| openstackgerrit | dekehn proposed a change to openstack/ironic: Adds Neutron support to Ironic https://review.openstack.org/66071 | 00:19 |
|---|---|---|
| *** rongze has joined #openstack-ironic | 00:35 | |
| *** zul has joined #openstack-ironic | 00:37 | |
| *** rongze has quit IRC | 00:40 | |
| *** blamar has joined #openstack-ironic | 01:10 | |
| * NobodyCam wanders afk | 01:13 | |
| *** jbjohnso has quit IRC | 01:19 | |
| *** harlowja has joined #openstack-ironic | 01:33 | |
| *** nosnos has joined #openstack-ironic | 01:35 | |
| *** rongze has joined #openstack-ironic | 01:46 | |
| *** sameer has quit IRC | 01:46 | |
| *** datajerk has joined #openstack-ironic | 01:48 | |
| *** datajerk has quit IRC | 01:58 | |
| *** rloo has quit IRC | 02:01 | |
| *** sameer has joined #openstack-ironic | 02:02 | |
| *** rloo has joined #openstack-ironic | 02:02 | |
| *** datajerk has joined #openstack-ironic | 02:02 | |
| *** rloo has quit IRC | 02:06 | |
| *** sameer has quit IRC | 02:06 | |
| *** rloo has joined #openstack-ironic | 02:06 | |
| *** rloo has quit IRC | 02:11 | |
| *** rloo has joined #openstack-ironic | 02:11 | |
| *** michchap has quit IRC | 02:24 | |
| *** michchap has joined #openstack-ironic | 02:25 | |
| *** rloo has quit IRC | 02:27 | |
| *** rloo has joined #openstack-ironic | 02:27 | |
| *** rloo has quit IRC | 02:29 | |
| *** vkozhukalov has joined #openstack-ironic | 02:30 | |
| *** rloo has joined #openstack-ironic | 02:30 | |
| *** rloo has quit IRC | 02:31 | |
| *** rloo has joined #openstack-ironic | 02:31 | |
| *** coolsvap_away has quit IRC | 02:33 | |
| *** rloo has quit IRC | 02:34 | |
| *** rloo has joined #openstack-ironic | 02:34 | |
| *** rloo has quit IRC | 02:41 | |
| *** rloo has joined #openstack-ironic | 02:42 | |
| *** rloo has quit IRC | 02:42 | |
| *** rloo has joined #openstack-ironic | 02:43 | |
| *** rloo has quit IRC | 02:46 | |
| *** rloo has joined #openstack-ironic | 02:46 | |
| *** epim has quit IRC | 02:56 | |
| openstackgerrit | A change was merged to openstack/ironic: Use oslo.rootwrap library instead of local copy https://review.openstack.org/66389 | 03:14 |
| openstackgerrit | A change was merged to openstack/ironic: Add missing "Filters" section to the ironic-images.filters https://review.openstack.org/66390 | 03:15 |
| openstackgerrit | A change was merged to openstack/ironic: Remove the absolute paths from ironic-deploy-helper.filters https://review.openstack.org/66400 | 03:16 |
| *** rloo has quit IRC | 03:22 | |
| *** harlowja is now known as harlowja_away | 03:54 | |
| *** datajerk has quit IRC | 04:03 | |
| *** coolsvap has joined #openstack-ironic | 04:08 | |
| *** rongze has quit IRC | 04:52 | |
| *** sameer has joined #openstack-ironic | 04:54 | |
| *** sameer has quit IRC | 05:19 | |
| *** sameer has joined #openstack-ironic | 05:19 | |
| *** coolsvap is now known as coolsvap_away | 05:23 | |
| *** rongze has joined #openstack-ironic | 05:23 | |
| *** rongze has quit IRC | 05:28 | |
| *** rongze has joined #openstack-ironic | 05:35 | |
| *** rongze has quit IRC | 05:40 | |
| *** sameer has quit IRC | 06:00 | |
| *** blamar has quit IRC | 06:08 | |
| *** aignatov_ is now known as aignatov | 06:17 | |
| *** rongze has joined #openstack-ironic | 06:17 | |
| *** coolsvap_away is now known as coolsvap | 06:19 | |
| *** blamar has joined #openstack-ironic | 06:23 | |
| *** mrda has quit IRC | 06:35 | |
| *** nosnos_ has joined #openstack-ironic | 06:43 | |
| *** nosnos has quit IRC | 06:43 | |
| *** mrda__ is now known as mrda_away | 06:51 | |
| *** vkozhukalov has quit IRC | 06:52 | |
| *** aignatov is now known as aignatov_ | 07:20 | |
| *** mdurnosvistov has joined #openstack-ironic | 07:29 | |
| *** jistr has joined #openstack-ironic | 07:43 | |
| GheRivero | morning Ironic! | 07:48 |
| *** aignatov_ is now known as aignatov | 07:54 | |
| Haomeng | GheRivero: morning:) | 07:55 |
| openstackgerrit | Jenkins proposed a change to openstack/ironic: Imported Translations from Transifex https://review.openstack.org/68024 | 07:57 |
