04:34
jog0gate uses 0.9.804:34
jog0because of
*** matsuhashi has joined #tripleo04:35
clarkbthough we can upgrade maybe if danpbs fix ends up in cloud archive04:36
lifelessjog0: oh god no04:36
lifelessjog0: 1.1.1-0ubuntu8.504:37
jog0lifeless: I thought 1.1.1 didn't pass tempest in gate04:43
jog0clarkb: how are you everwhere all the time?04:43
lifelessjog0: I don't know04:43
*** noslzzp has quit IRC04:44
uvirtbotLaunchpad bug 1228977 in nova "n-cpu seems to crash when running with libvirt 1.1.1 from ubuntu cloud archive" [High,Confirmed]04:45
jog0there are some deadlocks in the new libvirt04:46
clarkb1.1.1 is broken or was04:46
clarkbdanpb fixed it upstream04:46
*** ramishra_ has joined #tripleo04:47
clarkbnot sure how that falls downstream04:47
*** ramishra has quit IRC04:50
jog0lifeless: saucy?04:50
jog0clarkb lifeless: looks like the deadlock is still there04:51
*** yamahata_ has quit IRC04:51
jog0clarkb: trusty has the fix it looks like
*** akuznetsov has joined #tripleo04:56
*** yamahata_ has joined #tripleo04:56
lifelessjog0: ah04:56
lifelesswe should recommend to Ubuntu to include the fix in saucy, saucy is still supported04:57
jog0lifeless: is targeted for saucy05:03
uvirtbotLaunchpad bug 1228977 in nova "n-cpu seems to crash when running with libvirt 1.1.1 from ubuntu cloud archive" [High,Confirmed]05:03
lifelessjog0: the ci-overcloud is running this version and we didn't see issues during the few weeks it was under cloud05:03
jog0don't know hwhats going on05:03
jog0lifeless: have you run tempest against it?05:03
jog0turns out our gate is really stressful on OpenStack05:04
jog0the ci-overcloud probably is less busy and much  bigger05:04
lifelessnot yet05:04
lifelessjog0: well we're multinode05:05
lifelessjog0: 10 nodes, one client (nodepool)05:05
jog0lifeless: nodepool isn't running in a overcloud VM?05:05
lifelessjog0: its probably running in a rackspace vm05:05
jog0I mean the client05:05
lifelessjog0: what client05:05
jog010 nodes and one client (nodepool) ^05:06
lifelessnodepool is the client05:06
lifelessit makes nova api calls05:06
lifelesslike tempest would be a client05:06
lifelessthe script in works fine for me in a busy loop05:06
uvirtbotLaunchpad bug 1228977 in nova "n-cpu seems to crash when running with libvirt 1.1.1 from ubuntu cloud archive" [High,Confirmed]05:07
lifelessoccasional several second pauses05:07
lifelessthat mayjust be my wireless05:08
lifelessjog0: 1.1.1 maint tree is fixed05:10
lifelessyeah, needs an update05:11
clarkbis cloud archive fixed?05:12
lifelessclarkb: note
uvirtbotLaunchpad bug 1228977 in nova "n-cpu seems to crash when running with libvirt 1.1.1 from ubuntu cloud archive" [High,Confirmed]05:14
lifelessclarkb: looks like running with the package from -proposed would let use tell ubuntu its ok05:21
jog0horray progress05:22
jog0lifeless: thanks for updating the bug description05:24
lifelessI've pinged hallyn too05:25
lifelesshe probably needs to update it with the security fixes in .505:26
*** CaptTofu has joined #tripleo05:27
lifelessStevenK: your theory is looking likely in fact05:29
lifelessline 4299 or thereabouts in nova/virt/libvirt/driver.py05:29
jog0lifeless: thanks, looks like we may gate on libvirt 1.1x yet05:29
clarkblifeless: do we need a devstack change that does that then?05:29
lifelessclarkb: yes, theres a faq on how to test stuff from proposed05:31
lifelessclarkb: just need to do that in an experimental job05:31
*** CaptTofu has quit IRC05:32
clarkbexperimental job is a lot of work05:33
clarkbeasier to just propose change that overrides in devstack05:33
lifelessoh right, yes05:33
lifelessand tada05:34
lifelessit worked05:34
lifeless76 packets transmitted, 76 received, 0% packet loss, time 75063ms05:34
lifelessrtt min/avg/max/mdev = 0.243/0.416/0.827/0.079 ms05:34
StevenKlifeless: With VIR_MIGRATE_UNDEFINE_SOURCE?05:34
StevenKYou're welcome, I guess05:34
lifelessStevenK: thank you :)05:34
lifelessStevenK: I did end up still reading source to see05:34
lifelesswe poll libvirt, which is epic fail05:35
lifelessthere it is05:35
StevenKI was guessing based on obversed behaviour05:35
jog0lifeless: is that a live migrate working?05:35
lifelessjog0: yes05:35
jog0lifeless: most impressive05:36
lifelessStevenK: I know, strength of many eyes05:36
lifelessnow to propose a fix to nova to fix the defaults05:36
jog0lifeless: ++05:36
lifelesswait for 12 years for some old supported platform to go out of support05:36
lifelessand then fix it05:36
* jog0 looks at the time and signs off05:36
lifelessthe live migrate default also doesn't specify live migration05:37
lifelesswhich is bong, but a thing for another day05:38
* SpamapS awakens05:38
lifelessStevenK: if you're interested -05:43
clarkbI will try proposing devstack change tomorrow. I really want new libvirt05:46
SpamapSin soviet russia, new libvirt really want YOU05:49
clarkbwe can kill nova site packages venvs with it05:49
lifelessclarkb: oh, its pip installable?05:55
lifelessclarkb: why don't you just try that?05:55
clarkbthe python bindings are with newer C libvirt05:56
clarkbyou need the distro package first though05:56
StevenKlifeless: So, something about nodes: [] in the JSON.05:57
StevenKlifeless: What am I missing that we don't need defaults in the JSON?05:58
lifelessStevenK: where do we use the defaults05:58
SpamapSclarkb: the cloud archive beckons? ;)05:58
clarkbSpamapS after we convince ubuntu to backport the fix from proposed05:59
lifelessnot backport ;)05:59
*** coolsvap1 has joined #tripleo05:59
StevenKlifeless: _seed for setup-baremetal ; _testenv uses creates-nodes to get MAC address ; _undercloud also uses setup-baremetal05:59
SpamapSclarkb: ah, in that case, -proposed beckons!05:59
lifelessStevenK: testenv /makes/ the JSON06:00
SpamapSclarkb: is there a bug attached? Test it and post on the bug, that is usually all it takes to enable the copy to -updates06:00
openstackgerritRichard Su proposed a change to openstack/tripleo-incubator: Enable horizon on overcloud control node
