| jroll | lifeless: in your experience, how many nodes can nova-compute handle? | 00:05 |
|---|---|---|
| *** killer_prince has joined #openstack-ironic | 00:16 | |
| *** killer_prince is now known as lazy_prince | 00:17 | |
| *** Haomeng|2 has joined #openstack-ironic | 00:17 | |
| *** Haomeng has quit IRC | 00:19 | |
| *** jcoufal has joined #openstack-ironic | 00:21 | |
| lifeless | jroll: I have not tested beyond 50 yet | 00:41 |
| jroll | oh. | 00:41 |
| *** malini1 has quit IRC | 00:41 | |
| jroll | we're seeing the resource tracker task do about 22 nodes/second | 00:42 |
| jroll | or about 15 seconds for ~330 nodes | 00:42 |
| lifeless | let mee | 00:42 |
| lifeless | let me see | 00:42 |
| lifeless | 2014-07-21 06:31:18.190 24497 AUDIT nova.compute.resource_tracker [-] Auditing locally available compute resources | 00:42 |
| lifeless | 2014-07-21 06:31:18.470 24497 AUDIT nova.compute.resource_tracker [-] Auditing locally available compute resources | 00:43 |
| lifeless | so about 300ms to do one | 00:43 |
| lifeless | 2014-07-21 06:31:18.900 24497 AUDIT nova.compute.resource_tracker [-] Auditing locally available compute resources | 00:43 |
| lifeless | 400ms there | 00:43 |
| jroll | yeah | 00:44 |
| lifeless | much less than 22/second | 00:45 |
| lifeless | this is an idle node | 00:45 |
| lifeless | one-node cloud | 00:45 |
| lifeless | 24 cores, 64GB of ram | 00:45 |
| lifeless | top - 00:45:45 up 13 days, 1:46, 4 users, load average: 0.99, 1.11, 0.93 | 00:45 |
| lifeless | sorry, 96GB | 00:45 |
| jroll | ok so | 00:46 |
| jroll | standalone n-cpu, 4GB cloud server | 00:46 |
| jroll | slash vm, whatever you want to call it | 00:46 |
| * jroll greps | 00:47 | |
| jroll | I don't think that's the right loop, sec | 00:47 |
| JayF | jroll: note: we have the hack to use node/details | 00:47 |
| JayF | jroll: so ours might be faster already | 00:47 |
| JayF | jroll: and we should upstream that :) | 00:47 |
| jroll | getting to that... | 00:48 |
| lifeless | yeah :) | 00:48 |
| jroll | 2014-07-22 00:47:54.476 27177 DEBUG ironic.nova.virt.ironic.driver [-] Returning 332 available node(s): | 00:49 |
| JayF | oh! | 00:49 |
| jroll | 2014-07-22 00:48:42.409 27177 DEBUG ironic.nova.virt.ironic.driver [-] Returning 332 available node(s): | 00:49 |
| JayF | I thought he was saying 300ms/node, not 300ms/whole run | 00:50 |
| jroll | gah | 00:50 |
| JayF | that's impressive | 00:50 |
| jroll | wrong lines, again | 00:50 |
| jroll | 2014-07-22 00:48:09.336 27177 INFO nova.compute.resource_tracker [-] Compute_service record updated for ironic-compute | 00:50 |
| jroll | that's the last one... | 00:50 |
| * jroll puts things in order | 00:50 | |
| jroll | 2014-07-22 00:47:54.476 27177 DEBUG ironic.nova.virt.ironic.driver [-] Returning 332 available node(s): | 00:50 |
| jroll | 2014-07-22 00:47:54.476 27177 DEBUG ironic.nova.virt.ironic.driver [-] Returning 332 available node(s): | 00:50 |
| jroll | ^ that's the entire update loop | 00:51 |
| jroll | fuck | 00:51 |
| * JayF hands jroll a clipboard | 00:51 | |
| jroll | 2014-07-22 00:47:54.476 27177 DEBUG ironic.nova.virt.ironic.driver [-] Returning 332 available node(s): | 00:51 |
| jroll | 2014-07-22 00:48:09.336 27177 INFO nova.compute.resource_tracker [-] Compute_service record updated for ironic-compute | 00:51 |
| jroll | is the entire loop | 00:51 |
| jroll | so 15s there | 00:51 |
| jroll | lifeless: this is including a hack to only hit ironic-api once, at /nodes/details: https://gist.github.com/jimrollenhagen/8a8783ae107a79db090a | 00:53 |
| lifeless | jroll: can you git review that ;) | 00:55 |
| jroll | lifeless: definitely | 00:55 |
| jroll | needs to be cleaned up and tests, but that's planned | 00:55 |
| jroll | we don't believe in proprietary :) | 00:56 |
| mrda | jroll: +1 | 00:56 |
| jroll | (agent driver is my top priority right now, though) | 00:56 |
| lifeless | yeah - I meant, just get it up, folk might do that for you | 00:56 |
| JayF | ^ the only place we differ in philosphy is that we write code at about 2x the pace it merges :P | 00:56 |
| jroll | true | 00:56 |
| jroll | lol | 00:56 |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic: Query full node details and cache https://review.openstack.org/108545 | 01:03 |
| jroll | lifeless: ^ had some conflicts so no idea if that works :) | 01:03 |
| JayF | comstud: ^ probably wants to know you put that up too | 01:04 |
| JayF | in case he wants to clean it up | 01:04 |
| jroll | I told him :P | 01:05 |
| jroll | ok, I need to go, night y'all | 01:05 |
| lifeless | night | 01:05 |
| mrda | rloo: \o | 01:05 |
| mrda | jroll: \o | 01:06 |
| rloo | hi Mr mrda | 01:06 |
| mrda | hi rloo - sorry typing fail. | 01:06 |
| mrda | Thanks for the review comments btw | 01:06 |
| rloo | yw mrda. hope it helps. although i need to look at the baremetal refactoring now that I think I understand what's going on ;) | 01:06 |
| mrda | it's not trivial, but I hope it's clear now | 01:07 |
| *** eghobo has joined #openstack-ironic | 01:07 | |
| rloo | mrda: I was thinking that it was hard to know what the 'original' baremetal_host_manager was, but I guess it doesn't make sense to checkin the original file first, before refactoring. | 01:08 |
| rloo | anyway mrda, I'll look at that tomorrow. that's the plan anyway ;) | 01:09 |
| mrda | yeah, mikal provided a unified diff of the ironic_host_manager.py against the original | 01:09 |
| mrda | thanks rloo | 01:09 |
| mrda | much appreciated! | 01:09 |
| *** eghobo has quit IRC | 01:12 | |
| *** eghobo has joined #openstack-ironic | 01:12 | |
| *** jcoufal has quit IRC | 01:13 | |
| *** nosnos has joined #openstack-ironic | 01:40 | |
| *** Haomeng has joined #openstack-ironic | 01:45 | |
| *** Haomeng has quit IRC | 01:46 | |
| *** Haomeng|2 has quit IRC | 01:47 | |
| *** eguz has joined #openstack-ironic | 02:03 | |
| *** eguz has quit IRC | 02:04 | |
| *** eguz has joined #openstack-ironic | 02:04 | |
| *** rainya has joined #openstack-ironic | 02:06 | |
| *** eghobo has quit IRC | 02:07 | |
| *** rainya has quit IRC | 02:09 | |
| *** rainya has joined #openstack-ironic | 02:10 | |
| *** rloo has quit IRC | 02:12 | |
| *** eguz has quit IRC | 02:36 | |
| *** vinbs has joined #openstack-ironic | 02:59 | |
| mrda | dtantsur|afk: as per our conversation yesterday, here's a Nova patch you will be pleased with: https://review.openstack.org/#/c/108564/ | 03:00 |
| *** eghobo has joined #openstack-ironic | 03:18 | |
| *** rainya has quit IRC | 03:34 | |
| *** Poornima|mtg has joined #openstack-ironic | 03:34 | |
| *** Poornima|mtg is now known as Poornima_ | 03:34 | |
| *** nosnos has quit IRC | 03:40 | |
| *** eghobo has quit IRC | 03:55 | |
| *** eghobo has joined #openstack-ironic | 03:55 | |
| *** nosnos has joined #openstack-ironic | 04:06 | |
| *** shakamunyi has joined #openstack-ironic | 04:06 | |
| *** shakamunyi has quit IRC | 04:08 | |
| *** pcrews has quit IRC | 04:10 | |
| *** rameshg87 has joined #openstack-ironic | 04:11 | |
| *** lazy_prince is now known as killer_prince | 04:14 | |
| *** Nisha has joined #openstack-ironic | 04:14 | |
| *** krtaylor has quit IRC | 04:18 | |
| *** Poornima_ has quit IRC | 04:19 | |
| *** Poornima_ has joined #openstack-ironic | 04:30 | |
| *** Poornima_ has quit IRC | 04:39 | |
| *** krtaylor has joined #openstack-ironic | 05:03 | |
| *** Poornima|mtg has joined #openstack-ironic | 05:07 | |
| *** Poornima|mtg is now known as Poornima_ | 05:07 | |
| *** ramineni has joined #openstack-ironic | 05:17 | |
| *** romcheg1 has joined #openstack-ironic | 05:30 | |
| *** romcheg1 has left #openstack-ironic | 05:30 | |
| *** shausy has joined #openstack-ironic | 05:33 | |
| *** rameshg87 has left #openstack-ironic | 05:37 | |
| *** bvivek has joined #openstack-ironic | 05:37 | |
| *** killer_prince is now known as lazy_prince | 05:57 | |
| *** rakesh_hs has joined #openstack-ironic | 05:58 | |
| *** k4n0 has joined #openstack-ironic | 05:59 | |
| *** pradipta_away is now known as pradipta | 05:59 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic: Imported Translations from Transifex https://review.openstack.org/106948 | 06:12 |
| openstackgerrit | A change was merged to openstack/ironic: Use opportunistic approach for migration testing https://review.openstack.org/107053 | 06:24 |
| *** kpavel has joined #openstack-ironic | 06:31 | |
| *** eghobo has quit IRC | 06:36 | |
| *** Halacs has joined #openstack-ironic | 06:37 | |
| openstackgerrit | Haomeng,Wang proposed a change to openstack/ironic: Implements send-data-to-ceilometer https://review.openstack.org/72538 | 06:39 |
| *** vkdrao has joined #openstack-ironic | 06:42 | |
| *** vkdrao has quit IRC | 06:46 | |
| *** Mikhail_D_ltp has joined #openstack-ironic | 06:49 | |
| openstackgerrit | Haomeng,Wang proposed a change to openstack/ironic: Implements send-data-to-ceilometer https://review.openstack.org/72538 | 06:50 |
| *** Haomeng has joined #openstack-ironic | 06:51 | |
| *** bmahalakshmi has joined #openstack-ironic | 07:01 | |
| dtantsur|afk | mrda, cool, thank you, will have a look | 07:03 |
| *** dtantsur|afk is now known as dtantsur | 07:04 | |
| dtantsur | and good morning Ironic (said man trying to open his eyes) | 07:04 |
| *** harlowja is now known as harlowja_away | 07:04 | |
| mrda | dtantsur: that was me at the meeting 12 hours ago :) | 07:11 |
| dtantsur | heh, I can understand it | 07:11 |
| dtantsur | and I'm always feeling bad the morning after the meeting :( | 07:11 |
| *** rameshg87 has joined #openstack-ironic | 07:22 | |
| rameshg87 | dtantsur: hi | 07:27 |
| dtantsur | rameshg87, hi | 07:27 |
| rameshg87 | dtantsur: just one more question on https://review.openstack.org/#/c/107344 | 07:27 |
| dtantsur | sure | 07:27 |
| rameshg87 | dtantsur: in the spec we say "On next start-up the driver's __init__ code will set up the environment" | 07:28 |
| *** kpavel has quit IRC | 07:28 | |
| rameshg87 | dtantsur: but where does the driver tear down the environment so that nodes don't boot again ?? | 07:28 |
| rameshg87 | dtantsur: so that the pxe/tftp configuration to bootup the node to discovery ramdisk is removed off .. | 07:28 |
| rameshg87 | dtantsur: i think we might want this for an admin who might set discovery.enabled=false later on | 07:29 |
| dtantsur | rameshg87, good catch! I guess we need to introduce something like set_up_discovery and tear_down_discovery to DriverInterface... | 07:29 |
| dtantsur | what do you think? | 07:29 |
| rameshg87 | dtantsur: yes, we might need | 07:29 |
| *** bvivek has quit IRC | 07:30 | |
| rameshg87 | dtantsur: and regarding the nics i think your point is valid | 07:30 |
| rameshg87 | dtantsur: better to leave it to ramdisk which does discovery :-) | 07:31 |
| dtantsur | rameshg87, ack, thank you | 07:31 |
| rameshg87 | dtantsur: thanks | 07:31 |
| *** bvivek has joined #openstack-ironic | 07:32 | |
| rameshg87 | dtantsur: but when can tear_down_discovery happen ? on stopping the conductor ? | 07:34 |
| dtantsur | rameshg87, currently yes. Later - when it's disabled by any means. | 07:35 |
| rameshg87 | dtantsur: okay .. | 07:35 |
| rameshg87 | dtantsur: a problem when process dies off suddenly due to any issue :-( | 07:36 |
| *** malini has joined #openstack-ironic | 07:36 | |
| dtantsur | rameshg87, here is HA-related part of discovery: if it's still enabled, we need another conductor to carry on | 07:36 |
| dtantsur | so set_up_discovery must somehow clean up environment before it proceedes | 07:37 |
| rameshg87 | dtantsur: yes .. | 07:37 |
| rameshg87 | dtantsur: probably __init__() can try to clean up environment if discovery.enabled=false ?? | 07:37 |
| rameshg87 | dtantsur: there is no harm in trying once .. | 07:37 |
| dtantsur | rameshg87, if the only conductor dies and then the admin disabled discovery... | 07:38 |
| dtantsur | rameshg87, but what driver to call? we do not know, which driver used to do the discovery | 07:38 |
| rameshg87 | dtantsur: hmm.. yes | 07:38 |
| rameshg87 | dtantsur: it could have changed | 07:38 |