| *** Haomeng has quit IRC | 08:03 | |
| *** Haomeng has joined #openstack-ironic | 08:04 | |
| *** aignatov is now known as aignatov_ | 08:06 | |
| *** romcheg has joined #openstack-ironic | 08:21 | |
| *** romcheg1 has joined #openstack-ironic | 08:23 | |
| *** yuriyz has joined #openstack-ironic | 08:24 | |
| *** romcheg has quit IRC | 08:26 | |
| *** vkozhukalov has joined #openstack-ironic | 08:30 | |
| *** Alexei_987 has joined #openstack-ironic | 08:44 | |
| openstackgerrit | Ghe Rivero proposed a change to openstack/ironic: Remane and update ironic-deploy-helper rootwrap https://review.openstack.org/68340 | 09:01 |
| *** derekh has joined #openstack-ironic | 09:03 | |
| *** mdurnosvistov has quit IRC | 09:08 | |
| *** aignatov_ is now known as aignatov | 09:10 | |
| *** vetalll has joined #openstack-ironic | 09:12 | |
| *** derekh has quit IRC | 09:18 | |
| *** aignatov is now known as aignatov_ | 09:34 | |
| *** aignatov_ is now known as aignatov | 09:37 | |
| *** jistr has quit IRC | 09:37 | |
| *** jbjohnso has joined #openstack-ironic | 09:45 | |
| *** ndipanov has joined #openstack-ironic | 09:45 | |
| *** aignatov is now known as aignatov_ | 09:49 | |
| *** aignatov_ is now known as aignatov | 09:53 | |
| *** jistr has joined #openstack-ironic | 09:57 | |
| *** athomas has joined #openstack-ironic | 10:00 | |
| *** martyntaylor has joined #openstack-ironic | 10:08 | |
| *** martyntaylor has quit IRC | 10:12 | |
| *** mdurnosvistov has joined #openstack-ironic | 10:15 | |
| *** martyntaylor has joined #openstack-ironic | 10:27 | |
| *** derekh has joined #openstack-ironic | 10:35 | |
| *** max_lobur_afk is now known as max_lobur | 10:39 | |
| *** jbjohnso has quit IRC | 10:45 | |
| *** romcheg has joined #openstack-ironic | 10:45 | |
| GheRivero | what was the consensus about getting rid of the object dict behaviour? | 10:46 |
| *** romcheg1 has quit IRC | 10:46 | |
| max_lobur | Hi GheRivero | 10:54 |
| GheRivero | hi max_lobur | 10:54 |
| max_lobur | we decided to drop all the usages but not to drop the actual interface | 10:55 |
| max_lobur | because oslo object model still has it | 10:55 |
| max_lobur | and once we merge with it we will get it again | 10:55 |
| GheRivero | ok. thanks for the info | 10:56 |
| max_lobur | we probably will want to create an intermediate class between oslo object and our to make sure we're not using this interface (e.g. throw Exception in these methods) until it dropped from oslo | 10:56 |
| *** lucasagomes has joined #openstack-ironic | 10:57 | |
| max_lobur | morning lucasagomes :) | 10:58 |
| lucasagomes | morning max_lobur :) | 10:58 |
| max_lobur | take a look https://bugs.launchpad.net/ironic/+bug/1271317 | 10:59 |
| max_lobur | <devananda> tl;dr -- let's just drop XML support and be done with it | 10:59 |
| lucasagomes | max_lobur, btw will test the other problem about the traceback now in the morning, yesterday I got busy with other stuff | 11:04 |
| * lucasagomes clicks on the link | 11:05 | |
| max_lobur | lucasagomes, you can test, but deva confirmed that he cannot reproduce too | 11:05 |
| max_lobur | https://bugs.launchpad.net/ironic/+bug/1244747 | 11:05 |
| lucasagomes | max_lobur, ah nice :D so it might have been the debug flag | 11:05 |
| max_lobur | if you have time please test to be sure | 11:05 |
| max_lobur | but I think it's OK now | 11:06 |
| *** michchap has quit IRC | 11:06 | |
| lucasagomes | ah good, so it's already closed | 11:06 |
| max_lobur | yep | 11:06 |
| lucasagomes | :D | 11:06 |
| *** michchap has joined #openstack-ironic | 11:06 | |
| lucasagomes | max_lobur, great email about xml | 11:08 |
| lucasagomes | I agree with that | 11:08 |
| lucasagomes | (and with the bug as well, will cmment to express that) | 11:08 |
| max_lobur | yep | 11:08 |
| max_lobur | I'm totally agree too | 11:08 |
| max_lobur | xml is always a head pain | 11:08 |
| lucasagomes | heh +1 | 11:11 |
| lucasagomes | yea and supporting both is costly | 11:11 |