lifelesssetup-baremetal needs the node data, not defaults.06:00
SpamapSTechnically I still have powers to do that.06:00
SpamapSthough I'm inactive in ubuntu-sru and should probably self-remove06:00
lifelessthe seed architecture baked into the config should be pulled from the testenv06:00
clarkbSpamapS see sb :)06:00
StevenKlifeless: Right. Now I get it. testenv happens before seed06:01
StevenKlifeless: So, _testenv needs defaults, then?06:01
lifelessStevenK: yes06:01
clarkbSpamapS there is and I eill write a devstack change to test it06:01
StevenKBut they don't hit the JSON, only nodes: [] gets populated?06:01
lifelessStevenK: uhm, so I think you'll find corners, but we should work through them as you do06:02
*** e0ne has joined #tripleo06:02
lifelessStevenK: in general think about it this way06:02
lifelessStevenK: I have 50 machines, possibly heterogeneous06:03
lifelessStevenK: but (for now anyhow) same arch06:03
*** coolsvap has quit IRC06:03
lifelessStevenK: I want to capture all that data in a sane machine readable form06:03
lifeless[ipmi power mac memory disk etc]06:03
lifelessStevenK: and the run devtest and make a seed vm, use it to deploy an undercloud on one real machine and a virt cloud on the reset06:03
StevenKlifeless: Right.06:04
StevenKlifeless: I think this will require toci changes06:04
lifelessStevenK: so - specifically - flavor creation probably needs to become introspective and be based on the sizes in the json and perhaps make multiple flavors06:04
StevenKAnd probably a revert of the node-cpu stuff06:04
lifelessStevenK: sure, but lets start with incubator06:04
lifelessStevenK: in the sense of defining it06:04
lifelessStevenK: with the variables as backwards compat06:05
lifelessStevenK: changing toci is straight forward [in fact, toci needs to start calling testenv to make its environments]06:05
lifeless[in the host, not in the test job]06:05
StevenKRight, which depends on the -incubator changes06:06
StevenKlifeless: So I think I'm clear enough to make a start on this tomorrow morning06:06
lifelessthus doing it in incubator first :)06:06
lifelessStevenK: wseet06:06
StevenKI may write a handler to write out the jSON06:07
lifelessStevenK: cool06:07
lifelessStevenK: there is a python script already; it could be nicer:)06:07
SpamapSlifeless: FYI, I think I understand the push toward a json file vs. env vars now btw.06:07
lifelessSpamapS: cool06:08
SpamapSIt makes me want bash to have built in json06:08
lifelessSpamapS: I didn't realise you didn't :)06:08
lifelessSpamapS: jq :/06:08
SpamapSlifeless: well I go along to get along sometimes. :)06:08
SpamapSlifeless: but now I'm like "ok I want this everywhere now"06:08
*** vkozhukalov has joined #tripleo06:10
lifelessSpamapS: heh. What brought that on? The etherpad?06:11
lifelessSpamapS: there is something I could really use a hand with06:11
lifelessSpamapS: the first entry in this checklist
SpamapSlifeless: trying to run two on one box ... config files just make more sense.06:12
SpamapSoh.. hosts entries for all others.. interesting...06:12
lifelessSpamapS: yes, this is why I've been against env variables in the broad sense, but without good answers, for a while06:13
SpamapSlifeless: I wonder.. if we got behind Designate and pushed......06:16
*** cadenzajon has joined #tripleo06:20
lifelessSpamapS: we'd have yet another yak.06:21
lifelessSpamapS: one that talks back06:21
SpamapSlifeless: so having a hosts file with all nova nodes seems like a harder thing than having DNS updated...06:22
lifelessSpamapS: I mean I'm not against deploying designate in the undercloud06:22
lifelessSpamapS: but06:22
SpamapSlifeless: having a hosts file with _most_ nodes... that is more doable.06:22
lifelessSpamapS: whats the diff ?06:22
lifelessSpamapS: I just need a couple of str joins across the defined compute nodes06:22
SpamapSlifeless: we'd have to have wait conditions that wait for boxes to update hosts files before we can do any live migration.06:23
lifelessSpamapS: when a box checks in to say its ready, that would be included06:23
lifelessSpamapS: no special machinery is needed06:23
SpamapSlifeless: I'm thinking at scale this breaks down... I'm not sure we can ever assume all compute nodes are in the same stack.06:23
lifelessSpamapS: ok, so I'm thinking of doing no-downtime deploys today :)06:24
SpamapSanother option is to just scrape all of nova into a hosts file for everybody.06:24
lifelessSpamapS: I don't care if its global06:24
SpamapSbut yeah, we can just join up all the names and addresses06:26
lifelessSpamapS: here is what we need06:26
SpamapSand then when we have multiple stacks, accept a similar string to append..06:26
SpamapSand produce one as an output..06:26
lifeless10.10.16.147 overcloud-novacompute0-pvxacl6m7vvt.novalocal overcloud-novacompute0-pvxacl6m7vvt06:26
lifelessnote the two entries06:26
SpamapSYeah I think that is pretty doable06:26
*** cadenzajon has quit IRC06:26
lifelessSpamapS: can you do it ?06:26
SpamapSlifeless: yeah, do we already have the hosts file append thing worked out?06:27
lifelessthen we can close this ticket and I can move onto the control plane with jprovzn and greg06:27
lifelessSpamapS: yes06:27
SpamapSok I thought so..06:27
lifelessSpamapS: we use it to deploy ci-overcloud06:27
lifelessSpamapS: which needs the controller hostname to work for SSL06:27
*** e0ne has quit IRC06:28
lifelessSpamapS: have we landed dans' image building fix yet ?06:29
SpamapSlifeless: I don't see it in dib's review queue06:30
SpamapSoh wrong view06:30
SpamapS ?06:30
SpamapSlooks like slagle had some stuff to address06:31
lifelessI'll do that, sec06:31
lifelessjog0: oh you too -
*** xuhanp has joined #tripleo06:32
openstackgerritRichard Su proposed a change to openstack/tripleo-image-elements: Update selinux policies for mysqld
SpamapSlifeless: ugh.. with the scaling code in it is hard to do a for loop-like-thing elsewhere...06:33
openstackgerritlifeless proposed a change to openstack/tripleo-image-elements: Stop using deprecated libvirt_type setting.