| dtantsur | rameshg87, the more I think about it, the more I suspect we need to persist some information about discovery in DB | 07:40 |
| rameshg87 | dtantsur: do you mean persist discovery.driver in db ? | 07:40 |
| dtantsur | rameshg87, yes, store the last driver and status of discovery | 07:41 |
| dtantsur | rameshg87, so that on restart/failure this information can be recovered | 07:41 |
| *** jistr has joined #openstack-ironic | 07:42 | |
| dtantsur | rameshg87, we can cheat a bit and store it in the Conductor model, so that we do not create a new table | 07:42 |
| rameshg87 | dtantsur: yeah, i was just checking the same | 07:43 |
| rameshg87 | dtantsur: i don't think it's cheating because it truly belongs to conductor | 07:43 |
| rameshg87 | :-) | 07:43 |
| dtantsur | makes sense, yeah | 07:43 |
| *** foexle has joined #openstack-ironic | 07:43 | |
| rameshg87 | dtantsur: yeah, may be on startup of conductor, it can retrieve value from db, and call cleanup of discovery.driver in db | 07:44 |
| dtantsur | rameshg87, yeah. There is a corner case, when this driver is no longer enabled, but it's solvable | 07:44 |
| rameshg87 | dtantsur: and then call setup() for the new(or the same) discovery.driver if its enabled | 07:45 |
| rameshg87 | dtantsur: oh :-) | 07:45 |
| dtantsur | we can error-out, if we need to clean up, but the driver is disabled | 07:45 |
| rameshg87 | dtantsur: yeah, makes sense so that admin is informed | 07:45 |
| rameshg87 | dtantsur: then we might have covered every scenario :-) | 07:46 |
| rameshg87 | dtantsur: then for now it might not be required to do cleanup on stopping the conductor right ? | 07:46 |
| rameshg87 | dtantsur: or may be still worth attempting to prevent unexpected pxe bootings from the conductor node | 07:47 |
| dtantsur | rameshg87, I think it is | 07:47 |
| rameshg87 | dtantsur: okay | 07:47 |
| dtantsur | rameshg87, thanks for this discussion, helps a lot :) | 07:49 |
| rameshg87 | dtantsur: :-) | 07:49 |
| dtantsur | quick walk, be back in ~ 1 hour | 07:50 |
| *** k4n0 has quit IRC | 07:51 | |
| *** foexle has quit IRC | 07:54 | |
| openstackgerrit | A change was merged to openstack/ironic: IPMITool to use the new ManagementInterface https://review.openstack.org/86092 | 07:55 |
| openstackgerrit | Michael Davies proposed a change to openstack/ironic: Review fixes from Nova scheduler reviews https://review.openstack.org/107316 | 07:55 |
| *** foexle has joined #openstack-ironic | 07:56 | |
| *** k4n0 has joined #openstack-ironic | 08:04 | |
| viktors | GheRivero: hi! | 08:07 |
| *** kpavel has joined #openstack-ironic | 08:10 | |
| *** bvivek has quit IRC | 08:13 | |
| *** lucasagomes has joined #openstack-ironic | 08:17 | |
| yuriyz | morning Ironic | 08:19 |
| *** bvivek has joined #openstack-ironic | 08:21 | |
| *** k4n0 has quit IRC | 08:32 | |
| *** saripurigopi has joined #openstack-ironic | 08:41 | |
| saripurigopi | I'm trying to add blueprint, when I try to push the changes, it is failing with permission denied. | 08:42 |
| saripurigopi | do I need additional privileges to upload blueprint ? | 08:42 |
| *** malini has quit IRC | 08:42 | |
| saripurigopi | remote: Permission to openstack/ironic-specs.git denied to saripurigopi@gmail.com. fatal: unable to access 'https://github.com/openstack/ironic-specs/': The requested URL returned error: 403 | 08:42 |
| lazy_prince | saripurigopi | 08:43 |
| lazy_prince | are you using gerrit or git push..? | 08:43 |
| saripurigopi | git push for committing the spec. | 08:44 |
| lazy_prince | saripurigopi: you need to use git-review. | 08:44 |
| dtantsur | yuriyz, morning :) | 08:48 |
| *** k4n0 has joined #openstack-ironic | 08:49 | |
| dtantsur | lucasagomes, morning :) "IPMITool to use the new ManagementInterface" merged | 08:49 |
| lucasagomes | o///// | 08:49 |
| lucasagomes | thanks! | 08:49 |
| mrda | hi dtantsur and lucasagomes | 08:49 |
| lucasagomes | mrda, morning! | 08:49 |
| dtantsur | mrda, good evening :) | 08:50 |
| mrda | what time is it? | 08:50 |
| mrda | :) | 08:50 |
| lucasagomes | 9:50am | 08:50 |
| dtantsur | mrda, 10:50 (morning) here | 08:50 |
| lucasagomes | mrda, good evening there? | 08:50 |
| mrda | 6:20pm here, still going from the 4:30am meeting this morning :) | 08:50 |
| lucasagomes | woah! | 08:51 |
| dtantsur | lucasagomes, also we had a discussion with rameshg87 before you arrived. It resulted in following: 1. we need a set_up_discovery and tear_down_discovery on DriverInterface; 2. we do need to persist information about discovery (I'm finally convinced!), we can use Conductor model for it | 08:51 |
| lucasagomes | yup | 08:52 |
| lucasagomes | set_up...tear_down is the active/deactive functions | 08:52 |
| lucasagomes | dtantsur, so, it will be exposed on the API or config options? | 08:53 |
| lucasagomes | I was thinking about the config options, the problem with them is how to pass parameters | 08:53 |
| lucasagomes | hw to tell ironic what is the discovery ramdisk and kernel | 08:53 |
| lucasagomes | when it was on the api I was thinking about something like | 08:53 |
| lucasagomes | PUT -d {'discovery_ramdisk' ..., 'discovery_kernel':...} /drivers/pxe_*/discovery | 08:54 |
| lucasagomes | somehting like that | 08:54 |
| dtantsur | lucasagomes, too many problems with new API, as was discussed yesterday | 08:54 |
| dtantsur | lucasagomes, currently it will happen on start/stop the conductor | 08:55 |
| dtantsur | lucasagomes, we can store everything driver-specific in static configuration | 08:55 |
| lucasagomes | dtantsur, right | 08:55 |
| lucasagomes | that means that, this parameters should be at the driver_info before the conductor starts? | 08:56 |
| dtantsur | lucasagomes, in configuration. driver_info is for a node | 08:56 |
| lucasagomes | ah | 08:56 |
| lucasagomes | true | 08:56 |
| lucasagomes | hmm | 08:56 |
| dtantsur | lucasagomes, I am updating the spec right now, so we can discuss later, how it is shaping | 08:58 |
| *** igordcard has joined #openstack-ironic | 08:59 | |
| lucasagomes | ack | 09:01 |
| *** bvivek has quit IRC | 09:04 | |
| *** pradipta is now known as pradipta_away | 09:07 | |
| *** bvivek has joined #openstack-ironic | 09:13 | |
| *** athomas has joined #openstack-ironic | 09:18 | |
| *** k4n0 has quit IRC | 09:19 | |
| dtantsur | lucasagomes, putting any HA considerations there is a can of worms >_< | 09:19 |
| dtantsur | once you start, you can't finish... | 09:19 |
| lucasagomes | hah I can imagine | 09:20 |
| dtantsur | lucasagomes, I'm seriously thinking of 2 specs: first stating "only one conductor can do discovery at the same time", second about HA and friends (on which we can argue infinitely) | 09:21 |
| mrda | dtantsur: if that will help us make progress, I see no harm | 09:22 |
| lucasagomes | dtantsur, :/ | 09:22 |
| lucasagomes | well openstack wise that's not very well acceptable really | 09:22 |
| lucasagomes | people want it to scale | 09:22 |
| lucasagomes | 1 conductor hardly would be able to handle more than 5~10 machines booting at the same time with the normal PXE (tftp) | 09:23 |
| saripurigopi | I'm trying to submit spec review with git-review, now it is falling with "fatal: ICLA contributor agreement requires current contact information." | 09:23 |
| lucasagomes | if you look at the video I have for the discovery it's 3 machines, the network is just a bridge, and the TFTP speed is very slow | 09:24 |
| saripurigopi | I've the contact info updated in my profile. | 09:24 |
| dtantsur | lucasagomes, conductor without discovery support will boot 0 new machines :/ and that's where we are going | 09:24 |
| lazy_prince | saripurigopi: did you sign ICLA..? | 09:24 |
| rameshg87 | saripurigopi: you neeed to goto review.openstack.org -> top right bar Settings -> Agreements | 09:24 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Rename/update common/tftp.py to common/pxe_utils.py https://review.openstack.org/103595 | 09:24 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Add iPXE Installation Guide documentation https://review.openstack.org/106809 | 09:24 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Add iPXE support for Ironic https://review.openstack.org/99318 | 09:24 |
| saripurigopi | Yes, I've signed ICLA. | 09:25 |
| saripurigopi | VerifiedICLAOpenStack Individual Contributor License Agreement | 09:25 |
| rameshg87 | lucasagomes: just a question on ipxe cache in https://review.openstack.org/#/c/99318/11/ironic/drivers/modules/pxe.py | 09:25 |
| lucasagomes | rameshg87, yup I answered that | 09:25 |
| lucasagomes | rameshg87, and fixed the tests you pointed out | 09:25 |
| rameshg87 | lucasagomes: yeah the question is on the comment | 09:26 |
| lucasagomes | rameshg87, oh... lemme take a look I'm sorry | 09:26 |
| rameshg87 | lucasagomes: i am just about to ask :-) | 09:26 |
| lucasagomes | rameshg87, (2 comments) | 09:26 |
| lucasagomes | Just a couple of small issues. | 09:26 |
| rameshg87 | lucasagomes: when we have a conductor running ipxe only | 09:26 |
| rameshg87 | lucasagomes: wouldn't the admin need to make sure space is available for two mounts /tftproot and /http_root ? | 09:27 |
| rameshg87 | lucasagomes: if we continue to make use of TFTPImageCache | 09:27 |
| lucasagomes | rameshg87, yeah, as you said I still using TFTP for the master images | 09:27 |
| lucasagomes | I didn't want to touch that part of the code for now | 09:27 |
| lucasagomes | rameshg87, do you think I need to refactor that before? | 09:28 |
| dtantsur | lucasagomes, btw, if Neutron can't route new nodes to different conductors, it won't work anyway | 09:28 |
| lucasagomes | I don't really mind doing that, the TFTP on the names really annoys me | 09:28 |
| rameshg87 | lucasagomes: can't we just change the master dir to /http_root/master_images if ipxe is enabled ? | 09:28 |
| lucasagomes | rameshg87, yup we can | 09:29 |
| rameshg87 | lucasagomes: so that all the deploy_kr_images go inside /http_root | 09:29 |
| rameshg87 | lucasagomes: and we the admin doesn't need to bother about space in /tftp_root | 09:29 |
| lucasagomes | I will take a look at it | 09:29 |
| lucasagomes | rameshg87, +1 | 09:29 |
| rameshg87 | lucasagomes: okay :-) | 09:29 |
| rameshg87 | lucasagomes: thanks | 09:30 |
| lucasagomes | I will do in a separated patch that refacotr, cause it involves config options | 09:30 |
| saripurigopi | do I need additional permissions to access github.com/openstack/ironic-specs to submit review? | 09:30 |
| lucasagomes | tftp_master_dir | 09:30 |
| lucasagomes | etc... | 09:30 |
| dtantsur | lucasagomes, and if TFTP is slow for 3 machines, we just have to use iPXE. I don't believe a single node can't serve hundreds of files simultaneously | 09:30 |
| lucasagomes | dtantsur, yeah with iPXE it will work better | 09:30 |
| lucasagomes | I mean, way better | 09:30 |
| lucasagomes | rameshg87, btw... so what about making the master_dir generic? | 09:31 |
| lucasagomes | rameshg87, I mean... the op can choose the dir he wants to cache the master images | 09:31 |
| lucasagomes | doesn't matter if it's tftp or http | 09:31 |
| *** openstackgerrit has quit IRC | 09:31 | |
| lucasagomes | I prefer to avoid to tie the dir to cache image with the transport protocol, it makes little sense IMO | 09:32 |
| rameshg87 | lucasagomes: oh yeah | 09:32 |
| rameshg87 | lucasagomes: i guess it is already a config option | 09:32 |
| lucasagomes | rameshg87, yeah, with the "tftp_" on the name | 09:32 |
| rameshg87 | lucasagomes: then you were just referring to change the name ? | 09:32 |
| lucasagomes | I will try to work that out | 09:32 |
| lucasagomes | yeah, small refactor | 09:33 |
| lucasagomes | to make it less misleading | 09:33 |
| *** openstackgerrit has joined #openstack-ironic | 09:33 | |
| openstackgerrit | Michael Davies proposed a change to openstack/ironic: Review fixes from the Nova driver reviews https://review.openstack.org/107882 | 09:33 |
| rameshg87 | lucasagomes: oh okay..that makes sense .. i just wanted to see if we can easily make both the cache and the actual files in the same directory - whether it will be pxe or ipxe | 09:33 |
