Friday, 2015-12-04

*** penick has quit IRC00:00
*** smoriya_ has joined #openstack-ironic00:01
*** rajathagasthya has quit IRC00:03
*** sacharya has quit IRC00:04
*** penick has joined #openstack-ironic00:07
*** rajathagasthya has joined #openstack-ironic00:08
*** naohirot has joined #openstack-ironic00:09
*** chlong has joined #openstack-ironic00:17
*** penick has quit IRC00:27
*** devananda has quit IRC00:28
*** amotoki has joined #openstack-ironic00:28
*** Pavo has joined #openstack-ironic00:36
*** Pavo has left #openstack-ironic00:39
*** rajathagasthya has quit IRC00:40
*** rajathagasthya has joined #openstack-ironic00:42
*** amotoki has quit IRC00:44
*** jaybeale has quit IRC00:45
*** devananda has joined #openstack-ironic00:53
*** Sukhdev has joined #openstack-ironic00:55
*** Sukhdev has quit IRC00:59
*** sacharya has joined #openstack-ironic01:02
*** sacharya has quit IRC01:09
*** penick has joined #openstack-ironic01:13
lintanmorning mrda and NobodyCam01:20
*** _zouyee has quit IRC01:23
*** penick has quit IRC01:31
*** kkoski has joined #openstack-ironic01:31
*** mtanino has quit IRC01:32
*** baoli_ has quit IRC01:34
*** zhenguo has joined #openstack-ironic01:37
openstackgerritTan Lin proposed openstack/ironic: Follow-up Patch on I4d81001c01e7908100a6571b366cb296253f2fc1
mrdahey lintan01:51
*** sacharya has joined #openstack-ironic01:54
*** Haomeng has quit IRC02:00
*** _zouyee has joined #openstack-ironic02:02
*** sacharya has quit IRC02:05
*** rajathagasthya has quit IRC02:10
*** sinval has quit IRC02:11
*** amotoki has joined #openstack-ironic02:13
*** aginwala has quit IRC02:21
*** changzhi has joined #openstack-ironic02:29
openstackgerritZhenguo Niu proposed openstack/ironic: Add console feature to ssh driver
*** aginwala has joined #openstack-ironic02:35
*** hanchao has joined #openstack-ironic02:35
*** aginwala has quit IRC02:36
*** yonglihe has quit IRC02:41
*** baoli has joined #openstack-ironic02:44
changzhihi, all. Did anyone deploy bare metal machines by using Ironic?02:47
openstackgerritMerged openstack/ironic: Correct the db connection string in dev-quickstart
openstackgerritNaohiro Tamura proposed openstack/ironic-specs: Enhance Power Interface for Soft Power Off and Inject NMI
hanchaohello, did anyone deployed the ironic service on a individual node rather than located same on the controller node?02:56
*** gbraad has joined #openstack-ironic02:57
*** rloo has quit IRC03:00
openstackgerritYuiko Takada proposed openstack/ironic-specs: Add allowed transitions to API
*** thrash has quit IRC03:24
*** Piet has joined #openstack-ironic03:31
*** Nisha has joined #openstack-ironic03:34
*** e0ne_ has joined #openstack-ironic03:52
*** links has joined #openstack-ironic04:00
*** e0ne_ has quit IRC04:00
*** Nisha has quit IRC04:04
*** baoli has quit IRC04:08
*** e0ne has joined #openstack-ironic04:12
*** e0ne has quit IRC04:16
*** links has quit IRC04:23
*** Nisha has joined #openstack-ironic04:31
*** jaybeale has joined #openstack-ironic04:45
*** jaybeale has quit IRC04:59
*** amotoki_ has joined #openstack-ironic05:09
*** links has joined #openstack-ironic05:09
openstackgerritChangBo Guo(gcb) proposed openstack/ironic: test:clean up wrong value in method CONF.set_override
*** jaybeale has joined #openstack-ironic05:10
*** amotoki_ has quit IRC05:12
*** amotoki has quit IRC05:13
*** Haomeng has joined #openstack-ironic05:16
*** amotoki has joined #openstack-ironic05:17
*** amotoki has quit IRC05:17
*** mayflyman has joined #openstack-ironic05:18
*** jaybeale has quit IRC05:22
*** aginwala has joined #openstack-ironic05:34
*** Sukhdev has joined #openstack-ironic05:39
*** Sukhdev has quit IRC05:43
*** hoangcx has joined #openstack-ironic05:53
openstackgerritMerged openstack/bifrost: Adjusted line length to 79 characters
*** rcernin has joined #openstack-ironic06:13
*** mayflyman has quit IRC06:14
*** Sukhdev has joined #openstack-ironic06:19
*** jaybeale has joined #openstack-ironic06:22
*** Haomeng has quit IRC06:22
*** thanhnt-z has joined #openstack-ironic06:28
*** ChubYann has quit IRC06:30
*** amotoki has joined #openstack-ironic06:30
*** aginwala has quit IRC06:35
*** aginwala has joined #openstack-ironic06:37
*** jlk has joined #openstack-ironic06:41
*** jlk has quit IRC06:41
*** jlk has joined #openstack-ironic06:41
*** Haomeng|2 has joined #openstack-ironic06:49
*** jaybeale has quit IRC06:50
*** Haomeng has joined #openstack-ironic06:50
*** rcernin has quit IRC06:59
*** PradeepV has joined #openstack-ironic07:04
*** deray has joined #openstack-ironic07:04
*** rcernin has joined #openstack-ironic07:07
*** absubram has quit IRC07:15
*** chlong has quit IRC07:19
*** Sukhdev has quit IRC07:23
*** kozhukalov_ has joined #openstack-ironic07:29
openstackgerritTan Lin proposed openstack/ironic: Add a config option [upgrade_level]/condcutor_rpcapi_cap
openstackgerritNaohiro Tamura proposed openstack/ironic: Update python-scciclient version number
*** kozhukalov_ has quit IRC07:34
sergekMorning, Ironicers 😃07:46
*** _zouyee has quit IRC07:47
lintan morning sergek :)07:47
sergeklintan: o/07:47
thanhnt-zMorning :D07:48
*** kozhukalov_ has joined #openstack-ironic07:48
openstackgerritTan Lin proposed openstack/ironic: Correct NotImplemented to NotImplementedError in
thanhnt-zHi Haomeng07:51
openstackgerritKan proposed openstack/python-ironicclient: [WIP]Format the valid fields for list commands
*** kozhukalov_ has quit IRC08:03
openstackgerritTan Lin proposed openstack/ironic: Add a config option [upgrade_level]/condcutor_rpcapi_cap
*** changzhi has quit IRC08:06
openstackgerritNisha Agarwal proposed openstack/python-ironicclient: Add CLI support for RAID configuration
Haomeng|2thanhnt-z: morning:)08:11
Haomeng|2sergek: morning:)08:12
thanhnt-zHaomeng: related to the issue that I got some days ago, it was already reported as this bug:
openstackLaunchpad bug 1494605 in Ironic "shellinabox doesn't restart when ironic-conductor restarted" [Low,Triaged] - Assigned to takahiko nishiyama (tnishiya)08:13
Haomeng|2thanhnt-z: ok, let me check, thanks for your bug08:13
*** Haomeng|2 is now known as Haomeng08:14
thanhnt-zHaomeng: not mine :D, it was reported some months ago :|08:14
sergekHaomeng: o/08:14
Haomengthanhnt-z: ok, dont worry, but it is assigned to submitter already08:14
Haomengthanhnt-z: so understand your concern that you have same issue, and is how about the progress?08:15
thanhnt-zHaomeng: Yep, when I set node-set-console-mode is True, the shellinabox service is not restarted and not run this command: shellinaboxd -t -p 10101 --background=/tmp/ -s /:1000:1000:HOME:ipmitool -H baremetal_ip -I lanplus -U admin -f /tmp/ -v sol activate08:18
thanhnt-zHaomeng: so I have to run this command by manually08:18
*** kozhukalov_ has joined #openstack-ironic08:19
*** athomas has joined #openstack-ironic08:19
*** jaybeale has joined #openstack-ironic08:20
*** deray has quit IRC08:20
Haomengthanhnt-z: ok, so can you enable debug log and send me(via gmail?) the conductor's log, and let me check what happens when enabling the console?08:20
thanhnt-zHaomeng: But it did not orcuss on my co-worker's environment. So that's why I told that it seems not a Ironic Bug08:20
thanhnt-zHaomeng: sure that conductor-log got no error message08:21
*** romcheg2 has quit IRC08:21
Haomengthanhnt-z: if we run node-set-console-mode is true, and it returns as normal behavior, but no shellinabox process, it should be a bug08:21
Haomengthanhnt-z: ok, but I have no env to recreate now, so can you help to provide more details, such as the shellinabox process status08:23
*** romcheg1 has joined #openstack-ironic08:23
Haomengthanhnt-z: ps -ef|grep shellinabox08:23
Haomengthanhnt-z: if there is no such process at all, have to check conductor log and debug the code08:23
*** vsaienko has joined #openstack-ironic08:24
vsaienkoHello Ironic team, could you please have look at the devstack patches that enables ironic neutron integration testing08:25
*** MattMan has quit IRC08:25
Haomengthanhnt-z: got this log in code -
*** MattMan has joined #openstack-ironic08:25
Haomengthanhnt-z: so can you make sure if your log is in debug level ,and help to grep 'Running subprocess:'08:25