openstackgerritlifeless proposed a change to openstack/tripleo-image-elements: Enable live block migration for libvirt.
lifelessSpamapS: any reference to stuff that scales outside of the context of a scaling thing will also scale06:34
lifelessSpamapS: so a independent top level dict06:34
lifelessSpamapS: should get that list of names trivially06:34
*** matsuhashi has quit IRC06:35
SpamapSSo an Fn::Select: [ 0 , {Fn::Select: [ Fn::GetAtt: [ NovaCompute0, networks ] ] } ]  will be duplicated?06:35
lifelessIn that situation06:36
lifelessNovaCompute0 will be replicated06:36
lifelessI was assuming you'd define the list of things to lookup in one place06:36
lifelesswhich will expand to compute=[NovaCompute0,NovaCompute1, ...]06:37
*** matsuhashi has joined #tripleo06:37
SpamapSI don't think the functions can do that06:37
SpamapSlifeless: this has been my frustration with cfn language as a whole.. it does not have a notion of arrays06:37
lifelessthere's no map ?06:37
SpamapSnot that I can find06:38
SpamapSI've wanted such a thing for a while though06:38
SpamapSso it might be worth adding. :p06:38
lifelesstheres Fn::FindInMap06:38
SpamapSThose are for finding one in a map of many06:38
SpamapSwe need one for each part of a list06:39
SpamapSthe mappings are for making things portable06:39
lifelessah yes06:40
lifelesswe want to strjoin06:40
lifelessstarts ove06:40
lifelesswe want to06:40
lifelessa bunch of one line hosts definitions06:40
lifelesseach of which looks identical06:40
lifelessbut refers to a different novacomputeN06:40 will do this if there is a containing dict that matches NovaCompute006:41
SpamapSalso I believe the mappings part of CFN is pre-parsed.. so it can't refer to attributes06:41
SpamapSas I tried to do that early on with something similar06:42
SpamapSlifeless: it might not be that hard to have a map function really06:42
lifelessso really, I think a map is going to be the right way06:43
lifelessrefs or not, anything I come up with wants to map06:43
SpamapSwe can do it with this and then later have resourcegroup/instancegroup/autoscalegroup produce lists of members and that will allow the bits that use our map function to stay stable as we move to groups instead of scaling.06:43
lifelessso Join is a map of sorts06:44
SpamapSand if we can't land map in heat immediately we can fake it in but I will try really hard not do do that. ;)06:44
lifelessI'll leave it with you06:44
lifelessand turn my baleful eye on the control plane06:44
SpamapSOk.. I'm going to do a bit of waking up and breaking fast..06:44
lifelesswhich btw will need the same thing06:45
lifelessfor rabbit cluster definitions etc06:45
* lifeless tries to rember what rwsu did for that 06:45
openstackgerritlifeless proposed a change to openstack/diskimage-builder: Make max-online-resize an option.
*** rdopieralski has joined #tripleo06:50
openstackgerritlifeless proposed a change to openstack/diskimage-builder: Make max-online-resize an option.
lifelessSpamapS: fixed (IMO) ^06:51
rwsulifeless: I think that was jprovazn,
lifelessargh lol - thanks06:51
lifelessah no thats the instance side code06:52
lifelessI'm interested in the cluster definition06:52
rwsumaybe this?
lifelessrwsu: no, it would be in tripleo-heat-templates06:54
lifelesscan't see anything open there06:55
lifelessSpamapS will save the day06:55
*** e0ne has joined #tripleo06:59
*** e0ne has quit IRC07:00
*** e0ne has joined #tripleo07:01
*** mrunge has joined #tripleo07:02
*** mrunge has quit IRC07:02
*** mrunge has joined #tripleo07:04
*** rlandy has joined #tripleo07:04
*** e0ne has quit IRC07:06
*** ramishra_ has quit IRC07:12
*** ramishra has joined #tripleo07:13
*** jcooley_ has quit IRC07:17
*** jcooley_ has joined #tripleo07:17
*** jcooley_ has quit IRC07:22
*** e0ne has joined #tripleo07:24
*** ifarkas has joined #tripleo07:27
*** e0ne has quit IRC07:28
*** CaptTofu has joined #tripleo07:29
*** saju_m has joined #tripleo07:30
*** lazy_prince has joined #tripleo07:30
*** CaptTofu has quit IRC07:33
*** jtomasek has joined #tripleo07:35
*** bauzas has joined #tripleo07:47
*** e0ne has joined #tripleo07:56
*** e0ne has quit IRC08:00
*** jprovazn has joined #tripleo08:07
*** dko has joined #tripleo08:07
lifelessthat devtest run went south: 2014-02-19 08:09:35.670 5629 WARNING os_collect_config.cfn [-] Invalid URL u'http://:8000/v1/': No host supplied08:09
openstackgerritJan Provaznik proposed a change to openstack/tripleo-heat-templates: Add rabbit.cookie attribute
*** rpodolyaka has joined #tripleo08:15
*** saju_m has quit IRC08:21
jprovaznSpamapS: Hi08:22
*** ramishra has quit IRC08:24
*** jistr has joined #tripleo08:29
*** dko has quit IRC08:30
*** openstack has joined #tripleo08:42
*** jog0 has joined #tripleo08:45
*** rsacharya has quit IRC08:47
*** jcoufal has joined #tripleo08:52
*** ramishra has joined #tripleo08:57
*** nosnos_ has joined #tripleo08:59
*** nosnos has quit IRC08:59
*** cd-undercloud has joined #tripleo09:01
cd-undercloud************** overcloud complete status=1 ************09:01
*** cd-undercloud has quit IRC09:01
lifeless| 8ab277c8-62f4-4cc4-8dae-6d6db4cb143c | overcloud      | UPDATE_COMPLETE | 2014-02-19T00:24:20Z |09:02
lifelesswoner what the fail was09:02
lifelessstopped it to debug09:02
*** e0ne has joined #tripleo09:04
lifelessOriginal contents retained as /root/.ssh/known_hosts.old09:04
lifelessERROR: No server with a name or ID of 'demo' exists.09:04
lifelessIhad deleted that server09:04