| rameshg87 | lucasagomes: so we don't have to go and change anything :-) | 09:33 |
| lucasagomes | rameshg87, sure... I will take a look at it | 09:34 |
| lucasagomes | rameshg87, thanks for pointing that out | 09:34 |
| rameshg87 | lucasagomes: :-) | 09:34 |
| *** athomas has quit IRC | 09:36 | |
| saripurigopi | I'm trying to submit the review for first time. | 09:36 |
| *** k4n0 has joined #openstack-ironic | 09:38 | |
| rameshg87 | saripurigopi: no you don't need | 09:38 |
| lucasagomes | saripurigopi, https://wiki.openstack.org/wiki/Documentation/HowTo/FirstTimers#Step-by-step_tutorial | 09:38 |
| openstackgerrit | Victor Sergeyev proposed a change to openstack/ironic: Use metadata.create_all() to get database schema https://review.openstack.org/107629 | 09:44 |
| openstackgerrit | Victor Sergeyev proposed a change to openstack/ironic: Add a test case for DB schema comparison https://review.openstack.org/107628 | 09:44 |
| *** athomas has joined #openstack-ironic | 09:49 | |
| saripurigopi | I've the SSH keys in place and still it fails with "Please make sure you have the correct access rights and the repository exists." | 09:51 |
| mrda | Good night ironic, see you tomorrow! | 09:53 |
| *** mrda is now known as mrda-away | 09:53 | |
| *** NobodyCam has quit IRC | 09:54 | |
| *** Shrews has quit IRC | 09:54 | |
| *** jogo has quit IRC | 09:54 | |
| *** anteaya has quit IRC | 09:54 | |
| *** jogo has joined #openstack-ironic | 09:54 | |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic-specs: Generic Hardware Discovery Bits https://review.openstack.org/107344 | 09:55 |
| *** NobodyCam has joined #openstack-ironic | 09:55 | |
| dtantsur | lucasagomes, rameshg87, ok, I got something new ^^^ can we discuss it? | 09:55 |
| *** anteaya has joined #openstack-ironic | 09:55 | |
| *** Shrews has joined #openstack-ironic | 09:56 | |
| *** nosnos has quit IRC | 09:58 | |
| *** nosnos has joined #openstack-ironic | 09:58 | |
| rameshg87 | dtantsur: i am in ... | 10:01 |
| lifeless | dtantsur: hi, you pinged the other day? | 10:02 |
| *** nosnos has quit IRC | 10:03 | |
| dtantsur | lifeless, hi, I don't quite remember :) maybe it was about whether lucas needs a spec for his ramdisk or not... | 10:03 |
| lucasagomes | dtantsur, what is DriverInterface? | 10:03 |
| *** martyntaylor has joined #openstack-ironic | 10:04 | |
| dtantsur | lucasagomes, hmm.. I meant BaseDriver, no idea why I called it that way | 10:04 |
| lucasagomes | so the setup and tear down won't be linked with any interface? | 10:05 |
| dtantsur | lucasagomes, I guess no. As Deva stated yesterday, interfaces are related to nodes | 10:06 |
| lifeless | oh yeah, it was, nvm :) | 10:07 |
| rameshg87 | dtantsur: i have a question | 10:08 |
| rameshg87 | dtantsur: "For existing row, check if conductor is alive, using the hash ring.86 If it is not and the same driver is enabled on the current node,87 ask this driver to do tear down, than proceed with enabling the discovery." | 10:08 |
| rameshg87 | dtantsur: will tear down work if it is performed by someother conductor node ? | 10:08 |
| rameshg87 | dtantsur: for pxe, conductor A would have setup a tftp configuration in its own machine | 10:09 |
| dtantsur | rameshg87, it should at least to some extend. Imagine PXE or IPA driver will have to clean Neutron config | 10:09 |
| rameshg87 | dtantsur: conductor B can't tear this down | 10:09 |
| dtantsur | rameshg87, of course it won't clean local caches, but local caches are not going to break things :) | 10:09 |
| dtantsur | Neutron configuration will, on the contrary | 10:09 |
| rameshg87 | dtantsur: oh okay..it can do it with neutron..:-) | 10:09 |
| rameshg87 | dtantsur: caches won't affect it .. | 10:10 |
| rameshg87 | dtantsur: looks okay to me. i will have a detailed look to see if my eye can find something :-) | 10:11 |
| dtantsur | thanks! | 10:11 |
| *** romcheg1 has joined #openstack-ironic | 10:15 | |
| lucasagomes | so on the conductor that is serving the dhcp for the discovery, neutron will be configured to answer the dhcp request for all unkown NICs | 10:16 |
| lucasagomes | conductor where the discovery is enabled | 10:16 |
| lucasagomes | I can't see how this take over happens | 10:16 |
| * lucasagomes thinking.... | 10:17 | |
| lucasagomes | cause, let's say the conductor doing the previously discovery haven't died | 10:18 |
| lucasagomes | it was a network problem | 10:18 |
| lucasagomes | so hashring thought it was died | 10:18 |
| lucasagomes | now it's back | 10:18 |
| lucasagomes | and the dhcp server still running on it... | 10:19 |
| lucasagomes | maybe I'm not understand the detection model... how it will detect that the previous conductor died? will have a periodic task looking at the hash ring to see that? | 10:20 |
| lucasagomes | how we are going to elect a new conductor to be the one serving the discovery? | 10:20 |
| *** viktors is now known as viktors|afk | 10:22 | |
| lifeless | lucasagomes: I'm confused | 10:24 |
| lucasagomes | lifeless, me too | 10:24 |
| lucasagomes | totally | 10:24 |
| lifeless | lucasagomes: thats a neutron dhcp agent HA problem, not an ironic conductor problem | 10:24 |
| lifeless | lucasagomes: isn't it? | 10:24 |
| lucasagomes | lifeless, so this is the new spec https://review.openstack.org/107344 | 10:25 |
| lucasagomes | it has the discovery enabled in one conductor in the cluster (instead of all conductors that support that driver) | 10:26 |
| lucasagomes | so only one can receive the dhcp request for unkown NICs | 10:26 |
| lucasagomes | that's what I'm missing, how to tie conductor with the neutron dhcp server | 10:26 |
| dtantsur | lucasagomes, ehhhrr.. we'll never receive DHCP requests | 10:29 |
| dtantsur | please don't confuse | 10:29 |
| lucasagomes | dtantsur, right... so how we are going to PXE boot the image? | 10:29 |
| dtantsur | lucasagomes, neutron will point to the conductor, on which it's enabled | 10:29 |
| dtantsur | I mean, discovery is enabled | 10:30 |
| dtantsur | lucasagomes, so as the first step, it's matter of static configuration | 10:31 |
| lucasagomes | dtantsur, right, so neutron dhcp server will be running and configured to answer DHCP requests on the conductor where discovery is enabled | 10:31 |
| lucasagomes | right? | 10:31 |
| dtantsur | (which obviously sucks, but still) | 10:31 |
| lucasagomes | DHCP requests for unknown NICs* | 10:31 |
| dtantsur | lucasagomes, yes | 10:31 |
| lucasagomes | dtantsur, ok... | 10:31 |
| lifeless | lucasagomes: why are you saying 'on the conductor' | 10:31 |
| lucasagomes | lifeless, cause the PXE enviroment will be set up on that conductor | 10:31 |
| lifeless | lucasagomes: theres absolutely no reason that the neutron dhcp server would be on the conductor machine | 10:31 |
| lucasagomes | lifeless, the default PXE configuration file where the unregistered machines will boot from will be locally on that conductor | 10:32 |
| lifeless | lucasagomes: yes, the PXE file will be | 10:32 |
| lifeless | but neutron' DHCP server(s) may be elsewhere | 10:32 |
| *** Amar has joined #openstack-ironic | 10:32 | |
| *** Amar is now known as Guest6703 | 10:32 | |
| lifeless | lucasagomes: (not to mention that Neutron doesn't have wildcard DHCP support today (at least, AFAIK), so I'm not sure how the spec can be implemented without some discussion with Neutron about that. | 10:33 |
| dtantsur | lifeless, re authentication: it's a long-lasting topic, IPA is also discussing it and the only answer for J currently is: secure your network | 10:33 |
| dtantsur | and yes, it sucks | 10:33 |
| *** ndipanov has joined #openstack-ironic | 10:33 | |
| lucasagomes | lifeless, right.. ok so we just update_port and give the address of the node which contains the PXE config file... yeah makes sense | 10:33 |
| lifeless | dtantsur: If there is consensus on that, thats fine - but I think it needs to be explictly spelt out in this spec, not just 'oh its a problem alright' :) | 10:34 |
| lifeless | lucasagomes: not update_port - there is no port oject | 10:34 |
| lifeless | lucasagomes: its going to be a subnet change | 10:34 |
| dtantsur | lifeless, ok, I'll try to make it clearer and mention discussion with IPA folks also... | 10:34 |
| lucasagomes | lifeless, right, ok I think I'm understanding it better | 10:35 |
| dtantsur | lucasagomes, if neutron does not support it, we're in trouble | 10:35 |
| lucasagomes | dtantsur, yeah, let's check with marios | 10:36 |
| *** ndipanov has quit IRC | 10:38 | |
| *** marios has joined #openstack-ironic | 10:39 | |
| marios | dtantsur: hey | 10:39 |
| dtantsur | marios, hi! We're discussing hardware auto-discovery here and we need to know, if we can tell neutron: "for all unknown MAC addresses, point them to this IP for PXE boot" | 10:39 |
| marios | dtantsur: hmm, i don't know off hand, but i can try and find out. the l2 subnets will hand out dhcp info to vms coming up (so for 'unkown mac addressses') but don't know if there is something for pxe in neutron (well, for this arbitrary metadata) | 10:41 |
| marios | dtantsur: but, you may get better mileage asking this directly in #openstack-neutron | 10:42 |
| marios | dtantsur: is there any more context here? the machines aren't booted yet, so are they even managed by neutron at this point? | 10:44 |
| dtantsur | marios, no, they are not. It is for new machines enrolling. | 10:44 |
| marios | right. so then this is more of a tripleo metadata question no? | 10:44 |
| marios | (heat metadata) | 10:44 |
| marios | but i guess i am missing lots still here | 10:44 |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic-specs: Generic Hardware Discovery Bits https://review.openstack.org/107344 | 10:44 |
| dtantsur | lucasagomes, lifeless ^^^ updated | 10:45 |
| dtantsur | marios, the task is: the admin brought 10 new servers, plugged them in network and power | 10:45 |
| dtantsur | marios, now we need to boot a specific ramdisk on them, collect hardware inventory and report back to Ironic | 10:45 |
| marios | dtantsur: is the description in that spec ^^ | 10:45 |
| dtantsur | marios, more-or-less should be | 10:46 |
| marios | dtantsur: ok | 10:51 |
| marios | so if you have a subnet, where the default gateway is the pxe server, that could work | 10:53 |
| marios | dtantsur: still there? | 10:54 |
| dtantsur | yep | 10:54 |
| dtantsur | marios, can we do it with Neutron? | 10:55 |
| marios | dtantsur: as i said, i don't know :) i am trying to understand more of the problem | 10:55 |
| marios | dtantsur: so once ironic has this info, it gives it to ? tuskar-ui? | 10:55 |
| dtantsur | marios, it stores it and tuskar UI uses it later, yes | 10:56 |
| marios | dtantsur: ok... so tuskar defines the heat env for tripleo (e.g. will include those discovered mac addresses per host) | 10:58 |
| marios | dtantsur: so first thing, i don't know enough about pxe, but the hosts need to have already received an IP address first before they can boot right | 10:59 |
| dtantsur | marios, I am not sure | 10:59 |
| dtantsur | ok, lucasagomes, should we bring this to #openstack-neutron ? | 11:03 |
| marios | dtantsur: so, your PXE server needs to be on the same subnet as your hosts | 11:03 |
| marios | dtantsur: i think that's a very good idea | 11:03 |
| marios | let's go there now | 11:03 |
| lucasagomes | dtantsur, yes sounds like we do | 11:03 |
| *** Halacs has quit IRC | 11:04 | |
| *** athomas_ has joined #openstack-ironic | 11:04 | |
| *** athomas has quit IRC | 11:06 | |
| *** ramineni has quit IRC | 11:06 | |
| lifeless | NobodyCam: comstud: https://review.openstack.org/108640 | 11:10 |
| lifeless | marios: VMs coming up are known MACs | 11:10 |
| lifeless | marios: PXE server doesn't need to be on the same subnet :) | 11:10 |
| marios | lifeless: oh really? but it's basically using dhcp (are you talking about relay?) | 11:12 |
| lifeless | PXE is unicast | 11:12 |
| lifeless | or rather | 11:12 |
| lifeless | PXE is a combination of DHCP metadata and a TFTP server | 11:12 |
| lifeless | the TFTP is unicast | 11:13 |
| marios | right | 11:13 |
| lifeless | the DHCP is relayable yes | 11:13 |