| lucasagomes | that email does give a nice explanation/summary about the problems involved | 11:12 |
| Haomeng | max_lobur: morning:) | 11:12 |
| Haomeng | lucasagomes: morning:) | 11:12 |
| max_lobur | morning Haomeng ! (: | 11:12 |
| lucasagomes | morning Haomeng | 11:13 |
| Haomeng | lucasagomes: :) | 11:14 |
| Haomeng | max_lobur: :) | 11:14 |
| max_lobur | need to save that email somewhere to show to my future customers :) | 11:14 |
| lucasagomes | hah | 11:14 |
| Haomeng | do you know how our api content_types are controled, I debug with pecan, still have no idea:) | 11:15 |
| Haomeng | current the supported type(s): ['application/json', 'application/xml'] | 11:16 |
| max_lobur | Haomeng, I found this in our api v1.media_types = [MediaType('application/json', | 11:16 |
| max_lobur | <max_lobur> 'application/vnd.openstack.ironic.v1+json')] | 11:16 |
| max_lobur | I think thats it | 11:16 |
| max_lobur | 'application/xml' currently is not supported as content type | 11:16 |
| max_lobur | only as accept | 11:16 |
| *** lucasagomes is now known as lucas-afk | 11:17 | |
| Haomeng | max_lobur: I debuged, current we support ['application/json', 'application/xml'] | 11:18 |
| Haomeng | max_lobur: ok, let me check api.v1.media_types | 11:18 |
| max_lobur | hm | 11:19 |
| *** lucas-afk is now known as lucasagomes | 11:19 | |
| max_lobur | I tried to send xml requests and got server erros | 11:19 |
| Haomeng | Haomeng: I try to input invalid content_types http request such as 'application/abc' | 11:19 |
| Haomeng | max_lobur: we get the error - 2014-01-22 02:15:21.336 6377 ERROR pecan.core [-] Controller 'get_all' defined does not support content_type 'None'. Supported type(s): ['application/json', 'application/xml'] | 11:20 |
| Haomeng | max_lobur: can not find where we define 'application/xml' | 11:20 |
| max_lobur | hm | 11:23 |
| max_lobur | that's probably just a wrong message | 11:24 |
| max_lobur | try to figure out where it's starts | 11:24 |
| max_lobur | because if you try to send xml you should get the server error too | 11:24 |
| Haomeng | max_lobur: found some default content_types defined by wsexpose | 11:24 |
| Haomeng | max_lobur: yes | 11:24 |
| Haomeng | max_lobur: so I try to disable xml support | 11:25 |
| Haomeng | max_lobur: looks like it is defined by - http://nullege.com/codes/show/src@w@s@WSME-0.5b1@wsmeext@pecan.py/44/pecan.expose | 11:25 |
| Haomeng | max_lobur: but not sure how to change the settings to disable xml support, other project such as Glance, they support json only | 11:26 |
| Haomeng | Haomeng: however Glance use ownself wsgi py | 11:26 |
| *** aignatov is now known as aignatov_ | 11:26 | |
| *** rongze has quit IRC | 11:26 | |
| *** rongze has joined #openstack-ironic | 11:27 | |
| *** derekh has quit IRC | 11:29 | |
| *** rongze has quit IRC | 11:31 | |
| *** aignatov_ is now known as aignatov | 11:50 | |
| *** vetalll1 has joined #openstack-ironic | 12:05 | |
| *** vetalll has quit IRC | 12:07 | |
| *** rongze has joined #openstack-ironic | 12:09 | |
| *** rongze has quit IRC | 12:14 | |
| openstackgerrit | A change was merged to openstack/ironic: Add [keystone_authtoken] to ironic.conf.sample https://review.openstack.org/68256 | 12:14 |
| *** ifarkas has quit IRC | 12:16 | |
| *** ifarkas has joined #openstack-ironic | 12:17 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Support preserve_ephemeral https://review.openstack.org/68236 | 12:24 |
| *** aignatov is now known as aignatov_ | 12:26 | |
| *** rongze has joined #openstack-ironic | 12:39 | |
| *** jbjohnso has joined #openstack-ironic | 13:02 | |
| *** coolsvap has quit IRC | 13:03 | |
| *** aignatov_ is now known as aignatov | 13:11 | |
| *** yfujioka has joined #openstack-ironic | 13:35 | |
| *** derekh has joined #openstack-ironic | 13:35 | |
| *** rongze has quit IRC | 13:37 | |
| *** jdob has joined #openstack-ironic | 13:39 | |
| *** rongze has joined #openstack-ironic | 13:47 | |
| *** rloo has joined #openstack-ironic | 13:47 | |