*** dtantsur has joined #tripleo09:05
*** coolsvap1 has quit IRC09:06
*** rsacharya_ has joined #tripleo09:10
*** derekh has joined #tripleo09:13
*** rsacharya_ has quit IRC09:13
*** rsacharya_ has joined #tripleo09:15
*** lucasagomes has joined #tripleo09:16
*** hashar has joined #tripleo09:18
*** hashar_ has joined #tripleo09:19
*** hashar has quit IRC09:22
*** gcha has joined #tripleo09:26
*** CaptTofu has joined #tripleo09:29
*** CaptTofu has quit IRC09:34
*** coolsvap has joined #tripleo09:34
*** pbelanyi has joined #tripleo09:40
*** jp_at_hp has joined #tripleo09:42
*** jang1 has joined #tripleo09:43
*** matsuhashi has quit IRC09:46
*** nosnos_ has quit IRC09:46
*** nosnos has joined #tripleo09:47
*** matsuhashi has joined #tripleo09:47
*** pblaho has joined #tripleo09:50
*** saju_m has joined #tripleo09:58
*** pblaho has quit IRC10:07
*** pblaho has joined #tripleo10:08
*** xuhanp has quit IRC10:13
*** hashar_ is now known as hashar10:20
*** athomas has quit IRC10:21
*** martyntaylor has joined #tripleo10:27
*** akrivoka has joined #tripleo10:32
*** athomas has joined #tripleo10:33
*** cd-undercloud has joined #tripleo10:50
cd-undercloud************** overcloud complete status=1 ************10:50
*** cd-undercloud has quit IRC10:50
SpamapSweird.. pip is just... slow on these servers. :-P11:00
*** panda has quit IRC11:06
*** panda has joined #tripleo11:06
*** bauzas has quit IRC11:11
*** gcha has quit IRC11:12
*** CaptTofu has joined #tripleo11:22
*** CaptTofu has quit IRC11:26
*** matsuhashi has quit IRC11:28
*** matsuhashi has joined #tripleo11:31
*** CaptTofu has joined #tripleo11:36
*** bauzas has joined #tripleo11:38
*** ifarkas has quit IRC11:42
*** hashar is now known as hasharAW11:47
*** bauzas has quit IRC11:48
*** coolsvap has quit IRC11:57
*** kwwii has joined #tripleo11:57
*** hasharAW has quit IRC11:58
*** max_lobur has joined #tripleo12:09
*** cd-undercloud has joined #tripleo12:12
cd-undercloud************** overcloud complete status=1 ************12:12
*** cd-undercloud has quit IRC12:12
*** gcha has joined #tripleo12:13
*** ifarkas has joined #tripleo12:14
*** geerdest has joined #tripleo12:19
*** matsuhashi has quit IRC12:20
geerdestcomments in read like LTS is supported with  I'm getting conflicting info on this.  what is the story with 12.04 and ?12:21
*** matsuhashi has joined #tripleo12:25
derekhgeerdest: lifeless is saying ubuntu lts and centos is supported with devtest, I got no idea much there being used/tested though12:29
derekhanybody having problems building images on today on fedora, conflicting vim packages ? I guess some mirrors havn't updated12:30
geerdestclint is here saying it isn't supported12:33
*** hashar has joined #tripleo12:42
slaglederekh: yep, just ran into that this morning12:49
derekhslagle: thanks, its not just me so12:50
*** CaptTofu has quit IRC12:55
*** CaptTofu has joined #tripleo12:56
*** nosnos has quit IRC12:56
*** CaptTofu has quit IRC13:00
*** kwwii has quit IRC13:02
*** noslzzp has joined #tripleo13:05
slaglederekh: i filed it:
uvirtbotslagle: Error: Could not parse XML returned by HTTP Error 404: Not Found13:08
*** matsuhashi has quit IRC13:08
*** matsuhashi has joined #tripleo13:08
*** matsuhashi has quit IRC13:08
derekhslagle: cool13:13
*** gcha has quit IRC13:14
*** lucasagomes is now known as lucas-hungry13:14
*** gcha has joined #tripleo13:14
*** dprince has joined #tripleo13:19
*** gcha has quit IRC13:22
*** jdob has joined #tripleo13:23
*** kwwii has joined #tripleo13:24
*** jistr is now known as jistr|english13:31
*** kwwii has quit IRC13:31
*** rsachary- has joined #tripleo13:32
*** weshay has joined #tripleo13:33
*** cd-undercloud has joined #tripleo13:35
cd-undercloud************** overcloud complete status=1 ************13:35
*** cd-undercloud has quit IRC13:35
*** ramishra has quit IRC13:36
slaglehmm, openstack bot has gone silent13:38
*** CaptTofu has joined #tripleo13:45
*** hashar has quit IRC13:47
*** kwwii has joined #tripleo13:49
*** ftcjeff has joined #tripleo13:53
jprovaznSpamapS: ping when you are online13:55
*** CaptTofu has quit IRC13:56
SpamapSjprovazn: online now13:57
SpamapSjprovazn: (I'm in Ireland at the moment ;)13:57
SpamapSanybody else seeing broken undercloud heat.metadata_server_url ?13:57
jprovaznSpamapS: wow, this explains why I couldn't catch your usual time :)13:57
SpamapSjprovazn: :)13:57
*** CaptTofu has joined #tripleo13:58
SpamapSjprovazn: anyway, I've got a bunch of people here at the moment I need to work with .. but I'll answer in an hour or so if you have questions13:58
*** lucas-hungry is now known as lucasagomes13:58
jprovaznSpamapS: I need your heat expertise - short question (before I describe in more details): how can I get hostname of other instances in stack?13:59
SpamapSjprovazn: you can only get their addresses13:59
* jprovazn groans13:59
SpamapSjprovazn: I know.. I told you to use the hostnames. ;)13:59
SpamapSjprovazn: but you _CAN_ get their _nova_ names14:00
SpamapSjprovazn: which should be their hostnames..14:00
jprovaznSpamapS: yep, istance_name works14:00
jprovaznbut it's not their hostname14:00
SpamapSjprovazn: for that, just Fn::GetAtt: [ resource_name, name ]14:00
jprovaznit's "instance-xxxx"14:00
SpamapSno it should be the name that is used to make the hostname14:00
SpamapSanyway... bbiab14:00
*** hashar has joined #tripleo14:01
*** ramishra has joined #tripleo14:04
*** markmc has joined #tripleo14:10
*** saju_m has quit IRC14:11
*** gcha has joined #tripleo14:13
*** lblanchard has joined #tripleo14:15
*** panda has quit IRC14:15
jprovaznSpamapS: when you are back... because instance_name != hostname, I think it makes sense to set NODENAME for rabbitmq server - then any name can be used and it's up to heat template if it uses instance_name, user-defined name or anything else. What do you think?14:20
*** bcrochet has quit IRC14:21
*** e0ne has quit IRC14:23
*** julim has joined #tripleo14:23
*** bcrochet has joined #tripleo14:23
*** matty_dubs|gone is now known as matty_dubs14:27
*** lazy_prince has quit IRC14:30
*** e0ne has joined #tripleo14:31
*** jistr|english is now known as jistr14:36
*** panda has joined #tripleo14:37
*** jdob_ has joined #tripleo14:40
*** jdob has quit IRC14:41
*** jp_at_hp has quit IRC14:47
*** rsachary- has quit IRC14:48
*** jp_at_hp has joined #tripleo14:49
*** panda has quit IRC14:52
*** rsacharya_ has quit IRC14:52
*** jdob has joined #tripleo14:58
*** cd-undercloud has joined #tripleo14:58
cd-undercloud************** overcloud complete status=1 ************14:58
