Wednesday, 2014-03-19

« Tuesday, 2014-03-18 Index Thursday, 2014-03-20 »
adam_gdevananda, https://review.openstack.org/#/c/81315/ should be pushed out soon. is there a way to trigger the job short of proposing something?00:02
*** matsuhashi has joined #openstack-ironic00:20
*** rloo has quit IRC00:24
*** rpodolyaka1 has joined #openstack-ironic00:40
*** rpodolyaka1 has quit IRC00:44
greghaynesNobodyCam: Messing with the adding to _undercloud patch?00:48
*** nosnos has joined #openstack-ironic00:51
devanandaadam_g: "recheck no bug" or "check experimental", depending on whether it's a patch in ironic's pipe or in devstack/tempest/nova00:53
*** nosnos_ has joined #openstack-ironic00:53
*** nosnos has quit IRC00:57
NobodyCamgreghaynes: yes!01:06
NobodyCambut last test didn't go well01:06
greghaynesFun - did you see my review comments from last patch?01:06
greghaynesSeems like it was back where it was a while ago :/01:06
devanandamordred: re pbr / version stuff, any thoughts on the comments here - https://review.openstack.org/#/c/81327/5/doc/source/conf.py ?01:10
*** eguz_ has quit IRC01:15
*** Lotus907efi is now known as Lotus907efi_01:17
NobodyCamgreghaynes: yes01:21
NobodyCamI did been tring to test the fix01:22
NobodyCamor I should say a fix01:22
* adam_g scratches head01:36
adam_gdevananda, not sure why, libvirt-bin is not being intstalled as it should. http://logs.openstack.org/23/81123/1/check/check-tempest-dsvm-virtual-ironic/1f9b0c0/console.html01:36
adam_ghttps://review.openstack.org/#/c/81113/ should help there, but it should still be working01:37
adam_gwifes gonna be mad in 13 minutes.. gotta run01:38
*** rpodolyaka1 has joined #openstack-ironic01:40
*** rpodolyaka1 has quit IRC01:46
adam_ghttps://review.openstack.org/#/c/81416/  <- that will help too. now im really in trouble. cya02:04
devanandaadam_g: good catch with 81416!02:23
*** shakamunyi has joined #openstack-ironic02:28
*** coolsvap has quit IRC02:39
*** eghobo has joined #openstack-ironic03:00
*** fandi has joined #openstack-ironic03:04
*** matsuhashi has quit IRC03:13
*** matsuhashi has joined #openstack-ironic03:20
*** harlowja is now known as harlowja_away03:23
*** matsuhashi has quit IRC03:25
*** lazy_prince has quit IRC03:28
*** lazy_prince has joined #openstack-ironic03:33
*** lazy_prince is now known as killer_prince03:33
*** hemna has quit IRC03:38
*** nosnos_ has quit IRC03:41
*** rpodolyaka1 has joined #openstack-ironic03:44
*** hemna has joined #openstack-ironic03:46
*** rpodolyaka1 has quit IRC03:48
*** killer_prince has quit IRC03:54
*** vkozhukalov_ has joined #openstack-ironic03:55
*** matsuhashi has joined #openstack-ironic04:23
*** nosnos has joined #openstack-ironic04:35
*** rpodolyaka1 has joined #openstack-ironic04:45
*** rpodolyaka1 has quit IRC04:49
*** hemna has quit IRC04:59
*** rameshg87 has joined #openstack-ironic05:10
*** Killer_prince has joined #openstack-ironic05:37
*** rpodolyaka1 has joined #openstack-ironic05:45
*** rpodolyaka1 has quit IRC05:50
*** romcheg has joined #openstack-ironic05:50
*** pradipta_away is now known as pradipta05:51
*** romcheg1 has joined #openstack-ironic05:52
*** romcheg has quit IRC05:54
*** saju_m has joined #openstack-ironic05:58
*** matsuhashi has quit IRC05:59
*** matsuhashi has joined #openstack-ironic06:01
*** rpodolyaka1 has joined #openstack-ironic06:01
*** romcheg1 is now known as romcheg06:24
*** mrda is now known as mrda_away06:28
*** Slower has quit IRC06:40
*** max_lobur has joined #openstack-ironic06:42
*** matsuhashi has quit IRC06:47
*** matsuhashi has joined #openstack-ironic06:50
*** Slower has joined #openstack-ironic06:53
*** rpodolyaka1 has quit IRC07:17
*** eghobo has quit IRC07:26
*** matsuhashi has quit IRC07:30
*** matsuhashi has joined #openstack-ironic07:31
*** mdurnosvistov_lt has joined #openstack-ironic07:33
*** ifarkas has quit IRC07:58
*** athomas has joined #openstack-ironic08:04
*** shakamunyi has quit IRC08:07
GheRiveromorning Ironic!08:08
romchegMorning GheRivero!08:09
GheRiverohi romcheg08:09
romchegHow are you?08:11
dtantsurmorning Ironic08:15
*** martyntaylor has joined #openstack-ironic08:16
*** romcheg has left #openstack-ironic08:19
*** rpodolyaka1 has joined #openstack-ironic08:24
*** romcheg has joined #openstack-ironic08:24
*** martyntaylor1 has joined #openstack-ironic08:24
*** mdurnosvistov_lt has quit IRC08:25
*** martyntaylor has quit IRC08:26
*** martyntaylor has joined #openstack-ironic08:27
*** yuriyz has joined #openstack-ironic08:27
*** rpodolyaka1 has quit IRC08:28
*** martyntaylor1 has quit IRC08:30
*** max_lobur has quit IRC08:31
*** shakamunyi has joined #openstack-ironic08:33
*** shakamunyi has quit IRC08:38
rameshg87good morning everyone08:52
romchegMorning dtantsur rameshg87!08:53
*** vkozhukalov_ has quit IRC08:55
openstackgerritYuriy Zveryanskyy proposed a change to openstack/ironic: Update Neutron if mac address of the port changed  https://review.openstack.org/7793909:12
*** ndipanov has joined #openstack-ironic09:12
*** rpodolyaka1 has joined #openstack-ironic09:24
*** rpodolyaka1 has quit IRC09:28
*** jrist has joined #openstack-ironic09:32
*** shakamunyi has joined #openstack-ironic09:34
*** lucasagomes has joined #openstack-ironic09:36
mordreddevananda: looking09:38