| *** yfujioka has quit IRC | 13:52 | |
| *** yfujioka has joined #openstack-ironic | 13:53 | |
| *** aignatov is now known as aignatov_ | 13:57 | |
| *** rloo has quit IRC | 13:58 | |
| *** rloo has joined #openstack-ironic | 13:59 | |
| *** datajerk has joined #openstack-ironic | 14:07 | |
| openstackgerrit | A change was merged to stackforge/pyghmi: Add support for discrete sensors https://review.openstack.org/68279 | 14:08 |
| *** aignatov_ is now known as aignatov | 14:09 | |
| *** rloo has quit IRC | 14:09 | |
| *** yfujioka has quit IRC | 14:09 | |
| *** rloo has joined #openstack-ironic | 14:09 | |
| *** matsuhashi has joined #openstack-ironic | 14:14 | |
| *** rongze_ has joined #openstack-ironic | 14:14 | |
| *** rongze has quit IRC | 14:16 | |
| *** datajerk1 has joined #openstack-ironic | 14:17 | |
| *** rongze has joined #openstack-ironic | 14:17 | |
| *** rongze_ has quit IRC | 14:19 | |
| *** datajerk has quit IRC | 14:19 | |
| *** coolsvap has joined #openstack-ironic | 14:22 | |
| *** rongze_ has joined #openstack-ironic | 14:22 | |
| *** rongze has quit IRC | 14:25 | |
| *** rloo has quit IRC | 14:27 | |
| *** rloo has joined #openstack-ironic | 14:27 | |
| *** datajerk1 has quit IRC | 14:33 | |
| openstackgerrit | Yuriy Zveryanskyy proposed a change to openstack/ironic: Add parameter for filtering nodes by maintenance mode https://review.openstack.org/63937 | 14:35 |
| *** datajerk has joined #openstack-ironic | 14:35 | |
| *** rloo has quit IRC | 14:36 | |
| *** rloo has joined #openstack-ironic | 14:36 | |
| *** rloo has quit IRC | 14:38 | |
| *** rloo has joined #openstack-ironic | 14:38 | |
| *** rloo has quit IRC | 14:39 | |
| *** rloo has joined #openstack-ironic | 14:40 | |
| *** matty_dubs|gone is now known as matty_dubs | 14:41 | |
| *** rloo has quit IRC | 14:42 | |
| *** rloo has joined #openstack-ironic | 14:43 | |
| *** rloo has quit IRC | 14:43 | |
| *** rloo has joined #openstack-ironic | 14:43 | |
| *** matsuhashi has quit IRC | 14:45 | |
| *** rongze_ has quit IRC | 14:49 | |
| *** linggao has joined #openstack-ironic | 14:53 | |
| *** nosnos_ has quit IRC | 14:57 | |
| *** rloo has quit IRC | 14:59 | |
| *** rloo has joined #openstack-ironic | 15:00 | |
| *** rongze has joined #openstack-ironic | 15:05 | |
| NobodyCam | morning Ironic, says the man slowly waking up :) | 15:11 |
| rloo | morning NobodyCam. WAKE UP! :) | 15:12 |
| NobodyCam | lol morning rloo | 15:13 |
| NobodyCam | :) | 15:14 |
| rloo | btw NobodyCam, if you get a chance, maybe you can take a look at https://review.openstack.org/#/c/61859/. it is in response to a bug you filed. | 15:17 |
| NobodyCam | oh :) | 15:17 |
| yuriyz | morning rloo nobodycam and Ironic | 15:18 |
| rloo | NobodyCam, just that the 'soln' is wrong, and it might be useful to the assignee to get your opinion. not sure. | 15:18 |
| NobodyCam | morning yuriyz | 15:18 |
| rloo | afternoon yuriyz! | 15:18 |
| *** rongze has quit IRC | 15:22 | |
| max_lobur | morning Ironic :) | 15:23 |
| NobodyCam | morning max_lobur ;) | 15:23 |
| NobodyCam | rloo: interesting point about the multiple nodes, do you have a spicific example of where this could be a issue? | 15:27 |
| rloo | NobodyCam. No, just reading the code though, I can see it won't work. I think Ghe said he got an exception. | 15:28 |
| rloo | NobodyCam. I looked at that awhile ago. Just thought it'd be good to give the person feedback sooner rather than later. | 15:29 |
| linggao | Morning NobodyCam. | 15:29 |
| NobodyCam | morning linggao | 15:30 |
| linggao | Can you let me know where is the Ironic hypervisor (driver) located? | 15:30 |
| NobodyCam | linggao: https://review.openstack.org/#/c/51328/9 | 15:30 |
| NobodyCam | there? | 15:30 |
| *** lucasagomes is now known as lucas-hungry | 15:31 | |
| linggao | NobodyCam, yes it is! I thought it was merged already :-) | 15:31 |
| NobodyCam | no | 15:32 |
| NobodyCam | :( | 15:32 |
| NobodyCam | we need tp land the pxe / boot options patch from don so it can actually work | 15:33 |