*** cd-undercloud has quit IRC14:58
*** jdob_ has quit IRC14:58
*** CaptTofu has quit IRC15:01
*** CaptTofu has joined #tripleo15:01
*** CaptTofu has quit IRC15:02
*** CaptTofu has joined #tripleo15:02
*** e0ne_ has joined #tripleo15:03
*** e0ne has quit IRC15:07
*** openstackgerrit has joined #tripleo15:09
*** e0ne_ has quit IRC15:19
*** e0ne has joined #tripleo15:20
openstackgerritA change was merged to openstack/python-tuskarclient: Add initial Sphinx layout and documentation
openstackgerritLadislav Smola proposed a change to openstack/tuskar-ui: Showing configuration tab of deployed overcloud
*** sseago has joined #tripleo15:24
openstackgerritA change was merged to openstack/tuskar-ui: Add a deployment scaling dialog
openstackgerritA change was merged to openstack/tuskar: Fix tuskar docs building
derekhbnemec: thanks spotted it15:27
*** coolsvap has joined #tripleo15:33
openstackgerritDan Prince proposed a change to openstack/diskimage-builder: Add libvirt-bin -> libvirtd to map-services.
*** pbelanyi has quit IRC15:44
*** slagle has quit IRC15:46
*** slagle has joined #tripleo15:47
openstackgerritJiri Tomasek proposed a change to openstack/tuskar-ui: Fix overcloud action buttons disabling
openstackgerritRadomir Dopieralski proposed a change to openstack/tuskar-ui: Actually get node profiles from the API.
openstackgerritRadomir Dopieralski proposed a change to openstack/tuskar-ui: Update the deployment design page
*** athomas has quit IRC16:03
openstackgerritRadomir Dopieralski proposed a change to openstack/tuskar-ui: Actually get node profiles from the API.
*** jprovazn has quit IRC16:06
*** edmund has joined #tripleo16:09
openstackgerritAna Krivokapic proposed a change to openstack/tuskar-ui: Add pie charts to node overview and deployment overview
openstackgerritRadomir Dopieralski proposed a change to openstack/tuskar-ui: Actually get node profiles from the API
*** athomas has joined #tripleo16:13
openstackgerritJaromir Coufal proposed a change to openstack/tuskar-ui: Fixing few styling issues
*** kwwii has quit IRC16:16
openstackgerritA change was merged to openstack/tuskar-ui: Dealing with simplified Overcloud without workflow
openstackgerritA change was merged to openstack/tuskar-ui: Create the deployment configuration form dynamically
openstackgerritJaromir Coufal proposed a change to openstack/tuskar-ui: Switched menu items into correct order
*** cd-undercloud has joined #tripleo16:21
cd-undercloud************** overcloud complete status=1 ************16:21
*** cd-undercloud has quit IRC16:21
*** rdopieralski has quit IRC16:23
openstackgerritJaromir Coufal proposed a change to openstack/tuskar-ui: Fixing few styling issues
*** jcoufal has quit IRC16:28
openstackgerritA change was merged to openstack/diskimage-builder: Correct DIB long option parsing
*** jprovazn has joined #tripleo16:32
*** mrunge has quit IRC16:37
*** spzala has joined #tripleo16:48
*** jistr has quit IRC16:50
*** dshurtle has joined #tripleo16:53
devanandacody-somerville: ping16:55
*** matty_dubs is now known as matty_dubs|lunch16:57
*** e0ne has quit IRC16:58
cody-somervilledevananda: pong16:59
*** jcooley_ has joined #tripleo16:59
devanandacody-somerville: hi! wanted to make sure you and lucasagomes can touch base re: sprint attendance. lucasagomes pinged me a few minutes ago, so i'm pinging both of you :)17:00
lucasagomescody-somerville, hi17:00
cody-somervilledevananda: Just on a call at the moment but got your e-mail this morning (have window open with half written response).17:02
cody-somervillelucasagomes: Hey there :)17:02
*** derekh has quit IRC17:03
lucasagomescody-somerville, lemme know if I need to send you something to confirm my presence17:04
SpamapSwhat sort of hackery is needed to get our tripleo-rack-booted machines on the internet?17:04
*** jcooley_ has quit IRC17:05
*** markmc has quit IRC17:06
cody-somervillelucasagomes: You're good to go. I can confirm you're RSVPed.17:07
lucasagomescody-somerville, ah awesome! thank you :)17:07
*** cwolferh has joined #tripleo17:08
*** cwolferh_ has joined #tripleo17:09
*** jcooley_ has joined #tripleo17:10
*** cadenzajon has joined #tripleo17:13
*** dshurtle has quit IRC17:18
*** cwolferh_ has quit IRC17:22
*** jcoufal has joined #tripleo17:36
*** pblaho has quit IRC17:41
*** rollerj has quit IRC17:42
*** cd-undercloud has joined #tripleo17:43
cd-undercloud************** overcloud complete status=1 ************17:43
*** cd-undercloud has quit IRC17:43
*** ramishra has quit IRC17:44
*** rollerj has joined #tripleo17:44
*** ramishra_ has joined #tripleo17:47
*** hashar has quit IRC17:48
*** matty_dubs|lunch is now known as matty_dubs17:49
*** max_lobur is now known as max_lobur_afk17:50
*** ramishra_ has quit IRC17:51
*** gcha has quit IRC17:58
openstackgerritAna Krivokapic proposed a change to openstack/tuskar-ui: Add pie charts to node overview and deployment overview
openstackgerritA change was merged to openstack/tuskar: Getting correct count and attributes from database
lifelessgeerdest: SpamapS: precise as a deployment *target*, I rather think we have to support that18:14
lifelessgeerdest: precise as a devtest host - /might/ work with UCA enabled.18:14
SpamapSlifeless: yes I would agree18:17
*** ramishra has joined #tripleo18:18
lifelessSpamapS: so, do we have map now?18:19
*** jcooley_ has quit IRC18:20
openstackgerritDevananda van der Veen proposed a change to openstack/tripleo-incubator: Add Ironic bits to _undercloud