thanhnt-zHaomeng: Yep, I will send you an email related to this concern in next Monday, it's ok? Now I also dont have enviroment :| and I will rebuild on next monday08:26
Haomengthanhnt-z: ok, will check your mail, thank you08:26
thanhnt-zHaomeng: Sure that it set DEBUG level08:26
Haomengthanhnt-z: assume you enable log in debug level08:26
Haomengthanhnt-z: ok, that is fine08:26
Haomengthanhnt-z: inform me once you sent mail08:26
thanhnt-zHaomeng: sure08:27
Haomengthanhnt-z: will be next Mon, right? ok08:27
thanhnt-zHaomeng: yep08:27
Haomengthanhnt-z: ok08:27
Haomengthanhnt-z: but if there is no env to be recreated, it is diffcult to debug I think, will try:)08:28
thanhnt-zHaomeng: yes, I see. I tried to debug in that way some times before :D08:29
Haomengthanhnt-z: :)08:30
thanhnt-zHaomeng: have you tried to investigate about how to use nova console for Ironic node yet?08:32
*** ifarkas has joined #openstack-ironic08:32
Haomengthanhnt-z: no, and think it is not easy to support both nova novnc and ironic terminal console08:32
Haomengthanhnt-z: but maybe it can be integrated in horizon, need to horizon team guys support, ironic just pass the console url08:33
*** deray has joined #openstack-ironic08:33
thanhnt-zHaomeng: yes, it is not easy issue, I have just wanted to investigate more about them. Then, I will have a better look about ironic08:35
Haomengthanhnt-z: yes08:35
*** mgoddard has joined #openstack-ironic08:47
*** jcoufal has joined #openstack-ironic08:49
*** _degorenko|afk is now known as degorenko09:03
openstackgerritTan Lin proposed openstack/ironic: Add Code Contribution Guide for Ironic
*** dims_ has quit IRC09:05
*** jaybeale has quit IRC09:05
*** praneshp has quit IRC09:10
*** dims has joined #openstack-ironic09:11
*** mbound_ has joined #openstack-ironic09:13
openstackgerritTan Lin proposed openstack/ironic: Follow-up Patch on 1ab3c9217edc8ebed7643f8ded7b16c39dd4bc70
openstackgerritYuriy Zveryanskyy proposed openstack/ironic: Fix awake AMT unit test
*** yuanying has quit IRC09:22
yuriyzhi Ironic09:22
*** aginwala has quit IRC09:23
*** dims has quit IRC09:23
lintanhi yuriyz :)09:25
yuriyzhi lintan haomeng, more AMT stuff :)
lintancool, yuriyz, I will try your patch :)09:29
-openstackstatus- NOTICE: Tox tests are broken at the moment. From openstack-infra we are working to fix them. Please don't approve changes until we notify that tox tests work again.09:31
*** dims has joined #openstack-ironic09:34
*** dtantsur|afk is now known as dtantsur09:35
dtantsurMorning Ironic!09:35
dtantsurgate is broken, meaning...09:35
*** gbraad has quit IRC09:35
*** derekh has joined #openstack-ironic09:38
openstackgerritTan Lin proposed openstack/python-ironicclient: Correct node-port-list help info
*** ionutbalutoiu has joined #openstack-ironic09:45
yuriyzmorning dtantsur09:46
*** mdbooth has quit IRC09:47
aarefievmorning all!09:50
yuriyzdtantsur, looks like I'm member of ironic-inspector-specs cores09:51
dtantsuryuriyz, well, as all ironic-core's09:51
dtantsurironic-inspector-core contains ironic-core09:51
*** mdbooth has joined #openstack-ironic09:51
dtantsurusual rule applies: don't approve if you don't feel you should :)09:52
*** Marga_ has quit IRC09:54
yuriyzdatntsur good maybe I will look into Inspector, I prefer "-1" for reviews :)09:54
*** mkovacik_ has joined #openstack-ironic09:55
*** zhenguo has quit IRC09:59
*** kevinbenton has quit IRC10:01
*** naohirot has quit IRC10:02
openstackgerritTan Lin proposed openstack/ironic: Follow-up Patch on 1ab3c9217edc8ebed7643f8ded7b16c39dd4bc70
*** kevinbenton has joined #openstack-ironic10:04
dtantsuryuriyz, :)10:07
dtantsurmorning sergek10:07
*** Marga_ has joined #openstack-ironic10:07
*** Marga_ has quit IRC10:11
*** Marga_ has joined #openstack-ironic10:11
*** athomas has quit IRC10:15
*** pas-ha has joined #openstack-ironic10:19
*** hoangcx has quit IRC10:21
*** athomas has joined #openstack-ironic10:24
*** Marga_ has quit IRC10:28
*** yuanying has joined #openstack-ironic10:29
*** openstackgerrit has quit IRC10:32
*** yuanying has quit IRC10:32
*** yuanying has joined #openstack-ironic10:33
*** openstackgerrit has joined #openstack-ironic10:33
*** yuanying has quit IRC10:33
*** yuanying has joined #openstack-ironic10:34
*** electrofelix has joined #openstack-ironic10:35
openstackgerritMerged openstack/python-ironic-inspector-client: Fixed links to the new ironic-inspector documentation
sambetts-afkMorning all o/10:44
dtantsurmorning sambetts-afk10:46
*** sambetts-afk is now known as sambetts10:46
sambettso/ dtantsur hows it going?10:50
dtantsurit's Friday :)10:50
sambettshaha yes :D TGIF10:50
openstackgerritShuquan Huang proposed openstack/ironic: Move install-guide's troubleshooting to troubleshooting page
*** aginwala has joined #openstack-ironic10:54
*** PradeepV has quit IRC10:55
*** thanhnt-z has quit IRC10:57
*** aginwala has quit IRC10:58
*** deray has quit IRC11:09
openstackgerritVladyslav Drok proposed openstack/ironic: Fix awake AMT unit test
lucasagomesmorning all!11:13
sambettso/ lucasagomes!11:13
dtantsurmorning lucasagomes11:14
dtantsurfolks, if you haven't seen the notice, tox jobs are broken now, please don't approve anything11:14
dtantsuralso unit tests are broken for IPA
openstackLaunchpad bug 1522756 in ironic-python-agent "Unit tests fail with DeviceNotFoundByFileError: [Errno 2] No such file or directory: '/dev/sda'" [High,Confirmed] - Assigned to Dmitry Tantsur (divius)11:16
dtantsurtypical Friday, I would say11:16
sambettsyup :D everything breaks on Fridays its like even the machines know the weekend is coming11:19
*** Nisha has quit IRC11:21
dtantsurpyudev broke backward compatibility awesome11:22
sambettsyay! :D11:22
vdrokdtantsur, unittests in ironic too -
vdrokmorning everyone, tgif :D11:24
lucasagomesthe security fix didn't get in yet11:27
pas-hamorning Ironic11:27
dtantsurmorning pas-ha! join the celebration of broken gates11:28
lucasagomesdtantsur, yeah, oh friday11:28
pas-halucasagomes, I suspect this is related to problem being fixed in this patch
* lucasagomes looks11:30
lucasagomespas-ha, vdrok that's what causing the gate to break?11:32
* lucasagomes check more logs11:32
pas-hawell, the py27 error in cve patch is exactly in this testcase11:32
openstackgerritVladyslav Drok proposed openstack/ironic-specs: Add support for hardware composition
pas-haconcurrent unit tests run always bring interesting problems ^)11:34
*** ndipanov has joined #openstack-ironic11:35
openstackgerritDmitry Tantsur proposed openstack/ironic-python-agent: pyudev exception has changed for from_device_file
dtantsurdiscussing the issue on #openstack-stable too, if someone cares11:42
*** chlong has joined #openstack-ironic11:44
*** sinval-afk is now known as sinval11:47
sinvalhey hey Ironicers, good morning11:47
dtantsurmorning sinval11:47
sinvalquick question, is 4.3 released?11:48
dtantsurI don't think so11:49
sinvaldtantsur, thanks11:51
lucasagomessinval, morning11:51
sinvallucasagomes, o/11:52
openstackgerritMerged openstack/ironic: Fix awake AMT unit test
*** noqa_v_g1ovnie is now known as noqa_v_qoovnie11:58
* lucasagomes rechecks the security fix now12:03
*** yuanying_ has joined #openstack-ironic12:04
*** smoriya_ has quit IRC12:05
dtantsurlucasagomes, please don't recheck/approve anything until gate is fixed12:05
lucasagomesdtantsur, it seems fixed12:05
lucasagomesafter the AMT stuff12:05
dtantsurlucasagomes, yeah, but tox jobs do nothing12:05
lucasagomeswe gotta get this security fix in12:05
dtantsursee "[all][infra] All tox jobs broken, don't approve changes for now"12:05
lucasagomesfair enuff12:06
dtantsurI'd not do anything until a new notice from them12:06
lucasagomes++ yeah sorry for that12:06
sambettswow... can't they force them to -1 until they fix it to prevent accidental merges of borked code?12:06
*** hanchao has quit IRC12:06
lucasagomestbh, I want that security fix to get in regardless12:06
lucasagomesit should have been merged yesterday12:06
lucasagomessince now it's open to the public12:07
dtantsurlucasagomes, well, maybe it's fair. I'm just making sure you're aware of the situation :)12:08
lucasagomesdtantsur, yeah thanks for that12:08