| openstackgerrit | Max Lobur proposed a change to openstack/ironic: Fix JSONEncodedDict default values https://review.openstack.org/68413 | 15:37 |
| *** rloo has quit IRC | 15:46 | |
| *** rloo has joined #openstack-ironic | 15:46 | |
| *** aignatov is now known as aignatov_ | 15:49 | |
| GheRivero | morning all | 15:50 |
| NobodyCam | morning GheRivero :) | 15:50 |
| rloo | morning GheRivero. | 15:51 |
| NobodyCam | lucas-hungry: around? | 15:55 |
| *** datajerk has quit IRC | 15:56 | |
| NobodyCam | brb | 15:59 |
| *** datajerk has joined #openstack-ironic | 16:04 | |
| rloo | yuriyz: wrt https://review.openstack.org/#/c/63937/, patch 4, my comment about 'chassis_id' instead of 'chassis' was my bad, i forgot to delete the comment. Sorry. (I can't seem to add a comment to that now.) | 16:05 |
| yuriyz | ok :) | 16:05 |
| *** vkozhukalov has quit IRC | 16:07 | |
| *** hemna has quit IRC | 16:10 | |
| *** aignatov_ is now known as aignatov | 16:19 | |
| *** lucas-hungry is now known as lucasagomes | 16:21 | |
| lucasagomes | NobodyCam, morning | 16:21 |
| lucasagomes | morning rloo yuriyz linggao :D | 16:22 |
| rloo | afternoon lucasagomes! | 16:22 |
| NobodyCam | morning / afternoon lucasagomes | 16:22 |
| lucasagomes | :) I will take a look at the patch | 16:22 |
| linggao | morning lucasagomes, rloo, yuriyz, GheRivero and everyone. | 16:23 |
| rloo | morning linggao ;) | 16:26 |
| openstackgerrit | Yuriy Zveryanskyy proposed a change to openstack/ironic: Add parameter for filtering nodes by maintenance mode https://review.openstack.org/63937 | 16:26 |
| *** datajerk has quit IRC | 16:28 | |
| *** matty_dubs has quit IRC | 16:35 | |
| *** martyntaylor has quit IRC | 16:39 | |
| *** datajerk has joined #openstack-ironic | 16:40 | |
| *** aignatov is now known as aignatov_ | 16:42 | |
| *** vetalll1 has quit IRC | 16:43 | |
| *** matty_dubs has joined #openstack-ironic | 16:47 | |
| NobodyCam | lucasagomes: off the top of your head. do you recall if the nova driver will need to call the neutron update_port before issuing deploy? | 16:49 |
| lucasagomes | NobodyCam, I think no, I think the pxe module will do that | 16:49 |
| *** datajerk has quit IRC | 16:50 | |
| lucasagomes | NobodyCam, https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/pxe.py#L478 | 16:50 |
| NobodyCam | :) that was what I recalled was thinking... | 16:50 |
| NobodyCam | that last sentence made no sense | 16:50 |
| NobodyCam | lol | 16:50 |
| NobodyCam | :) that was what I recalled... | 16:51 |
| lucasagomes | xD | 16:51 |
| lucasagomes | I can try to make a patch for that if u want to | 16:51 |
| lucasagomes | that stub is already being called as part of the deploy | 16:51 |
| NobodyCam | let see what I get in testing today | 16:52 |
| *** yuriyz has quit IRC | 16:52 | |
| lucasagomes | ack | 16:53 |
| *** datajerk has joined #openstack-ironic | 16:53 | |
| *** matty_dubs has quit IRC | 16:55 | |
| *** datajerk has quit IRC | 16:58 | |
| *** mdurnosvistov has quit IRC | 16:59 | |
| *** datajerk has joined #openstack-ironic | 16:59 | |
| *** datajerk has quit IRC | 17:05 | |
| *** datajerk has joined #openstack-ironic | 17:07 | |
| NobodyCam | bbt brb | 17:08 |
| *** matty_dubs has joined #openstack-ironic | 17:09 | |
| devananda | morning, all | 17:09 |
| NobodyCam | morning devananda | 17:11 |
| *** matty_dubs is now known as matty_dubs|lunch | 17:11 | |
| lucasagomes | morning devananda | 17:11 |
| rloo | morning devananda | 17:11 |
| *** matty_dubs|lunch has quit IRC | 17:15 | |
| devananda | awesome job landing i2 fixes yesterday guys | 17:15 |
| devananda | i thin we got ~5 in at the last minute :) | 17:16 |
| devananda | https://launchpad.net/ironic/+milestone/icehouse-2 is great | 17:16 |
| *** ndipanov has quit IRC | 17:16 | |
| NobodyCam | woo hoo :) | 17:16 |
| rloo | yay! Are you/we going to try to land more changes for I2? | 17:17 |
| devananda | rloo: nope. it's being cut now() | 17:17 |
| rloo | ooo. I thought tomorrow was the deadline. | 17:17 |