*** ramishra has quit IRC18:23
*** geerdest has quit IRC18:35
*** jp_at_hp has quit IRC18:42
*** jang1 has quit IRC18:43
*** lucasagomes has quit IRC18:43
*** e0ne has joined #tripleo18:44
*** spzala has quit IRC18:44
*** spzala has joined #tripleo18:45
*** newell has joined #tripleo18:51
*** robsparker has quit IRC18:52
*** slagle has quit IRC18:54
*** kwwii has joined #tripleo18:56
*** slagle has joined #tripleo18:56
*** kwwii has quit IRC18:56
*** killer_prince is now known as lazy_prince18:59
*** e0ne has quit IRC19:00
*** ccrouch1 has quit IRC19:02
*** dkehn_ has joined #tripleo19:05
*** cd-undercloud has joined #tripleo19:06
cd-undercloud************** overcloud complete status=1 ************19:06
*** cd-undercloud has quit IRC19:06
*** ccrouch has joined #tripleo19:12
lifeless| NovaCompute0                    | 1bd056ca-df5b-4872-b2c3-49f9a33daec2 | ClientException: The server has either erred or is incapable of performing the requested operation. (HTTP 500) (Request-ID: req-197b9b03-c09e-4706-a788-ad98bb8983c7) | UPDATE_FAILED      | 2014-02-19T09:54:23Z |19:13
lifelessdeleting stack, starting over19:13
*** julim has quit IRC19:15
*** ramishra has joined #tripleo19:19
*** e0ne has joined #tripleo19:21
*** ramishra has quit IRC19:24
openstackgerritJames Slagle proposed a change to openstack/tripleo-image-elements: Move 99-neutronclient under neutron-source-install
openstackgerritJames Slagle proposed a change to openstack/tripleo-image-elements: Add missing +x
openstackgerritJames Slagle proposed a change to openstack/tripleo-image-elements: Add create-dir service for neutron ovs agent
openstackgerritJames Slagle proposed a change to openstack/tripleo-image-elements: Install lvm2 package for cinder
openstackgerritJames Slagle proposed a change to openstack/tripleo-image-elements: Common cinder install code
openstackgerritJames Slagle proposed a change to openstack/tripleo-image-elements: Use os-svc-restart for cinder-api
openstackgerritJames Slagle proposed a change to openstack/tripleo-image-elements: Remove unneeded chown of /var/run/nova
*** martyntaylor has left #tripleo19:28
*** sballe has quit IRC19:31
*** sballe has joined #tripleo19:32
*** sballe has quit IRC19:33
*** sballe has joined #tripleo19:34
openstackgerritAna Krivokapic proposed a change to openstack/tuskar-ui: Add MAC addresses to node details page
*** jrist has quit IRC19:35
*** sballe has quit IRC19:38
openstackgerritA change was merged to openstack/tripleo-incubator: Messaging backend selection
*** jcooley_ has joined #tripleo19:47
*** sballe has joined #tripleo19:48
wendarlifeless: got a second for fact-checking before I send some first drafts off to David Graves?19:49
lifelessof course19:49
wendarGeneric statements about TripleO components.19:50
wendarIs it (at least abstractly) accurate to say that the Tuskar UI and APIs in the undercloud substitute for Horizon and related APIs in the overcloud?19:50
wendaror "serve the same functional purpose"19:51
lifelesstuskar UI *is* horizon19:52
wendarI know.19:52
lifelessbut it doesn't substitute in any way, it complements19:52
wendarI'm trying to explain what Tuskar is.19:52
wendari.e. Tuskar UI is an expanded version of Horizon19:52
tzumainnwendar, Tuskar exists to help create an overcloud deployment19:53
lifelessTuskar (API) replaces the shell scripts we're using from the cnosole19:53
lifelessin principle anyhow19:53
tzumainn(sorry to jump in)19:53
lifelessit's an API for deploying a cloud using Heat.19:53
lifelessin the coarsest possible description19:53
lifelesstzumainn: jumping in is fine :)19:54
openstackgerritA change was merged to openstack/tripleo-image-elements: Install syslinux also in nova-baremetal
wendarPerhaps it would be helpful to see the abstract diagram I modified.19:54
tzumainnlifeless, lol, okay19:54
lifelesswendar: diagrams would help; is it on the internal wiki somewhere?19:54
wendartzumainn: always welcome to jump in :)19:54
lifelessI've just signed into the vpn19:54
wendarlifeless: I don't even have access to the wiki yet, so no :)19:54
lifelesswendar: mail me? Ouch.19:54
wendarlifeless: we're kicking things around in email still19:54
wendarlifeless, yup, will do...19:55
lifelessSpamapS: heat bug ?19:55
lifelessSpamapS: what #?19:55
wendarlifeless: draft message away, including the full text I'm looking to fact check19:59
*** lblanchard has quit IRC20:01
*** julim has joined #tripleo20:01
lifelesswendar: ok so diagram 120:01
lifelessyou say 'Tuskar and APIs' on the bottom - I'd say 'Horizon and APIs'20:01
wendarlifeless: the one I annotated dg-original?20:01
*** jrist has joined #tripleo20:01
*** dtantsur has quit IRC20:01
wendarah, ar-revised20:01
wendarlifeless: would you put Tuskar as a separate component?20:02
wendari.e. integrated with Horizon in the undercloud?20:02
*** lblanchard has joined #tripleo20:03
Ngdid /query mordred20:03
Ngwow, I even mistyped derp20:03
wendarlifeless: hmmm... call Tuskar a "Horizon plugin for the undercloud"?20:04
wendaror simply an incubated addition to Horizon, planned to ultimately be integrated into the main Horizon?20:05
openstackgerritA change was merged to openstack/tripleo-incubator: Switch over to oslosphinx
openstackgerritA change was merged to openstack/tripleo-image-elements: Set default_ephemeral_format to ext4 for seed
*** akrivoka has quit IRC20:08
*** jcooley_ has quit IRC20:08
mordredjog0: ping20:09
*** hashar has joined #tripleo20:09
*** jcooley_ has joined #tripleo20:09
*** jcooley_ has quit IRC20:13
hewbroccawendar: it's definitely a Horizon plugin20:19
hewbroccaSpecifically, it provides a specialized user experience20:19
wendarlifeless: or, Tuskar is actually the deployment APIs, and the project is developing a plugin for Horizon to interface with the Tuskar APIs20:19
hewbroccathat is tailored to the creation and management of the Overcloud app20:19
hewbroccathis is Tuskar UI, specifically20:19
hewbroccaThe main API the Tuskar UI calls is actually the Heat API20:20