*** yuanying has quit IRC12:08
*** dtantsur is now known as dtantsur|brb12:09
*** mattymo has left #openstack-ironic12:12
*** kozhukalov_ has quit IRC12:13
*** masterbound has joined #openstack-ironic12:15
*** mbound_ has quit IRC12:16
*** mbound_ has joined #openstack-ironic12:17
*** masterbound has quit IRC12:17
*** yuanying_ has quit IRC12:23
*** kozhukalov_ has joined #openstack-ironic12:26
*** yuanying has joined #openstack-ironic12:26
*** yuanying has quit IRC12:27
*** yanghy has joined #openstack-ironic12:30
yanghyHello, sorry to bother you guys...could somebody take this patch, it's been there for a long time and no progress
*** trown|outttypeww is now known as trown12:36
lucasagomesyanghy, will take a look12:37
yanghylucasagomes, thanks a lot!12:38
lucasagomesyanghy, there's a request there for unittests12:38
liliarsgood morning everyone12:41
lucasagomesliliars, morning12:42
yanghylucasagomes, yeah, so you all think it's a blocker right? then I will take some time to look at this, because I'm not quite familiar with ironic code now...this bug was originally found in magnum through code review, and I happen to see that ironic code also has this problem...12:43
sambettsMorning lucasagomes12:43
sambettsmorning liliars12:43
lucasagomesyanghy, not a blocker. It's ideal to have unittests tho. If you don't mind I can create the unittests for you and upload the code12:44
lucasagomesyanghy, at the moment there's a problem with tox so we are not approving patches until the infra fix it12:44
yanghylucasagomes, that will be great!~ it takes time to look into the code though :)12:44
lucasagomesyanghy, cool12:44
* lucasagomes does it while gate is broken anyway12:45
yanghythanks a lot lucasagomes !12:45
lucasagomesyanghy, thank YOU for fixing it12:46
lucasagomesand reportng the problem12:46
*** dtantsur|brb is now known as dtantsur12:46
liliarslucasagomes, sambetts o/12:46
*** thrash has joined #openstack-ironic12:47
*** thrash has quit IRC12:47
*** thrash has joined #openstack-ironic12:47
liliarsoh I see we have a not-so-nice-broken-gate Friday :(12:47
dtantsurmorning liliars, yes, we do :)12:48
openstackgerritPavlo Shchelokovskyy proposed openstack/ironic-inspector-specs: CMDB integration
yanghylucasagomes, it's my pleasure :)12:51
*** thrash has quit IRC12:55
lucasagomesliliars, yeah :-( happens from time to time12:55
lucasagomeseither friday or monday12:55
*** mbound_ has quit IRC12:57
*** thrash has joined #openstack-ironic13:04
*** thrash has quit IRC13:04
*** thrash has joined #openstack-ironic13:04
*** yanghy has quit IRC13:04
*** mbound_ has joined #openstack-ironic13:05
*** dprince has joined #openstack-ironic13:07
liliarsmorning dtantsur :)13:09
liliarslucasagomes, well now we have it on a Friday close to release :(13:10
TheJuliaGood morning13:10
openstackgerritLucas Alvares Gomes proposed openstack/python-ironicclient: Fix exceptions.from_response() parameter
lucasagomesliliars, lol13:12
liliarsmorning TheJulia13:12
*** thrash has quit IRC13:15
*** moorryan has joined #openstack-ironic13:19
*** thrash has joined #openstack-ironic13:19
*** thrash has quit IRC13:19
*** thrash has joined #openstack-ironic13:19
liliarsdo we have a date for spec-freeze already?13:19
lucasagomesliliars, nah, I don't think we have a spec freeze13:20
lucasagomessince we have a diff release model13:20
dtantsurliliars, we never freeze specs. neither do we guarantee when a patch implementing a spec will land13:20
dtantsurtreat specs repo as a backlog for project, not an attempt to plan releases13:21
* lucasagomes needs to write a spec for the json fields13:23
lucasagomesdo it now13:23
liliarsdtantsur, "neither do we guarantee when a patch implementing a spec will land" yeah that part I knew13:24
liliarsdtantsur, lucasagomes oh well, deadlines help me organize, so I was hoping.. ;p but thanks guys! will plan differently13:24
dtantsurliliars, previously we had a deadline around M3, so you can use it for self-organizing :)13:24
dtantsurif you don't merge your spec by M3, chances are very low it will get to Mitaka13:25
*** sambetts has quit IRC13:25
liliarsdtantsur, thanks, that's a direction :) will do!13:25
*** sambetts has joined #openstack-ironic13:26
*** degorenko is now known as degorenko|afk13:26
*** mbound_ has quit IRC13:27
*** Piet has quit IRC13:36
*** mbound_ has joined #openstack-ironic13:38
*** david-lyle has quit IRC13:43
openstackgerritNaohiro Tamura proposed openstack/ironic: Add oslo_config.Opt support in Ironic config generator
*** pas-ha has quit IRC13:48
*** pas-ha has joined #openstack-ironic13:49
*** liliars is now known as liliars-afk13:56
*** caiobo_ has quit IRC13:57
*** liliars-afk has quit IRC13:57
*** baoli_ has joined #openstack-ironic13:58
*** marcusrafael has quit IRC13:59
*** mbound_ has quit IRC13:59
*** degorenko|afk is now known as degorenko13:59
*** sergek has quit IRC14:00
jrollmornin y'all14:00
*** sinval has quit IRC14:00
dtantsurmorning jroll! see above, we have Friday gate fun14:01
jrollso I see tox is fixed now yeah?14:01
dtantsurjroll, I don't see an official notice14:01
jrolland now it's just IPA unit tests broken?14:01
*** sergek has joined #openstack-ironic14:01
*** mbound_ has joined #openstack-ironic14:01
dtantsurjroll, I assume tox is broken until we get a confirmation from infra14:01
dtantsurjroll, and yes, IPA unit tests14:02
jrolldtantsur: oops, I thought it was a topic change and saw normal topic was there14:02
*** links has quit IRC14:02
dtantsurit's weird they didn't update topics..14:03
jrollI guess I should just back away slowly then :)14:03
openstackgerritVladyslav Drok proposed openstack/ironic: Add proxy related parameters in agent driver
*** zhenguo has joined #openstack-ironic14:13
lucasagomesjroll, morning14:21
*** [1]cdearborn has joined #openstack-ironic14:27
lucasagomesjroll, around? For the JSON fields stuff14:27
lucasagomesjroll, do we expect all json fields to be indexable? I think that things like extra, driver_internal_info and so on they could remain as json14:28
*** Marga_ has joined #openstack-ironic14:28
lucasagomesanyone has opnions on that ^ ?14:28
lucasagomes(that's about the way the are stored in the db, in the API they will remain as JSON ofc)14:28
dtantsurI'd start with properties/capabilities, and solve other separately, if we ever feel like14:29
lucasagomesdtantsur, right, driver_info ?14:29
lucasagomesdtantsur, I want to have a separated spec for capabiltiies, I would do properties first and then separate capability into another spec and field in the API14:29
dtantsurwe in ironic-inspector would love to see driver_info searchable, but I'm afraid it's only us..14:29
lucasagomesdtantsur, that's a valueable use case14:30
lucasagomesif we are doing 1 field, doing another is not a big deal14:30
lucasagomesI will propose the spec to do properties and driver_info14:30
*** absubram has joined #openstack-ironic14:31
*** alexpilotti has quit IRC14:32
lucasagomesdtantsur, yeah well actually I will think about it after the call14:32
NobodyCamgood morning Ironicers14:37
*** absubram has quit IRC14:37
vdrokmorning TheJulia jroll and NobodyCam !14:39
NobodyCammorning lucasagomes dtantsur jroll vdrok sambetts jlvillal  :)14:39
dtantsurmorning NobodyCam!14:40
*** absubram has joined #openstack-ironic14:41
*** rkharya has joined #openstack-ironic14:41
dtantsurrkharya, hi, so what's the problem?14:45
rkharyaopenstack-ironic-discoverd.service logs has this error - Dec 04 09:34:20 overcloud-compute-0.localdomain ironic-discoverd[16749]: ERROR:ironic_discoverd.utils:Could not find a node for attributes {'bmc_address': u'', 'mac': [u'00:25:b5:77:00:2c']}14:46
dtantsursambetts, is this IPMI address expected for UCS ^^?14:47
rkharyapxe_ucs driver does not rely on bmc address, it directly talks to UCSM via Python-ucsSDK for power driver functionality14:47
dtantsurrkharya, then you have to use port (bring it back, if you deleted it after our previous conversation), and make sure you start introspection by its CLI, not just by powering on the machine14:48 did that as you mentioned. cleaned up and started afresh via 'openstack introspection start <node-id>' command14:49
rkharyabefore that did the node registration with all the details14:50
dtantsurrkharya, do you have a port for MAC 00:25:b5:77:00:2c?14:51
rkharyadtantsur, not yet, expecting it to get created once instrospection completes. because if i create it first, then i get that mac-address listed in iptables drop list as soon i execute 'ironic port-create <>' command14:53