| marios | (so much wikipedia told me, as i said i really don't know a lot about pxe) | 11:13 |
| openstackgerrit | Gopi Krishna S proposed a change to openstack/ironic-specs: Add HTTP GET support for vendor_passthru API https://review.openstack.org/108646 | 11:13 |
| marios | i see. but do we have dhcp-relay in neutron, i'm not sure we do today | 11:13 |
| marios | lifeless: ^^^ | 11:13 |
| *** nosnos has joined #openstack-ironic | 11:13 | |
| marios | lifeless: but in anycase, aren't we considering the simple case first, where they are on the same subnet? | 11:13 |
| marios | lifeless: isn't out baremetal network single flat, right now. or are we already talking about subnetting the undercloud network? | 11:14 |
| lifeless | possibly, I don't think it affects our code | 11:14 |
| lifeless | marios: those are separate concerns | 11:14 |
| lifeless | marios: point is the DHCP server doesn't need to be in the same subnet | 11:15 |
| lifeless | it could be on a different subnet, our flat network code still works | 11:15 |
| marios | lifeless: which dhcp server though, the neutron one, or the server running on your real world router | 11:15 |
| marios | lifeless: since in neutron dhcp is enabled by subnet (is what i mean) | 11:16 |
| dtantsur | marios, what do you mean by "dhcp enabled by subnet" | 11:17 |
| dtantsur | aren't subnets property of IP? | 11:17 |
| marios | dtantsur: yes they are, that's why i was asking about neutron vs real | 11:17 |
| lifeless | marios: a neutron managed one, which doesn't preclude it being on the router:) | 11:17 |
| marios | lifeless: i see :) | 11:17 |
| lifeless | point is we don't need to think about that | 11:18 |
| lifeless | its a separate concern | 11:18 |
| marios | lifeless: ok, so i only just got pulled into this so am trying to understand the problem still | 11:18 |
| lifeless | we tell neutron what we want, and if the network admins have set it up with relay vs same-l2-domain, its not our concern | 11:19 |
| *** bmahalakshmi has quit IRC | 11:19 | |
| marios | lifeless: so, i think lucasagomes is right that we may be able to use extra-dhcp-options (since it has an example for pxe specifically) | 11:20 |
| lifeless | its per port | 11:20 |
| lifeless | you would need to create a port for every possible MAC :) | 11:20 |
| marios | AHH... this conversation is split between two channels, i have a headache | 11:20 |
| marios | lifeless: right, why is that a problem | 11:20 |
| lucasagomes | hah yeah let's stick to this channel? | 11:20 |
| marios | lifeless: you then use those ports to boot | 11:20 |
| marios | lifeless: one port for each host | 11:21 |
| lifeless | so as I said, AFAIK neutron has no support for wildcard DHCP today | 11:21 |
| marios | nova boot <server_name> --image <image> --flavor <flavor> --nic port-id=<port1_id> | 11:21 |
| *** lazy_prince has quit IRC | 11:21 | |
| marios | i am still missing something then | 11:21 |
| *** romcheg1 has quit IRC | 11:22 | |
| lifeless | yes | 11:22 |
| marios | haha, thanks | 11:22 |
| lifeless | the problem lucasagomes is trying to solve is booting unknown machiens | 11:22 |
| lifeless | nova boot boots known machines | 11:22 |
| lucasagomes | yes, so those dhcp extra options needs to be set in a more "global way" not for a specific port | 11:23 |
| lifeless | extra-dhcp-options on ports was added by dkehn to support nova-bm and Ironic, so we didn't have a separate DHCP server (which caused all sorts of issues) | 11:23 |
| lifeless | they have to be set per-port for booting known machines | 11:23 |
| marios | ok | 11:23 |
| lifeless | because they are unique per machine | 11:23 |
| lifeless | changing that stuff would be bad | 11:23 |
| lifeless | we're talking an entirely different thing here | 11:23 |
| lifeless | wildcard DHCP | 11:23 |
| lifeless | which is where the DHCP server answers a request with a just-in-time lease allocation | 11:24 |
| lifeless | all Neutron DHCP servers today are static allocation | 11:24 |
| lifeless | (done by API calls) | 11:24 |
| lifeless | so to PXE boot unknown macs we need 'wildcard DHCP' *and* we need the ability to set extra DHCP options just for the wildcard responses | 11:25 |
| *** athomas_ has quit IRC | 11:25 | |
| lifeless | we can't sanely do it by running a parallel server, because wildcard and Neutron would race, so it has to be a Neutron feature | 11:25 |
| marios | OK. thanks for taking the time to give all this context, i am reading and thinking | 11:26 |
| marios | lifeless: so the 'set extra dhcp options' is tripleo-code. the initial bit is in fact completely missing in neutron today | 11:26 |
| lifeless | AIUI yes | 11:27 |
| marios | lifeless: i.e. some kind of extension for neutron-dhcp-agent | 11:27 |
| *** ryanpetrello has quit IRC | 11:27 | |
| *** athomas has joined #openstack-ironic | 11:27 | |
| lifeless | yes | 11:27 |
| *** k4n0 has quit IRC | 11:28 | |
| dtantsur | so it's still unlikely that neutron supports it? | 11:28 |
| dtantsur | (my research also didn't show anything similar) | 11:29 |
| *** Guest6703 has quit IRC | 11:29 | |
| dtantsur | lifeless ^^^ | 11:29 |
| lucasagomes | lifeless, thanks for the detailed explanantion of what's needed | 11:30 |
| dtantsur | ok, how can we work around it? | 11:30 |
| marios | dtantsur: yes i think that's true. BUT I still think it would be worth sending a mail out with this context to [#neutron]. we will need to write up a spec for 'wildcard dhcp' if this is something we need to pursue in neutron | 11:30 |
| marios | err openstack-dev [neutron] | 11:31 |
| dtantsur | marios, that will be definitely for Kilo, right? | 11:31 |
| marios | dtantsur: 100000% | 11:31 |
| marios | dtantsur: so if that is a factor, we need another way | 11:31 |
| dtantsur | so if we want something in J, we only can try to work around? | 11:31 |
| marios | right | 11:31 |
| marios | lifeless: what about os-net-config | 11:31 |
| marios | oh i am still confuging thigs, this is before they boot right? | 11:32 |
| dtantsur | yes, before | 11:32 |
| *** ryanpetrello has joined #openstack-ironic | 11:32 | |
| lifeless | marios: as dtantsur says | 11:32 |
| lucasagomes | the only workaround I can think of is having a external DHCP server... but idk what that may cause, it may overlap with neutron | 11:33 |
| marios | lucasagomes: and we'd need dhcp relay for that, which we also don't have (in neutron, afaik) | 11:34 |
| dtantsur | lucasagomes, it probably will overlap... | 11:34 |
| lucasagomes | :/ | 11:34 |
| lucasagomes | aight | 11:34 |
| dtantsur | lucasagomes, I'm out of ideas, honestly. This discovery is adding me too much gray hair :/ | 11:35 |
| lucasagomes | dtantsur, heh many missing bits and pieces on both ironic and now neutron | 11:36 |
| lucasagomes | dtantsur, but it's good if we can at least have a plan out of it | 11:36 |
| lucasagomes | find out the missing parts etc | 11:36 |
| lucasagomes | that's all good | 11:36 |
| dtantsur | lucasagomes, I agree, I just can't accept, that it's barely possible for J | 11:37 |
| lucasagomes | marios, hey, thank you very much for the conversation! | 11:37 |
| *** shausy has quit IRC | 11:37 | |
| *** shausy2 has joined #openstack-ironic | 11:37 | |
| dtantsur | yeah, marios, lifeless, thanks a lot! | 11:37 |
| lucasagomes | dtantsur, yeah | 11:37 |
| lucasagomes | yeah, thanks guys | 11:38 |
| lifeless | lucasagomes: that workaround won't work | 11:39 |
| lifeless | lucasagomes: see my comment about races | 11:39 |
| Nisha | dtantsur: lifeless lucasagomes :) could you consider https://review.openstack.org/#/c/100951/ for review for J | 11:39 |
| lifeless | 23:25 < lifeless> we can't sanely do it by running a parallel server, because wildcard and Neutron would race, | 11:39 |
| lifeless | so it has to be a Neutron feature | 11:39 |
| *** nosnos has quit IRC | 11:40 | |
| dtantsur | Nisha, left my comments already | 11:40 |
| Nisha | dtantsur: thanks | 11:40 |
| marios | dtantsur: np, wasn't really very useful ;) | 11:40 |
| marios | lucasagomes: same matey, ^^^ | 11:41 |
| lucasagomes | lifeless, :( yeah | 11:41 |
| lucasagomes | marios, it was useful yes | 11:41 |
| lucasagomes | identified what's missing is always useful :) | 11:42 |
| lucasagomes | identifying* | 11:42 |
| * marios still trying to get his head around the sequence of ops | 11:42 | |
| Nisha | dtantsur: Thanks. Minor comments will address them and post patch | 11:43 |
| marios | lucasagomes: do you have any more specs that have any of this written down (i.e. 'once the h/w info is retrieved by Ironic, we need to boot the host and it should contact our pxedchp' ...) | 11:43 |
| lucasagomes | marios, I'm afraid not | 11:43 |
| Nisha | lifeless: lucasagomes please could you guys also review the spec as it is also on dicovery | 11:43 |
| Nisha | discovery* | 11:43 |
| dtantsur | marios, after retrieving the information, we work as usual: create Ironic node, give it to Nova etc | 11:44 |
| marios | lifeless: if we could define the mac address for a port, that could also help right (but still, ponies and rainbows if we could get it into neutron asap) | 11:45 |
| lifeless | marios: *blink* - how? | 11:46 |
| marios | lifeless: define the neutron port with mac and then dhcp works as usual. oh but its the pxe that is missing. | 11:46 |
| marios | k | 11:46 |
| marios | (pxe dhcp) | 11:46 |
| marios | lifeless: (i.e. port with mac and use that to boot the server) | 11:47 |
| marios | nm | 11:47 |
| lucasagomes | marios, well if you define the neutron port with the mac you can set the extra_dhcp_opts for pxe | 11:48 |
| lucasagomes | but the idea of the discovery is to find that information (like mac address) | 11:48 |
| lucasagomes | that's why we need the wild cards | 11:48 |
| lucasagomes | like tell neutron to boot everything that is "00-FF-*" | 11:48 |
| marios | lucasagomes: but the mac we get from the ramdisk right | 11:49 |
| lucasagomes | and set extra dhcp options for that wildcard configuration ^ | 11:49 |
| lucasagomes | marios, yes, but we need to boot it first | 11:49 |
| lucasagomes | and for that we need a dhcp server | 11:49 |
| lucasagomes | for unknown nics | 11:50 |
| marios | lol. i'll get there eventually lucasagomes :) | 11:50 |
| lucasagomes | :D | 11:50 |
| lucasagomes | marios, https://www.youtube.com/watch?v=gZGYeo7sJhQ << with an external dhcp server | 11:51 |
| *** jcoufal has joined #openstack-ironic | 11:51 | |
| lucasagomes | just to give an idea of how it would looks like | 11:51 |
| lucasagomes | marios, on that example, I'm not using neutron and I've a dnsmasq started with: sudo dnsmasq --conf-file= --port=69 --enable-tftp --tftp-root=/tftpboot --dhcp-boot=pxelinux.0 --bind-interfaces --pid-file=/var/run/dnsmasq.pid --interface=ens9 --dhcp-range=192.0.2.32,192.0.2.48 | 11:52 |
| lifeless | marios: its the mac that is missing | 11:54 |
| lifeless | marios: if you have the mac, you've already discovered the machine | 11:54 |
| lifeless | lucasagomes: that will race with neutron and cause Ironic boots to fail | 11:54 |
| lucasagomes | lifeless, yeah yeah I know, it was just the first stab | 11:54 |
| openstackgerrit | Gopi Krishna S proposed a change to openstack/ironic-specs: Updated the spec file format. https://review.openstack.org/108657 | 11:54 |
| lucasagomes | lifeless, I had an external dhcp server to test the DIB element for the discovery ramdisk | 11:55 |
| lifeless | sure | 11:55 |
| *** vinbs_ has joined #openstack-ironic | 11:58 | |
| *** vinbs has quit IRC | 12:01 | |
| *** vinbs_ is now known as vinbs | 12:01 | |
| saripurigopi | If I submit modified spec for review, does it create new review request ? | 12:03 |
| *** kpavel has quit IRC | 12:04 | |
| *** kpavel has joined #openstack-ironic | 12:05 | |
| lucasagomes | saripurigopi, if ur using the same Change-Id in the commit message, no | 12:05 |
| lucasagomes | it will just create a new patch-set on that review | 12:05 |
| *** chuckC has quit IRC | 12:06 | |
| lucasagomes | (well assuming "new review request" == a new patch) | 12:06 |
