| *** achanda has quit IRC | 00:02 | |
| *** romcheg has quit IRC | 00:12 | |
| *** naohirot has joined #openstack-ironic | 00:13 | |
| *** mtanino_ has joined #openstack-ironic | 00:24 | |
| *** davideagnello has quit IRC | 00:24 | |
| *** mtanino has quit IRC | 00:25 | |
| *** saripurigopi has joined #openstack-ironic | 00:26 | |
| openstackgerrit | Devananda van der Veen proposed openstack/ironic: Open Liberty development https://review.openstack.org/171274 | 00:27 |
|---|---|---|
| mrda | \o/ | 00:35 |
| *** ijw has quit IRC | 00:35 | |
| *** zhenguo has joined #openstack-ironic | 00:39 | |
| openstackgerrit | jxiaobin proposed openstack/ironic-specs: Mount config drive as loop device to supply data to cloud-init https://review.openstack.org/173142 | 00:59 |
| *** meghal has quit IRC | 01:02 | |
| *** jmank has quit IRC | 01:02 | |
| *** dttocs has quit IRC | 01:13 | |
| *** kkoski has joined #openstack-ironic | 01:15 | |
| openstackgerrit | jxiaobin proposed openstack/ironic-specs: Mount config drive as loop device to supply data to cloud-init https://review.openstack.org/173142 | 01:23 |
| *** chenglch has joined #openstack-ironic | 01:25 | |
| *** jxiaobin has quit IRC | 01:29 | |
| *** kkoski has quit IRC | 01:33 | |
| *** kkoski has joined #openstack-ironic | 01:38 | |
| *** wuhg has joined #openstack-ironic | 01:53 | |
| *** jerryz has joined #openstack-ironic | 02:02 | |
| *** mtanino_ has quit IRC | 02:04 | |
| *** chenglch|2 has joined #openstack-ironic | 02:04 | |
| *** chenglch has quit IRC | 02:07 | |
| *** davideagnello has joined #openstack-ironic | 02:10 | |
| openstackgerrit | Haomeng,Wang proposed openstack/python-ironicclient: (WIP)validate input uuid from client side https://review.openstack.org/172618 | 02:13 |
| *** achanda has joined #openstack-ironic | 02:21 | |
| *** saripurigopi has quit IRC | 02:21 | |
| *** thrash is now known as thrash|g0ne | 02:26 | |
| openstackgerrit | Haomeng,Wang proposed openstack/python-ironicclient: validate input uuid from client side https://review.openstack.org/172618 | 02:36 |
| *** davideagnello has quit IRC | 02:41 | |
| *** ramineni has joined #openstack-ironic | 02:44 | |
| *** rloo has quit IRC | 02:53 | |
| *** wuhg has quit IRC | 03:06 | |
| *** ndipanov has quit IRC | 03:06 | |
| *** sambetts has quit IRC | 03:06 | |
| *** arif-ali has quit IRC | 03:06 | |
| *** wuhg has joined #openstack-ironic | 03:13 | |
| *** ndipanov has joined #openstack-ironic | 03:13 | |
| *** sambetts has joined #openstack-ironic | 03:13 | |
| *** arif-ali has joined #openstack-ironic | 03:13 | |
| *** achanda has quit IRC | 03:14 | |
| *** dttocs has joined #openstack-ironic | 03:19 | |
| openstackgerrit | Haomeng,Wang proposed openstack/python-ironicclient: validate input uuid from client side https://review.openstack.org/172618 | 03:20 |
| openstackgerrit | Merged openstack/ironic: Open Liberty development https://review.openstack.org/171274 | 03:21 |
| *** saripurigopi has joined #openstack-ironic | 03:22 | |
| openstackgerrit | Haomeng,Wang proposed openstack/python-ironicclient: validate input uuid from client side https://review.openstack.org/172618 | 03:23 |
| openstackgerrit | Merged openstack/ironic: Fix typos in vendor-passthru.rst https://review.openstack.org/172934 | 03:39 |
| openstackgerrit | Merged openstack/ironic: Update session headers during initialization of AgentClient https://review.openstack.org/172154 | 03:39 |
| *** harlowja is now known as harlowja_away | 03:58 | |
| openstackgerrit | jxiaobin proposed openstack/ironic-specs: Mount config drive as loop device to supply data to cloud-init https://review.openstack.org/173142 | 04:08 |
| *** chenglch|2 has quit IRC | 04:21 | |
| *** davideagnello has joined #openstack-ironic | 04:29 | |
| *** davideagnello has quit IRC | 04:34 | |
| *** krtaylor has quit IRC | 04:35 | |
| *** krtaylor has joined #openstack-ironic | 04:38 | |
| * naohirot white board cannot be loaded due to error https://etherpad.openstack.org/p/IronicWhiteBoard | 04:40 | |
| * naohirot it seems all etherpad cannot be loaded? I couldn't load it too, https://etherpad.openstack.org/p/IronicReviewDay | 04:43 | |
| mrda | :( | 04:44 |
| mrda | naohirot: both work for me | 04:46 |
| naohirot | mrda: really? It would be my proxy or cache problem, I'll try to check. | 04:48 |
| mrda | Could be platform? Are you on Linux? What browser? | 04:48 |
| naohirot | mrda: error said "Error: Attempt to load undefined module. in https://etherpad.openstack.org/static/js/require-kernel.js (line 1)" | 04:48 |
| mrda | Both work for me right away | 04:49 |
| mrda | And https://etherpad.openstack.org/static/js/require-kernel.js loads for me too | 04:50 |
| pshige | me to. I can access both pages. | 04:51 |
| naohirot | mrda: It worked on another PC, so something wrong in my desktop pc. | 04:51 |
| mrda | Sorry to be the bearer of bad news... | 04:51 |
| *** achanda has joined #openstack-ironic | 04:52 | |
| *** jcoufal has joined #openstack-ironic | 04:53 | |
| naohirot | mrda: thanks! I could notice earlier :) now I'm going to quickly fill in my status. | 04:54 |
| mrda | naohirot: np | 04:57 |
| devananda | o/ | 04:57 |
| mrda | hey devananda! | 04:57 |
| devananda | anyone around for the meeting? :) | 04:57 |
| mrda | of course :) naohirot and I are here | 04:58 |
| naohirot | devananda: good evening :) | 04:58 |
| devananda | k k | 04:58 |
| *** rameshg87 has joined #openstack-ironic | 04:58 | |
| devananda | i'm hanging on to wakefulness ;) | 04:58 |
| rameshg87 | good morning ironic | 04:58 |
| mrda | hi rameshg87 | 04:58 |
| rameshg87 | hello mrda | 04:58 |
| pshige | devananda: good evening :) | 04:59 |
| *** oomichi has joined #openstack-ironic | 05:13 | |
| pshige | oomichi: weekly irc meeting is now on #openstack-meeting-3 | 05:15 |
| *** dttocs has quit IRC | 05:15 | |
| *** dttocs has joined #openstack-ironic | 05:23 | |
| * devananda goes to bed | 05:25 | |
| rameshg87 | good night devananda | 05:26 |
| pshige | devananda: good night | 05:26 |
| *** lintan_ has joined #openstack-ironic | 05:26 | |
| openstackgerrit | Yuiko Takada proposed stackforge/ironic-discoverd: Use keystonemiddleware auth credentials https://review.openstack.org/170782 | 05:42 |
| Haomeng | devananda: good night:) | 05:47 |
| openstackgerrit | Haomeng,Wang proposed openstack/python-ironicclient: validate input uuid from client side https://review.openstack.org/172618 | 05:56 |
| openstackgerrit | Haomeng,Wang proposed openstack/python-ironicclient: validate input uuid from client side https://review.openstack.org/172618 | 06:00 |
| lintan_ | Hi Haomeng | 06:04 |
| Haomeng | lintan_: pong:) | 06:04 |
| lintan_ | I am a little confused about the state machine, do we still have zapping or not? | 06:05 |
| *** chenglch has joined #openstack-ironic | 06:05 | |
| lintan_ | Haomeng, or we only have cleaning state? | 06:06 |
| Haomeng | lintan_: I think so - https://github.com/openstack/ironic/blob/master/ironic/common/states.py#L270 | 06:09 |
| Haomeng | lintan_: :) | 06:10 |
| *** devananda has quit IRC | 06:10 | |
| *** krtaylor has quit IRC | 06:11 | |
| lintan_ | thanks :) Haomeng | 06:11 |
| Haomeng | lintan_: welcome:) | 06:11 |
| *** devananda has joined #openstack-ironic | 06:17 | |
| *** krtaylor has joined #openstack-ironic | 06:22 | |
| *** dttocs has quit IRC | 06:29 | |
| *** dttocs has joined #openstack-ironic | 06:30 | |
| *** Marga_ has quit IRC | 06:33 | |
| pshige | Haomeng: we already have both cleaning and zapping state, but we have no real implementation of zapping at this moment, I think. Is it right? | 06:33 |
| *** chenglch has quit IRC | 06:34 | |
| Haomeng | pshige: I understand we define *clean* status now, and have no such *zapping* state defination, so we handle both cleaning/zapping for same logic code | 06:34 |
| *** dttocs has quit IRC | 06:35 | |
| pshige | Haemeng: Thank you | 06:36 |
| Haomeng | pshige: welcome | 06:37 |
| pshige | and there are no real implementation of zapping, not even HP iLO. | 06:39 |
| *** kozhukalov has joined #openstack-ironic | 06:39 | |
| pshige | I wonder which of them is lintan's question ... | 06:42 |
| Haomeng | pshige: :) | 06:44 |
| *** ukalifon has joined #openstack-ironic | 06:50 | |
| openstackgerrit | Tan Lin proposed openstack/ironic-specs: Bare Metal Trust Using Intel TXT https://review.openstack.org/133902 | 06:58 |
| *** andreykurilin has joined #openstack-ironic | 07:00 | |