jrolllucasagomes: yeah, I agree with dtantsur, start with properties and driver_info, and go from there14:53
lucasagomesjroll, fair enuff14:53
*** dansmith is now known as superdan14:53
rkharyadtantur, that prevents bare metal node to get the dhcp address14:53
dtantsurrkharya, it will get whitelisted after you start the introspection. without both BMC address and MAC ironic-inspector wont' be able to lookup your node14:53
*** alexpilotti has joined #openstack-ironic14:54
rkharyadtantsur, ok let me try with this again14:54
rkharyadtantsur, one question though out of BMC and MAC will it work if only MAC is available?14:55
*** alex_xu is now known as alexus14:56
dtantsurrkharya, IIRC yes14:56
* dtantsur double-checks14:56
dtantsuryeah, should be fine14:57
rkharyadtantsur, ok thanks let me try with manual port creation and get back14:57
*** alexpilotti has quit IRC14:58
openstackgerritSergii Turivnyi proposed openstack/python-ironicclient: Add tests for testing node-delete command
*** mbound_ has quit IRC15:10
*** Marga_ has quit IRC15:12
*** ir2ivps9 has quit IRC15:12
dtantsurwe got broken by pyudev, requirements repo is broken by neutron, and all this is broken by something in infra. could it be better?15:16
* dtantsur rage quits and runs to the forest to live with owls15:16
openstackgerritSergii Turivnyi proposed openstack/python-ironicclient: Add tests for testing node-delete command
NobodyCamdtantsur: NOOOOOOOOO!!!!!15:17
*** Nisha has joined #openstack-ironic15:17
*** HimanshuGarg has quit IRC15:17
*** Marga_ has joined #openstack-ironic15:19
*** kozhukalov_ has quit IRC15:20
*** dprince has quit IRC15:21
*** gabriel has quit IRC15:22
*** Nisha_away has joined #openstack-ironic15:23
*** Nisha has quit IRC15:24
sambettsdtantsur: -_- my irssi client made me post a random message in openstack-meeting3 by accident...15:26
dtantsurI hope it was not too offensive :)15:26
sambettsdtantsur: I thought the tripleo/OSP thing should have created the port in ironic manaually?15:26
*** Marga_ has quit IRC15:27
sambettsdtantsur: because it had the wrong channel selected but wasn't telling me my auto complete wasn't working so I thought you'd actually rage quit15:27
dtantsursambetts, it should, but we recommended deleting it last time :)15:27
*** degorenko has quit IRC15:27
sambettsdtantsur: ah15:28
sambettswhoops haha15:28
lucasagomessambetts, hehe now your message is registered forever :-)15:29
sambettslucasagomes: yep...15:29
*** mtanino has joined #openstack-ironic15:31
*** degorenko has joined #openstack-ironic15:31
sambettsdtantsur: Did you see the response to your naming ML email?15:32
*** Nisha_away has quit IRC15:32
dtantsursambetts, I see one response now15:32
sambetts:-P its good15:33
*** Nisha_away has joined #openstack-ironic15:33
dtantsursambetts, by the way. I was told by the release team that we are no longer *required* to have milestone tracking in launchpad. I'm inclined to stop doing that and drop information about all milestones15:33
dtantsuronly use launchpad to track bugs and blueprints15:33
sambettsdtantsur: because reno?15:34
dtantsurthat puts unnecessary burden on me now, so I'd avoid it15:35
dtantsurjroll, do you plan to continue using milestones in launchpad for ironic?15:35
*** mbound_ has joined #openstack-ironic15:36
jrolldtantsur: ehhhh, dunno, probably not since we aren't using it for bugs15:36
sambettsHow do we track what we're targeting for a specfic release now?15:36
sambettsor is it just when we feel we've done enough15:36
dtantsursambetts, do we really need to? or rather: did we *really* do it previously?15:36
jrollI was doing it for completed stuff and as a thing to indicate I'd like to finish a thing for a release15:37
jrollbut yeah, that's where the 'dunno' comes in for me15:37
openstackgerritSergii Turivnyi proposed openstack/python-ironicclient: Tests for testing node-set-power-state command
sambettsthe things we want to finish is the part I think that would be missing15:38
sambettscompleted stuff gets a reno note so thats fine and tracked by that15:38
sambettsbut the list of things we want to complete before we release isn't anywhere else is it?15:39
openstackgerritSergii Turivnyi proposed openstack/python-ironicclient: Add tests for testing node-delete command
jrollnot really, no15:40
jrollbut we don't do a very good job of that today15:41
jrollmostly because launchpad is terrible15:41
*** jcoufal has quit IRC15:41
* lucasagomes call's finished... will grab a quick lunch15:41
sambettstrue :/15:41
*** lucasagomes is now known as lucas-hungry15:41
dtantsurjroll, sambetts, IMO ideally we should use the importance field in launchpad for priorities15:42
dtantsure.g. "critical" means NOW, "high" means this release15:42
sambettsyeah, I think that would be nice, I think thats why neutron have ditched BPs now and use bugs for everything15:43
sambettseven for BP things they use a bug with a special tag on it15:43
dtantsurbugs + specs? sounds not so bad..15:43
dtantsurwe do it in Red Hat with our bugzilla15:44
*** jcoufal has joined #openstack-ironic15:44
sambettsI think it works much better than BPs do15:44
jrollI'm more wondering when phabricator will be available for use15:44
jrollif soon, I'd rather just wait for that15:44
sambettsthats does it everything bugs right?15:44
dtantsurjroll, the same time as previous "solve-all-the-problems" solution? :D how was it called?15:44
*** yonglihe has joined #openstack-ironic15:44
jrollheh, storyboard15:45
jrollbut storyboard was "let's make our own", this is just deploying an existing thing15:45
*** mgould has joined #openstack-ironic15:45
jrollI do kind of like neutron's RFE thing15:46
sambettsneutron also do specless RFEs if its straight forward enough15:46
dtantsurme too. I never found LP blueprints convenient.. and now that we also have specs, I see little sense in them15:46
dtantsurfair enough, yeah15:46
sambettsits like levels of work, bug, RFE, RFE + Spec15:47
dtantsuralso, we already have bugs in ironic that are essentially RFE's for small additions15:47
dtantsurshould I bring it to the ML and then to the meeting?15:49
jrolldtantsur: sounds good, maybe the december 14 meeting to give folks time15:50
*** dprince has joined #openstack-ironic15:50
* dtantsur writes yet another email15:50
*** jerrygb has joined #openstack-ironic15:52
*** alexpilotti has joined #openstack-ironic16:01
*** jerrygb_ has joined #openstack-ironic16:03
*** jerrygb has quit IRC16:06
*** jerrygb_ has quit IRC16:07
*** mgould has quit IRC16:09
*** jerrygb has joined #openstack-ironic16:10
openstackgerritMerged openstack/bifrost: Initial support for installation of ironic-inspector
openstackgerritMerged openstack/bifrost: ironic-inspector install doc polish
*** lucas-hungry is now known as lucasagomes16:19
*** mgould has joined #openstack-ironic16:22
*** mbound_ has quit IRC16:25
*** mbound_ has joined #openstack-ironic16:26
*** alexpilotti has quit IRC16:28
*** ifarkas has quit IRC16:28
dtantsurFYI: tox is ok again, we can approve patches (to everything except for IPA)16:30
dtantsurlucasagomes, jroll ^^16:30
lucasagomesdtantsur, w00t!16:31
* lucasagomes recheck security fix16:31
jrollthanks dtantsur16:31
*** alexpilotti has joined #openstack-ironic16:31
jrollI just did16:31
mariojvhi - quick question about code review. 2 x +2s and 1 x +1 does not mean workflow +1 (and thus a merge), right?16:32
lucasagomesfolks this had a +2 already but I added unittests (talked to the owner of the patch on IRC today)16:32
lucasagomescan we have some eyes on that?16:32
lucasagomesit's a simple patch16:32
mariojvnever mind, checked contributor guidelines... :) i've got 2 +2s on (notifications)16:33
*** mbound_ has quit IRC16:38
*** mbound_ has joined #openstack-ironic16:38
devanandamorning, all16:39
*** mbound_ has quit IRC16:40
dtantsurmorning devananda16:40
*** jerrygb_ has joined #openstack-ironic16:40
*** mbound_ has joined #openstack-ironic16:40
lucasagomesdevananda, morning16:41
* dtantsur has sent a email about milestones and blueprints16:42
*** jerrygb has quit IRC16:42
openstackgerritVladyslav Drok proposed openstack/ironic: Add documentation for proxies usage with IPA