| lucasagomes | saripurigopi, I think the best is to take a look here https://wiki.openstack.org/wiki/Gerrit_Workflow | 12:07 |
| saripurigopi | lucasagomes, ok, I'll correct it. | 12:07 |
| *** Poornima_ has quit IRC | 12:07 | |
| *** dtantsur is now known as dtantsur|lunch | 12:07 | |
| *** krtaylor has quit IRC | 12:08 | |
| *** vinbs_ has joined #openstack-ironic | 12:09 | |
| *** vinbs has quit IRC | 12:11 | |
| *** vinbs_ has quit IRC | 12:14 | |
| openstackgerrit | Gopi Krishna S proposed a change to openstack/ironic-specs: Add HTTP GET support for vendor_passthru API https://review.openstack.org/108646 | 12:18 |
| *** lucasagomes is now known as lucas-hungry | 12:19 | |
| *** jdob has joined #openstack-ironic | 12:27 | |
| *** i159 has joined #openstack-ironic | 12:27 | |
| *** romcheg1 has joined #openstack-ironic | 12:27 | |
| i159 | Hi! I have a problem with requirements installation of Ironic: IOError: [Errno 2] No such file or directory: '.tox/py26/build/nova/setup.py'. Branch master. Is it OK? How to solve it? | 12:29 |
| *** killer_prince has joined #openstack-ironic | 12:31 | |
| *** killer_prince is now known as lazy_prince | 12:32 | |
| *** saripurigopi has quit IRC | 12:34 | |
| *** derekh_ has joined #openstack-ironic | 12:49 | |
| devananda | morning, all | 12:55 |
| *** rameshg87 has left #openstack-ironic | 12:56 | |
| *** matty_dubs|gone is now known as matty_dubs | 12:59 | |
| *** pcrews has joined #openstack-ironic | 13:01 | |
| *** jcoufal has quit IRC | 13:03 | |
| *** athomas has quit IRC | 13:03 | |
| *** greghaynes has joined #openstack-ironic | 13:05 | |
| *** dtantsur|lunch is now known as dtantsur | 13:05 | |
| dtantsur | devananda, morning | 13:06 |
| *** rloo has joined #openstack-ironic | 13:07 | |
| *** athomas has joined #openstack-ironic | 13:07 | |
| *** rwsu has joined #openstack-ironic | 13:08 | |
| *** lucas-hungry is now known as lucasagomes | 13:11 | |
| lucasagomes | devananda, morning | 13:11 |
| lucasagomes | i159, requirements installation? are you following http://docs.openstack.org/developer/ironic/dev/dev-quickstart.html? | 13:11 |
| lucasagomes | i159, tox -evenv -- echo 'done' ; to create a new env | 13:12 |
| i159 | lucasagomes: I use tox -re py26 | 13:14 |
| *** krtaylor has joined #openstack-ironic | 13:15 | |
| i159 | lucasagomes: It worked well a few days ago. | 13:15 |
| lucasagomes | i159, right I'm not tox expert... so I mostly follow the guidelines at dev-quickstart and it works for me | 13:15 |
| NobodyCam | good morning Ironic | 13:16 |
| lucasagomes | NobodyCam, morning | 13:16 |
| NobodyCam | moring lucasagomes :) | 13:16 |
| i159 | lucasagomes: me too, have you tried to run tox -re with Ironic lately (today)? | 13:17 |
| lucasagomes | i159, nop, lemme give it a go | 13:17 |
| *** jgrimm has joined #openstack-ironic | 13:17 | |
| dtantsur | NobodyCam, morning :) | 13:17 |
| rloo | i159: I haven't scrolled back, but yesterday I had to redo my tox environment because I think the requirements or something changed. | 13:18 |
| lucasagomes | rloo, I think it was the oslo.db | 13:18 |
| rloo | i159: ie, I deleted .tox and reran tox -evenv -- echo 'done'. | 13:18 |
| lucasagomes | you would have to delete the pyc | 13:18 |
| lucasagomes | left by the openstack/common/db/* | 13:18 |
| NobodyCam | morning dtantsur and rloo too:) | 13:19 |
| rloo | morning NobodyCam and all ironickers. | 13:19 |
| i159 | lucasagomes: we have tried on two Ubuntu machines, and it falls down on both. | 13:19 |
| rloo | lucasagomes: yeah, might have been that. | 13:19 |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic: Add newly_discovered column to Node object https://review.openstack.org/107389 | 13:19 |
| lucasagomes | i159, right I will take a look | 13:19 |
| i159 | lucasagomes: thanks! | 13:20 |
| lucasagomes | i159, oh yeah I see it | 13:22 |
| lucasagomes | i159, it seems to be something with the nova tarball tho | 13:22 |
| lucasagomes | http://paste.openstack.org/show/87606/ < on my machine | 13:22 |
| *** jcoufal has joined #openstack-ironic | 13:23 | |
| jroll | mornin | 13:23 |
| lucasagomes | i159, yeah the nova tarball seems to be corrupted | 13:24 |
| dtantsur | jroll, morning | 13:24 |
| lucasagomes | tar: Skipping to next header | 13:24 |
| lucasagomes | gzip: stdin: invalid compressed data--crc error | 13:24 |
| lucasagomes | gzip: stdin: invalid compressed data--length erro | 13:24 |
| i159 | lucasagomes: yep this is it, I saw that the problem with some nova stuff, but I don't know nothing about Ironic, and it was happened on master branch | 13:24 |
| rloo | lucasagomes, i159: I see it too now :-( | 13:24 |
| * jroll scrolls up | 13:24 | |
| lucasagomes | i159, we download nova because we right now have the nova driver in our tree | 13:24 |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic: Fix leaking DB details to API on error https://review.openstack.org/73121 | 13:24 |
| lucasagomes | i159, so nova is a requirement for us to be able to run the tests on the driver | 13:24 |
| i159 | rloo lucasagomes: good, I mean good, that you see it too =) Can you fix it, or I should ping Nova guys? | 13:26 |
| devananda | what's wrong with the nova tests? | 13:26 |
| lucasagomes | i159, yeah I just brought it to the nova channel to let them know | 13:26 |
| lucasagomes | devananda, nothing... the tarball at http://tarballs.openstack.org/nova/nova-master.tar.gz is corrupted | 13:26 |
| i159 | lucasagomes: yep, good... | 13:27 |
| *** bvivek has quit IRC | 13:27 | |
| devananda | lucasagomes: corrupted? it opens fine for me | 13:27 |
| lucasagomes | devananda, if u try to extarct with "tar xvf" | 13:27 |
| lucasagomes | it fails in the middle | 13:27 |
| lucasagomes | and the setup.py is not there | 13:27 |
| lucasagomes | devananda, probably it failed when it was creating the tar ball, that's why there's some missing files | 13:28 |
| devananda | ahh | 13:28 |
| devananda | you're right | 13:28 |
| rloo | does jenkins use the same nova tarball? | 13:29 |
| lucasagomes | devananda, just brought it to the #openstack-infra should be fixed soon | 13:29 |
| devananda | lucasagomes: great, thanks | 13:29 |
| rloo | thx lucasagomes. Guess I'll do some reviews in the meantime ;) | 13:30 |
| lucasagomes | rloo, :) | 13:30 |
| devananda | rloo: i believe so. which means our tests will be broken atm | 13:30 |
| NobodyCam | good morning devananda | 13:30 |
| NobodyCam | mornign jroll | 13:30 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Import fixes from Nova scheduler reviews https://review.openstack.org/107316 | 13:31 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Import fixes from Nova scheduler reviews https://review.openstack.org/107316 | 13:33 |
| *** rakesh_hs has quit IRC | 13:33 | |
| devananda | lucasagomes: fixed a typo in the base host class just now too | 13:34 |
| lucasagomes | devananda, yeah just revoted after saw ur comment | 13:34 |
| lucasagomes | I mean removed my vote | 13:34 |
| jroll | holy scrollback | 13:34 |
| lucasagomes | devananda, the hard thing with that typo is that it's not exercises in our tree (and shouldn't) | 13:34 |
| * jroll stabs neutron | 13:34 | |
| devananda | actually it's not going to be exercised anywhere | 13:35 |
| lucasagomes | x.x | 13:35 |
| devananda | since nothing will instantiate the base class | 13:35 |
| lucasagomes | yeah baremetal classes | 13:35 |
| devananda | and there are no unit tests being proposed for this | 13:35 |
| devananda | i hate everything about that patch, fwiw, but am tired of arguing with Nova about hwo to land our code | 13:35 |
| devananda | so it gets my +2 | 13:35 |
| lucasagomes | yeah, I figured that... sharing code with nova bm | 13:35 |
| lucasagomes | not good | 13:35 |
| lucasagomes | I mean... I wanted nova bm gone really | 13:36 |
| lucasagomes | and not have to refactor ironic code when nova bm time arrives | 13:36 |
| devananda | lucasagomes: gonna be a year before nova deletes taht code, apparently | 13:37 |
| lucasagomes | devananda, 2 releases?! | 13:37 |
| * lucasagomes thought it was 1 | 13:38 | |
| devananda | i thought so too | 13:38 |
| devananda | just quoting ... | 13:38 |
| lucasagomes | oh well... | 13:38 |
| NobodyCam | :( | 13:39 |
| lucasagomes | devananda, patch lgtm, I will go ahead and approve it, objections? | 13:39 |
| devananda | ++ | 13:39 |
| devananda | going to fix up 107882 too now | 13:45 |
| openstackgerrit | Chris Krelle proposed a change to openstack/ironic: Add option to allow soft power off https://review.openstack.org/107778 | 13:48 |
| jroll | dtantsur: that devstack patch is just a WIP... it's going to be fully configurable etc, don't worry :) | 13:50 |
| jroll | I didn't put it up for reviews, I put it up so that others can 'git review -d' | 13:51 |
| dtantsur | jroll, I could not resist, sorry ;) | 13:51 |
| jroll | lol | 13:51 |
| jroll | get those review points ;) | 13:51 |
| NobodyCam | :-p | 13:51 |
| *** rainya has joined #openstack-ironic | 13:52 | |
| *** bvivek has joined #openstack-ironic | 13:52 | |
| *** igordcard has quit IRC | 13:53 | |
| *** Nisha has quit IRC | 13:54 | |
| NobodyCam | lucasagomes: got a second | 13:54 |
| lucasagomes | NobodyCam, yo sure | 13:55 |
| NobodyCam | :) | 13:55 |
| NobodyCam | wanted to quickly chatt about 107778 | 13:56 |
| NobodyCam | thats my soft power off patch | 13:56 |
| NobodyCam | still very much a wip | 13:56 |
| NobodyCam | but | 13:56 |
| NobodyCam | I think my patch is the incorrect way to do this | 13:57 |
| NobodyCam | I like Ng's way (107864) much better, thou it too may need some clean up | 13:57 |
| lucasagomes | NobodyCam, will take a look at Ng's | 14:01 |
| *** romcheg1 has quit IRC | 14:01 | |
| NobodyCam | :) | 14:01 |
| lucasagomes | NobodyCam, yeah well it was just confusing, the loop logics there | 14:01 |
| lucasagomes | could be simplyfied, that was my point | 14:01 |
| NobodyCam | and only send soft / off one time each? | 14:02 |
| jroll | lucasagomes: you think it's ok for the agent to use a bunch of pxe options? (tftproot etc) or should we re-define those? | 14:03 |
| lucasagomes | jroll, I was talking to lifeless while ago... we want to introduce a new interface for drivers, like boot manager interface | 14:05 |
| lucasagomes | which will be separate of the deploy interface | 14:05 |
| lucasagomes | they are 2 diff things | 14:05 |
| jroll | well | 14:05 |
| jroll | yes | 14:05 |
| lucasagomes | jroll, but I'm afraid that for J | 14:05 |
| lucasagomes | idk if we should go for that much refactor | 14:05 |
| jroll | but we want to land ipa driver in the meantime :) | 14:05 |
| lucasagomes | jroll, sure, ofc | 14:06 |
| jroll | yeah, there's no way we can land that split in juno | 14:06 |
| lucasagomes | jroll, that's why I think that yeah using that options is not ideal, but... I would do it for the first version if u guys don't mind | 14:06 |
| lucasagomes | exactly | 14:06 |
| jroll | right, so | 14:06 |
| jroll | I don't mind because I won't be using pxe | 14:06 |
| jroll | but the drivers should be able to coexist | 14:06 |
| jroll | and was trying to think if that's a problem, if people are using both | 14:06 |
| lucasagomes | jroll, oh that's a very good point | 14:07 |
| jroll | I'm inclined to say its ok... can't think of a huge problem | 14:08 |
| lucasagomes | jroll, right, 1 min... I'm in a hangout | 14:08 |
| jroll | there is tftp_root, tftp_master_path, and tftp_server | 14:08 |
| jroll | yeah no worries | 14:08 |
| *** kpavel has quit IRC | 14:09 | |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Import fixes from the Nova driver reviews https://review.openstack.org/107882 | 14:11 |
| openstackgerrit | Chris Krelle proposed a change to openstack/ironic: Add option to allow soft power off https://review.openstack.org/107778 | 14:13 |
| devananda | dtantsur, rloo: on when to use _L* -- http://docs.openstack.org/developer/oslo.i18n/guidelines.html#examples | 14:19 |
| devananda | tldr; still use _() for any user (API) facing messages | 14:19 |
| devananda | like exceptions | 14:19 |
| rloo | thx devananda. | 14:19 |
| dtantsur | devananda, oh crazy :( | 14:19 |