| *** yuriyz has joined #openstack-ironic | 07:00 | |
| *** jamielennox is now known as jamielennox|away | 07:02 | |
| *** chenglch|2 has joined #openstack-ironic | 07:05 | |
| *** Nisha has joined #openstack-ironic | 07:09 | |
| *** dmellado has quit IRC | 07:10 | |
| *** a1exhughe5 has joined #openstack-ironic | 07:12 | |
| *** andreykurilin has quit IRC | 07:13 | |
| *** dmellado has joined #openstack-ironic | 07:24 | |
| *** achanda has quit IRC | 07:26 | |
| *** chlong has quit IRC | 07:27 | |
| *** jistr has joined #openstack-ironic | 07:29 | |
| *** dttocs has joined #openstack-ironic | 07:30 | |
| *** dtantsur|afk is now known as dtantsur | 07:35 | |
| dtantsur | Morning Ironic | 07:35 |
| pshige | dtantsur: morning | 07:35 |
| *** dttocs has quit IRC | 07:35 | |
| mrda | hey dtantsur | 07:40 |
| *** ndipanov has quit IRC | 07:41 | |
| dtantsur | o/ | 07:41 |
| *** ndipanov has joined #openstack-ironic | 07:42 | |
| *** ifarkas has joined #openstack-ironic | 07:42 | |
| openstackgerrit | chenglch proposed openstack/ironic-specs: Add console log support https://review.openstack.org/164586 | 07:49 |
| *** pas-ha has joined #openstack-ironic | 07:50 | |
| *** yog__ has joined #openstack-ironic | 07:56 | |
| *** mgoddard has joined #openstack-ironic | 07:57 | |
| *** yog__ has quit IRC | 07:57 | |
| *** wuhg has quit IRC | 07:59 | |
| *** jcoufal_ has joined #openstack-ironic | 08:06 | |
| *** jcoufal has quit IRC | 08:09 | |
| *** derekh has joined #openstack-ironic | 08:10 | |
| *** kan__ has joined #openstack-ironic | 08:14 | |
| *** athomas has joined #openstack-ironic | 08:16 | |
| *** lintan_ has quit IRC | 08:17 | |
| openstackgerrit | Ramakrishnan G proposed openstack/ironic-specs: Add new boot interface in Ironic https://review.openstack.org/168698 | 08:22 |
| *** lucasagomes has joined #openstack-ironic | 08:26 | |
| *** edcragg has joined #openstack-ironic | 08:26 | |
| *** dttocs has joined #openstack-ironic | 08:32 | |
| *** yog__ has joined #openstack-ironic | 08:34 | |
| openstackgerrit | Ramakrishnan G proposed openstack/ironic-specs: Add generic RAID configuration spec for liberty https://review.openstack.org/173214 | 08:37 |
| *** mgoddard1 has joined #openstack-ironic | 08:38 | |
| *** dttocs has quit IRC | 08:39 | |
| *** mgoddard has quit IRC | 08:39 | |
| *** romcheg has joined #openstack-ironic | 08:45 | |
| openstackgerrit | Ramakrishnan G proposed openstack/ironic-specs: Add inband RAID configuration spec for liberty https://review.openstack.org/173218 | 08:55 |
| openstackgerrit | Haomeng,Wang proposed openstack/python-ironicclient: validate input uuid from client side https://review.openstack.org/172618 | 08:55 |
| *** dtantsur is now known as dtantsur|brb | 09:06 | |
| *** pelix has joined #openstack-ironic | 09:10 | |
| kejiahu | morning all | 09:23 |
| a1exhughe5 | morning! | 09:24 |
| kan__ | morning | 09:24 |
| kan__ | @lucasagomes. Hi, lucas. I have a issue 1443003. It is about the usage of 'node_uuid' as the sort key in port list command. Could you please take a little time to have a look? | 09:25 |
| kan__ | here is the link:https://bugs.launchpad.net/ironic/+bug/1443003 | 09:25 |
| openstack | Launchpad bug 1443003 in Ironic "port-list --detail using 'Node UUID' as sort key and node-list --detail using 'Chassis UUID' as sort key fail" [Undecided,In progress] - Assigned to Kan (kansks) | 09:25 |
| kan__ | thx | 09:25 |
| lucasagomes | kan__, hi there, oh... I will take a look soonish | 09:25 |
| lucasagomes | thanks for reporting it | 09:25 |
| kan__ | thx :-) | 09:25 |
| *** mgoddard1 has quit IRC | 09:29 | |
| *** mgoddard has joined #openstack-ironic | 09:32 | |
| edcragg | hi, does anyone have any experience setting the boot device for HP moonshot cartridges using the ironic ipmitool driver? power control works fine using double bridged ipmitool commands, but i haven't seen standard IPMI commands be able to set the boot device for individual cartridges in a chassis, even though ipmitool `chassis bootdev disk` returns with a relatively sensible 'Set Boot Device to d | 09:33 |
| edcragg | isk'. i know this is an ipmi issue rather than ironic, but wondered if anyone had any experience | 09:33 |
| *** yuanying has quit IRC | 09:35 | |
| *** dttocs has joined #openstack-ironic | 09:35 | |
| *** dttocs has quit IRC | 09:42 | |
| *** ramineni1 has joined #openstack-ironic | 09:54 | |
| *** lifeless1 has joined #openstack-ironic | 09:54 | |
| *** ramineni has quit IRC | 09:57 | |
| *** naohirot has quit IRC | 09:58 | |
| *** lifeless has quit IRC | 10:02 | |
| *** harlowja_away has quit IRC | 10:02 | |
| *** dtantsur|brb is now known as dtantsur | 10:09 | |
| *** chenglch|2 has quit IRC | 10:13 | |
| *** saripurigopi has quit IRC | 10:14 | |
| *** Nisha has quit IRC | 10:18 | |
| tiagogomes_ | morning ironic, I am seeing this warning on nova-compute "760 WARNING nova.virt.ironic.driver [-] cpu_arch not defined for node '1956aca6-1f68-462c-9386-f92a2933411f'", however it is on the properties of the node http://paste.openstack.org/show/203833/ | 10:22 |
| tiagogomes_ | ah wait, the keyname shouldn't have the 'properties' prefix | 10:23 |
| tiagogomes_ | I guess | 10:23 |
| dtantsur | tiagogomes_, morning, it should be just cpu_arch etc | 10:24 |
| lucasagomes | tiagogomes_, yeah, did you define it in the flavor as well? | 10:33 |
| tiagogomes_ | lucasagomes yes, it is working fine now | 10:33 |
| lucasagomes | cool | 10:34 |
| *** Haomeng|2 has joined #openstack-ironic | 10:36 | |
| *** dttocs has joined #openstack-ironic | 10:38 | |
| *** Haomeng has quit IRC | 10:39 | |
| *** ijw has joined #openstack-ironic | 10:43 | |
| *** ijw has quit IRC | 10:43 | |
| *** dttocs has quit IRC | 10:44 | |
| *** ijw has joined #openstack-ironic | 10:44 | |
| openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic: Troubleshoot: Do not power off node if deployment fail https://review.openstack.org/172932 | 10:48 |
| TheJulia | ohh | 10:53 |
| *** ramineni1 has quit IRC | 10:57 | |
| *** lsmola_ has joined #openstack-ironic | 10:58 | |
| *** ijw has quit IRC | 11:01 | |
| *** ijw has joined #openstack-ironic | 11:01 | |
| tiagogomes_ | I tried to provision a machine using nova, but I got the following error ""message": "No valid host was found. ", "code": 500, "details": " File \"/usr/lib/python2.7/site-packages/nova/conductor/manager.py\", line 614, in build_instances" | 11:03 |
| tiagogomes_ | The ironic node was created using the ssh_pxe driver | 11:03 |
| tiagogomes_ | Any ideas? | 11:04 |
| dtantsur | tiagogomes_, 1. something wrong with Ironic nodes (check status); 2. mismatch between flavor and node properties | 11:06 |
| *** thrash|g0ne is now known as thrash | 11:14 | |
| *** saripurigopi has joined #openstack-ironic | 11:15 | |
| *** ijw_ has joined #openstack-ironic | 11:15 | |
| *** ijw has quit IRC | 11:16 | |
| tiagogomes_ | dtantsur, those things look fine | 11:20 |
| openstackgerrit | Gopi Krishna S proposed openstack/ironic-specs: Add pxe_ucs driver spec for liberty. https://review.openstack.org/173271 | 11:21 |
| openstackgerrit | Gopi Krishna S proposed openstack/ironic-specs: Add pxe_ucs driver spec for liberty https://review.openstack.org/173271 | 11:24 |
| * rameshg87 goes home | 11:26 | |
| *** rameshg87 has quit IRC | 11:26 | |
| tiagogomes_ | I am not using the nova-conductor, is this required? | 11:27 |
| dtantsur | nova-conductor? not sure, but I would support it's something needed. lucasagomes ^^^? | 11:28 |
| lucasagomes | dtantsur, tiagogomes_ hmm nova-conductor is the guy talking to the db. I've never tried use nova without it... | 11:29 |
| lucasagomes | so I don't know | 11:29 |
| lucasagomes | I can give it a go tho | 11:29 |
| lucasagomes | tiagogomes_, nothing in the logs? | 11:30 |
| openstackgerrit | Gopi Krishna S proposed openstack/ironic-specs: Add pxe_ucs driver spec for liberty https://review.openstack.org/173271 | 11:31 |
| tiagogomes_ | just a warning: WARNING nova.compute.manager [-] Bandwidth usage not supported by hypervisor. | 11:31 |
| TheJulia | nova conductor performs the scheduling | 11:31 |
| dtantsur | TheJulia, morning | 11:32 |
| TheJulia | dtantsur: good morning | 11:32 |
| tiagogomes_ | 0 WARNING nova.scheduler.driver [-] [instance: 2ddd4a5b-b143-4d6e-ac71-5c29e6398d95] NoValidHost exception with message: 'No valid host was found.' | 11:32 |