*** shakamunyi has quit IRC09:38
*** martyntaylor has quit IRC09:40
*** martyntaylor1 has joined #openstack-ironic09:41
*** lucasagomes has quit IRC09:41
*** max_lobur has joined #openstack-ironic09:45
*** max_lobur1 has joined #openstack-ironic09:46
*** mdurnosvistov_lt has joined #openstack-ironic09:48
*** max_lobur has quit IRC09:50
*** lynxman has quit IRC10:04
*** lynxman has joined #openstack-ironic10:05
mdurnosvistovMorning all! =)10:18
*** rpodolyaka1 has joined #openstack-ironic10:25
*** rpodolyaka1 has quit IRC10:29
*** vkozhukalov_ has joined #openstack-ironic10:33
*** shakamunyi has joined #openstack-ironic10:35
*** shakamunyi has quit IRC10:39
*** mdurnosvistov_lt has quit IRC10:44
*** mdurnosvistov_lt has joined #openstack-ironic10:44
*** mdurnosvistov_lt has quit IRC10:44
*** derekh has joined #openstack-ironic10:54
*** rameshg87 has left #openstack-ironic11:08
*** matsuhashi has quit IRC11:22
*** matsuhashi has joined #openstack-ironic11:24
*** rpodolyaka1 has joined #openstack-ironic11:25
*** lucasagomes has joined #openstack-ironic11:29
*** rpodolyaka1 has quit IRC11:29
*** fandi has quit IRC11:30
*** matsuhashi has quit IRC11:37
*** lucasagomes has quit IRC11:53
*** jrist has quit IRC11:53
*** pradipta is now known as pradipta_away12:14
*** rpodolyaka1 has joined #openstack-ironic12:26
*** rpodolyaka1 has quit IRC12:30
*** shakamunyi has joined #openstack-ironic12:36
*** Killer_prince has quit IRC12:37
*** shakamunyi has quit IRC12:41
agordeevgood morning, Ironic!12:44
agordeevdoes anybody know is it true that hardware discovering capabilities are out of ironic's scope?12:50
agordeevdiscovery ramdisk blueprint was superseeded by utility ramdisk, but the latter has no reference to hw discovery features12:52
agordeevhttps://blueprints.launchpad.net/ironic/+spec/discovery-ramdisk12:52
agordeevhttps://blueprints.launchpad.net/ironic/+spec/utility-ramdisk12:53
*** jrist has joined #openstack-ironic12:58
*** Manishanker has joined #openstack-ironic13:07
*** jbjohnso_ has joined #openstack-ironic13:14
*** nosnos has quit IRC13:24
*** rpodolyaka1 has joined #openstack-ironic13:26
*** jrist has quit IRC13:27
*** Manishanker has quit IRC13:28
*** rpodolyaka1 has quit IRC13:31
openstackgerritYuriy Zveryanskyy proposed a change to openstack/ironic: Fix 'spacing' parameters for periodic tasks  https://review.openstack.org/8152413:31
*** rloo has joined #openstack-ironic13:32
*** jrist has joined #openstack-ironic13:36
*** lucasagomes_ has joined #openstack-ironic13:45
*** jgrimm has joined #openstack-ironic13:53
dtantsurGot devstack + patches by adam_g (thanks!) finish on F20 \o/13:56
dtantsurThe remaining problem is SELinux vs SSH port 222213:56
dtantsurah, yes and missing net-tools package13:57
agordeevdtantsur: sounds great!13:59
dtantsurI think I'll try to make a patch for SELinux13:59
NobodyCamGood morning Ironic, says the man makin coffee14:02
*** romcheg has quit IRC14:05
*** jrist has quit IRC14:05
*** vkozhukalov_ has quit IRC14:05
*** lucasagomes_ has quit IRC14:06
agordeevNobodyCam: g'morning :)14:06
NobodyCammorning agordeev :)14:07
agordeevNobodyCam: you might know something about hardware discovery capabilities in ironic agent, don't you?14:07
*** vkozhukalov_ has joined #openstack-ironic14:08
NobodyCamwe have hareware discovery capabilities?14:08
NobodyCam:)14:08
* NobodyCam notes coffe is almost ready14:09
agordeevNobodyCam: nope, but new ironic agent has some kind of it14:09
NobodyCamif I recall correctly it was plugable for things like that14:10
NobodyCambrb14:15
*** shakamunyi has joined #openstack-ironic14:18
NobodyCami'm back w/ coffee :)14:21
*** rpodolyaka1 has joined #openstack-ironic14:27
*** rpodolyaka1 has quit IRC14:31
*** shakamunyi has quit IRC14:52
*** shakamunyi has joined #openstack-ironic14:52
*** jrist has joined #openstack-ironic14:59
*** ndipanov is now known as ndipanov_gone14:59
*** linggao has joined #openstack-ironic15:15
openstackgerritYuriy Zveryanskyy proposed a change to openstack/ironic: Fix traceback hook for avoid duplicate traces  https://review.openstack.org/8155515:23
*** mdickson has joined #openstack-ironic15:26
*** rpodolyaka1 has joined #openstack-ironic15:28
*** dwalleck has joined #openstack-ironic15:29
*** rpodolyaka1 has quit IRC15:32
NobodyCamhumm AMQP server on 127.0.0.1:5672 is unreachable: [Errno 111] ECONNREFUSED15:33
jrollmornin!15:36
jrollagordeev: do you have questions about hardware discovery?15:37
NobodyCammorning jroll :)15:37
dwalleckGoing for my Ironic + devstack install exploration fun day /fingers-crossed15:39
jrolldwalleck: if you follow that wiki page, it should work out of the box :)15:40
jrolldwalleck: https://wiki.openstack.org/wiki/Ironic/Devstack15:40
dwalleckjroll: Hoping so, but I'm trying it on the RAX cloud, which I thought still might have some problems15:41
dwalleckjroll: Not https://etherpad.openstack.org/p/IronicAndDevstackAgain? I thought this one was the latest?15:41
jrolldwalleck: hmm15:42
jrollI'm not sure15:43
jrollfwiw the first one worked for me in our dev lab15:43
jrollwith our agent stuff15:43
agordeevjroll: morning, yes i have.15:44
agordeevjroll: teeth-agent has pluggable hardware-manager, but it's not available throught rest API. I might be wrong.15:45
Shrewsdwalleck: you are correct. IronicAndDevstackAgain is the latest. however, adam_g has a doc change up for review (https://review.openstack.org/81123) with instructions. if you could follow that and give feedback, that would be most helpful15:45