| dtantsur | thanks | 14:19 |
| NobodyCam | :-p | 14:19 |
| *** russellb has joined #openstack-ironic | 14:20 | |
| rloo | dtantsur: ha ha. I was refraining from commenting... | 14:20 |
| russellb | devananda: o/ just checking in on juno-2 status whenever you have a few minutes | 14:20 |
| *** romcheg1 has joined #openstack-ironic | 14:20 | |
| devananda | russellb: hi! i was just updating it :) | 14:20 |
| devananda | russellb: should be done in a few minutes | 14:21 |
| russellb | sounds good, ping when you're done then | 14:21 |
| devananda | rloo: what's the status of your get-required-info patches? looks like they got rebased and should just be re approved? | 14:22 |
| rloo | devananda: yeah. rebased. although i was wondering if it needed to be rebased again today, but since tox isn't working, i can't really check/test. | 14:23 |
| *** rainya has quit IRC | 14:24 | |
| devananda | dtantsur, lucasagomes: either of you have a few to review https://review.openstack.org/#/c/107096/ ? | 14:26 |
| dtantsur | devananda, we're on the call, I'll review afterwards | 14:26 |
| jroll | I can grab it | 14:27 |
| devananda | cheers, thanks | 14:27 |
| rloo | dtantsur, devananda: i've been staring at the 2nd example and I don't get it: http://docs.openstack.org/developer/oslo.i18n/guidelines.html#examples | 14:27 |
| rloo | dtantsur, devananda: it seems to contradict the first example | 14:28 |
| devananda | rloo: which one? | 14:28 |
| rloo | # RIGHT | 14:29 |
| rloo | LOG.exception(_LE('There was an error.')) | 14:29 |
| rloo | raise LocalExceptionClass(_('An error occured.')) | 14:29 |
| rloo | devananda: the first example was msg = _(), LOG.exception(msg), raise Exception(msg) | 14:29 |
| jroll | rloo: so... if I update a driver and restart my conductor, then I need to restart api servers to get the updated properties? | 14:29 |
| devananda | rloo: difference between that and the first example -- the i18n'd string is different for the LOG vs the raise | 14:29 |
| rloo | jroll: yup. | 14:29 |
| devananda | jroll: yep | 14:30 |
| jroll | rloo: and we're ok with that? :) | 14:30 |
| rloo | jroll: yup the spec was approved. | 14:30 |
| jroll | I'm ok with that... just making sure | 14:30 |
| jroll | right | 14:30 |
| * jroll keep sreading | 14:30 | |
| devananda | jroll: spec outlined several alternatives. this was achievable :) | 14:30 |
| rloo | devananda: right, there's a diff. but for both, the exception msg is seen by the user. so why the diff with the LOG in the 2nd example? | 14:30 |
| jroll | devananda: yep, no worries | 14:31 |
| devananda | rloo: the string seen by the user is not _LE in either case | 14:31 |
| devananda | rloo: in the second example, the LOG is _LE but the exception is not | 14:31 |
| rloo | devananda. so why in the first case, the LOG is not _LE, but in the 2nd case, it is _LE? | 14:32 |
| devananda | rloo: because in the first case, the string is the same. in the second case, the string is different | 14:32 |
| rloo | devananda: ie, when do you use _LE for the log.exception and when don't you use it? | 14:32 |
| devananda | "A common pattern is to define a single message object and use it more than once, for the log call and the exception. In that case, use _() because the message is going to appear in an exception that may be presented to the user." | 14:32 |
| rloo | devananda: $@% got it. I didn't realize that in the 2nd case, the strings were diff. need more caffeine... thx. | 14:33 |
| openstackgerrit | David Shrewsbury proposed a change to openstack/ironic: Implement retry on NodeLocked exceptions https://review.openstack.org/107710 | 14:33 |
| devananda | :) | 14:33 |
| rloo | devananda: so for logging exceptions and logging errors, we use _LE(). | 14:34 |
| rloo | devananda: or maybe for LOG.exception we use whatever _L*() that seems right for that situation. Sigh. | 14:35 |
| devananda | LOG.warning(_LW('words')) | 14:35 |
| dtantsur | IOError: [Errno 2] No such file or directory: '/home/jenkins/workspace/gate-ironic-pep8/.tox/pep8/build/nova/setup.py' | 14:35 |
| dtantsur | now in the gate (just FYI) | 14:35 |
| rloo | dtantsur: yeah. we're waiting for a fix... (I updated the whiteboard status) | 14:36 |
| lucasagomes | devananda, will take a look, was in a call | 14:36 |
| devananda | AIUI, LOG.error and LOG.exception should be _LE'd *unless* the same message is raised in such a way that it may be emitted by the API | 14:36 |
| lucasagomes | dtantsur, the nova tarball is corrupted, infra will fix it | 14:36 |
| rloo | devananda: ok, so _LE for both error & exception. thx. | 14:36 |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic: Updated from global requirements https://review.openstack.org/106569 | 14:38 |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic-python-agent: Updated from global requirements https://review.openstack.org/88722 | 14:38 |
| jroll | rloo: just one question... https://review.openstack.org/#/c/107096/4/ironic/common/hash_ring.py | 14:39 |
| rloo | jroll: the answer is yes :-) | 14:40 |
| jroll | actually, that just clicked | 14:40 |
| jroll | yy | 14:40 |
| jroll | yay* | 14:40 |
| rloo | jroll: ha ha. | 14:40 |
| jroll | approved | 14:40 |
| jroll | devananda: ^ | 14:40 |
| dtantsur | btw, http://docs.openstack.org/developer/oslo.i18n/guidelines.html#avoid-forcing-the-translation-of-translatable-variables <-- this is often encountered in our code | 14:40 |
| rloo | dtantsur: yeah, I saw that too. I'm sure someone will have a patch to fix all that... | 14:41 |
| *** MattMan has quit IRC | 14:44 | |
| *** MattMan has joined #openstack-ironic | 14:55 | |
| *** jgrimm has quit IRC | 14:56 | |
| *** shausy2 has quit IRC | 14:57 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Expose {set,get}_boot_device in the API https://review.openstack.org/90151 | 14:57 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: SSH virsh to use the new ManagementInterface https://review.openstack.org/89884 | 14:57 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: SeaMicro to use the new ManagementInterface https://review.openstack.org/86328 | 14:57 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: IPMINative to use the new ManagementInterface https://review.openstack.org/86588 | 14:57 |
| *** ramineni has joined #openstack-ironic | 15:02 | |
| devananda | jroll: have you applied for a nova spec freeze exception for the configdrive changes? | 15:02 |
| devananda | jroll: fwiw, I'd prefer we punt on that work until after hte ironic driver lands, as that is going to be hard enough as it is | 15:03 |
| jroll | devananda: no... should I? | 15:03 |
| jroll | yeah | 15:03 |
| devananda | jroll: nova spec freeze deadline has passed | 15:03 |
| jroll | I'm still going to write and deploy the code, I don't mind so much when it lands :) | 15:03 |
| jroll | right, I know | 15:03 |
| devananda | jroll: right, k | 15:03 |
| jroll | I'm not rushed | 15:03 |
| *** rainya has joined #openstack-ironic | 15:04 | |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Import fixes from the Nova driver reviews https://review.openstack.org/107882 | 15:06 |
| jroll | 07:36:40 lucasagomes | dtantsur, the nova tarball is corrupted, infra will fix it <- how aware are they? | 15:06 |
| lucasagomes | jroll, I pinged the channel | 15:06 |
| lucasagomes | lemme ping again | 15:06 |
| jroll | heh ok | 15:07 |
| *** mdorman has joined #openstack-ironic | 15:08 | |
| devananda | lucasagomes: management interface patches on track to be landable (jenkins issues notwithstanding) today or tomorrow? | 15:08 |
| lucasagomes | devananda, yup, I've been addressing the comments and rebasing it as quickly as I can | 15:08 |
| rloo | devananda: qq line 125: https://review.openstack.org/#/c/107882/5/ironic/nova/virt/ironic/driver.py | 15:09 |
| devananda | yea, i've seen. good stuff | 15:09 |
| devananda | rloo: ack | 15:09 |
| lucasagomes | devananda, btw, now we have pagination and the cache thing on the client | 15:10 |
| *** mdorman has quit IRC | 15:10 | |
| lucasagomes | devananda, we might want to make a new release of it | 15:10 |
| devananda | lucasagomes: yep! thanks for the ping | 15:11 |
| devananda | rloo: replied | 15:11 |
| devananda | getting food -- bbiab | 15:11 |
| lucasagomes | devananda, there's one patch in the queue adding a vendor passthru to the client, will review it | 15:11 |
| lucasagomes | we may want to included that also | 15:11 |
| *** mdorman has joined #openstack-ironic | 15:11 | |
| rloo | lucasagomes, devananda: are we in a hurry to do a new client pkg? I just remembered, the API to get driver properties has a change there too, but it is WIP until the code in ironic is merged. | 15:12 |
| devananda | client releases can be done as often as we like | 15:13 |
| devananda | it's not like doing one today prevents doing one next week (or tomorrow) -- i've just been waiting until we had actually landed those fixes | 15:14 |
| *** ChuckC has joined #openstack-ironic | 15:14 | |
| *** lazy_prince is now known as killer_prince | 15:15 | |
| devananda | russellb: j2 milestone status page is accurate. we have patches up for all the specs targeted there, should be landable in the next ~48 hrs | 15:17 |
| * devananda really gets food now... bbs | 15:17 | |
| russellb | devananda: thanks for the update! if you could aim for tagging by EOD Wednesday, that'd be best. Thursday should really be reserved for emergencies. | 15:17 |
| russellb | good news is it's just an intermediate milestone, so just a checkpoint, not a huge deal to defer things | 15:18 |
| russellb | just needs to be a in a good workable state, at least | 15:18 |
| dtantsur | lucasagomes, As I understand the last discussion, if you reuse log message as exception message, you should not use _LE, right? | 15:21 |
| lucasagomes | dtantsur, yes, that's how I understand it as well | 15:21 |
| lucasagomes | dtantsur, the _L* are for logs | 15:21 |
| lucasagomes | only | 15:21 |
| dtantsur | lucasagomes, ok, you have one place in seamicro patch | 15:21 |
| dtantsur | lucasagomes, ah no, seems like I'm wrong | 15:22 |
| openstackgerrit | David Shrewsbury proposed a change to openstack/ironic: Implement retry on NodeLocked exceptions https://review.openstack.org/107710 | 15:22 |
| dtantsur | nevermind | 15:22 |
| lucasagomes | dtantsur, ack | 15:22 |
| openstackgerrit | Ghe Rivero proposed a change to openstack/ironic: Raise MissingParameterValue instead of invalid https://review.openstack.org/108455 | 15:25 |
| dtantsur | devananda, https://review.openstack.org/#/c/107882/7/ironic/nova/virt/ironic/client_wrapper.py _LE is imported, but does not seem to be used. Why is that? | 15:26 |
| rloo | dtantsur: ugh, that should be removed. It was there cuz _LE had been used but was removed. | 15:30 |
| dtantsur | enough for now, maybe will be back in the evening | 15:38 |
| *** dtantsur is now known as dtantsur|afk | 15:39 | |
| lucasagomes | urgh now the nova tarball is affecting our gate | 15:40 |
| rloo | lucasagomes: yup. A long list of FAILUREs :-( | 15:40 |
| lucasagomes | yeah I just pinged sean directly | 15:41 |
| lucasagomes | telling on the channel wasn't giving me any feedbacks | 15:41 |
| rloo | thx lucasagomes | 15:42 |
| *** ramineni has quit IRC | 15:42 | |
| *** foexle has quit IRC | 15:53 | |
| *** Nisha has joined #openstack-ironic | 15:53 | |
| *** i159 has quit IRC | 15:54 | |
| openstackgerrit | Ramakrishnan G proposed a change to openstack/ironic-specs: Move code to cleanup ImageCache to a common place https://review.openstack.org/107996 | 15:55 |
| *** eghobo has joined #openstack-ironic | 15:56 | |
| *** eghobo has quit IRC | 16:00 | |
| *** eghobo has joined #openstack-ironic | 16:01 | |
| *** jgrimm has joined #openstack-ironic | 16:06 | |
| *** hemna has joined #openstack-ironic | 16:13 | |
| openstackgerrit | Ghe Rivero proposed a change to openstack/ironic: Raise MissingParamenterValue when validating glance info https://review.openstack.org/108456 | 16:17 |
| openstackgerrit | David Shrewsbury proposed a change to openstack/ironic: Implement retry on NodeLocked exceptions https://review.openstack.org/107710 | 16:24 |