| TheJulia | tiagogomes_: I suspect that might be enough for it to not schedule on the hosts. Perhaps if we could see the information for the node your attempting to provision in nova and the information on one of the ironic nodes that nova should have scheduled the host on to? | 11:33 |
| tiagogomes_ | TheJulia, http://paste.openstack.org/show/203839/ | 11:37 |
| *** dttocs has joined #openstack-ironic | 11:40 | |
| *** saripurigopi has quit IRC | 11:41 | |
| *** yog__ has quit IRC | 11:41 | |
| TheJulia | tiagogomes_: Honestly haven't looked at nova hypervisor-show in a really long time, but checkout your memory settings, that doesn't seem quite right to me | 11:43 |
| tiagogomes_ | The I didn't set the ram_allocation_ration and some other memory setting, but the error doesn't _appear_ to be related with that | 11:44 |
| *** davideagnello has joined #openstack-ironic | 11:44 | |
| *** kan__ has quit IRC | 11:45 | |
| lucasagomes | tiagogomes_, btw, http://docs.openstack.org/developer/ironic/deploy/install-guide.html#configure-compute-service-to-use-the-bare-metal-service | 11:45 |
| lucasagomes | some tweaks we have to do on nova scheduler | 11:46 |
| tiagogomes_ | lucasagomes, yes, I followed that guide, except for ram_allocation_ratio and reserved_host_memory_mb | 11:46 |
| tiagogomes_ | I'll try with the memory settings | 11:47 |
| *** dttocs has quit IRC | 11:47 | |
| *** jistr is now known as jistr|class | 11:47 | |
| lucasagomes | ok | 11:48 |
| lucasagomes | let us know if that works for u | 11:48 |
| * lucasagomes will grab a sandwich, bbl | 11:49 | |
| *** davideagnello has quit IRC | 11:49 | |
| lucasagomes | TheJulia, morning :-) | 11:49 |
| *** lucasagomes is now known as lucas-hungry | 11:49 | |
| TheJulia | lucas-hungry: good morning | 11:49 |
| tiagogomes_ | nope, same error | 11:51 |
| *** trown|outttypeww is now known as trown | 11:53 | |
| *** Marga_ has joined #openstack-ironic | 11:57 | |
| *** ukalifon has quit IRC | 11:58 | |
| TheJulia | tiagogomes_: if memory serves, nova does compare all of the parameters, have you tried turning up the logging level at all? | 12:00 |
| *** romcheg has quit IRC | 12:01 | |
| *** romcheg has joined #openstack-ironic | 12:02 | |
| *** ukalifon has joined #openstack-ironic | 12:03 | |
| tiagogomes_ | TheJulia, trying.... | 12:03 |
| *** pas-ha has quit IRC | 12:04 | |
| *** dprince has joined #openstack-ironic | 12:07 | |
| *** oomichi has quit IRC | 12:12 | |
| tiagogomes_ | 2015-04-14 14:24:08.000 17623 INFO nova.filters [req-ccb3a40f-2378-43cb-8336-d3e26fe87d5e None] Filter AggregateInstanceExtraSpecsFilter returned 0 hosts | 12:16 |
| tiagogomes_ | I assume this happens when the flavor specs don't match the ironic node properties | 12:17 |
| tiagogomes_ | but they look the same to me http://paste.openstack.org/show/203839/ | 12:18 |
| *** pas-ha has joined #openstack-ironic | 12:18 | |
| TheJulia | Hmm, i forgot the scheduler was a separate process | 12:27 |
| TheJulia | tiagogomes_: Seems like it still might be memory related, the hypervisor shows half it's ram in use, which means it can't schedule the requested flavor on to the machine | 12:29 |
| Haomeng|2 | TheJulia: did you set ram_allocation_ratio =1 | 12:29 |
| TheJulia | tiagogomes_: ^^^ | 12:29 |
| Haomeng|2 | TheJulia: I see your ironic hypervisor memory_mb | 1024, but memory_mb_used | 512 | 12:29 |
| tiagogomes_ | TheJulia, yes | 12:29 |
| tiagogomes_ | ah, sorry | 12:29 |
| tiagogomes_ | that's old | 12:30 |
| TheJulia | ohh | 12:30 |
| Haomeng|2 | TheJulia: after change nova.conf, have to restart nova process | 12:30 |
| TheJulia | tiagogomes_: ^^^ | 12:30 |
| Haomeng|2 | Haomeng|2: and check nova compute log to see if the ironic hypervisor is updated | 12:30 |
| tiagogomes_ | http://paste.openstack.org/show/203846/ | 12:30 |
| tiagogomes_ | yes, I restarted every nova process | 12:31 |
| openstackgerrit | Haomeng,Wang proposed openstack/python-ironicclient: validate input uuid from client side https://review.openstack.org/172618 | 12:31 |
| Haomeng|2 | tiagogomes_: now looks fine, still not working now? | 12:31 |
| tiagogomes_ | Haomeng|2 no | 12:31 |
| Haomeng|2 | tiagogomes_: now | memory_mb_used | 1024 | 12:32 |
| Haomeng|2 | tiagogomes_: so need to delete nova instance | 12:32 |
| Haomeng|2 | tiagogomes_: to release the resource | 12:32 |
| tiagogomes_ | mm | 12:32 |
| Haomeng|2 | tiagogomes_: I guess nova instance is in error status | 12:33 |
| *** ijw_ has quit IRC | 12:33 | |
| Haomeng|2 | tiagogomes_: delete all nova instance which is in error status, which will still use the ironic hypervisor resource | 12:33 |
| tiagogomes_ | ok | 12:33 |
| tiagogomes_ | it still fails :( | 12:36 |
| tiagogomes_ | | memory_mb | 1024 | | 12:37 |
| tiagogomes_ | | memory_mb_used | 0 | | 12:37 |
| *** edcragg has quit IRC | 12:41 | |
| *** a1exhughe5 has quit IRC | 12:41 | |
| *** tiagogomes_ has quit IRC | 12:41 | |
| *** edcragg has joined #openstack-ironic | 12:41 | |
| *** a1exhughe5 has joined #openstack-ironic | 12:41 | |
| *** tiagogomes has joined #openstack-ironic | 12:42 | |
| *** dttocs has joined #openstack-ironic | 12:43 | |
| *** lucas-hungry is now known as lucasagomes | 12:46 | |
| *** dttocs has quit IRC | 12:47 | |
| *** dttocs has joined #openstack-ironic | 12:47 | |
| Haomeng|2 | tiagogomes_: can you share the nova compute log to see details why can not match | 12:47 |
| tiagogomes | Haomeng|2, it doen't say the details that don't match: http://paste.openstack.org/show/203849/ | 12:49 |
| tiagogomes | that's nova-scheduler log | 12:50 |
| tiagogomes | nova-compute: http://paste.openstack.org/show/203850/ | 12:52 |
| *** ukalifon has quit IRC | 12:57 | |
| openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic: Only sync power state if node is ACTIVE (DO NOT MERGE) https://review.openstack.org/173299 | 12:59 |
| lucasagomes | jroll, can't add you as reviewer due ur launchpad problem ^ | 13:00 |
| *** mitz has quit IRC | 13:01 | |
| lucasagomes | I just want to trigger some discussions about it | 13:01 |
| *** kkoski has joined #openstack-ironic | 13:01 | |
| *** jistr|class is now known as jistr | 13:03 | |
| *** mitz has joined #openstack-ironic | 13:04 | |
| openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic: Only sync power state if node is ACTIVE (DO NOT MERGE) https://review.openstack.org/173299 | 13:06 |
| trown | lucasagomes: +1 on daredevil references | 13:06 |
| lucasagomes | trown, heh :-) | 13:06 |
| lucasagomes | good series, I'm following it now | 13:06 |
| TheJulia | Agreed, although it is surprisingly dark | 13:07 |
| lucasagomes | +1 | 13:08 |
| lucasagomes | after miller daredevil became pretty dark | 13:08 |
| *** romcheg has quit IRC | 13:14 | |
| *** kozhukalov has quit IRC | 13:20 | |
| *** romcheg has joined #openstack-ironic | 13:21 | |
| tiagogomes | after adding some debug prints, I think that my problem is due cpu_arch not being on the metadata dictionary | 13:33 |
| *** coolsvap|afk is now known as coolsvap | 13:34 | |
| tiagogomes | I am not sure how to add it though | 13:34 |
| *** MattMan has joined #openstack-ironic | 13:34 | |
| tiagogomes | do why need to create an aggregate in addition to a flavor as well? | 13:36 |
| jroll | mornin' | 13:43 |
| jroll | lucasagomes: my thought is that the power state in the DB should always be (close to) up to date, because IPMI is slow | 13:43 |
| jroll | tiagogomes: the aggregate stuff is only if you want to run virt and baremetal in the same environment, afaik, it directs things to the correct compute | 13:44 |
| jroll | there's other ways to solve it like cells | 13:44 |
| trown | jroll: have you tried out cells? | 13:46 |
| lucasagomes | jroll, so the other way around then? | 13:46 |
| jroll | trown: yes, we run them in prod | 13:46 |
| jroll | lucasagomes: ? | 13:47 |
| trown | ah, sweet, I will need to look into that | 13:47 |
| * tiagogomes notes that is using Juno | 13:47 | |
| JayF | devananda: I did not make nor am I aware of an ironic-spec-core meeting | 13:47 |
| lucasagomes | jroll, if the node is off, instead of sync. Just update the db saying it;s off | 13:47 |
| JayF | devananda: I think BadCub was trying to organize something; that's probably it? | 13:47 |
| jroll | lucasagomes: oh, right. so we run with force_power_state_during_sync=False | 13:48 |
| lucasagomes | jroll, right, yeah that's fine | 13:48 |
| jroll | lucasagomes: I think we should always run the periodic task, I think operators can decide whether to force the power state | 13:48 |