*** degorenko is now known as _degorenko|afk16:44
*** kozhukalov_ has joined #openstack-ironic16:47
*** david-lyle has joined #openstack-ironic16:49
*** alexpilotti has quit IRC16:52
*** rcernin has quit IRC16:52
dtantsurleaving for a week now \o/ can someone please keep an eye on ?16:53
-openstackstatus- NOTICE: The earlier JJB bug which disrupted tox-based job configurations has been reverted and applied; jobs seem to be running successfully for the past two hours.16:53
dtantsursee you16:54
sambettso/ dtantsur16:54
JayFo/ dtantsur16:54
*** dtantsur is now known as dtantsur|afk16:54
*** Piet__ has joined #openstack-ironic16:55
mgouldo/ dtantsur|afk16:56
*** mbound_ has quit IRC17:01
openstackgerritLucas Alvares Gomes proposed openstack/ironic-specs: Add indexable JSON fields spec
lucasagomesdevananda, jroll ^ it's a succinct spec, but hope the whole idea is there ^17:03
lucasagomesdtantsur|afk, have a good weekend!17:07
lucasagomesse eya17:07
mgouldhi everyone17:10
mgouldI'm gradually unsticking myself on
mgouldbut I have a question about the state machine17:10
mgouldlooking at, it looks like a node can go enroll -> verifying -> enroll17:11
mgouldwhich signals a failure17:11
mgould1) dtantsur|afk's initial patch only makes sense if a node can go enroll -> manageable -> enroll on failure; can this ever happen?17:12
mgould2) are we committed to this? It seems a very unreliable way to signal failure17:12
mgouldcheck node: "enroll"; sleep; check node: "enroll"17:12
*** david-lyle has quit IRC17:13
mgouldyou don't know if it's failed or just not started transitioning yet17:13
JayFI believe in that cast17:13
JayFnode[last_error] would be populated17:13
JayFFor our internal dashboards, we usually key off a combo of state, target state, and last error17:13
JayFto suss out error situations17:13
lucasagomesyeah, the last_error should indicate what went wrong when a transition actually failed17:14
devanandaJayF: as an aside, I've found several cases where last_error gets set by one process and cleared by another17:14
lucasagomessince when in enroll/manageable the node is not used yet I'm not sure if maintenace flag would make sense yet17:14
JayFdevananda: usually the power status loop in my observation17:15
devanandawe more and more need an event/error log // notifiation bus17:15
devanandaJayF: yah17:15
JayFI kinda like the way Nova does it17:15
JayFwith instance /errors/17:15
JayFwe could similarly have node errors in a separate table, perhaps17:15
mgouldOK, so my set_node_state function should check node[last_error], try to set the state, then loop until last_error changes or the node is in the target state?17:15
devanandaclient could ask for last error, or all errors since $ref17:16
mgouldeven better17:16
mgouldwait, is that available now, or something you're proposing?17:16
JayFmgould: node[last_error] always has the most recent node error17:17
mgouldyes, that sounds useful :-)17:17
JayFmgould: devananda and I have always hated that we only keep the most recent error, so just daydreaming about fixing it :)17:17
mgouldJayF, gotcha17:17
mgouldis last_error timestamped?17:17
*** cemason has joined #openstack-ironic17:17
*** rkharya has quit IRC17:17
*** alexpilotti has joined #openstack-ironic17:17
lucasagomesJayF, ++ having the concept of instance17:18
JayFmgould: no17:18
lucasagomesclaims is somewhat similar17:19
mgouldI'm wondering if I can tell "this happened before I started, but the process completed fine" from "this error happened before, and it happened again when you tried it"17:19
* sambetts is calling it a day 17:19
JayFlucasagomes: well, I was thinking more a "node errors" table that maps to the idea of Nova's instance errors table17:19
sambettsnight all o/17:19
JayFmgould: you can clear last error as a client, iirc17:19
mgouldsambetts, good night!17:19
lucasagomesJayF, ah right, yeah I was thinking differently17:19
JayFmgould: and for nodes in ENROLL state, not a whole lot (including power status loop) will be running on them yet17:19
*** sambetts is now known as sambetts-afk17:19
mgouldJayF, OK, that'll work17:19
mgouldJayF, good to know17:20
mgouldbut I also have to deal with available nodes17:20
mgould...which apparently can't fail to transition to manageable, so that's OK17:20
lucasagomesI was thinking more about having /instance (which is similar to claim) e.g, create an instance with those properties (mapped to the nodes) and deploy it, so we can track the errors at /instance/<instance uuid> and so on17:20
lucasagomesit's a big change17:21
lucasagomessambetts-afk, g'night!17:21
mgouldlucasagomes, that sounds like a Good Thing, but I don't know how hard it would be17:21
lucasagomesyeah changes our API a lot17:21
lucasagomesso probably it's a no for now17:21
lucasagomesbut v2 who knows we can better architect things17:21
*** shakamunyi has quit IRC17:21
lucasagomesnow we have experience dealing with this, when we first created the API was hard because we didn't know exactly how everything would work17:22
JayFlucasagomes: I don't even care about having huge in-depth instance information17:23
JayFlucasagomes: just simply keeping a historical "all the instance uuids ever on this node" would make some operability things 100x easier17:23
JayFlucasagomes: you can ask Nova that question today; but only via db and it's obtuse17:24
lucasagomesright, yeah17:24
lucasagomesthe instance_uuid field in ironic is very hmm "soft" idk how to say17:24
lucasagomescause for the standalone case it doesn't even eed to be populated17:24
lucasagomesbut yeah can be worked out17:25
openstackgerritLucas Alvares Gomes proposed openstack/ironic-specs: Add indexable JSON fields spec
* lucasagomes fixes line's too long errors17:25
*** kozhukalov_ has quit IRC17:29
lucasagomesdevananda, hi there, when you get some free time mind taking a peek at see if it's aligned with the suggestion you gave?17:31
*** vishwanathj has quit IRC17:35
openstackgerritLucas Alvares Gomes proposed openstack/ironic-specs: Add indexable JSON fields spec
*** cemason has quit IRC17:38
NobodyCamnoight sambetts-afk17:38
NobodyCamnighgt dtantsur|afk17:39
NobodyCamnight even17:39
NobodyCammorning devananda, JayF, mgould :)17:39
lucasagomesNobodyCam, morning17:40
mgouldNobodyCam, morning :-)17:41
*** alexpilotti has quit IRC17:44
*** alexpilotti has joined #openstack-ironic17:44
*** derekh has quit IRC17:45
NobodyCamam I remembering correctly that closes-bug tag reguires reno notes?17:46
mgouldNobodyCam, I think so17:46
*** alexpilotti has quit IRC17:48
*** cemason has joined #openstack-ironic17:49
openstackgerritMerged openstack/bifrost: Initial addition of an inspection module
JayFlucasagomes: not sure I grok what node_driver_info and node_properties have that are different from driver_info and properties (in your indexable fields spec)17:52
*** cemason has quit IRC17:53
lucasagomesJayF, driver_info and properties are today a column in the node's table. In that spec each one would be a separated table17:53
lucasagomesso one could query stuff like17:53
JayFI grok that part17:53
JayFbut why replace the two fields with 4 tables?17:53
lucasagomesSELECT node_id WHERE property_id=(SELECT id FROM properties WHERE field_name='memory_mb') AND value='1024'17:54
lucasagomesJayF, so we can query on it17:55
lucasagomeswhere having a blob makes it impossible17:55
lucasagomeswell not impossible, but inefficient17:55
JayFOH, because we need one big flat table for all nodes17:56
JayFso you have to index what maps to what17:56
JayFTook me a second to understand, thanks for spelling it out17:56
JayFI've interacted with a lot of API + SQL schemas, just now learning how to build them. That makes a lot of sense, thanks17:57
lucasagomesJayF, that spec is just plumbing work. After that we can go ahead and put an API on top of it17:57
lucasagomeswhich then can be used for claims as well17:57
lucasagomesJayF, no problem17:57
lucasagomesbut yeah keep in mind that, that spec alone doesn't bring any apparent benefit. Because it's all internal stuff17:58
mgouldso each node can have many properties; each property has a name and a JSON value; most property names are common to many nodes?17:58
lucasagomesmgould, exactly17:58
lucasagomesmgould, pretty much all nodes have cpus, memory_mb, local_gb, capabilities and so on17:58