jrollagordeev: are you thinking about hardware discovery, or doing things to the hardware?15:46
dwalleckShrews: Absolutely! I'll do that15:46
Shrewsexcellent15:47
agordeevjroll: right, i'm thinking at least of two different thing. Discovery and utility. Regarding hardware discovery what is the way to get it? by introducing discovery mode ?15:48
NobodyCamanyone seen this error before?15:48
NobodyCamroot@ubuntu:/var/log/rabbitmq# cat startup_log15:48
NobodyCam{error_logger,{{2014,3,19},{15,47,59}},"Too short cookie string",[]}15:48
*** lucasagomes_ has joined #openstack-ironic15:48
Shrewsi hate when my string of cookies is too short15:49
NobodyCamlol me too :)15:49
ShrewsNobodyCam: though, no, i've not seen that15:49
NobodyCami've made strings of popcorn but never cookies15:50
ShrewsNobodyCam: though google says: http://jinntech.blogspot.com/2012/02/rabbitmq-startup-and-too-short-cookie.html15:50
jrollagordeev: so there's this list_hardware_info() method: https://github.com/rackerlabs/teeth-agent/blob/master/teeth_agent/hardware.py#L8815:50
jrollagordeev: there's no API endpoint for it; we don't necessarily need it but I'm open to adding it15:51
agordeevjroll: if we going to implement discovery by introducing new one mode to agent, will it be the most preferred way for the current agent's architecture?15:51
NobodyCamShrews: no fair you asked google :-p ... Thank you :)15:51
ShrewsNobodyCam: I'm not afraid to admit that google knows way more than I do  :)15:51
jrollagordeev: how we (plan to) do hardware info is that the agent sends it to the ironic api on boot, and ironic registers the agent url to a node to be able to issue commands15:52
agordeevjroll: yes, i saw it. It used for polling external service with the fresh hw info in heartbeat manner :)15:52
jrollyep15:52
NobodyCamhehehe :)15:52
jrollI would rather not add a new mode to the agent, but just have an API endpoint to discover hardware that can be accessed in either mode15:52
jrollsomething like /v1/hardware/ to list all, and maybe something like /v1/hardware/disks/ to list just one type of hardware15:53
jrolland maybe even somehow assign a UUID to each hardware device to get only one, but I don't think that's necessary15:54
*** coolsvap has joined #openstack-ironic15:54
jrolllast week I re-architected the API stuff, so you could add a module at api.v1.hardware15:54
jrolldoes that make sense?15:54
devananda morning, all15:55
jrollheya dev15:55
NobodyCamgood morning devananda :)15:55
Shrewsmorning devananda15:56
agordeevjroll: yes, makes sense. That should be nice15:56
agordeevdevananda: morning :)15:56
russell_hjroll: I think we should eliminate the idea of "modes" in the agent and just have drivers or something15:56
jrollrussell_h: yeah, was just thinking about that15:57
jrollrussell_h: what sort of drivers? api extensions or?15:57
russell_hwell, either API extensions or just the ability to register new command handlers15:57
russell_hor both15:57
russell_hI'm +1 to putting hardware discovery in the core API15:58
agordeevjroll: new hardware api enpoint will definitely expose all the power of pluggable hardware manager15:58
jrollrussell_h: sure15:58
russell_hjroll: basically "decom" and "standby" should be drivers (or even merged into one)15:59
russell_hthe agent can load more than one at a timme15:59
russell_hand you can add more using stevedore15:59
jrollagordeev: as in manipulating hardware, as well?15:59
jrollrussell_h: yeah15:59
russell_hthe only thing with these is that ideally you would always use the HardwareManager interface to actually do stuff16:00
russell_hI guess thats a little limiting16:00
russell_hheading into the office, bbiab16:00
mdicksonHardware discovery meaning discovering the BMC or do an inventory on the server (sorry if I'm rehashing.. New to the irc dicussions)16:00
jrollmdickson: hardware inventory. disks, NICs, etc16:01
mdicksonok makes sense16:01
mdicksonIf you dont mind me asking where does stuff like sensor data and so on live.  I've been poking at Ceilometer also..16:02
jrollI'm not sure we've talked about this much16:02
jrollin our agent model, I think we would just expose an API for that16:02
jrollbut personally I haven't had this discussion, not sure about the core guys16:03
agordeevjroll: russell_h: have you ever thought about adding additional modes to agent like utility and discovery?16:03
mdicksonnods, ok. Its secondary really. Getting the discovery and provisioning cases should come first.16:03
jrollagordeev: I'd like to remove the "modes" concept16:03
russell_hagordeev: yeah, I want to kill "modes", but definitely add more functionality like that16:04
russell_hagordeev: and make it extensible via drivers16:05
* russell_h actually heads out now16:05
*** eghobo has joined #openstack-ironic16:06
*** eghobo has quit IRC16:08
*** eghobo has joined #openstack-ironic16:11
agordeevrussell_h: jroll: i think all of that deserves additional discussion. Could you represent your vision of new agent architecture to ML/etherpad/wiki?16:16
jrollagordeev: I don't think we've fully thought it through yet16:19
jrollagordeev: we'll chat today and I'll put something up by end of day thursday16:19
jrolltry to put something up* :P16:19