| lucasagomes | +1 | 13:49 |
| jroll | I think if a user powers off the machine, that's fine if that's what they want to do | 13:49 |
| lucasagomes | my point on that patch is, when does it make sense to sync power state | 13:49 |
| lucasagomes | say node is in zapping state. Should we still run sync on that state? | 13:49 |
| jroll | well, let's be clear on words | 13:49 |
| jroll | by 'sync' do you mean sync real power state to db, or sync db state to the node and change that real power state | 13:50 |
| lucasagomes | sync the db state to the node and change that real power state | 13:50 |
| jroll | ok | 13:50 |
| jroll | so I personally think we should never do that :) | 13:50 |
| lucasagomes | (which is what that force_power_state_during_sync does if it's True | 13:50 |
| lucasagomes | jroll, reason? I mean I see some benefits on having it, say the node is deployed and someone just turned it off accidentaly | 13:51 |
| lucasagomes | so ironic bringing it on again seems fair | 13:52 |
| jroll | lucasagomes: then they call nova start or whatever | 13:52 |
| jroll | I don't want my computers turning on by themselves | 13:52 |
| *** r-daneel has joined #openstack-ironic | 13:52 | |
| lucasagomes | right, yeah nova start is fair :-/ | 13:53 |
| * lucasagomes wonders even more about that option now | 13:53 | |
| jroll | unless you mean someone in the DC turned it off, in which case I have some four letter words | 13:53 |
| lucasagomes | lol | 13:53 |
| *** absubram has quit IRC | 13:54 | |
| lucasagomes | as I stated in the patch. Right now that config option tries to sync the state (real state on the node) for almost all states | 13:54 |
| lucasagomes | except DEPLOYWAIT, or if it's maintenance | 13:54 |
| lucasagomes | now... it seems wrong. And we can: 1) remove that option, not force sync state anymore. | 13:55 |
| lucasagomes | 2) constraint the scope of that, to only act upon some certain states | 13:55 |
| lucasagomes | that patch does 2) | 13:55 |
| jroll | right | 13:55 |
| lucasagomes | so idk :-) that's why I put it up there. To trigger some discussion | 13:56 |
| jroll | I guess I don't see why other states are a problem? | 13:56 |
| jroll | when ironic changes the state it updates the db | 13:56 |
| jroll | so really the only reason to sync is if someone is messing with power state outside of ironic | 13:56 |
| lucasagomes | yeah it seems ok. It just that we were talking about it downstream and I decided to bring it upstream see if more people has opnions on that | 13:57 |
| lucasagomes | other states like enroll, may not even have right credentials | 13:57 |
| lucasagomes | no reason why try to sync it | 13:57 |
| jroll | yeah, true | 13:57 |
| jroll | I guess I just don't have a strong opinion | 13:57 |
| * lucasagomes same | 13:57 | |
| lucasagomes | anyhoo... just food for thought | 13:58 |
| jroll | yeah :) | 14:00 |
| *** mtanino has joined #openstack-ironic | 14:04 | |
| NobodyCam | morning Ironicers | 14:07 |
| TheJulia | Good morning! | 14:08 |
| NobodyCam | morning TheJulia :) | 14:08 |
| dtantsur | morning jroll, NobodyCam | 14:08 |
| NobodyCam | morning dtantsur jroll :) | 14:08 |
| * TheJulia needs a large amount of coffee() | 14:09 | |
| *** dttocs has quit IRC | 14:12 | |
| BadCub | JayF: Yes, I am trying to organize the meeting. I put it on the agenda for folks to discuss and say yes or no to | 14:13 |
| jroll | morning NobodyCam TheJulia dtantsur and BadCub :) | 14:15 |
| BadCub | morning jroll | 14:15 |
| tiagogomes | I am still struggling with the same error. Is any scheduler_default_filters that I need to use for Ironic | 14:15 |
| NobodyCam | hey hey jroll | 14:15 |
| jroll | tiagogomes: what error? | 14:17 |
| jroll | tiagogomes: there's a scheduler_use_baremetal_filters that may help | 14:17 |
| *** zhenguo has quit IRC | 14:17 | |
| tiagogomes | jroll, http://paste.openstack.org/show/203875/ | 14:18 |
| jroll | tiagogomes: have scheduler and compute logs? | 14:19 |
| kbs1 | NobodyCam, et. al.: FWIW, I found the source of the auth issue with IPA that I ping'd about yesterday. IPA is looking for the image in swift but it doesn't exist since my glance isn't configured that way. Kicks out an auth error when the object doesn't exist. | 14:19 |
| jroll | kbs1: :/ | 14:20 |
| tiagogomes | jroll, http://paste.openstack.org/show/203876/, http://paste.openstack.org/show/203878/ | 14:20 |
| NobodyCam | the new google pony express Doodle is like a complete game! | 14:21 |
| TheJulia | goodmorning BadCub | 14:22 |
| BadCub | morning TheJulia | 14:22 |
| NobodyCam | kbs1: sounds like Ironic could use better error messages around that | 14:22 |
| jroll | tiagogomes: looks like your flavor needs extra_specs that match your aggregate | 14:22 |
| tiagogomes | jroll, which extra_specs does it need? I already added cpu_arch to the flavor | 14:24 |
| tiagogomes | but I didn't create any aggregate | 14:24 |
| jroll | tiagogomes: it's failing on AggregateInstanceExtraSpecsFilter, so it sounds like you have an aggregate? | 14:26 |
| jroll | I really don't know nova well enough to say :( | 14:26 |
| tiagogomes | ~ # nova aggregate-list | 14:26 |
| tiagogomes | +----+------+-------------------+ | 14:26 |
| tiagogomes | | Id | Name | Availability Zone | | 14:26 |
| tiagogomes | +----+------+-------------------+ | 14:26 |
| tiagogomes | +----+------+-------------------+ | 14:26 |
| tiagogomes | I am following the Ironic installation guide, and there it doesn't talk about aggregates | 14:26 |
| jroll | tiagogomes: right, ok, so I think you should set scheduler_use_baremetal_filters=True | 14:27 |
| kbs1 | NobodyCam: Yeah. Not obvious. Would be good to have ironic vet the url before handing it out. It happily drops the image ID in the url string and returns it for propagation to the agent. Early failure would be good here, IMO. | 14:27 |
| tiagogomes | jroll, it didn't work | 14:30 |
| jroll | tiagogomes: define didn't work | 14:30 |
| tiagogomes | jroll, sorry, exactly same error | 14:30 |
| jroll | tiagogomes: which scheduler filter returned 0 hosts, then? | 14:31 |
| *** zz_jgrimm is now known as jgrimm | 14:32 | |
| tiagogomes | oh, a different one : 2015-04-14 16:39:05.002 4540 INFO nova.filters [req-3ab6b1c8-f7a1-4f90-99b9-0b0d11dbd8e8 None] Filter RetryFilter returned 0 hosts | 14:32 |
| jroll | ok, good, we're getting somewhere | 14:33 |
| tiagogomes | :) | 14:33 |
| jroll | I think that means it actually tried to build, can you check compute logs for errors? | 14:33 |
| tiagogomes | no errors on nova-compute | 14:34 |
| jroll | does it look like it tried to build something? | 14:34 |
| *** Marga_ has quit IRC | 14:34 | |
| jroll | oh, I wonder if just 0 hosts are going through the filters in the first place | 14:35 |
| jroll | can you run 'nova hypervisor-stats'? | 14:35 |
| tiagogomes | jroll, http://paste.openstack.org/show/203881/ | 14:37 |
| tiagogomes | There was this on `nova show` before failing: | 14:37 |
| tiagogomes | | OS-EXT-STS:task_state | spawning | | 14:37 |
| tiagogomes | | OS-EXT-STS:vm_state | building | 14:37 |
| jroll | tiagogomes: hrm, so it must have tried to build, at least | 14:38 |
| jroll | that means it got to the compute | 14:38 |
| tiagogomes | jroll, cool, it did. I was just looking at the tail of the log : http://paste.openstack.org/show/203882/ | 14:39 |
| jroll | tiagogomes: cool, so you're missing those keys in driver_info | 14:40 |
| jroll | this is juno, correct? | 14:40 |
| tiagogomes | yep, pxe_deploy_ramdisk pxe_deploy_kernel are empty fields | 14:41 |
| tiagogomes | yes Juno | 14:41 |
| jroll | yeah, so those are the kernel and ramdisk for your deploy ramdisk | 14:41 |
| jroll | whether that's the DIB ramdisk or IPA | 14:41 |
| *** saripurigopi has joined #openstack-ironic | 14:44 | |
| tiagogomes | | provision_state | deploying | 14:49 |
| tiagogomes | :) | 14:49 |
| tiagogomes | thanks jroll | 14:49 |
| jroll | \o/ | 14:49 |
| jroll | np tiagogomes :) | 14:49 |
| *** absubram has joined #openstack-ironic | 14:53 | |
| *** dttocs has joined #openstack-ironic | 14:56 | |
| *** rloo has joined #openstack-ironic | 14:59 | |
| *** coolsvap is now known as coolsvap|afk | 15:00 | |
| *** pas-ha has quit IRC | 15:02 | |
| *** mtanino has quit IRC | 15:03 | |
| *** saripurigopi has quit IRC | 15:03 | |
| *** saripurigopi has joined #openstack-ironic | 15:05 | |