| devananda | i only had to defer 3 bug fixes though | 17:17 |
| devananda | ah. no. yesterday was the deadline | 17:17 |
| rloo | that explains the long queue! | 17:18 |
| devananda | mp-cut is tuesdays. there's a day or two for backports, thursday is the final | 17:18 |
| devananda | rloo: https://wiki.openstack.org/wiki/ReleaseTeam/How_To_Release#MP_cut_.28Tuesday.29 | 17:19 |
| rloo | cool. so anything that lands now, is for i3. | 17:19 |
| devananda | basically, yep | 17:19 |
| rloo | the ironic driver didn't land, so ironic will be in i2, but not quite usable? | 17:20 |
| *** datajerk has quit IRC | 17:21 | |
| lucasagomes | nice | 17:21 |
| devananda | rloo: the nova driver is a separate thing entirely. a) we dont control when it lands, b) ironic is functional without it | 17:22 |
| rloo | devananda. ok :) | 17:22 |
| devananda | rloo: i mean, it's much better WITH it. but it's a separate thing in a separate project. nova gets to decide how important that is to their release cycle | 17:23 |
| *** ndipanov has joined #openstack-ironic | 17:23 | |
| *** matty_dubs|lunch has joined #openstack-ironic | 17:24 | |
| rloo | devananda: got it. thx! | 17:24 |
| devananda | also guys, our plans for i3 are a bit more aggressive :) https://launchpad.net/ironic/+milestone/icehouse-3 | 17:25 |
| devananda | and please, if you see a bug or bp that you expect in Icehouse and it's not targeted, update it or talk to me | 17:25 |
| NobodyCam | devananda: rloo and GheRivero raised some very good points on https://bugs.launchpad.net/bugs/1259346 | 17:26 |
| NobodyCam | I need to look in to them a bit more | 17:27 |
| rloo | it would be very nice if we could land the dict* changes and ()-instead-of-\ changes cuz they are PITAs. | 17:27 |
| rloo | in case you have nothing to do. but i think ()-instead-of-\ isn't ready yet (well, my opinion anyway). | 17:29 |
| *** hemnafk is now known as hemna | 17:30 | |
| NobodyCam | post bbt walkies... bbiafm | 17:31 |
| *** jistr has quit IRC | 17:32 | |
| *** lucasagomes has quit IRC | 17:40 | |
| devananda | NobodyCam, rloo - re: dont pass node // task has >1 node | 17:40 |
| *** blamar_ has joined #openstack-ironic | 17:40 | |
| devananda | yes, this is something we may start thinking about | 17:40 |
| *** datajerk has joined #openstack-ironic | 17:41 | |
| rloo | devananda: i don't believe I've seen a case where a task has > 1 node, but I assume we may want that functionality? | 17:43 |
| *** blamar has quit IRC | 17:43 | |
| *** blamar_ is now known as blamar | 17:43 | |
| devananda | here's a thought. what if (in the long run) we restructure the ConductorManager and TaskManager code to someting more like this... | 17:45 |
| devananda | http://paste.openstack.org/show/61693/ | 17:45 |
| devananda | actually this has been my plan all along. but we have to walk before we can run, as they say | 17:45 |
| rloo | ha ha. I don't think I've ever heard that wrt coding! | 17:47 |
| NobodyCam | in the above example nodes is a [list] of node_uuids or node object? | 17:47 |
| rloo | i think Nodes | 17:48 |
| rloo | unless it is cheap to get the Node from the uuid | 17:49 |
| devananda | bigger example | 17:50 |
| devananda | http://paste.openstack.org/show/61694/ | 17:50 |
| devananda | list of node uuids,a ctually | 17:50 |
| devananda | because task_manager.acquire() will create the node objs when it instantiates the task | 17:50 |
| devananda | we couldchange that | 17:51 |
| devananda | task_manager.acquire could accept node objects and just call obj.refresh() | 17:51 |
| *** datajerk has quit IRC | 17:51 | |
| devananda | so, also, relevant discussion is what tier the group-awareness should be | 17:52 |
| rloo | ah sorry NobodyCam. node uuids is probably better. | 17:52 |
| devananda | things like Heat expect that they operate on discrete things | 17:52 |
| devananda | not on groups of things | 17:52 |
| rloo | cuz until task_manager does the acquire, any Node could be out of sync. | 17:53 |