agordeevjroll: it would be great. thanks in advance :)16:20
jrollnp :)16:20
*** athomas has quit IRC16:21
jrollrussell_h: let's chat about this sometime today ^16:21
*** lucasagomes_ has quit IRC16:21
NobodyCambbt...brb16:22
vkozhukalov_+1 to remove the concept of modes, having just drivers will be great16:24
*** jrist has quit IRC16:25
devanandajroll: hmm, long scroll back. would you mind summarizing?16:25
jrolldevananda: agordeev was asking about exposing an api for the hardware manager, which we're going to do. also planning to remove the concept of "modes" (currently decom/standby), and instead making those drivers, loaded through stevedore, and more than one can be loaded at once16:27
jrolldevananda: we're going to formalize this and post it somewhere today or tomorrow16:27
*** rpodolyaka1 has joined #openstack-ironic16:28
devanandajroll: ++ to writing it up16:29
jroll:)16:30
jrolldevananda: in other things, I'm fixing up this instance info thing. if nova generates the configdrive and we put it in instance info - any thoughts how large that db column should be?16:31
jrollin terms of VARCHAR(n)16:31
JayFConfigdrive can contain arbitrarily included files, right?16:32
devanandayep16:32
*** martyntaylor1 has quit IRC16:32
vkozhukalov_jroll, the suggestion is that agordeev and I will think about architecture as well and will write up our thoughts and then it'll be possible to synchronize our views16:32
devanandajroll: stashing the configdrive *content* in Ironic's DB is not the right way to go, if that's what you're thinking16:32
jrollagordeev: sure16:32
*** rpodolyaka1 has quit IRC16:32
jrolldevananda: ok, yeah, we talked about putting it in swift, too. I'll go that route16:32
devananda++16:33
devanandajroll: how is nova handling user-uploaded files?16:33
jrolldevananda: I'll need to look unless comstud knows off-hand16:33
jrolldevananda: I think they go into the configdrive itself though16:33
devanandathat's my understanding as well16:34
devanandaand once n-cpu has built the configdrive, it can be shipped off to ironic16:34
jrollyeah, they must, I don't think it works any other way16:34
jrollyeah16:34
devanandai see two avenues16:36
devananda- nova.virt.ironic.driver uploads cd somewhere and passes ref to ir-api, which passes ref to util ramdisk, which downloads CD and writes to disk16:37
devananda- nova.virt.ironic.driver uploads cd to ir-api, which uploads it somewhere and passes ref to util ramdisk, ...16:37
jrollright16:37
NobodyCamquestion on a odd seed vm error.. anyone encountering a strange error with rabbit saying it started but nothing listening on port 5672?16:37
NobodyCamgah ww16:37
devanandajroll: i think the former is better. ironic's not in the business of accepting file uploads -- there are other services which already do this, and we don't need to duplicate effort16:39
devanandaso swift ++16:39
jrolldevananda: I think I'm indifferent to which method, I'm leaning toward the first, so we aren't sending a ton of data through the API16:39
jrollyeah, nova driver uploads to swift and we pass a ref around16:40
jrollcool16:40
*** shakamunyi has quit IRC16:40
jrolldevananda: so back to the original question, does VARCHAR(512) seem reasonable to you for the instance_info column?16:42
jrollmaybe even 102416:43
devanandajroll: text is used today for the other JSONEncodedDict fields16:45
jrollis it?16:47
jrollhttps://github.com/openstack/ironic/blob/master/ironic/db/sqlalchemy/models.py#L51-5416:47
jrolloh, huh, the initial migration has them as text16:49
*** dwalleck has quit IRC16:49
devanandayea, i was just looking at that16:50
devanandaa bit inconsistent :(16:50
jrollyeah :/16:50
devanandajroll: the fields are not index(ed|able) and TEXT provides more efficient off-page storage in InnoDB16:50
jrollright, I totally agree16:51
*** epim has joined #openstack-ironic16:51
jrollthe reason I made a new type in my original patch was because I thought the existing JSON types were varchar16:51
jrolland wanted better storage16:51
jroll:P16:51
devanandayep :)16:51
jrollI'll go ahead and submit another patch for the impl16:51
jrollassuming that makes sense16:52
devanandawant to correct /ironic/db/sqlalchemy/models.py#L51-54 while you're in there?16:53
devanandaseparate patch. but yes, both make sense16:53
jrollright, that's what I meant by "for the impl"16:53
vkozhukalov_jroll, russell_h what about not publishing discovery url from agent to ir-api but instead getting ip leases from dhcp and sending discovery requests to those IPs?16:53
jrollfixing up my instance_info stuff first16:53
devanandavkozhukalov_: if i follow you, that puts more dependency on the dhcp service, which i'd like to avoid16:54
jrollvkozhukalov_: how do you get which IPs have been given out over DHCP?16:54
JayFI agree. I don't want a dependency on a particular DHCP server.16:54
jrollindeed16:55
devanandaone option that i'd like ironic to eventually support is to have a fixed dhcp config -- always serve the same image for anything that PXE boots16:55
jroll+116:55
devanandaramdisk spins up, POSTs to Ironic saying "i'm here", then Ironic decides what to do. A fully provisioned node would be configured to boot-from-disk rather than -from-net16:55
russell_hthats exactly what we're trying to do16:56