| *** mtanino has joined #openstack-ironic | 15:14 | |
| *** pas-ha has joined #openstack-ironic | 15:14 | |
| *** maurosr has quit IRC | 15:21 | |
| *** jgrimm has quit IRC | 15:21 | |
| *** jmank has joined #openstack-ironic | 15:24 | |
| openstackgerrit | Chris Krelle proposed openstack/ironic: Add check for usable tftp directory to PXE driver https://review.openstack.org/167451 | 15:26 |
| NobodyCam | brb | 15:26 |
| devananda | morning, all | 15:34 |
| BadCub | morning devananda | 15:34 |
| BadCub | brb | 15:34 |
| lucasagomes | devananda, NobodyCam BadCub good morning | 15:35 |
| lucasagomes | devananda, https://review.openstack.org/#/c/173379/ | 15:35 |
| lucasagomes | when you get some time, this is for nova | 15:35 |
| lucasagomes | we've discussed it a bit in the feature freeze | 15:36 |
| devananda | lucasagomes: why put networking data in node.extra['configdrive_metadata'] and not, say, in ports? | 15:38 |
| *** jerryz has quit IRC | 15:39 | |
| devananda | also, I'd like to point out how the keys used to store data in our JSON fields (like node.properties and node.extra) are, in fact, just part of the API | 15:39 |
| lucasagomes | devananda, that is one example. But the idea is to have 1 endpoint where nova looks to get the data from | 15:40 |
| lucasagomes | not multiples, which might complicate stuff | 15:40 |
| lucasagomes | having in ports is like useless to ironic because that's about instance configuration, not ports properties really | 15:40 |
| jroll | morning devananda :) | 15:40 |
| lucasagomes | devananda, right, yes, I would like to document it as such | 15:41 |
| jroll | devananda: this is a fun one, would love to see your opinion: https://review.openstack.org/#/c/173142/ | 15:41 |
| lucasagomes | the same way we document the capabilities in node.properties for example | 15:41 |
| NobodyCam | morning devananda | 15:41 |
| lucasagomes | we could document the extra['configdrive_metadata'] | 15:41 |
| devananda | lucasagomes: document it - sure. my point is a side bar to the spec itself. we've gone and created non-discoverable portions of our API to "extend" it. | 15:42 |
| lucasagomes | devananda, oh... I see what you mean | 15:42 |
| devananda | there is no way to discover, from the REST API, that the node.properties resource should somehow contain a "capabilities" key to trigger some other special behavior | 15:42 |
| devananda | ditto for your proposal here with configdrive_metadata | 15:43 |
| lucasagomes | devananda, yeah that's true, but again, nova also looks into the capabilties which is also in the json field. So I just keep the same | 15:43 |
| jroll | we could break those out into tables | 15:43 |
| jroll | e.g. flavor extra_specs in nova is a separate table | 15:43 |
| devananda | the difference is that, at least in principle, ironic can auto-populate the capabilities field | 15:43 |
| lucasagomes | devananda, right, we could have it as an attribute for the node | 15:43 |
| devananda | so the operator never has to actually know about it | 15:43 |
| devananda | (in principle) | 15:43 |
| devananda | but for this, it sounds like something the operator needs to set in ironic directly | 15:44 |
| lucasagomes | right | 15:44 |
| *** jgrimm has joined #openstack-ironic | 15:44 | |
| lucasagomes | that's a fair point | 15:44 |
| lucasagomes | but again, the same for cpus, memory_mb, disk_gb etc | 15:44 |
| lucasagomes | all json field, requires operators to put it directly | 15:45 |
| lucasagomes | non-discoverable and nova will looks into it | 15:45 |
| devananda | indeed -- but again, at least in principle, ironic could auto-populate that info via inspection | 15:45 |
| *** maurosr_ has joined #openstack-ironic | 15:45 | |
| lucasagomes | yeas and the port example we do via discover too | 15:45 |
| devananda | yep | 15:45 |
| *** dprince has quit IRC | 15:46 | |
| lucasagomes | we add it to the configdrive_metadata, if the iface A is connect to the data network we set it to be renamed to "X" | 15:46 |
| devananda | hmm. so again, it sounds like that should be stored on the port | 15:46 |
| devananda | nova driver, if using configdrive, should be able to collate the information ironic has about the hardware. no? | 15:47 |
| lucasagomes | devananda, right, but then the nova driver will need to look at multiple parts of our api to get the data | 15:47 |
| lucasagomes | and sometimes we may not have a representation of that data in our api | 15:48 |
| lucasagomes | say disks, same example. But say we want to name the disk devices accordingly | 15:48 |
| lucasagomes | we don't represent disks in our api | 15:48 |
| dtantsur | morning devananda and g'night to everyone else :) | 15:48 |
| *** dtantsur is now known as dtantsur|afk | 15:48 | |
| lucasagomes | but I still would like to have consistent naming for such devices | 15:48 |
| *** ifarkas has quit IRC | 15:48 | |
| *** romcheg has quit IRC | 15:48 | |
| lucasagomes | the idea is be to generic | 15:49 |
| *** romcheg has joined #openstack-ironic | 15:49 | |
| *** romcheg has quit IRC | 15:49 | |
| lucasagomes | and the data there is not relevant for Ironic itself, since we don't care whether the port 1 will be renamed to X when deploying a node | 15:50 |
| jroll | lucasagomes: is this something --user-data could solve? | 15:50 |
| lucasagomes | (that's why I put on the extra field) | 15:51 |
| lucasagomes | jroll, not really, because it's operators that set it | 15:51 |
| *** jistr has quit IRC | 15:51 | |
| lucasagomes | or discover, not users | 15:51 |
| jroll | lucasagomes: right, ok, guess I'm curious why operators care | 15:51 |
| *** saripurigopi has quit IRC | 15:51 | |
| lucasagomes | jroll, we care about the network naming for e.g | 15:51 |
| lucasagomes | so that we now that, across all isntances in my deployment | 15:51 |
| lucasagomes | the iface name X is connected to the data network for e.g | 15:52 |
| lucasagomes | some might have other use cases | 15:52 |
| jroll | hmm | 15:52 |
| *** saripurigopi has joined #openstack-ironic | 15:52 | |
| jroll | I mean, I agree it could be useful, I just don't see a use case | 15:52 |
| jroll | I might be missing something | 15:53 |
| lucasagomes | jroll, consistently naming devices is useful | 15:54 |
| devananda | lucasagomes: isn't that part of a much larger problem of 1) how ironic interacts with more complex networks, 2) that nova assumes 1 network == 1 port, 3) that we're not storing any information about the physical links, aside from MAC, to supply to neutron? | 15:54 |
| * lucasagomes otherwise udev perisistent rules wouldn't exist | 15:54 | |
| jroll | lucasagomes: I mean, we do that today without this feature | 15:55 |
| lucasagomes | jroll, how ? | 15:55 |
| * lucasagomes looks for alternatives | 15:55 | |
| jroll | lucasagomes: it might just be that we're bonding both interfaces together | 15:55 |
| jroll | which doesn't help you | 15:55 |
| jroll | sorry, in a call at the same time | 15:55 |
| lucasagomes | devananda, yes but that is for the deployment right? that data is post configuration | 15:56 |
| lucasagomes | jroll, right, yeah... | 15:56 |
| * jroll needs to step away for a bit | 15:57 | |
| devananda | lucasagomes: both | 15:57 |
| *** krtaylor has quit IRC | 15:57 | |
| devananda | lucasagomes: in non-flat network environments, we need to be able to supply the instance with enough information for its networks to be configured properly | 15:58 |
| BadCub | demolition crew just arrived. Will be bouncing in-and-out for the rest of the day | 15:58 |
| devananda | BadCub: mmm, fun. ttyl! | 15:58 |
| NobodyCam | :-p | 15:58 |
| TheJulia | BadCub: The archiological dig? | 15:58 |
| * BadCub wonders if there is a new definition of "fun" lol | 15:58 | |
| BadCub | TheJulia: nope. Trees and walls | 15:59 |
| TheJulia | BadCub: Enjoy! Oh and yes, there are many definitions of "fun" | 15:59 |
| lucasagomes | devananda, right, but this is something that is already passed to the config drive by nova? we pass the network_info to it | 15:59 |
| lucasagomes | BadCub, heh enjoy it! | 15:59 |
| BadCub | TheJulia: lucasagomes as long as they don;t damage the power lines, we are good hehehehe | 16:00 |
| lucasagomes | +1 :D | 16:00 |
| TheJulia | devananda: provide network info to what? Neutron/Infrastucture? or the host it's self? | 16:01 |