| devananda | rloo: right | 17:53 |
| devananda | rloo: hence either task_manager sets up the Node obj, or it must call refresh() on it | 17:53 |
| rloo | i like the clean look of the new code. | 17:53 |
| devananda | it moves a lot of logic out of ConductorManager and into TaaskManager | 17:54 |
| NobodyCam | refresh() seems kinda hackish to me? | 17:54 |
| *** datajerk has joined #openstack-ironic | 17:54 | |
| rloo | i think when I looked before, it was only the operation that modified a node, that passed a Node object (with the changes) to the ConductorManager. | 17:54 |
| devananda | NobodyCam: how so? objects are designed to be long-lived and refreshed when needed, eg, after being passed over RPC | 17:54 |
| *** lucasagomes has joined #openstack-ironic | 17:55 | |
| *** lucasagomes has quit IRC | 17:55 | |
| NobodyCam | devananda: true, guess I just like the get it when you need approch | 17:55 |
| *** lucasagomes has joined #openstack-ironic | 17:55 | |
| devananda | lifeless: if you're around, this mockup may be interesting. we talked about this ~6mo ago, and we're still probably another few months from doing anything with it. but worth your input on layer-effects: http://paste.openstack.org/show/61694/ | 17:56 |
| *** matty_dubs|lunch is now known as matty_dubs | 17:56 | |
| devananda | rloo: iirc, only update_node expects a changed node object | 17:57 |
| devananda | rloo: the rest of the CM methods expect a uuid | 17:57 |
| devananda | tangent: we should probably pull an update of nova's object code before icehouse release | 17:58 |
| devananda | so we dn't slip further out of syn | 17:58 |
| devananda | sync | 17:58 |
| rloo | devananda. ah yes, we already made that change to use node_ids. | 17:59 |
| devananda | it's more efficient for RPC when we really don't need to either manipulate the node_obj in the API, or pass the whole obj over RPC | 18:00 |
| *** datajerk has quit IRC | 18:00 | |
| rloo | to get back to the bug that started this discussion -- ?? | 18:01 |
| devananda | yes? | 18:03 |
| rloo | what to do with it? should the person continue working on it? | 18:04 |
| rloo | (or down the path they've taken, which doesn't seem right.) | 18:05 |
| *** vkozhukalov has joined #openstack-ironic | 18:05 | |
| devananda | hmm | 18:11 |
| *** datajerk has joined #openstack-ironic | 18:12 | |
| devananda | i think the question is, do we want drivers to be aware of groups of nodes (in which case they need to receive a task) | 18:13 |
| devananda | or just operate on one node at a time (in which case they should receive the node) | 18:14 |
| rloo | even if groups of nodes, unless they need something other than the list of nodes from the task, they don't need the task. | 18:15 |
| devananda | but they do | 18:15 |
| devananda | require_exclusive_lock decorator | 18:16 |
| devananda | that currently pulls the task from the *args | 18:16 |
| rloo | oh... I'm in over my head now... | 18:16 |
| devananda | i suppose we could try to pul lit from the stack instead? | 18:16 |
| *** derekh has quit IRC | 18:16 | |
| devananda | we are also accessing task.context in a few places, but we could move that to *args | 18:16 |
| *** mdurnosvistov has joined #openstack-ironic | 18:18 | |
| devananda | so instea of eg: def tear_down(self, task, node) it would be def tear_down(self, context, node) | 18:18 |
| rloo | just did a grep, pxe.py uses task to eg get resources | 18:22 |
| *** epim has joined #openstack-ironic | 18:23 | |
| devananda | yes. which is a short-cut to avoid doing a DB query | 18:23 |
| devananda | to fetch the Ports associatd with a node | 18:24 |
| devananda | since those are already bundled in the Task | 18:24 |
| devananda | that could be refactored | 18:24 |
| rloo | I am wondering whether it is worth deferring changes here, until we have more end-to-end working. | 18:26 |
| *** Alexei_987 has quit IRC | 18:30 | |
| *** harlowja_away is now known as harlowja | 18:32 | |
| lifeless | devananda: morning | 18:36 |