lucasagomesit's duplicated across all nodes, so we only need to register it once17:59
mgouldand currently we have a "properties" field in the nodes table, which is a string containing JSON?17:59
lucasagomesmgould, exactly17:59
lucasagomesit's a JSON blob stored in the node's table17:59
mgouldso an alternative would be to make that field a JSON field rather than a VARCHAR, and use MySQL's JSON-indexing features18:00
lucasagomesmgould, yeah well the value has to be json because we currently support things like nested dicts in those fields18:01
lucasagomescause they are json fields18:01
mgouldsure, yes18:01
lucasagomesso, I want to keep the same18:01
lucasagomesoslo.db does have a JSONType that I will use for that18:01
JayFMySQL has a native JSON field type now?18:01
JayFTIL; I thought only Postgres did18:01
mgouldJayF, apparently so:
mgouldwait, sorry, my link's fine18:02
mgouldlucasagomes, I'm talking about the node table having a native-JSON field containing eg {'cpus': 3, 'capabilities': ['things', 'stuff', 'more things']}18:03
mgouldie, what we have now but using MySQL's native JSON type rather than VARCHAR18:03
lucasagomesoh I see18:03
lucasagomesso we wouldn't need to dumps() loads()18:03
mgouldyeah, I think so18:04
mgouldand searching can be fast because the DB holds it all in an indexed form18:04
lucasagomesright so we wouldn't even need to change our db model18:04
mgouldNB I have never actually used this feature :-)18:04
mgouldwhat version of MySQL do we require?18:04
lucasagomeswe use an oRM18:04
mgouldlucasagomes, right18:04
*** praneshp has joined #openstack-ironic18:05
mgouldwe'd need to migrate the field type18:05
lucasagomesso we abstract the backend18:05
lucasagomesmgould, yeah18:05
mgouldah, OK18:05
*** raildo has joined #openstack-ironic18:05
lucasagomesmgould, that's the problem for us now18:05
mgouldso if the ORM doesn't support native JSON types, my plan won't work18:05
raildopas-ha: ping, are you around?18:06
lucasagomesmgould, I will look into it see if I can find something18:06
mgouldlucasagomes, cool18:06
*** Piet__ has quit IRC18:06
lucasagomesbut I don't think it has something like that because we use things like sqllite for tests and so on18:07
mgouldah, right18:07
mgouldI'll leave a comment on your patch saying "this could be an alternative if the ORM supports it" - would that make sense?18:07
lucasagomesmgould, totally18:07
mgouldhang on, you're specifying the type of field_value as JSON18:08
mgouldwill that work if the ORM doesn't support it?18:08
*** jcoufal has quit IRC18:08
*** cemason has joined #openstack-ironic18:08
mgouldI suppose it might support fetch and update of JSON types, but not querying18:08
lucasagomesmgould, this is the oslo.db type18:09
lucasagomesbut that dumps() and loads()18:09
*** aginwala has joined #openstack-ironic18:09
lucasagomesmgould, that's wrong in the spec actually, it should be TEXT18:09
* lucasagomes changes it18:09
mgouldoh well18:10
mgouldPSA: native JSON types exist in both MySQL and Postgres now :-)18:10
* lucasagomes will wait for mgould comment so he can add the alternative thing18:11
lucasagomesthanks a lot18:12
mgouldyeah, looks like JSON hasn't made it into SQLAlchemy's MySQL backend yet18:13
mgouldthough the Postgres backend supports Postgres' JSON type :-)18:14
*** aginwala has quit IRC18:14
mgouldright, I have to go to the first ever Red Hat Scotland christmas party :-)18:15
mgouldat which I will meet most of the Scottish Red Hatters for the first time18:15
mgouldhave a good weekend, everyone!18:15
*** rajathagasthya has joined #openstack-ironic18:16
lucasagomesmgould, thanks for the suggestions/comments! Enjoy the weekend18:17
lucasagomesenjoy the party as well18:17
*** mgould has quit IRC18:18
openstackgerritLucas Alvares Gomes proposed openstack/ironic-specs: Add indexable JSON fields spec
*** trown is now known as trown|lunch18:24
openstackgerritJulia Kreger proposed openstack/bifrost: WIP: Canary test commit **DO NOT MERGE**
*** kkoski has quit IRC18:28
*** vishwanathj has joined #openstack-ironic18:32
*** Marga_ has joined #openstack-ironic18:33
*** Marga_ has quit IRC18:34
*** Marga_ has joined #openstack-ironic18:35
*** vishwanathj has quit IRC18:36
*** penick has joined #openstack-ironic18:37
*** jaybeale has joined #openstack-ironic18:38
*** cemason has quit IRC18:40
*** kozhukalov_ has joined #openstack-ironic18:41
*** aginwala has joined #openstack-ironic18:46
*** kkoski has joined #openstack-ironic18:46
*** Nisha_brb has joined #openstack-ironic18:48
*** Nisha_away has quit IRC18:48
openstackgerritMerged openstack/ironic: Fix bug where clean steps do not run
JayFfinally \o/18:50
*** aginwala has quit IRC18:52
*** ChubYann has joined #openstack-ironic18:53
*** aginwala has joined #openstack-ironic18:54
*** aginwala has quit IRC18:58
*** electrofelix has quit IRC19:00
*** aginwala has joined #openstack-ironic19:01
*** penick has quit IRC19:05
*** david-ly_ has joined #openstack-ironic19:06
*** HimanshuGarg has joined #openstack-ironic19:08
*** pas-ha has quit IRC19:12
*** HimanshuGarg has quit IRC19:13
devanandatime to tag that release, eh? jroll - you around, or shall I do it?19:13
*** trown|lunch is now known as trown19:14
*** jerrygb has joined #openstack-ironic19:16
*** jerrygb_ has quit IRC19:18
jrolldevananda: go for it, I'm about to hit the road. thank you!19:28
jrolldon't forget liberty :P19:28
*** penick has joined #openstack-ironic19:29
devanandajroll: have a great weekend :)19:29
jrollthanks, you too!19:30
* jroll off to STL19:30
*** Sukhdev has joined #openstack-ironic19:33
*** HimanshuGarg has joined #openstack-ironic19:33
lucasagomesjroll, enjoy!19:35
NobodyCamhave a good weekend jroll19:39
*** david-ly_ is now known as david-lyle19:39
*** rajathagasthya has quit IRC19:41
Nisha_brbdevananda, i was testing manual cleaning through curl19:41
NobodyCammgoddard: around? wanted to check if your still working on ???19:42
Nisha_brbi am facing an issue, could you/anyone else tell me what am i doing wrong19:43
Nisha_brbHere is the paste19:43
Nisha_brbNobodyCam, ^^^19:43
openstackgerritJohn L. Villalovos proposed openstack/ironic: Documentation points to official release notes
*** rajathagasthya has joined #openstack-ironic19:49
*** jaybeale has quit IRC19:50
*** mgarza_ has joined #openstack-ironic19:51
openstackgerritLucas Alvares Gomes proposed openstack/ironic-inspector-specs: test
*** Marga_ has quit IRC19:53
*** mgarza_ has quit IRC19:53
*** rajathagasthya has quit IRC19:53
*** mjturek1 has quit IRC19:53
openstackgerritMilan Kováčik proposed openstack/ironic-inspector-specs: High Availability for Ironic Inspector
raildopas-ha: ping, are you around? it's about two bugs that you're working :) #1494776 and #142263219:59
*** ChuckC has joined #openstack-ironic20:05
NobodyCamNisha_brb: maybe
lucasagomesfolks calling it a day20:08
lucasagomeshave a great night everyone! Enjoy tehe weekend20:08
NobodyCamlucasagomes: have a great weekend :)20:08
*** baoli_ has quit IRC20:09
lucasagomesNobodyCam, you too! See ya20:10
*** lucasagomes is now known as lucas-afk20:10
*** mkovacik_ has quit IRC20:10
*** Nisha_brb has quit IRC20:12
*** shakamunyi has joined #openstack-ironic20:13
*** aginwala has quit IRC20:13
*** aginwala has joined #openstack-ironic20:17
*** mgoddard has quit IRC20:25
*** mgoddard has joined #openstack-ironic20:26
*** alexpilotti has joined #openstack-ironic20:26
mgoddardNobodyCam: I'm around briefly20:27
NobodyCamhey hey mgoddard20:27
NobodyCamI was looking at
mgoddardI'm not actively working on that bug20:28
*** aginwala has quit IRC20:28
NobodyCambut you implatmented a solution based on that for your own devices?20:28
mgoddardI proposed a solution, and was hoping for feedback but didn't get much20:28
mgoddardyes we're using a solution based on the same idea20:29
NobodyCamya was reading theu the bug20:29
mgoddardwould be great if the thread could be awoken - it seems like a fairly big limitation of Ironic IMHO20:30
*** aginwala has joined #openstack-ironic20:31
mgoddardif you're willing and able to nudge Nova reviews along then would appreciate a vote (for or against) on
*** vishwanathj has joined #openstack-ironic20:33
*** alexpilotti has quit IRC20:33
NobodyCamnot sure if I can influnce nova review but will take a look in a few20:36