| lucasagomes | devananda, although I feel that we are going a bit out of scope here. What I'm looking at is https://github.com/openstack/nova/blob/master/nova/virt/ironic/driver.py#L605-L606 | 16:01 |
| lucasagomes | how me as an operator of Ironic can give my instances some information about the hardware I've so that it get it configured correctly | 16:01 |
| devananda | lucasagomes: right. and I'm questioning whether you, as an operator, should be telling ironic configdrive-specific formatted data, or whether you should be telling ironic about your hardware | 16:03 |
| devananda | lucasagomes: perhaps an example of what you'd propose to store in node.extra['configdrive_metadata'] would help | 16:03 |
| devananda | TheJulia: both :) | 16:03 |
| lucasagomes | devananda, right, we do it here https://github.com/stackforge/ironic-discoverd/blob/master/ironic_discoverd/plugins/edeploy.py#L136-L138 | 16:04 |
| lucasagomes | I can get a sample of that | 16:04 |
| TheJulia | For the node deploying, I somehow suspect that people should be able to do whatever they really want/need, since they can format the configdrive how they wish and inject whatever they desire, I suspect the more important portion is the infrastructure | 16:05 |
| devananda | jroll: lucasagomes: re the "mount configdrive as loopback" proposal, I don't dislike your proposal (make an extension to IPA) too much, but it does move ironic into the realm of injecting files into instances, which I dislike out of hand. | 16:05 |
| *** ijw has joined #openstack-ironic | 16:05 | |
| lucasagomes | TheJulia, yes, configdrive already supports adding extra metadata | 16:05 |
| jroll | devananda: right, I don't think we should touch the images, at all | 16:05 |
| lucasagomes | that interface is an api to do it | 16:05 |
| lucasagomes | that spec* | 16:05 |
| jroll | devananda: I linked that to you because I thought we were all on board with that opinion, but maybe not | 16:06 |
| lucasagomes | devananda, yeah :/ I hated nova when it injected network information | 16:06 |
| lucasagomes | devananda, tho I see some usefulness on that spec | 16:06 |
| lucasagomes | and since it's optional as well makes me happier | 16:06 |
| lucasagomes | the fact that IPA is trying to create a primary partition (where you can have only 4) is a real problem | 16:07 |
| * jroll wonders what happens when the image is ntfs or jayfs or something I've never heard of | 16:07 | |
| lucasagomes | cause we don't specify how the disk images should be built | 16:07 |
| lucasagomes | jroll, it won't find fstab | 16:07 |
| lucasagomes | and won't do anything I assume | 16:07 |
| jroll | maybe we do something weird, but we deploy coreos with a configdrive partition just fine | 16:07 |
| lucasagomes | fail deployment probably | 16:07 |
| jroll | lucasagomes: right, and so it's broken | 16:08 |
| jroll | yeah | 16:08 |
| * TheJulia begins pondering juliafs where everything will be represented by delicious food | 16:08 | |
| lucasagomes | cause it can't inject the data | 16:08 |
| jroll | TheJulia++ | 16:08 |
| lucasagomes | jroll, but again, we are doing some assumptions that breaks too | 16:08 |
| lucasagomes | the room for an extra primary partition | 16:08 |
| lucasagomes | and MBR | 16:08 |
| openstackgerrit | John L. Villalovos proposed openstack/ironic: Check temp dir is usable for ipmitool driver https://review.openstack.org/160383 | 16:08 |
| jroll | true | 16:08 |
| jroll | idk, just that touching user images seems really scary to me | 16:09 |
| lucasagomes | it does, yeah more discussion is needed whether we really should go that way | 16:09 |
| lucasagomes | but at least, in that constraint scope, it seems a fair addition IMO | 16:09 |
| jlvillal | NobodyCam, ^^^ I did minor changes to your patch. I hope you don't mind. | 16:09 |
| jlvillal | NobodyCam, And +1 from me on it | 16:10 |
| NobodyCam | jlvillal: I don't mind, however I think you rebased on an older version of 160383 | 16:10 |
| NobodyCam | :-p | 16:10 |
| jlvillal | NobodyCam, really? | 16:11 |
| *** jcoufal_ has quit IRC | 16:11 | |
| NobodyCam | Jenkins says the depends on is out of date | 16:11 |
| *** krtaylor has joined #openstack-ironic | 16:12 | |
| jlvillal | NobodyCam, I did: git fetch https://review.openstack.org/openstack/ironic refs/changes/83/160383/24 && git checkout FETCH_HEAD | 16:12 |
| *** Marga_ has joined #openstack-ironic | 16:12 | |
| NobodyCam | but its at rev 25 :-p | 16:12 |
| jlvillal | NobodyCam, That was my push :) | 16:12 |
| jlvillal | NobodyCam, Where does it say that? I'm looking on Gerrit | 16:12 |
| NobodyCam | oh never mind | 16:12 |
| * NobodyCam needs more coffee | 16:12 | |
| NobodyCam | I was looking at the wrong patch | 16:13 |
| NobodyCam | :-p | 16:13 |
| jlvillal | NobodyCam, Whew! :) | 16:13 |
| jlvillal | NobodyCam, You were making me nervous :D | 16:13 |
| *** Marga_ has quit IRC | 16:14 | |
| jlvillal | NobodyCam, I basically just added some 'autospec=True' lines and change 'Mb' to 'MiB'. I think of 'Mb' as mega-bits. | 16:14 |
| *** Marga_ has joined #openstack-ironic | 16:15 | |
| NobodyCam | jlvillal: ++ | 16:15 |
| NobodyCam | Thank you | 16:15 |
| jlvillal | NobodyCam, Now let's get that merged! :D | 16:16 |
| lucasagomes | TheJulia, thanks the review, I will add some docs in the troubleshoot section | 16:16 |
| *** mtanino has quit IRC | 16:17 | |
| *** saripurigopi has quit IRC | 16:17 | |
| *** saripurigopi_ has joined #openstack-ironic | 16:18 | |
| *** saripurigopi_ is now known as saripurigopi | 16:18 | |
| *** Marga_ has quit IRC | 16:18 | |
| TheJulia | lucasagomes: excellent! | 16:19 |
| *** Marga_ has joined #openstack-ironic | 16:19 | |
| * devananda pushes a client release fix | 16:20 | |
| devananda | also - holy crap folks, that's a lot of bugs ya'll fixed in the last few weeks :) https://launchpad.net/ironic/kilo/kilo-rc1 | 16:22 |
| openstackgerrit | Chris Krelle proposed openstack/ironic: Add check for usable tftp directory to PXE driver https://review.openstack.org/167451 | 16:22 |
| NobodyCam | :) | 16:22 |
| *** Marga_ has quit IRC | 16:23 | |
| *** ijw has quit IRC | 16:26 | |
| lucasagomes | :-) | 16:27 |
| *** Nisha has joined #openstack-ironic | 16:30 | |
| *** a1exhughe5 has quit IRC | 16:32 | |
| *** saripurigopi has quit IRC | 16:33 | |
| BadCub | devananda: yes, the gang pushed pretty hard on the bug list :-) | 16:34 |
| BadCub | devananda: NobodyCam: jroll: lucasagomes; et al. did we all decide if tomorrow was good to get the Spec Core team together? Or would a different date/time be better? | 16:36 |
| lucasagomes | BadCub, hmm what time is it? | 16:37 |
| * lucasagomes didn't know it was tomrrow | 16:37 | |
| lucasagomes | I think it works to me, but thursday would be ideal | 16:37 |
| BadCub | I put a tentative for 1500hrs UTC. But wanted folks to decide if that was good or not. | 16:38 |
| BadCub | I am available pretty much any day/time, so what ever works best for the whole team works for me :) | 16:38 |
| devananda | JayF: ^ work for you? | 16:39 |
| *** maurosr_ is now known as maurosr | 16:41 | |
| *** derekh has quit IRC | 16:43 | |
| lucasagomes | BadCub, me and dtantsur|afk has a meeting exactly at this time | 16:45 |
| lucasagomes | but it shouldn't be long | 16:45 |
| lucasagomes | so after that I think we are good | 16:45 |
| lucasagomes | have* | 16:45 |
| *** meghal has joined #openstack-ironic | 16:46 | |
| BadCub | lucasagomes: okay. I am at y'all's disposal, so what ever works best, I will be there! | 16:46 |
| lucasagomes | BadCub, thanks for that! | 16:46 |
| *** Marga_ has joined #openstack-ironic | 16:46 | |
| *** Marga_ has quit IRC | 16:46 | |
| BadCub | lucasagomes: anytime :-) I like things to be easy lol | 16:47 |
| *** ijw has joined #openstack-ironic | 16:47 | |
| *** meghal1 has joined #openstack-ironic | 16:47 | |
| BadCub | I have put up a pad for everyone to gander at. I made some observations during the end of K and put up some ideas for discussion on how we might make things easier too. | 16:48 |
| rloo | BadCub: did you check with dtantsur|afk, wrt the core spec meeting? | 16:48 |
| BadCub | rloo: I put it out last week for folks to discuss and had it on the meeting agenda :) | 16:48 |
| rloo | BadCub: I can verify that I saw it mentioned in the meeting minutes yesterday, although I think it wasn't clear whether it was ON at that date/time, or whether it was up for discussion :-) | 16:50 |
| *** meghal has quit IRC | 16:50 | |