hewbroccaTuskar API itself is in flux as we determine how much of it we actually need20:20
hewbroccajdob: ^^^ correct me if I'm wrong here...20:20
wendarI can also say that Tuskar effectively doesn't exist as a distinct component.20:20
wendarIt is a UI plugin to Horizon, that interfaces with Heat APIs.20:20
hewbroccaThat's probably the most accurate thing you can say20:21
hewbroccaWe think there will be a place for Tuskar-specific API, but I don't think we're entirely sure yet20:21
hewbroccacertainly we will need a CLI that lets operators script overcloud management actions20:22
dprincepleia2: I think this should fix the Fedora ssh issue:
dprincepleia2: I dev tested it locally and that covered the case where sshd was up... but the user had not yet been created20:23
*** cadenzajon has quit IRC20:26
lifelesssorry, ELOCAL20:26
lifelesscynthia now @ kindy :)20:26
lifelesswendar: ok so20:27
lifelesswendar: there are two bits to tuskar; a horizon plugin that extends horion in much the same way the nova console extends horizon - it lets you do 'tuskar' things with horizon20:28
lifelesswendar: but 'tuskar things' are 90% heat stack operations20:28
lifelesswendar: so ui is a plugin is the sum of it, adds calls to tuskar-api and some screens20:28
lifeless(and obviously calls to nova etc etc for the bits of tripleo that are using nova, neutron, whatnot.20:29
wendarlifeless: I guess the key distinction for my arch view is whether there is a running service somewhere in the system serving up tuskar-api responses that is distinct from Heat20:30
lifelesswendar: tuskar-api itself is a thin business layer that accentuates nova/neutron/heat/keystone20:30
wendari.e. what kind of moving parts are we dealing with...20:30
lifelesswendar: there is today, its not clear that there needs to be20:30
lifelesswendar: which is what hewbrocca means by20:30
lifeless09:20 < hewbrocca> Tuskar API itself is in flux as we determine how much of it we actually need20:30
lifelesswendar: so the second change I'd make in your diagram is to break out the 'BM's box20:31
lifelessbecause unlike VM's they aren't 'contained within' nova20:31
wendarlifeless: so currently (not necessarily in the future), in the undercloud, Horizon talks to a thin Tuskar API, as well as directly with nova, neutron, heat, etc?20:31
lifelessas you have Ironic there (which makes sense) I would create a stacked box, label it BM's, and an ownership arrow from Ironic to it.20:32
lifelesswendar: yes, and also to the APIs of the overcloud(s)20:32
lifelesswendar: well, i say that but really20:32
lifelesswhat we do is give the user a link to the horizon running in the overcloud20:32
wendarlifeless: yeah, it's a bit odd, since in overcloud it's really the KVM driver that "owns" the VMs, and Ironic is the bare metal equivalent to the KVM driver20:32
lifelesstuskar-API may end up doing some calls to the overcloud API during bringup20:33
lifelessthe overcloud machines - all of them- also have an ongoing functional relationship with undercloud services:20:33
lifeless - neutron for DHCP20:33
lifeless - heat for configuration metadata20:33
lifeless - nova-api for bootstrap configuration20:33
lifelessI think that needs to be represented somehow20:34
lifelessperhaps the dotted line you have is sufficient20:34
wendarso, the overcloud and undercloud aren't running separate instances of those services?20:34
lifelessbut maybe annotate it?20:34
lifelessthey are separate20:34
lifelessbut to boot a given overcloud service - say the overcloud horizon20:34
lifelessthe machine it's own is powered on over IPMI by Ironic20:34
*** dprince has quit IRC20:35
lifelessit DHCPs (served by undercloud neutron)20:35
wendarwe'll have a much more heavily annotated digram later, with explicit relationships between each components20:35
wendarthis is just a generic "overview"20:35
lifelessit makes EC2 metadata requests (served by undercloud nova API)20:35
lifelessit makes heat requests (served by undercloud heat API)20:35
wendarOk, so the BM makes use of the undercloud versions of neutron, heat, and nova-api.20:35
wendarBut, the VM in the overcloud doesn't?20:36
lifelessa VM deployed in the overcloud makes use of the matching services delivered by the overcloud20:36
lifelesse.g. it DHCPs off of Neutron20:36
lifeless(overcloud Neutron)20:36
lifelesswhat else20:37
wendarmakes sense20:37
lifelessthe uses-lines from from 'tuskar and apis' don't connect to ironic or swift, but this is probably incorrect20:37
wendarso, I think the abstraction is good enough for this diagram, but I'll make it more explicit in the full diagram20:37
lifelessIronic is talked to to register machines20:37
lifelessswift would be talked to to check storage capacity (for images, but still)20:37
wendaroh, swift was a mistake20:37
wendarI was unsure about ironic20:38
lifelesshow so ?20:38
openstackgerritJan Provaznik proposed a change to openstack/tripleo-image-elements: Update openstack config files to support rabbitmq cluster
openstackgerritJan Provaznik proposed a change to openstack/tripleo-image-elements: Fix rabbitmq env file name and log path
openstackgerritJan Provaznik proposed a change to openstack/tripleo-image-elements: Add cluster support for rabbitmq element
lifelessoh, not linking it ?20:38
wendarI mean, I meant to connect swift.20:38
wendarbut I wasn't sure about ironic20:38
wendardoes horizon talk to it directly?20:38
lifelessreading the text now20:38
wendaror, only through Nova and Heat?20:38
lifelessI believe horizon has a swift module20:38
lifelessoh thats a potential source of confusion in your prose20:39
wendarI mean does horizon talk directly to ironic?20:39
lifelessthere is no such thing as a Horizon API20:39
lifelessit will20:39
*** jcooley_ has joined #tripleo20:39
wendaryes, I already edited out Horizon APIs from the new text20:39