devanandathat's not the only model we should support, but it makes a lot of sense for many people, and requires full decoupling from the DHCP service16:56
devanandarussell_h: right :)16:56
russell_hwe have a hard requirement _not_ to depend on manipulating DHCP16:56
russell_hbut yeah, I'd love to support both models16:57
devanandafwiw, there are folks who _want_ to manipulate DHCP16:57
russell_hyep16:57
vkozhukalov_ok, afaiu, the discovery flow is supposed to be as follows: nodes boots via PXE, then agent starts and sends put/post into ir-api, informing it that this node exists and then ir-cond sends discovery request on that url and gets hardware info. right?16:57
russell_hI actually want to as well :)16:57
russell_hwe just can't today16:57
jrollvkozhukalov_: correct16:57
jrollactually16:57
jrollvkozhukalov_: hardware is sent with the "node exists" POST16:58
jrollor at least enough hardware info to identify the node16:58
jrollthe node is already in the database at this point16:58
russell_hvkozhukalov_: ^ this. Because the server doesn't know what node it is, we're sending in a hardware inventory that we use to try to find the right node16:58
vkozhukalov_should agent send "i'm here" requests periodically?16:58
russell_heventually we could create a node if it doesn't exist, at least in some partially-filled-out state16:58
russell_halso that16:59
russell_hwe heartbeat periodically16:59
devanandajroll: at min, that should include MAC(s), cpu_arch, RAM, etc. it'd be great if it could pull in JTAG, SN, etc, too ....16:59
devanandaas i'm sure you can imagine, some deployers have a CMDB and want to easily relate that to the info in ironic17:00
jrolldevananda: I'm not sure cpu_arch and RAM help uniquely identify a machine, but I agree in general17:00
devanandare: node state, fault tracking, etc17:00
jrolland we would love to pull serial numbers etc17:00
devanandajroll: yea, those are not uniquely identifying, but worth validating17:00
jrollsure17:00
russell_hyeah, we should focus on submitting a comprehensive inventory17:00
jroll+117:00
devanandajroll: eg, if the JTAG is the same but RAM doesn't match, tell someone17:00
russell_hconductor-side, we can use the bits we care about17:00
devanandarussell_h: ++17:00
russell_hyep17:00
jrollah yeah, makes sense17:01
* russell_h gotta run again17:01
jrollhave fun17:01
*** eguz has joined #openstack-ironic17:04
*** eguz has quit IRC17:04
*** eguz has joined #openstack-ironic17:04
*** eghobo has quit IRC17:07
*** derekh has quit IRC17:07
*** dwalleck has joined #openstack-ironic17:08
*** harlowja_away is now known as harlowja17:09
openstackgerritJim Rollenhagen proposed a change to openstack/ironic: Add Node.instance_info field  https://review.openstack.org/7946617:10
*** dwalleck_ has joined #openstack-ironic17:11
*** dwalleck_ has quit IRC17:11
openstackgerritJim Rollenhagen proposed a change to openstack/ironic: Change JsonEncodedType.impl to TEXT  https://review.openstack.org/8158317:13
jrolldevananda: that's that patch for https://github.com/openstack/ironic/blob/master/ironic/db/sqlalchemy/models.py#L51-5417:13
devanandaack17:13
*** vkozhukalov_ has quit IRC17:14
*** dwalleck has quit IRC17:14
*** rloo has quit IRC17:16
*** Lotus907efi_ has quit IRC17:17
NobodyCampost bbt walkies...brb17:18
*** harlowja has quit IRC17:26
NobodyCamack17:27
NobodyCamback even17:28
*** rpodolyaka1 has joined #openstack-ironic17:29
*** rameshg87 has joined #openstack-ironic17:29
*** max_lobur1 has quit IRC17:31
*** rpodolyaka1 has quit IRC17:33
*** harlowja has joined #openstack-ironic17:36
mdicksonNew to the chat and looking for a roadmap and/or architectural overview.  Can someone shoot me a pointer to something like that?17:41
JoshNangmdickson: http://docs.openstack.org/developer/ironic/17:42
JoshNangSystem Architecture should give you a decent overview17:42
mdicksonTY17:42
JoshNangand the rest of the links are very helpful too :)17:42
NobodyCammdickson: welcone and there is this https://wiki.openstack.org/wiki/Ironic17:42
mdicksonTYVM, starts reading :-)17:42
jrolland welcome!17:42
NobodyCam:)17:43
rameshg87hello everyone ..17:46
rameshg87i encountered a small issue which i feel is a bug ..17:46
NobodyCamhey rameshg87 :) good morning17:46
rameshg87good morning NobodyCam:17:47
NobodyCamrameshg87: what is the issue17:47
rameshg87i added a node in which i gave a wrong IP due to which the next sync_power_state failed17:47
openstackgerritJim Rollenhagen proposed a change to openstack/ironic: Add Node.instance_info field  https://review.openstack.org/7946617:47
rameshg87it set the power status of the node to error17:47
NobodyCamrameshg87: real hardware or vm?17:48
rameshg87now i can't delete the node due to this check in conductor/manager.py17:48
rameshg87655             if node.power_state not in [states.POWER_OFF, states.NOSTATE]:17:48
rameshg87656                 msg = (_("Node %s can't be deleted because it's not "17:49
rameshg87NobodyCam: real hardware17:49
NobodyCamrameshg87: have you tried ironic node-set-power-state $IRONIC_NODE_ID off17:49
rameshg87NobodyCam: i guess even that will fail because i have provided the wrong IP17:50
rameshg87the IP is unreachable :-)17:50
rameshg87i accidently gave a wrong IP while adding the node17:50
NobodyCamoh the power management ip17:50
NobodyCamfor ipmi17:51
NobodyCam?17:51