mgoddardgreat, thanks20:36
*** Marga_ has joined #openstack-ironic20:37
*** vishwanathj has quit IRC20:38
*** alexpilotti has joined #openstack-ironic20:40
*** kozhukalov_ has quit IRC20:43
*** cemason has joined #openstack-ironic20:48
*** cemason has quit IRC20:53
*** alexpilotti has quit IRC20:53
*** raildo is now known as raildo-afk20:54
*** alexpilotti has joined #openstack-ironic20:55
*** aginwala has quit IRC20:57
JayF#success Cleaning works in Ironic again.20:58
openstackstatusJayF: Added success to Success page20:58
*** ionutbalutoiu has quit IRC20:59
*** ionutbalutoiu has joined #openstack-ironic20:59
*** rajathagasthya has joined #openstack-ironic20:59
*** penick has quit IRC21:01
*** mjturek1 has joined #openstack-ironic21:02
*** mbound_ has joined #openstack-ironic21:03
*** Sukhdev has quit IRC21:04
*** penick has joined #openstack-ironic21:06
NobodyCamoh nice!21:07
*** baoli has joined #openstack-ironic21:07
*** alexpilotti has quit IRC21:09
*** aginwala has joined #openstack-ironic21:14
*** vishwanathj has joined #openstack-ironic21:20
*** boris-42_ has joined #openstack-ironic21:27
*** alexpilotti has joined #openstack-ironic21:30
JayFdevananda: if you get a chance, mariojv's notifications spec has two +2s on it (one of them mine). You had comments on it earlier so wanted to give you a shot to review it again before it landed (my hope is you'll make it 3x+2 and land it)21:32
mariojvthanks JayF. here's a link to the spec:
devanandaJayF: awesome21:34
devanandamariojv: JayF: oh, good point about shipping the Node obj around. it can potentially be LARGE21:38
devanandaalso contain sensitive info21:38
JayFwe should probably strip instance_info as well as any creds21:38
devanandaeg, it can contain a base64 configdrive21:38
JayFHeh, yep21:38
JayFso it should be sanitized some21:38
devanandastrip instance_info, driver_info21:38
devanandaactually, why not just pass the node UUID?21:39
devanandadon't assume that the event receiver has privileges sufficient to access Ironic21:39
mariojvwe might want to retain some information about the node in the notification, like its state and target state at the time the notification had to be sent21:39
devanandamariojv: for an event related to a state transition - yes, send that information21:39
devanandafor an event related to node creation / deletion, dont send that information21:40
mariojvhere's an example of what nova sends if anyone's curious:
JayFmariojv: it just occurred to me why you picked up the notifications piece -- you worked with nova notifications a lot in your pre-onmetal project at Rackspace, didn't you?21:42
mariojvyes, mostly from a consumer standpoint of using it in the context of metrics21:43
JayFthat probably means you have a good perspective on what those notifications should include21:43
JayFI mean, it sounds like devananda is arguing it shouldn't be "one size fits all" (i.e. a node object with a few things stripped) but maybe sending along the relevant fields instead?21:44
* JayF not 100% sure he followed but that's what I thought was meant21:44
NobodyCamdevananda: just wanted to check and see if you've put any more thought in to
mariojvwell, i'd like to send just the relevant fields, but that's not always trivial to figure out for something like a .error event21:44
JayFError might be the case where you ship as much stuff as possible :)21:45
JayF(implied: barring sensitive info)21:45
mariojvJayF: for example, downstream we have the idea of "moods" where there are known combinations of things that are symptoms of problems in our environment21:45
JayFmariojv: fwiw a versino of that dash was open sourced for one of JoshNang's talks (in Vancouver)21:46
mariojvyeah, couldn't find the link21:46
JayFmariojv: I'd suspect, though, we'd want those notifications to have all the info we'd use in a dash like that to key off "what happened"21:46
mariojvsometimes i use downstream to mean non-openstack21:46
JayFbut that's generally: 1) What flavor is it? 2) What state is it in and targetting? 3) Is it locked? 4) What does last_error say? 5) Is it maint'd21:47
mariojvsometimes power state, instance uuid, time since last heartbeat as well - but yes they're finite21:48
JayFI mean, I think there are two key pieces of info, right?21:49
mariojvgenerally for each mood21:49
*** aginwala has quit IRC21:49
*** alexpilotti has quit IRC21:49
mariojvin either case i don't want to presume that the fields we're interested in are the ones that everyone is interested in21:49
JayFStuff about the node itself (power state, node.properies)21:49
devanandaJayF: yes, your understanding my my point is correct21:49
JayFand stuff about what was happening (states, requested transition, etc)21:50
mariojvso i'm more prone to the idea that we should ship more fields rather than less, but we shouldn't send fields that don't have any limit on their size21:50
*** mgoddard has quit IRC21:50
JayFI think that's a pretty reasonable course of action, actually21:50
devanandasending data which is relates to the node, but not the event notificatoin itself, is not helpful -- it might be stale by the time it is received and processed21:50
JayFdevananda: I throughly disagree21:51
devanandawe should only send data related to the event21:51
devanandaJayF: great! why?21:51
JayFdevananda: in troubleshooting strange stuff, I've oft wished I had a snapshot of what the node itself looked like at the time21:51
JayFdevananda: I guess the question is; do we expect the notification into to be used for troubleshooting? If so, I want as much info as possible, because the point-in-time snapshot is a very useful input to that troubelshooting21:51
mariojvwe use notifications as a data store for systems that help us "debug" our cloud21:51
JayFI mean, we just stuff notifications into elasticsearch and query them in fun ways21:52
JayFfor nova21:52
mariojvwe also do other things with them21:52
JayFmariojv: oh, I'm sure you know of a million more than I do :)21:52
mariojvfor performance monitoring, for example21:52
devanandaJayF: yea, that's fair. if you're using notifications as a PIT snapshot so you can reconstruct the changes a node went through21:52
JayFdevananda: that's /exactly/ what I had in mind when saying that21:53
devanandathen it makes sense to put as much data, even sensitive data, as possible21:53
devanandawhat if the password changed and that caused the BMC to be unreachable and trigger maintenance mode?21:53
JayFI mean, IDK about sensitive data though?21:53
devanandasanitizing the payload would prevent debugging that21:53
devanandait's terrible from a security POV21:53
JayFdevananda: you have no idea how close to a bullseye on a recent problem we had you are :)21:53
*** aginwala has joined #openstack-ironic21:53
JayFI mean, it's a continuum, right?21:54
JayFMost of the data is not-sensitive and helpful, so including it seems sensible21:54
mariojvyeah. i see this going towards a more per-use-case sort of configurability21:54
devanandaback to the premise -- what is the notification bus for?21:54
mariojvthis is something we said we wanted to avoid at the summit, and have it standardized21:54
*** vishwana_ has joined #openstack-ironic21:54
devanandaa) debugging / recreating a series of events21:54
devanandab) informing other services of specific changes in state, to enable those services to take an action21:54
mariojvb) sending notifications to inform external services of when ironic ...  that ^21:54
JayFWhy can't we have both?21:55
devanandamy point above -- we should not be sending the whole Node object for (b)21:55
devanandawe should only send the relevant portions for that notification21:55
*** Sukhdev has joined #openstack-ironic21:55
devanandabecause the service which acts upon that notification probably needs to know the CURRENT state of hte node, when it processes that notification21:55
devananda"hey, this node was added, and it had 0GB of RAM at enrollment" --> "the node now has 64GB of RAM"21:56
JayFWhat, other than a debugging tool, might you want to act on these notifications?21:56
devananda"this node failed to transition to power_off" --> the node is currently off, so don't do anything21:56
devanandaJayF: uh, Nova and Neutron21:56