| * devananda is back | 16:24 | |
| devananda | lucasagomes: did you get any traction with infra? | 16:25 |
| lucasagomes | devananda, clarkb seems to be looking at it | 16:25 |
| lucasagomes | <mordred> clarkb: the second thing | 16:26 |
| lucasagomes | <mordred> clarkb: what is the problem? | 16:26 |
| lucasagomes | <clarkb> nova tarball corrupted | 16:26 |
| lucasagomes | <mordred> awesome | 16:26 |
| lucasagomes | * annegent_ (~annegentl@50.57.17.244) has joined #openstack-infra | 16:26 |
| lucasagomes | <clarkb> two tarball jobs raced each other | 16:26 |
| lucasagomes | <clarkb> logs are linked about 12 minutes ago | 16:26 |
| devananda | cool | 16:26 |
| *** bvivek has quit IRC | 16:30 | |
| *** romcheg1 has quit IRC | 16:39 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic: Updated from global requirements https://review.openstack.org/106569 | 16:40 |
| *** jistr has quit IRC | 16:41 | |
| openstackgerrit | Ghe Rivero proposed a change to openstack/ironic: Avoid calling _parse_deploy_info twice https://review.openstack.org/108442 | 16:42 |
| *** killer_prince has quit IRC | 16:43 | |
| *** jgrimm has quit IRC | 16:44 | |
| *** Mikhail_D_ltp has quit IRC | 16:47 | |
| *** jcoufal has quit IRC | 16:49 | |
| *** jcoufal has joined #openstack-ironic | 16:49 | |
| *** jcoufal has quit IRC | 16:50 | |
| *** jcoufal has joined #openstack-ironic | 16:50 | |
| *** pelix has quit IRC | 16:51 | |
| *** killer_prince has joined #openstack-ironic | 16:52 | |
| *** killer_prince is now known as lazy_prince | 16:52 | |
| *** martyntaylor has quit IRC | 16:56 | |
| *** jgrimm has joined #openstack-ironic | 16:57 | |
| *** jogo has left #openstack-ironic | 17:01 | |
| *** harlowja_away is now known as harlowja | 17:19 | |
| *** tatyana has joined #openstack-ironic | 17:28 | |
| openstackgerrit | Nisha Agarwal proposed a change to openstack/ironic-specs: Discover node properties at node-create/node-update https://review.openstack.org/100951 | 17:32 |
| mordred | devananda: sorry I've been mostly flaky recently - are your 2.6 tests passing again? | 17:32 |
| mordred | or whatever the massive fail you had yesterday | 17:33 |
| rloo | mordred: it was failing today too. do you know if a new nova tarball was generated recently? | 17:34 |
| devananda | mordred: i dont know if the 2.6 jobs were fixed, because ALL our jobs broke today. ^ | 17:35 |
| rloo | (oops, forgot about the 2.6 failures.) | 17:36 |
| lucasagomes | aight ppl I will call it a day | 17:36 |
| lucasagomes | have a good night everyone! | 17:36 |
| rloo | night lucasagomes! | 17:36 |
| *** lucasagomes is now known as lucas-dinner | 17:37 | |
| NobodyCam | nigt lucas-dinner | 17:37 |
| jroll | night lucas | 17:38 |
| *** tatyana has quit IRC | 17:39 | |
| openstackgerrit | Nisha Agarwal proposed a change to openstack/ironic-specs: Discover node properties for iLO driver https://review.openstack.org/103007 | 17:47 |
| *** Mikhail_D_ltp has joined #openstack-ironic | 17:52 | |
| *** jcoufal has quit IRC | 17:59 | |
| ryanpetrello | hey ironic folks | 17:59 |
| ryanpetrello | https://review.openstack.org/#/c/105583/1 this backport passed | 17:59 |
| ryanpetrello | merging it should fix ironic stable/icehouse tests :) | 17:59 |
| devananda | ryanpetrello: cheers! adam_g ^ | 18:01 |
| * jroll tried to +2 that | 18:05 | |
| *** chuckC_ has joined #openstack-ironic | 18:05 | |
| devananda | i can't +2 it either, fwiw | 18:05 |
| devananda | only members of stable-maint team can approve it | 18:05 |
| *** chuckC_ has quit IRC | 18:06 | |
| *** romcheg1 has joined #openstack-ironic | 18:11 | |
| *** jcoufal has joined #openstack-ironic | 18:15 | |
| *** rainya_ has joined #openstack-ironic | 18:27 | |
| devananda | messing with graphite .... gate status graph short link: bit.ly/1qyzwG2 | 18:28 |
| devananda | still messing around -- bit.ly/1o6uKwn and bit.ly/1nj3lqc | 18:35 |
| *** tatyana has joined #openstack-ironic | 18:45 | |
| Shrews | Run: 509 in 1475.027208 sec. | 18:48 |
| Shrews | - Passed: 381 | 18:48 |
| Shrews | - Skipped: 66 | 18:48 |
| Shrews | - Failed: 64 | 18:48 |
| Shrews | devananda: adam_g: ^^^ smoke test results | 18:48 |
| *** rainya has quit IRC | 18:49 | |
| *** rainya has joined #openstack-ironic | 18:50 | |
| rloo | fwiw, the nova tarball seems to be fixed. I can run tox in my dev environment. Not sure about jenkins though -- has anyone seen anything pass recently? | 18:50 |
| devananda | rloo: i just rechecked something to see | 18:51 |
| *** shakamunyi has joined #openstack-ironic | 18:56 | |
| *** tatyana has quit IRC | 18:57 | |
| *** chuckC_ has joined #openstack-ironic | 18:59 | |
| Shrews | devananda: adam_g: ignore those numbers... for some reason, i was on the wrong tempest branch and didn't have all of our recent changes included | 19:00 |
| rloo | I have a patch that passed 2 gates so Jenkins is working fine now. (I need to rebase, the story of our lives...) | 19:02 |
| *** jcoufal has quit IRC | 19:04 | |
| *** jcoufal has joined #openstack-ironic | 19:05 | |
| *** chuckC_ has quit IRC | 19:06 | |
| openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Add drivers.base.BaseDriver.get_properties() https://review.openstack.org/107092 | 19:17 |
| *** shakamunyi has quit IRC | 19:42 | |
| *** jbjohnso has joined #openstack-ironic | 19:44 | |
| openstackgerrit | Ghe Rivero proposed a change to openstack/ironic: Fix tear_down a node with missing info https://review.openstack.org/103685 | 19:47 |
| *** tatyana has joined #openstack-ironic | 19:49 | |
| openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Implement API to get driver properties https://review.openstack.org/107096 | 19:57 |
| *** Nisha has quit IRC | 20:06 | |
| *** rloo has quit IRC | 20:10 | |
| *** fausto has joined #openstack-ironic | 20:13 | |
| *** rloo has joined #openstack-ironic | 20:15 | |
| openstackgerrit | A change was merged to openstack/ironic: Import fixes from Nova scheduler reviews https://review.openstack.org/107316 | 20:28 |
| jroll | NobodyCam: I hear there's an openstack NW meetup in portland tonight... has voodoo donuts. | 20:32 |
| jroll | devananda: ^ | 20:33 |
| devananda | jroll: openstack birthday party @ OSCON, yah? | 20:33 |
| devananda | jroll: except we're in Raleigh | 20:33 |
| jroll | I think so | 20:33 |
| jroll | oh shit | 20:33 |
| jroll | right | 20:33 |
| jroll | so confused | 20:34 |
| jroll | so many people in portland lately :P | 20:34 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Use auth_token from keystonemiddleware https://review.openstack.org/107197 | 20:34 |
| * devananda wishes he was in portland right now | 20:34 | |
| jroll | :) | 20:35 |
| Shrews | devananda: there is an OS b-day party at the local meetup this thursday. Would require a vehicle to get there, though | 20:35 |
| jroll | there's one in sf this week, too :P | 20:35 |
| devananda | Shrews: and I'll be on a plane at that time | 20:37 |
| *** tatyana has quit IRC | 20:43 | |
| devananda | what do ya'll think of the client printing a big warning message any time someone calls a vendor_passthru method? | 20:47 |
| devananda | I'm reviewing https://review.openstack.org/#/c/101297/ and realized why I'm hesitating to approve it, even thoug the code is fine | 20:47 |
| devananda | -- it validates *using* the vendor_passthru API in client code and bash scripts | 20:48 |
| jroll | wat | 20:48 |
| devananda | when I intend to change that API and expressly don't want to support any guarantee of compatibility (what a vendor does is up to them) | 20:48 |
| devananda | *intend to move things in that API into the common/standard/core API in the future | 20:49 |
| jroll | I don't see this validation you speak of | 20:49 |
| devananda | not a programatic validation, a perceptual one | 20:49 |
| devananda | having it easily scriptable in the client enables folks to use it without realizign why they should NOT use it | 20:50 |
| jroll | ohhhh, I see | 20:50 |
| jroll | right, I still don't know how I feel about it being in the client :/ | 20:50 |
| devananda | it makes it possible for me to write a bash script that relies on some particular feature in the $foo driver's API | 20:50 |
| devananda | someone's going to put it in some client somewhere | 20:51 |
| jroll | yep, I don't know if I like that :) | 20:51 |
| jroll | I'm going to write *a* client for *my* driver... | 20:51 |
| jroll | idk what vendor passthru methods people would want to use from a bash script, though :/ | 20:51 |
| devananda | jroll: discovery | 20:52 |
| devananda | jroll: downstream (proprietary) drivers | 20:52 |
| jroll | :| | 20:52 |
| devananda | jroll: it's the only case I can think of | 20:52 |
| devananda | and they want support in the common client to use that functionailty | 20:52 |
| jroll | I'm starting to hate the idea of discovery, fwiw | 20:52 |
| devananda | heh | 20:52 |
| jroll | it's like, really really cool | 20:52 |
| jroll | but really complicated and totally unneccessary if you can convince your vendor to give you a spreadsheet full of mac addresses :P | 20:53 |
| jroll | and hardware specs etc | 20:53 |
| devananda | jroll: except for when that spreadsheet is almost always wrong | 20:53 |
| jroll | hmmm this would be a good blog post | 20:53 |
| jroll | validation | 20:53 |
| jroll | or something | 20:53 |
| jroll | homogeneous hardware | 20:53 |
| devananda | using discovery as a way to validate that spreadsheet? | 20:53 |
| devananda | yah | 20:54 |
| jroll | (yes, I know that's not an option) | 20:54 |
| jroll | ehhhhhhh | 20:54 |
| jroll | I'm not questioning that it's useful, I'm questioning how much deployer work it saves :) | 20:55 |
| jroll | (compared to how much dev work it creates) | 20:55 |
| *** jdob has quit IRC | 20:57 | |
| Shrews | devananda: at least a warning seems like a good idea | 20:58 |
| *** mrda-away is now known as mrda | 21:00 | |
| mrda | Morning Ironic! | 21:00 |
| rloo | so if anyone has some time, these need to be re-reviewed and approved: 107092 & 107096 | 21:02 |
| Shrews | morning mrda | 21:02 |
| rloo | they also happen to conflict with lucas-dinner's new ManagementInterface changes. So every time one of his patches makes it in, the above need to be rebased/fixed. | 21:03 |
| mrda | rloo: I'll take a look | 21:03 |
| mrda | Shrews: \o | 21:03 |
| *** eghobo has quit IRC | 21:03 | |
| rloo | thx mrda. They've already been approved once, so I really need +2 powers ;) | 21:03 |
| Shrews | rloo: looking | 21:04 |
| mrda | rloo: I understand, but at least I can do the sanity check | 21:04 |
| devananda | Shrews: lifeless had an interesting point. Simply that the word "vendor" is in the CLI command is probably sufficient to scare people a bit :) | 21:05 |
| * devananda approves it | 21:05 | |
| rloo | yes, I trust that you're sane mrda :-) (Didn't meeeerda anyone recently have you?) | 21:05 |
| devananda | mrda: hi! I fixed up several things in your patches to Ironic | 21:06 |
| mrda | rloo: *groan* | 21:06 |
| mrda | devananda: thanks I'll take a look | 21:06 |
| rloo | Mr. mrda, you're the one that mentioned it to me :-) | 21:06 |
| *** athomas has quit IRC | 21:07 | |
| mrda | devananda: if you have a few cycles, Dan Prince's config options questions might be worth your opinion (that's the outstanding thing from the nova-side of these reviews) | 21:07 |
| devananda | mrda: alraedy done :) | 21:07 |
| mrda | rloo: I need to make you forget | 21:07 |
| mrda | devananda: ta | 21:07 |
| devananda | mrda: see latest version of https://review.openstack.org/#/c/107882/ | 21:07 |
| devananda | i'm looking into the jenkins failure | 21:08 |
| devananda | also i dont know that I agree with dtantsur|afk's objections | 21:08 |
| devananda | oh. error is trivial, duh. http://logs.openstack.org/82/107882/7/check/check-tempest-dsvm-virtual-ironic/5edd30f/logs/screen-n-cpu.txt.gz?level=INFO | 21:08 |
| devananda | err | 21:09 |
| devananda | hah | 21:09 |
| zer0c00l | Hi rloo | 21:09 |
| rloo | hi zer0c00l | 21:10 |
| mrda | devananda: oh | 21:10 |
| zer0c00l | I am Arun from Yahoo's Linux team. | 21:11 |