rameshg87NobodyCam: i am using ilo power driver, but i guess ipmi will have the same result17:51
rameshg87if the power management ip is wrong, any power operation will set the power state to error and we won't be able to delete that node17:52
NobodyCamrameshg87: and you'runable to update the ip to the correct one?17:53
rameshg87NobodyCam: ah yes, i think i will be able to .. :-)17:54
NobodyCam:)17:54
rameshg87NobodyCam: but i think in my current case, i don't need that machine now as it is down17:54
rameshg87NobodyCam: the IP is correct infact17:54
rameshg87NobodyCam: isn't this a general problem that if the machine becomes unreachable for some reason, the sync power state will fail and then we won't be able to delete that node until it is reachable again :-)17:55
NobodyCamrameshg87: I think we should tlk about that. would you check for / file a bug17:57
rameshg87NobodyCam: sure, let me just confirm and if yes i will file a bug17:57
rameshg87NobodyCam: meanwhile i just noticed one of my reviews is sitting idle https://review.openstack.org/#/c/7891217:58
rameshg87please take a look at it when you are free17:58
JoshNangare config option help messages supposed to use i18n?17:59
JayFhttps://review.openstack.org/#/c/79088/ is still in need of reviews. Infra change to integrate teeth-agent into openstack18:01
NobodyCamrameshg87: ack,18:01
NobodyCamJoshNang: humm good question, what are other project doing with there help messages18:02
JoshNangnot sure, i'll check a couple.18:02
JoshNangdoesn't look like nova or swift do18:03
*** epim has quit IRC18:09
*** rpodolyaka1 has joined #openstack-ironic18:09
*** pradipta_away is now known as pradipta18:13
*** rloo has joined #openstack-ironic18:22
*** rloo has quit IRC18:23
*** rloo has joined #openstack-ironic18:23
*** rloo has quit IRC18:40
*** rloo has joined #openstack-ironic18:40
*** derekh has joined #openstack-ironic19:02
NobodyCamundercloud | CREATE_IN_PROGRESS & undercloud-undercloud-vtdzoh2xs7ai | BUILD  | spawning .... so far so good.19:06
NobodyCamundercloud deployed from seed via Ironic using USE_IRONIC=1; devtest.sh ... patches to follow after lunch19:18
NobodyCam:)19:18
jrollNobodyCam: I thought that second message was a log message for a moment19:20
jrollincluding the "after lunch" bit19:20
NobodyCamlol19:20
NobodyCamnope: http://paste.openstack.org/show/vykcN27wlPuhrrVV3aov/19:20
jrollI figured it out myself :P19:20
NobodyCamhehehehe :)19:20
*** saju_m has quit IRC19:21
*** rpodolyaka1 has quit IRC19:22
*** saju_m has joined #openstack-ironic19:22
*** vkozhukalov_ has joined #openstack-ironic19:23
*** rpodolyaka1 has joined #openstack-ironic19:25
*** pradipta is now known as pradipta_away19:42
*** epim has joined #openstack-ironic20:06
*** ndipanov_gone has quit IRC20:08
*** saju_m has quit IRC20:08
*** eguz has quit IRC20:08
*** eghobo has joined #openstack-ironic20:08
*** saju_m has joined #openstack-ironic20:08
*** ndipanov_gone has joined #openstack-ironic20:09
*** eguz has joined #openstack-ironic20:14
*** eghobo has quit IRC20:17
*** rpodolyaka1 has quit IRC20:20
*** rpodolyaka1 has joined #openstack-ironic20:24
*** rpodolyaka1 has quit IRC20:28
*** rameshg87 has left #openstack-ironic20:34
adam_gany ideas why ironic would not be under-reporting my 3 powered-off VM nodes' resources? http://paste.ubuntu.com/7121949/20:41
NobodyCamadam_g: what does node-list show?20:44
adam_gNobodyCam, oh, i see. yeah--they have instance associations20:44
*** mrda_away is now known as mrda20:50
*** coolsvap has quit IRC20:51
*** harlowja is now known as harlowja_away20:52
*** rpodolyaka1 has joined #openstack-ironic20:53
*** vkozhukalov_ has quit IRC20:54
*** rpodolyaka1 has quit IRC20:57
*** epim has quit IRC20:58
*** epim has joined #openstack-ironic21:03
*** lazy_prince has joined #openstack-ironic21:06
*** lazy_prince is now known as killer_prince21:07
*** killer_prince has quit IRC21:07
*** killer_prince has joined #openstack-ironic21:08
*** martyntaylor has joined #openstack-ironic21:12
*** blamar is now known as blamar-away21:12
*** jbjohnso_ has quit IRC21:13
openstackgerritRuby Loo proposed a change to openstack/ironic: Replace 'nova' with 'ironic'  https://review.openstack.org/8163121:14
*** harlowja_away is now known as harlowja21:14
*** max_lobur has joined #openstack-ironic21:15
NobodyCamoh good catch rloo :)21:18
rlooNobodyCam: thx. Although after I sent that out, I had this thought, what if someone meant it to be 'nova'. I hope we don't pass it to nova...21:19
*** derekh has quit IRC21:24
*** saju_m has quit IRC21:28
NobodyCambrb quick walkies21:32
devanandarussell_h: would you mind filing a bug related to https://review.openstack.org/#/c/80187 and updating the commit msg?21:37
russell_hdevananda: sure thing21:37
devanandarussell_h: or, if it's not fixing a bug, we should defer it21:37
russell_hdevananda: yeah, I was going to ask about that.. I guess it depends what you consider a bug21:38
russell_hdevananda: like, drivers should probably have links, thats sort of a bug21:38
russell_hand IMO if the API allows listing, it should allow fetching21:38
russell_hbut those are both stretching a little21:38
russell_hmy real interest was in enabling another change that I haven't submitted yet, which definitely is a feature21:39
devanandarussell_h: I'd agree that, where the API allows listing, it should allow fetching21:39