mariojvi'm kind of interested in that21:56
JayFdevananda: OH!21:56
JayFdevananda: So no more polling?21:56
mariojvwill we assume the message bus is lossy?21:56
devanandaJayF: another inventory management system21:56
devanandaJayF: exactly!!21:56
devanandamariojv: yes21:56
*** vishwanathj has quit IRC21:56
devananda[CMDB] receives notification that Node N provision_state transition to active has started21:57
devananda[CMDB] quickly changes local property "managed: False", then starts up a watchdog to keep an eye on the server21:58
devanandathings like that21:58
JoshNangmariojv: i think if you're going to have nova/neutron actions depending on bus notifications, you're going to need to ensure it isn't lossy21:58
devanandaJoshNang: o/21:59
mariojvi understand why some of the fields would be unnecessary for b), but i don't understand how being more verbose would prevent the notification bus from being consumed by nova or neutron21:59
JayFJoshNang: ++ I was thinking the same21:59
mariojvJoshNang: hi!21:59
JayFand if we can ensure, for instance, API->Conductor bus messages aren't lossy21:59
JoshNangpretty sure most message busses will support "ensure at least once"..."ensure exactly once" isn't doable iirc21:59
JayFwhy can't we do the same for notifications?21:59
devanandaJoshNang: we'll still need to poll periodically to catch anything that fell through the cracks21:59
devanandaJayF: we can't ensure that21:59
mariojvdevananda: ++21:59
JayFdevananda: How do we handle a lost message between an API and a Conductor?22:00
NobodyCamhey JoshNang welcome back :)22:00
devanandaJayF: we currently abuse rabbit horribly in Ironic22:00
JoshNangNobodyCam: o/ :D22:00
JoshNangsaw a mention and found the conversation interesting22:00
devanandaJayF: right now, we're using rabbit for RPC -- not as a message bus -- and so we send an ACK for every message22:00
mariojvFWIW, oslo.messaging supports multiple drivers22:00
devanandaif that ACK isn't receivedin X time, we raise an error22:00
JayFdevananda: Ahh, of course, that makes sense22:01
devanandaJayF: we are not using it as a queue at all22:01
devanandabut for notifications, we need to22:01
devanandawe won't (and can't) know who all the receivers are22:01
mariojvnot a lot though:
devanandanow, there are durable lossless message queues out there22:02
*** penick has quit IRC22:02
devanandabut scaling those to run at the volume of messages openstack passes around? yea, fun.22:02
*** trown is now known as trown|outttypeww22:02
devanandait would need to be a different message bus than the rabbit bus we use between services22:03
mariojvoslo messaging provides a separate notifications endpoint config option (separate from rpc)22:05
*** penick has joined #openstack-ironic22:06
JayFmariojv: so it sounds like maybe we should talk more specifically about the use case(s) in the spec and change the payload accordingly22:09
mariojvi really think that is something that should be hashed out in the implementation22:10
mariojvworried about scope creep where we define each notification payload for everything we'd want in the future in the spec22:10
JayFI maybe agree with you w/r/t that22:10
JayFbut the use cases need to be explicitly spelled out22:11
*** thrash is now known as thrash|wknd22:11
JayFbecause there may be an explicit decision, based on some of the chatter in here, that we could reject the "debug" use case22:11
mariojvi think problem description talks about both, but it could have more detail on our specific case22:11
JayFI mean, either way, they are clearly two cases that call for radically different contents of the notification22:12
JayFunless we want to "pile on" the whole node obj minus a few things22:12
JayFwhich the more I think of it seems like the type of thing we may regret in a couple of years as the node object may grow22:12
mariojvi agree. that's why they're versioned, so we can deprecate things :D22:14
mariojv(kind of kidding, but we're avoiding a huge problem nova had for a while)22:14
JayF(in 2 years) <mariojv> I have to deprecate things D:22:15
JayFwhat's the huge problem nova had for a while22:15
*** jerrygb has quit IRC22:15
mariojvnot having versioned or standardized notifications22:15
mariojvi don't know of specific cases where that broke people, but it was very annoying22:15
mariojvsorry about link split22:16
*** aginwala_ has joined #openstack-ironic22:20
*** aginwala_ has quit IRC22:21
*** aginwala has quit IRC22:22
*** aginwala_ has joined #openstack-ironic22:22
mariojvso, let me double check it's clear what should be added to the spec, and maybe what should be left for later22:22
mariojv1) definitely add more info about the two uses cases we've been discussing, since people who read the spec weren't clear on it, it needs more detail22:23
mariojv2) for the details of how use cases are isolated from each other, either a) add a config option for verbosity, or b) leave that for implementation22:23
mariojvi prefer 2b since more use cases may come out that might not fit with 1 config option22:24
JayFI suspect we aren't going to get the spec approved without figuring out the verbosity ahead of time, but we'll see :)22:24
mariojvdevananda: JayF: does that cover it, or is there more? ^22:24
JayFmariojv: I'd also suggest putting this question on the agenda for the ironic meeting business-tomorrow22:24
mariojvJayF: is that something i need to request to have added as a topic myself, or something for open discussion?22:26
mariojvi see on the wiki that we need 2 days in advance to have a new topic, not sure if that's something i just request in irc or if there's a form or something22:27
JayFmariojv: I generally just edit the wiki and add things to the agenda. I don't think there's an approval process.22:27
JayFmariojv: I'd add it to "open discussion" if that makes you feel strange :)22:27
JayFmariojv: if you weren't supposed to, just remember how harsh of a master jroll is22:27
*** david-ly_ has joined #openstack-ironic22:29
mariojv /nick mar:P22:31
mariojv"Erroneous nickname" :(22:31
*** david-ly_ is now known as david-lyle_22:31
JayFmar[tab]: :P22:31
*** david-lyle has quit IRC22:32
*** rvasilets___ has joined #openstack-ironic22:32
*** dprince has quit IRC22:37
openstackgerritMerged openstack/bifrost: Additional inspection role and adding role to NV job script
*** ionutbalutoiu has quit IRC22:40
*** kkoski has quit IRC22:43
devanandamariojv: no form. I'll add it22:44
mariojvdevananda: already added it to the wiki, but feel free to make any modifications. thanks22:44
devanandamariojv: yea, that doesn't follow our layout template22:45
devanandamariojv:      Example topic: (devananda) Let's talk about zebras. Reference:
devanandaput a line like that under the Discussion heading, pls22:45
*** aginwala_ has quit IRC22:46
devanandaactually - this qualifies as a stuck spec22:46
devanandastuck since you need moar input22:46
JayFI think also some decisioneering needs to happen too22:46
JayFi.e. I'm not sure we can talk our way out of some folks wanting bigger notifications vs folks thinking we need smaller22:46
JayFbut idk22:46
*** penick has quit IRC22:48
*** aginwala has joined #openstack-ironic22:49
openstackgerritArun S A G proposed openstack/ironic-specs: Add spec for deploying OS tarballs
JayFOoh, I bet that's interesting22:51
devanandaJayF: IPA already can write an MBR, so ^ isn't much of a stretch22:54
JayFOh absolutely22:54
devanandait comes very close to the area traditionally held by installers, but I think it's still on this side -- there's a binary image (of files, not a filesystem)22:55
JayFI have no problem whatsoever with it in theory22:55
*** jaybeale has joined #openstack-ironic22:55
*** [1]cdearborn has quit IRC22:57
*** rajathagasthya has quit IRC23:05
*** dims has quit IRC23:10
*** Sukhdev has quit IRC23:12
*** ionutbalutoiu has joined #openstack-ironic23:20
*** rajathagasthya has joined #openstack-ironic23:28
*** mbound_ has quit IRC23:31
*** openstack has joined #openstack-ironic23:38
*** Sukhdev has joined #openstack-ironic23:45
*** penick has joined #openstack-ironic23:45
*** dims has joined #openstack-ironic23:47
*** rajathagasthya has quit IRC23:51
*** penick has quit IRC23:51
*** penick has joined #openstack-ironic23:53

Generated by 2.14.0 by Marius Gedminas - find it at!