wendarmaybe I should just drop those boxes down to "Horizon"?20:39
wendarAren't all the APIs provide by other components already in the diagram?20:40
lifelessright, all the components have APIs20:41
lifelessthey also often have agents/engines separated out20:41
lifelesswendar: actually, I was referring to the text in your email in my 'source of confusion' reference - where you say 'though the APIs for Horizon vs Tuskar are different.'20:42
wendarlifeless: yeah deleted.20:42
wendarlifeless: is it more accurate to say "Horizon interfaces with different APIs in the overcloud and undercloud"?20:43
wendarwell, some of the same apis, but some different20:43
wendaractually, the only thing different is the addition of Ironic APIs, and (for now) Tuskar APIs20:43
lifelessYeah. Perhaps, for the avoidance of doubt, Each Horizon provides a GUI to use the APIs available in the matching cloud.20:43
lifelesswendar: and cinder20:43
lifelesswendar: possibly ceilometer etc; point is that all horizon is is a GUI - it doesn't add or delete capabilities or functionality, and it doesn't have privileged keys to access the APIs (AIUI)20:44
lifelesshewbrocca: ^ since you're around - sanity check me ?20:44
openstackgerritJaromir Coufal proposed a change to openstack/tuskar-ui: Fixing few styling issues
*** jcooley_ has quit IRC20:49
openstackgerritA change was merged to openstack/tuskar-ui: Switched menu items into correct order
*** ramishra has joined #tripleo20:50
hewbroccayes, I believe that's exactly right20:51
hewbroccathe Tuskar UI is merely an optimization of existing Horizon capabilities that make it easier for an operator to create and manage an overcloud20:51
hewbroccait's like if you were a cloud user, but you had a Heat stack/app so weird and specialized20:52
hewbroccathat you decided to write your own special Horizon plugin to make it easier to manage20:52
openstackgerritA change was merged to openstack/tripleo-image-elements: Remove rabbitmq-server from boot-stack's element-deps
*** jcooley_ has joined #tripleo20:55
wendargreat, thanks all. That's enough for a first draft. I'll do another round once I get into the detailed interface connections between all the components.20:55
hewbroccawendar: cool, can't wait to see it. Does ^^^ more or less make sense?20:56
wendarhewbrocca: it does indeed. Thanks!20:56
lifelesshewbrocca: wendar's working on an internal doc - but it probably makes sense to make a generic variant of it as well - what do you think wendar ?20:56
wendarlifeless: yes, definitely20:57
hewbroccavery cool20:57
wendarlifeless: 99.9999% is just an architecture overview of TripleO20:57
* matty_dubs would be interested in that as well21:01
matty_dubsMight help us describe/explain it more clearly upstream21:01
*** bauzas has joined #tripleo21:04
*** CaptTofu has quit IRC21:06
*** CaptTofu has joined #tripleo21:06
*** CaptTofu has quit IRC21:11
*** weshay has quit IRC21:12
*** cadenzajon has joined #tripleo21:13
*** jcooley_ has quit IRC21:17
*** jcooley_ has joined #tripleo21:18
openstackgerritDan Prince proposed a change to openstack/tripleo-incubator: Add NeutronPublicInterface arg for the undercloud
*** jcooley_ has quit IRC21:22
openstackgerritA change was merged to openstack/tripleo-incubator: Fix devtest_variables doc formatting
*** e0ne has quit IRC21:28
openstackgerritA change was merged to openstack/tuskar-ui: Fixing few styling issues
*** jcooley_ has joined #tripleo21:30
*** shakayumi has joined #tripleo21:35
openstackgerritA change was merged to openstack/tripleo-image-elements: Glance: make 'workers' configurable
*** spzala has quit IRC21:37
*** hashar has quit IRC21:38
*** jprovazn has quit IRC21:41
*** cwolferh has quit IRC21:56
*** rbrady is now known as rbrady-bbl21:56
*** jdob has quit IRC21:57
*** julim has quit IRC21:58
*** ramishra has quit IRC21:58
*** cwolferh has joined #tripleo22:02
*** cadenzajon has quit IRC22:09
*** CaptTofu has joined #tripleo22:10
*** vkozhukalov has quit IRC22:12
*** cadenzajon has joined #tripleo22:13
*** jtomasek has quit IRC22:27
*** jomara has quit IRC22:29
*** jomara has joined #tripleo22:29
*** jcooley_ has quit IRC22:31
*** jcooley_ has joined #tripleo22:31
*** jcooley_ has quit IRC22:35
*** lblanchard has quit IRC22:36
*** jcooley_ has joined #tripleo22:36
*** jcooley_ has quit IRC22:37
*** jcooley_ has joined #tripleo22:38
*** e0ne has joined #tripleo22:39
openstackgerritGregory Haynes proposed a change to openstack/tripleo-heat-templates: Add scaling size metadata to scaled mergable_types
*** jcooley_ has quit IRC22:42
*** e0ne has quit IRC22:42
*** rlandy has quit IRC22:52
openstackgerritBen Nemec proposed a change to openstack/diskimage-builder: Generalize install-packages for yum
openstackgerritBen Nemec proposed a change to openstack/diskimage-builder: Add redhat-common element
openstackgerritBen Nemec proposed a change to openstack/diskimage-builder: Use redhat-common in fedora element
openstackgerritBen Nemec proposed a change to openstack/diskimage-builder: Add RHEL 7 image element
*** julim has joined #tripleo23:08
*** matty_dubs is now known as matty_dubs|gone23:21
*** epim has joined #tripleo23:22
*** CaptTofu has quit IRC23:24
*** CaptTofu has joined #tripleo23:24
*** CaptTofu has quit IRC23:29
openstackgerritBen Nemec proposed a change to openstack/diskimage-builder: Add RHEL 7 image element
openstackgerritBen Nemec proposed a change to openstack/diskimage-builder: Add redhat-common element
openstackgerritBen Nemec proposed a change to openstack/diskimage-builder: Use redhat-common in fedora element
*** slagle has quit IRC23:31
*** CaptTofu has joined #tripleo23:32
*** slagle has joined #tripleo23:33
*** openstack has joined #tripleo23:34
*** ccrouch has quit IRC23:35
*** kfox1111 has joined #tripleo23:36
*** bauzas has quit IRC23:48
*** ftcjeff has quit IRC23:58

Generated by 2.14.0 by Marius Gedminas - find it at!