openstackgerritRussell Haering proposed a change to openstack/ironic: Expose API for fetching a single driver  https://review.openstack.org/8018721:43
devanandarussell_h: thanks. sorry for the extra process overhead right now ...21:46
*** krtaylor has quit IRC21:46
russell_hdevananda: no worries, I'll appreciate it when I'm trying to run an RC in production :)21:46
openstackgerritJosh Gachnang proposed a change to openstack/ironic: Adding swift temp url support  https://review.openstack.org/8139121:47
*** dhellmann is now known as dhellmann_21:48
openstackgerritRuby Loo proposed a change to openstack/ironic: Install guide for Ironic  https://review.openstack.org/8164521:52
*** krtaylor has joined #openstack-ironic21:53
*** rpodolyaka1 has joined #openstack-ironic21:53
rloodevananda: if you get a chance, my stab at the install guide: https://review.openstack.org/8164521:54
rloodevananda: i don't recall the urgency (or not) of the docn. Will touch base with you next week about it.21:55
*** rpodolyaka1 has quit IRC21:57
*** epim has quit IRC21:59
*** epim has joined #openstack-ironic22:02
openstackgerritJenkins proposed a change to openstack/ironic: Updated from global requirements  https://review.openstack.org/7933422:04
jroll2014-03-19 22:05:16.042 11950 TRACE nova.openstack.common.periodic_task AttributeError: 'IronicHostManager' object has no attribute 'run_periodic_tasks'22:06
jroll^ is that a known thing in the scheduler driver?22:06
*** martyntaylor has quit IRC22:08
*** epim has quit IRC22:09
*** branen has joined #openstack-ironic22:13
NobodyCami need to dbl check I haven't forgotten anything but: https://review.openstack.org/#/c/72969 https://review.openstack.org/#/c/81627 https://review.openstack.org/#/c/8164322:15
NobodyCamare now up to be poked at22:15
jroll2014-03-19 22:05:16.042 11950 TRACE nova.openstack.common.periodic_task AttributeError: 'IronicHostManager' object has no attribute 'run_periodic_tasks'22:23
jroll^ is that a known thing in the scheduler driver?22:24
jrollNobodyCam, have you been seeing that?22:24
NobodyCamoh no22:26
jrollhrm22:26
jroll(I might be doing it wrong(TM) to be fair)22:26
jrollnova.conf has this: scheduler_driver=ironic.nova.scheduler.ironic_host_manager.IronicHostManager22:26
NobodyCamthat seems right22:27
NobodyCamyou running in a venv?22:27
jrollyeah22:27
jrollwell, let me verify22:27
jrollbut 99% sure22:28
NobodyCamis ironic and ironic client installed in the venv?22:28
jrollironic is, not the client22:28
jrollbut this box is only running the scheduler22:28
jrollwhich afaict doesn't need the client22:28
NobodyCamI just landed this the other day for OoO22:29
NobodyCamhttps://review.openstack.org/#/c/80725/2/elements/nova-ironic/post-install.d/81-nova-ironic-driver22:29
NobodyCamcd ..22:32
NobodyCamdoh22:32
devanandarloo: hi!22:35
devanandarloo: looking at your feedback on 81327 -- i'm not sure why you're objecting. these are the same thing, aren't they?22:35
*** epim has joined #openstack-ironic22:43
devanandarloo: reading through your doc patch. leaving comments but overall looks good!22:44
devanandarloo: thanks!!!22:44
jrollNobodyCam: btw, still happening with the client installed22:45
jrolland yes, in a venv22:45
NobodyCamhumm22:45
* jroll goes nova source diving22:45
NobodyCamjroll: just because I have not hit the error yet,  does not me the IronicHostManager could still at fault :)22:46
NobodyCams/me/mean/22:47
jrollyeah22:47
jrollmmm I think my config is wrong22:47
NobodyCams/could still at/could not still be at/22:47
jrollis there a separate config option for the host manager?22:47
NobodyCamyes22:48
jrollit looks like I'm plugging into nova's scheduler/driver22:48
jrollok22:48
jrollthat will do it22:48
jrollI had scheduler_driver set22:48
NobodyCamhttps://github.com/openstack/tripleo-image-elements/blob/master/elements/nova/os-apply-config/etc/nova/nova.conf#L12722:49
jrollyeah, found it22:50
NobodyCam:) I was slow with the link :-p22:50
jroll:P22:50
jrollnow I just need to remember what scheduler_driver was set to...22:51
russell_his this "check-tempest-dsvm-virtual-ironic" failure a known/expected thing?22:51
jrolllooks like it22:51
jrollit's non-voting, so it doesn't count as a fail for the patch22:52
russell_hI'd still rather not break things22:52
russell_hbut yeah, its apparently failing on other patches too, so I don't think its me22:52
jrolljudging by the logs, it's not a code issue22:53
*** rpodolyaka1 has joined #openstack-ironic22:54
jrollthat was it, thanks NobodyCam22:57
*** harlowja is now known as harlowja_away22:59
*** rpodolyaka1 has quit IRC22:59
openstackgerritMichael Davies proposed a change to openstack/ironic: Change set_console_mode() to use greenthreads  https://review.openstack.org/8135223:02
devanandaadam_g: http://logs.openstack.org/45/81645/1/check/check-tempest-dsvm-virtual-ironic/1c9265a/logs/devstacklog.txt.gz23:05
devananda2014-03-19 22:08:27.392 | chown: invalid group: `stack:libvirtd'23:05
devanandaadam_g: any thoughts?23:05
adam_gdevananda, thats the same issue that https://review.openstack.org/#/c/81416/ solves23:05
adam_gyou'll notice the experimental check there gets much further23:05
adam_gand fails on something else: http://logs.openstack.org/16/81416/1/experimental/check-tempest-dsvm-virtual-ironic-nv/2555317/logs/error.txt.gz23:06