| BadCub | Yeah, I intended to be there last night to bring it up for discussion but passed out :( | 16:51 |
| *** Marga_ has joined #openstack-ironic | 16:51 | |
| rloo | BadCub: wonder if we should re-evaluate that meeting time then. Is it working for folks? Isn't it 9pm for you? | 16:51 |
| BadCub | rloo: it is 10pm for me | 16:52 |
| devananda | chrome chrashes again :-/ | 16:52 |
| BadCub | devananda: ack :( | 16:52 |
| jlvillal | One hour earlier (9pm PT) would be better for me :) | 16:52 |
| rloo | BadCub: oh yeah, due to the time change. midnight-1am for me, all the same. I don't attend :-) | 16:53 |
| BadCub | rloo: I try, but sometimes I am already dead sleeping at 10PM. Depends on when I take meds :p | 16:53 |
| BadCub | brb | 16:55 |
| *** Marga_ has quit IRC | 16:58 | |
| *** harlowja has joined #openstack-ironic | 16:59 | |
| lucasagomes | devananda, btw re the configdrive_metadata. R you OK with the idea? It's most the interface that concerns? | 16:59 |
| devananda | lucasagomes: totally | 17:00 |
| devananda | lucasagomes: the loopback one, i'm less OK with even the idea, but if its done as an extensionto IPA, it starts to be a grey area where i'm maybe ok with someone doing it even though I dont like it | 17:01 |
| lucasagomes | devananda, right... yeah leave the loopback :-) | 17:01 |
| lucasagomes | devananda, the configdrive... Would it be better to have a field in the node for that? | 17:01 |
| *** ijw has quit IRC | 17:01 | |
| lucasagomes | I understand that putting on the ports for the network case makes a bit of sense. But I would like to have it to be generic | 17:02 |
| *** ijw has joined #openstack-ironic | 17:02 | |
| devananda | lucasagomes: generic how? to me, putting information on th eports *is* generic | 17:02 |
| lucasagomes | it's not, it's related to the ports | 17:02 |
| devananda | that you want to bundle it up in a special configdrive format is the special case :) | 17:03 |
| lucasagomes | or should be | 17:03 |
| lucasagomes | configdrive supports extra data, I'm just using that interface | 17:03 |
| lucasagomes | the code in nova that generates the config drive supports passing a extra_md= parameter to it | 17:03 |
| lucasagomes | fwiw we do use it, as other drivers, to inject the root password | 17:04 |
| lucasagomes | admin password* | 17:04 |
| lucasagomes | https://github.com/openstack/nova/blob/master/nova/virt/ironic/driver.py#L701-L703 | 17:04 |
| *** dttocs_ has joined #openstack-ironic | 17:05 | |
| *** davideagnello has joined #openstack-ironic | 17:06 | |
| lucasagomes | devananda, btw http://paste.openstack.org/show/175855/ | 17:08 |
| lucasagomes | it's from Feb 13 | 17:08 |
| *** Marga_ has joined #openstack-ironic | 17:08 | |
| *** dttocs has quit IRC | 17:08 | |
| * lucasagomes thinks more people will do similar things | 17:09 | |
| devananda | lucasagomes: so I understand the problem | 17:11 |
| devananda | lucasagomes: and I agree that passing in some "hints" via configdrive is an acceptable solution | 17:12 |
| devananda | lucasagomes: but I think the canonical representation of the information that Ironic needs to record (how the NICs should be configured within the instance so that they work within the physical network outside the instance) should be done differently | 17:13 |
| lucasagomes | devananda, right and this is something that http://specs.openstack.org/openstack/ironic-specs/specs/kilo/network-provider.html should handle correct? | 17:15 |
| devananda | TheJulia and/or mordred has been doing some work, IIRC, on a thing to replace cloudinit that would do just this piece of it | 17:15 |
| devananda | lucasagomes: correct | 17:15 |
| lucasagomes | devananda, the "hints" is extra that ^ | 17:16 |
| lucasagomes | we use it for consistent for e.g | 17:16 |
| mordred | yes | 17:16 |
| lucasagomes | so scripting can be easier across all instances | 17:16 |
| mordred | https://gitorious.org/mordred-temp/glean | 17:16 |
| *** Nisha has quit IRC | 17:16 | |
| mordred | I'll be sucking it into an infra repo later today | 17:16 |
| lucasagomes | we not creating/removing networks or anything of that sort, and I hope people won't do that too | 17:17 |
| mordred | it handles reading static network config on boot from config drive | 17:17 |
| mordred | we've got a dib element that installs it properly here: https://review.openstack.org/#/c/167003/ | 17:17 |
| *** dprince has joined #openstack-ironic | 17:18 | |
| mordred | I'm currently walking through all of the distros and verifying that it does the right thing | 17:18 |
| lucasagomes | cool | 17:18 |
| devananda | also, this is odd... why are there two of them? | 17:18 |
| devananda | http://specs.openstack.org/openstack/ironic-specs/specs/kilo/network-provider.html | 17:18 |
| devananda | http://specs.openstack.org/openstack/ironic-specs/specs/kilo-archive/network-provider.html | 17:18 |
| mordred | the basic logic is - for every interface that /sys reports as being a real interface, see if there is network config in config drive - if there is, use it, if there is not, write a dhcp config. if there is already a config in place, do nothing | 17:18 |
| lucasagomes | devananda, no idea | 17:19 |
| mordred | it also, just for convenience, supports ssh keypairs in config drive | 17:19 |
| mordred | but it does _not_ support any other feature of cloud-init | 17:19 |
| lucasagomes | mordred, does it support renaming network devices? | 17:19 |
| mordred | it does not support any additional management after first boot | 17:19 |
| BadCub | devananda: that is odd. | 17:19 |
| mordred | it assumes that you are using this to boot something and that you'll manage the system with something else in an ongoing manner | 17:19 |
| lucasagomes | mapping interfaces names to some correspoding abstract name? | 17:20 |
| mordred | it does a mapping between mac address and interface name based on what's reported by the sys filesystem | 17:20 |
| lucasagomes | cool | 17:20 |
| mordred | so what it cares about is what network you've said a particualr mac address shuld be connected to | 17:20 |
| lucasagomes | mordred, not sure if you looked or care... but os-net-config does some of it | 17:20 |
| mordred | it does - but os-net-config has too many depends | 17:20 |
| mordred | this has zero depends other than pure python | 17:21 |
| mordred | because I dont' want to install tons of libraries in the base of the system that I personally will be suing to then test openstack installs | 17:21 |
| mordred | the depends step on each other | 17:21 |
| lucasagomes | +1 sounds good | 17:21 |
| devananda | BadCub: you have time to investiate that // clean up the old ones? | 17:21 |
| mordred | lucasagomes: we could add a flag to it to always blow away existing config | 17:21 |
| devananda | I think jayF did the initial clean up / split, but it seems something got left behind | 17:22 |
| mordred | lucasagomes: so that if you know you're runnign in an env where config-drive will always be the source of truth | 17:22 |
| *** ndipanov has quit IRC | 17:22 | |
| mordred | lucasagomes: you could have it skip the part where it avoids blowing away existing network config | 17:22 |
| lucasagomes | mordred, I will def take a look into it | 17:22 |
| BadCub | devananda: yeah, I can tackle that | 17:23 |
| * devananda steps afk to food | 17:23 | |
| mordred | lucasagomes: please do - and let me know if there's anything it can do to make your life better - I've been trying to keep ironic in mind while working on it | 17:23 |
| mordred | since I think ironic installs are in a similar boat in terms of what you want and don't want out of this | 17:24 |
| *** Nisha_brb has joined #openstack-ironic | 17:24 | |
| lucasagomes | devananda, but anyway... my reasoning about using extra. Is as it can be any key/pair data, I don't wanna attach it to any resource in our api. Perhaps the mac naming example we could move to ports yes, but that's is 1 use case only | 17:24 |
| openstackgerrit | John L. Villalovos proposed openstack/ironic: Fix PEP8 E129 errors https://review.openstack.org/173427 | 17:25 |
| *** trown is now known as trown|lunch | 17:26 | |
| lucasagomes | mordred, cool thanks much! I know you want to keep dependencies low but you may want to take a look at augeas for many of the configurations you generate | 17:26 |
| lucasagomes | (it does have python binds) | 17:27 |
| *** Nisha_brb has quit IRC | 17:27 | |
| *** Nisha has joined #openstack-ironic | 17:27 | |