| devananda | mrda: your import wrapper is failing, fwiw | 21:11 |
| mrda | yup, looking now | 21:11 |
| rloo | zer0c00l: are you going to help out on ironic? | 21:11 |
| devananda | mrda: it's assuming that the 'ironicclient' module has an 'client' and 'exception' member objects,w hih it does not | 21:11 |
| zer0c00l | Why not? | 21:12 |
| zer0c00l | I know little bit of python | 21:12 |
| zer0c00l | :) | 21:12 |
| rloo | zer0c00l: that would be GREAT | 21:12 |
| zer0c00l | I had questions about ironic-python-agent before | 21:12 |
| zer0c00l | It has lot of features that overlap with an OS installer | 21:13 |
| rloo | zer0c00l: I'm not the person to ask about IPA. jroll et al are. What kind of questions. | 21:13 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Import fixes from the Nova driver reviews https://review.openstack.org/107882 | 21:13 |
| devananda | mrda: lets see if that passes | 21:13 |
| zer0c00l | Anaconda has code for partitioning the disk and stuff , it would be nice to look at that code. May be reuse some of it | 21:14 |
| zer0c00l | Anaconda is from RedHat | 21:14 |
| harlowja | whats redhat, haha | 21:14 |
| harlowja | :-P | 21:14 |
| zer0c00l | and it is mostly python | 21:14 |
| devananda | zer0c00l: you should talk with jroll | 21:14 |
| zer0c00l | yeah | 21:14 |
| mrda | devananda: thanks for pushing a new version! | 21:14 |
| *** foexle has joined #openstack-ironic | 21:14 | |
| zer0c00l | sure! | 21:15 |
| zer0c00l | jroll: hi | 21:15 |
| devananda | JoshNang: status of the swiftclient changes? any news? | 21:17 |
| zer0c00l | I think lucas-dinner may know little bit more about anaconda. | 21:17 |
| zer0c00l | https://git.fedorahosted.org/cgit/anaconda.git/ | 21:17 |
| JoshNang | devananda: merged and packaged! | 21:17 |
| devananda | JoshNang: woot! thought I heard that. So. we should proceed with the changes here, ya? | 21:17 |
| JoshNang | python-swiftclient 2.2.0 has it. i have a patch to global reqs to bump it up and i'm working on the ironic patch right now | 21:18 |
| devananda | JoshNang: awesome. so not this week. possibly next. definitely before J3 | 21:18 |
| rloo | zer0c00l: might try dtantsur|afk wrt anaconda, he's a RH person too. | 21:19 |
| Shrews | rloo: reviewed 107092. not sure i'll get to the other one today | 21:19 |
| rloo | thx Shrews. The first is the important one - that's the one that clashes with lucas' new ManagementInterface stuff. | 21:19 |
| JoshNang | definitely before J3. I mean tests will pass before the global reqs get bumped, and -infra seemed to imply we could go ahead and merge before the bump, but that doesn't seem right to me. | 21:20 |
| Shrews | rloo: i'll +2 it if you agree to the recommended fix and push a new one soon | 21:20 |
| Shrews | brb | 21:20 |
| rloo | thx Shrews. good catch! | 21:21 |
| openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Add drivers.base.BaseDriver.get_properties() https://review.openstack.org/107092 | 21:23 |
| fausto | Hi all @here : ) | 21:24 |
| jroll | zer0c00l: hello! | 21:26 |
| * jroll reads up a little | 21:26 | |
| fausto | I'd like to give some help with Ironic, if possible | 21:26 |
| devananda | fausto: hi and welcome :) | 21:27 |
| fausto | devananda, thanks, I think we are collegues | 21:28 |
| fausto | :) | 21:28 |
| *** matty_dubs is now known as matty_dubs|gone | 21:29 | |
| fausto | opsauto | 21:29 |
| jroll | zer0c00l: does anaconda have a python API? it's unclear to me if we do anything advanced enough to want to depend on anaconda | 21:30 |
| jroll | devananda, JoshNang, josh's patch is a blocker for landing the ipa driver. so... I'd really like to land it this week :| | 21:30 |
| mrda | devananda: you meantioned that you didn't agree with dtantsur's comments on patch 7 of 107882. i.e interpolation and exceptions? | 21:30 |
| devananda | jroll: ack. I fully support that if it can be done | 21:31 |
| jroll | devananda: I think it can be done :) | 21:31 |
| devananda | jroll: ie, it depends on landing a change in global-requirements, which we don't control | 21:31 |
| * jroll +A's :P | 21:31 | |
| jroll | did that not land? | 21:31 |
| jroll | boo | 21:32 |
| devananda | 21:18:06 < JoshNang> python-swiftclient 2.2.0 has it. i have a patch to global reqs to bump it up and i'm working on the ironic patch right now | 21:32 |
| jroll | we just need to bug infra... | 21:32 |
| jroll | but I think we should review josh's patch on the next rev | 21:32 |
| jroll | +2 it all over the place | 21:32 |
| jroll | and land immediately after global req | 21:32 |
| *** shakamunyi has joined #openstack-ironic | 21:33 | |
| devananda | jroll: ++ | 21:33 |
| openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Implement API to get driver properties https://review.openstack.org/107096 | 21:34 |
| zer0c00l | jroll: It does not exactly have an API | 21:36 |
| jroll | zer0c00l: hmmmm, I don't love shelling out :| | 21:37 |
| zer0c00l | Dracut parses kickstart -> Starts anaconda -> Partitions -> installation -> Bootloader reboot | 21:37 |
| jroll | we do this already... but to single-purpose programs | 21:37 |
| jroll | hmmm | 21:37 |
| *** foexle has quit IRC | 21:39 | |
| zer0c00l | I think ironic-python-agent will be an installer with an API :) | 21:40 |
| zer0c00l | But the code/ideas can be copied easily from the existing installer (guessing) | 21:40 |
| zer0c00l | Like figuring out whether it is a bios boot/ uefi boot | 21:41 |
| zer0c00l | handling grub2 vs grub1 | 21:41 |
| devananda | zer0c00l: ironic should not (and does not) run installers | 21:41 |
| devananda | zer0c00l: it writes a machine image to a disk | 21:41 |
| JayF | devananda: jroll: I tried to get some urgency behind that global-reqs merge req, but ClarkB kept saying we should just merge the ironic patch before the global req patch hits because pip will install the newest version by default | 21:41 |
| zer0c00l | i understand, but as per correct spec there is a plan to support partitions ? | 21:42 |
| JayF | devananda: jroll: I do not /agree/ with this assessment however, but wanted to be sure it was stated :) | 21:42 |
| devananda | zer0c00l: no. Ive blocked and pushed back on that discussion. arbitrary partitioning, support for LVM, etc, should not, IMNSHO, be in Ironic | 21:42 |
| devananda | zer0c00l: what spec are you referring to? | 21:42 |
| zer0c00l | i see | 21:42 |
| devananda | zer0c00l: and has it been approved? | 21:43 |
| zer0c00l | let me check | 21:43 |
| devananda | JayF: gotcha. clarkb seldom gives bad advice when it comes to test infra ... | 21:43 |
| devananda | JayF: so perhaps we should proceed | 21:43 |
| jroll | devananda, zer0c00l: we'll need to support partitions to some extent, but not arbitrary partitions | 21:43 |
| zer0c00l | i think that is a slippery slope :) | 21:44 |
| devananda | zer0c00l: also, what does partitioning have to do with the "be an installer with an API"? | 21:44 |
| JayF | devananda: I would, and if it breaks, we'll tell him so he won't issue that advice in the future :) | 21:45 |
| zer0c00l | how do you define parition type/lvm | 21:45 |
| jroll | right... I don't think anaconda is very relevant here, unfortunately | 21:45 |
| devananda | there's a functional difference between traditional installers and taking a pre-built cloud image and booting it | 21:45 |
| devananda | that difference is tremendous | 21:45 |
| zer0c00l | yes | 21:45 |
| jroll | zer0c00l: nova/ironic should define the partition scheme, based on the user/flavor options | 21:45 |
| devananda | ironic is not an installer with an API | 21:45 |
| *** romcheg1 has quit IRC | 21:45 | |
| devananda | though I would like to know what gave you taht impression. Possibly our messaging about what ironic *is* could use some clarification :) | 21:46 |
| openstackgerrit | A change was merged to openstack/python-ironicclient: Add ironic cli support for vendor-passthru https://review.openstack.org/101297 | 21:47 |
| *** ChuckC has quit IRC | 21:48 | |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Push the image cache ttl way up https://review.openstack.org/106906 | 21:48 |
| jroll | devananda: IPA still has some cruft left over in launchpad from agordeev and friends :| | 21:48 |
| zer0c00l | devananda: https://wiki.openstack.org/wiki/Ironic-python-agent | 21:49 |
| * jroll is curious what happened to those folks | 21:49 | |
| *** Badcub01 has joined #openstack-ironic | 21:49 | |
| zer0c00l | The motivation, seems like a installer project to me. | 21:49 |
| zer0c00l | Sorry i might have misunderstood | 21:49 |
| zer0c00l | this whole thing | 21:49 |
| *** john___ has joined #openstack-ironic | 21:49 | |
| jroll | well... IPA does a lot of "write an image to disk", yes | 21:50 |
| devananda | zer0c00l: one motivation: treat hardware like cattle. use a cloud API to provision physical machines, just like any other instance in a cloud. | 21:50 |
| devananda | jroll: feel free to clean up cruft where you find it. should go without saying :) | 21:51 |
| zer0c00l | ok | 21:51 |
| jroll | devananda: ofc, I just haven't gone looking for it | 21:51 |
| *** chuckC has joined #openstack-ironic | 21:55 | |
| *** rainya has quit IRC | 21:55 | |
| *** john___ has quit IRC | 21:55 | |
| *** Badcub01 has quit IRC | 21:55 | |
| *** shakamunyi has quit IRC | 21:56 | |
| Shrews | rloo: i lied. i did get to the other review :) | 21:59 |
| *** chuckC has quit IRC | 21:59 | |
| *** jcoufal has quit IRC | 21:59 | |
| rloo | thx Shrews. I'll take a look. I suspect that they will both have to be rebased anyway. | 22:00 |
| *** derekh_ has quit IRC | 22:00 | |
| *** rwsu has quit IRC | 22:02 | |
| openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Implement API to get driver properties https://review.openstack.org/107096 | 22:04 |
| *** BadCub02 has joined #openstack-ironic | 22:06 | |
| devananda | getting food, bbl | 22:09 |
| *** eghobo has joined #openstack-ironic | 22:11 | |
| openstackgerrit | Ghe Rivero proposed a change to openstack/ironic: Raise MissingParameterValue instead of invalid https://review.openstack.org/108455 | 22:12 |
| *** jgrimm has quit IRC | 22:17 | |
| *** Mikhail_D_ltp has quit IRC | 22:23 | |
| openstackgerrit | A change was merged to openstack/ironic: Rename/update common/tftp.py to common/pxe_utils.py https://review.openstack.org/103595 | 22:23 |
| openstackgerrit | A change was merged to openstack/ironic: Imported Translations from Transifex https://review.openstack.org/106948 | 22:23 |
| *** mdorman has quit IRC | 22:24 | |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic: Adding swift temp url support https://review.openstack.org/81391 | 22:31 |
| *** eghobo has quit IRC | 22:38 | |
| openstackgerrit | Michael Davies proposed a change to openstack/ironic: Import fixes from the Nova driver reviews https://review.openstack.org/107882 | 22:38 |
| *** eghobo has joined #openstack-ironic | 22:50 | |
| openstackgerrit | Ghe Rivero proposed a change to openstack/ironic: Raise MissingParamenterValue when validating glance info https://review.openstack.org/108456 | 22:55 |
| openstackgerrit | Ghe Rivero proposed a change to openstack/ironic: Avoid calling _parse_deploy_info twice https://review.openstack.org/108442 | 22:56 |
| *** lucas-dinner has quit IRC | 22:56 | |
| *** fausto has quit IRC | 22:57 | |
| *** eghobo has quit IRC | 23:02 | |
| *** eghobo has joined #openstack-ironic | 23:02 | |
| openstackgerrit | A change was merged to openstack/ironic: SeaMicro to use the new ManagementInterface https://review.openstack.org/86328 | 23:03 |
| *** dkehnx has joined #openstack-ironic | 23:06 | |
| *** jcoufal has joined #openstack-ironic | 23:15 | |
| *** rainya has joined #openstack-ironic | 23:18 | |
| mrda | Thanks to all who reviewed 107316 - I've just proposed that code back into Nova now as an update to 103165 \o/ | 23:46 |
| * mrda goes back to the driver now that the scheduler changes are in Nova's court | 23:47 | |
| *** radsy has joined #openstack-ironic | 23:54 | |
| *** radsy has joined #openstack-ironic | 23:54 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!