adam_ghttps://review.openstack.org/#/c/81611/ should fix that23:06
NobodyCamdevananda: I'm going to a major overhall on 72969 to rebase it off 7846723:07
devanandaadam_g: looking23:07
NobodyCamgah I swear this thing is dropping keys. s/to a/to do a/23:08
NobodyCamloads a little high Load Avg: 2.49, 2.60, 2.6123:09
devanandaadam_g: so, DEVSTACK_GATE_NEUTRON=1 is going to break things23:13
adam_gdevananda, how so?23:14
devanandaadam_g: I explicitly removed that from our gating jobs because, if it's there, our gate blocks if neutron starts to fail23:14
devanandawhich was pretty often23:14
adam_gdevananda, hmm. but neutron is a requirement of the ironic+devstack?23:14
devanandaadam_g: yes23:15
devanandaadam_g: it should definitely be in the virtual-ironic job23:15
devanandaadam_g: which is non-voting for now23:15
adam_gOH!23:15
adam_gi gotcha23:15
*** derekh has joined #openstack-ironic23:15
devanandaadam_g: one sec23:16
adam_gdevananda, what are the other 2 jobs doing / testing exactly? (tempest-dvsm-ironic$branch + tempest-ironic-postgres$branch)23:17
devanandaadam_g: those cause devstack to start the ironic services with the "fake" driver, then tempest runs CRUD tests23:18
devanandaadam_g: so I suspect we can enable DEVSTACK_GATE_NEUTRON in our gate now. it's present in Nova's gate jobs too at this point23:18
devanandaadam_g: but I'd rather we add it to the -nv job first and make sure :)23:19
adam_gdevananda, yeah thats fine. ill update that review23:19
devanandaif 81611 lands now, it'll start running al lthe neutron tests in our gate23:19
devanandaadam_g: also, the way tempest is set up, if a service is enabled, it will run *all* thetests23:19
devanandafor that service23:19
devanandaso if we start neutron, tempest is going to do all the neutron tests23:19
devanandai have no idea if that's goig to work right now in our environment23:20
adam_gdevananda, yeah thats fine. ill update that review23:22
adam_ger23:23
adam_gdevananda, one other thing thats unclear to me in my infra newbishness--in the other 2 non-experimental gate checks, whats instructing devstack to enlist nodes /w the fake driver as opposed to pxe_ssh23:23
adam_gi dont see anything from JJB -> devstack-gate*.sh that would switch between the two23:23
devanandagood question :)23:24
adam_gdevananda, well, i see the tempest CRUD tests do it23:26
adam_gbut im curious how devstack (with our newly merged ironic stuff) will succeed w/o neutron23:26
*** max_lobur has quit IRC23:27
*** max_lobur has joined #openstack-ironic23:29
*** epim has quit IRC23:30
devanandaadam_g: fake driver doesn't call out to neutron23:31
devanandaadaso if tempest is creating nodes with driver=fake, then it'll work fine w/o neutron23:31
*** max_lobur has quit IRC23:34
adam_gdevananda, right but devstack is run before tempest starts, to setup ironic iteslf23:34
devanandayes23:34
adam_gbut i see, looking thru d-s-g + JJB side-by-side... the current gate only sets DEVSTACK_GATE_IRONIC, but never sets the VIRT_DRIVER=ironic23:35
devanandaadam_g: but those jobs don't do the vm creation & enrollment23:35
NobodyCamlol lots of echo "Ironic not supported..." in the OoO23:35
adam_gthe experimental check does23:35
devanandaadam_g: it also doesn't set BAREMETAL_BASIC_OPS23:35
devanandaadam_g: so the !virtal jobs don't create VMs23:35
devanandaadam_g: what do you think of rebasing 81416 on 81113, then running check experimental on 81416? think taht'll get further than either patch alone?23:36
adam_gdevananda, right, tempest just ends up with a standard nova+libvirt setup, with ironic living off in its own world23:36
devanandaadam_g: right23:36
*** epim has joined #openstack-ironic23:37
adam_gdevananda, 81113 really only helps if you're running fedora23:38
NobodyCamsun hitting work area, make it a tad warm... going to step away for a bit.. I will be back later, but don't expect to have this rework ready until tomorrow :)23:38
*** harlowja_away is now known as harlowja23:38
devanandaadam_g: eventually we should converge and all our d-g tests should be with VIRTDRIVER=ironic, etc23:39
adam_gdevananda, 81416 should get much closer (hopefully there!) once neutron is enabled in the config23:39
rloodevananda: wrt 81327. I took a quick look yest at pbr.version.VersionInfo. the .release_string() and .version_string() calls will set .release/.version the first time they are called. So seems safer to make the calls?23:41
rloodevananda: i'll add more comments to 81327.23:41
*** epim has quit IRC23:42
devanandaadam_g: gotcha23:44
mrdahey devananda, can I discuss youre review comments on 81352 briefly?23:50
*** epim has joined #openstack-ironic23:51
Shrewsmrda: are you going to finish that out then? i had planned to do it this evening23:51
mrdaShrews: I'm happy to23:52
mrdait's just a few last comments23:52
* mrda just trying to get this bug closed23:52
Shrewsmrda: ok23:52
*** rpodolyaka1 has joined #openstack-ironic23:55
rloomrda: was just checking my emails. you OK with deva's comments for 81352? Or should I actually look?23:59
mrdahey rloo23:59
rloomrda: since I think the comments were wrt my comments ;)23:59
*** rpodolyaka1 has quit IRC23:59
mrdarloo: So just on the check for enabled (i.e. L774)23:59
« Tuesday, 2014-03-18 Index Thursday, 2014-03-20 »

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!