| *** Nisha has quit IRC | 17:28 | |
| *** Nisha has joined #openstack-ironic | 17:28 | |
| mordred | lucasagomes: taht doens't look too crazy, actually - and I don't mind C-based command line tools that have distro packages | 17:29 |
| mordred | since they don't tend to actually conflict with things | 17:30 |
| mordred | lucasagomes: I'll play with shelling out to it and seeing if it can do some of the file output we're doing | 17:30 |
| lucasagomes | mordred, yeah, augeas is pretty awesome. And writing lens is not complicated | 17:30 |
| lucasagomes | if you really need some custom configs | 17:30 |
| lucasagomes | mordred, cool stuff! | 17:30 |
| *** pelix has quit IRC | 17:32 | |
| *** Marga_ has quit IRC | 17:35 | |
| *** Marga_ has joined #openstack-ironic | 17:35 | |
| *** Marga_ has quit IRC | 17:36 | |
| *** meghal1 has quit IRC | 17:36 | |
| *** Marga_ has joined #openstack-ironic | 17:36 | |
| lucasagomes | aight I will call it a day folks | 17:42 |
| lucasagomes | have a good night! | 17:42 |
| NobodyCam | night lucasagomes | 17:42 |
| *** lucasagomes is now known as lucas-dinner | 17:42 | |
| lucas-dinner | NobodyCam, you too! see ya tomorrow | 17:42 |
| *** pas-ha has quit IRC | 17:44 | |
| *** edcragg has quit IRC | 17:47 | |
| *** meghal has joined #openstack-ironic | 17:50 | |
| *** dttocs_ has quit IRC | 17:50 | |
| *** achanda has joined #openstack-ironic | 17:51 | |
| *** dttocs has joined #openstack-ironic | 17:54 | |
| *** dttocs has quit IRC | 17:56 | |
| *** jamielennox|away is now known as jamielennox | 18:05 | |
| *** krtaylor has quit IRC | 18:12 | |
| openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic: Troubleshoot: Do not power off node if deployment fail https://review.openstack.org/172932 | 18:12 |
| aweeks | devananda: jroll: I'm updating the metrics spec, and wondering about the feasibility of emitting metrics for node state changes. Initially, I considered just instrumenting the FSM, but it doesn't have information about how long a node was in a given state (for timing deploys, cleans, etc.) but it looks like it may be possible to calculate that by looking at | 18:19 |
| aweeks | Node.provision_updated_at, when Node.provision_state is changed | 18:19 |
| aweeks | however, that might require some relatively invasive changes to the Node/IronicObject classes | 18:20 |
| aweeks | I'm curious if you have any alternative ideas | 18:20 |
| *** Marga_ has quit IRC | 18:29 | |
| *** meghal has quit IRC | 18:32 | |
| *** trown|lunch is now known as trown | 18:32 | |
| *** meghal has joined #openstack-ironic | 18:32 | |
| *** meghal has joined #openstack-ironic | 18:33 | |
| *** meghal has quit IRC | 18:33 | |
| *** meghal has joined #openstack-ironic | 18:34 | |
| *** Marga_ has joined #openstack-ironic | 18:50 | |
| *** dttocs has joined #openstack-ironic | 18:57 | |
| * devananda finishes a phone call then goes to lunch | 18:59 | |
| *** krtaylor has joined #openstack-ironic | 19:04 | |
| *** meghal has quit IRC | 19:16 | |
| *** mtanino has joined #openstack-ironic | 19:18 | |
| *** andreykurilin has joined #openstack-ironic | 19:32 | |
| *** kkoski has quit IRC | 19:35 | |
| *** kkoski has joined #openstack-ironic | 19:35 | |
| *** kkoski has quit IRC | 19:36 | |
| *** mtanino has quit IRC | 19:49 | |
| *** achanda has quit IRC | 20:01 | |
| *** jgrimm is now known as zz_jgrimm | 20:11 | |
| *** andreykurilin has quit IRC | 20:13 | |
| *** lucas-dinner has quit IRC | 20:17 | |
| *** Haomeng|2 has quit IRC | 20:18 | |
| *** Haomeng has joined #openstack-ironic | 20:19 | |
| *** ijw has quit IRC | 20:24 | |
| *** achanda has joined #openstack-ironic | 20:25 | |
| *** ijw has joined #openstack-ironic | 20:25 | |
| *** romcheg has joined #openstack-ironic | 20:25 | |
| *** Marga_ has quit IRC | 20:28 | |
| *** ijw has quit IRC | 20:29 | |
| *** ijw has joined #openstack-ironic | 20:33 | |
| *** lifeless1 is now known as lifeless | 20:37 | |
| rloo | hi devananda and anyone else that -2'd patches until liberty was open -- time to undo your -2's :-) | 20:40 |
| *** Nisha has quit IRC | 20:52 | |
| *** trown is now known as trown|outttypeww | 21:05 | |
| devananda | rloo: all: query string to find anything you've blocked in a project: is:reviewer status:open project:openstack/ironic label:Code-Review=-2 | 21:08 |
| devananda | not perfect, but close | 21:09 |
| rloo | devananda: cool, thx. | 21:09 |
| *** dttocs has quit IRC | 21:12 | |
| rloo | devananda: it shows all the patches that are -2'd, that I've ever reviewed, even if I didn't -2 it :-) | 21:12 |
| devananda | oh hah | 21:13 |
| devananda | you're right. it's not a great query, but it just so happens that almost all of my first page is just my own -2's so it looked like it worked :p | 21:14 |
| *** dprince has quit IRC | 21:14 | |
| rloo | devananda: as long as it works for you. I didn't -2 any so don't need to look ;) | 21:15 |
| openstackgerrit | Alex Weeks proposed openstack/ironic-specs: Add pluggable metrics backend for Ironic and IPA https://review.openstack.org/137171 | 21:17 |
| aweeks | ^ that change addresses some, but not all of the comments so far | 21:19 |
| *** dttocs has joined #openstack-ironic | 21:29 | |
| *** dttocs has quit IRC | 21:41 | |
| *** dttocs has joined #openstack-ironic | 21:44 | |
| mrda | Morning Ironic | 21:46 |
| NobodyCam | morning mrda | 21:46 |
| mrda | o/ | 21:47 |
| BadCub | hiya mrda :-) | 21:49 |
| mrda | hey | 21:50 |
| *** kkoski has joined #openstack-ironic | 21:58 | |
| *** dttocs_ has joined #openstack-ironic | 22:06 | |
| *** dttocs has quit IRC | 22:10 | |
| openstackgerrit | Clif Houck proposed openstack/ironic: Add support for image caching for the agent driver https://review.openstack.org/161832 | 22:12 |
| devananda | anyone know if kevin fox took this proposal anywhere else? https://review.openstack.org/#/c/137404/ | 22:18 |
| devananda | rloo: you'll be pleased to know, i have removed all my procedural blocks :) | 22:20 |
| *** Marga_ has joined #openstack-ironic | 22:21 | |
| *** kkoski1 has joined #openstack-ironic | 22:23 | |
| *** kkoski has quit IRC | 22:23 | |
| BadCub | w00t | 22:23 |
| *** ijw has quit IRC | 22:32 | |
| *** Sukhdev has joined #openstack-ironic | 22:32 | |
| *** achanda has quit IRC | 22:36 | |
| *** kkoski has joined #openstack-ironic | 22:41 | |
| *** kkoski1 has quit IRC | 22:41 | |
| *** achanda has joined #openstack-ironic | 22:41 | |
| *** romcheg has quit IRC | 22:43 | |
| *** dttocs_ has quit IRC | 22:44 | |
| *** Marga_ has quit IRC | 22:49 | |
| *** kkoski has quit IRC | 22:50 | |
| rloo | devananda: thx, the whole world is pleased :-) | 22:57 |
| *** yuanying has joined #openstack-ironic | 22:57 | |
| *** absubram has quit IRC | 22:59 | |
| *** Marga_ has joined #openstack-ironic | 23:07 | |
| *** chlong has joined #openstack-ironic | 23:10 | |
| *** ijw has joined #openstack-ironic | 23:11 | |
| openstackgerrit | jxiaobin proposed openstack/ironic-specs: Mount config drive as loop device to supply data to cloud-init https://review.openstack.org/173142 | 23:13 |
| *** Sukhdev has quit IRC | 23:13 | |
| *** Sukhdev has joined #openstack-ironic | 23:13 | |
| rwsu | what should notification_driver be set to in /etc/ironic/ironic.conf? I've tried "messaging", but no messages are going through the exchange | 23:20 |
| Haomeng | rwsu: I think notification_driver = messaging should be fine, we set it for ceilometer which can consume the ironic ipmi sensor data | 23:24 |
| rwsu | Haomeng: are there any other options that need to be set in /etc/ironic/ironic.conf? | 23:27 |
| Haomeng | rwsu: for that time I test the ipmi sensor data sending to ceilometer, one option notification_driver = messaging is enough, but not sure your case | 23:28 |
| rwsu | Haomeng: ok thanks, yeah there is something else wrong with my setup | 23:29 |
| Haomeng | rwsu: ok, welcome:) | 23:30 |
| *** Sukhdev has quit IRC | 23:35 | |
| *** zer0c00l has quit IRC | 23:39 | |
| *** zer0c00l has joined #openstack-ironic | 23:39 | |
| *** jerryz has joined #openstack-ironic | 23:42 | |
| *** chlong has quit IRC | 23:43 | |
| *** chlong has joined #openstack-ironic | 23:45 | |
| *** Sukhdev has joined #openstack-ironic | 23:46 | |
| *** zhenguo has joined #openstack-ironic | 23:51 | |
| *** dttocs has joined #openstack-ironic | 23:53 | |
| *** dttocs has quit IRC | 23:56 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!