| devananda | lifeless: mornin! | 18:37 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Handle multiple exceptions raised by jsonpatch https://review.openstack.org/68457 | 18:46 |
| NobodyCam | devananda: I think the issue lsmola was / is having is due to incorrect (imcomplete) instance_type_extra_specs in his nova conf | 18:48 |
| NobodyCam | note is nova conf (last line) http://paste.openstack.org/show/61696/ | 18:49 |
| NobodyCam | s/is/his/ | 18:49 |
| *** athomas has quit IRC | 18:50 | |
| NobodyCam | brb | 18:51 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Enable driver methods without explicit 'task' arg https://review.openstack.org/68463 | 19:14 |
| devananda | rloo: ^ | 19:14 |
| rloo | ask and ye shall receive ;) thx devananda! | 19:15 |
| lifeless | devananda: ok, so that paste, a little more context would be good | 19:17 |
| devananda | lifeless: operate on a group of nodes at once | 19:17 |
| devananda | lifeless: the conversation from ~6mo ago where we talked about whether grouping should be done /only/ at a higher layer, or whether we should allow grouping within ironic as well | 19:17 |
| lifeless | oh, ok | 19:17 |
| lifeless | mmm | 19:18 |
| lifeless | I remember now | 19:18 |
| devananda | this is a code sketch of how we could enable it | 19:18 |
| lifeless | I was arguing that the Ironic API shouldn't group | 19:18 |
| devananda | yep | 19:18 |
| lifeless | that you should coalesce instead w/in Ironic | 19:18 |
| devananda | ah | 19:18 |
| lifeless | which the code you've pasted doesn't show | 19:18 |
| *** max_lobur is now known as max_lobur_afk | 19:19 | |
| devananda | well, true, but if we coalesced adjacent API requests within some midle tier, this code would perform the action | 19:19 |
| devananda | on taht group | 19:19 |
| lifeless | and I argued this because you'll get much bigger wins if you don't require every possible client to be fixed | 19:19 |
| lifeless | sure | 19:19 |
| devananda | as a code structure for "do something on group of nodes" i think this is pretty clean | 19:19 |
| lifeless | ok so that code specifically, when coalescing you need to preserve the original requests so you can raise seperate exceptions | 19:19 |
| devananda | but, yes, there's issues like that any time you coalesce separate requests | 19:20 |
| devananda | that also raises the sync vs async discussion | 19:20 |
| devananda | we're moving more towards async | 19:20 |
| lifeless | also you probably need to try all the nodes even if one fails | 19:20 |
| lifeless | and when raising signal x,y and z failed, a,b,c were all ok | 19:20 |
| devananda | esp with max_lobur_afk's threading proposal | 19:20 |
| lifeless | I haven't seen his threading proposal | 19:20 |
| *** mdurnosvistov has quit IRC | 19:20 | |
| lifeless | what else, you're not raising at the end of your except | 19:21 |
| lifeless | so noone will know if you succeeded or failed | 19:21 |
| lifeless | I have a couple of thoughts here | 19:21 |
| devananda | https://review.openstack.org/#/c/66368/1 | 19:21 |
| lifeless | firstly, I know some vendors have multiple-machine APIs; your code is actually a layer up from that | 19:22 |
| lifeless | they will need to do their own coalescing | 19:22 |
| lifeless | per-conductor or even cross-conductor | 19:22 |
| lifeless | so, I'd say - why both coalescing /here/, the driver seems like a more obvious place to do it | 19:23 |
| lifeless | secondly | 19:23 |
| lifeless | the points where coalescing is useful are (not exclusively, but I'd argue primarily) around high latency operations | 19:24 |
| *** epim has quit IRC | 19:24 | |
| lifeless | e.g. imagine multicast deploys | 19:24 |
| lifeless | for a single deploy you want to cache the image, start streaming to the group, power the node, wait for complete, stop streaming to the group | 19:24 |
| *** epim has joined #openstack-ironic | 19:24 | |
| *** datajerk has quit IRC | 19:25 | |
| lifeless | for a second lter deploy, you want to use that cached image, start streaming, power the node, wait for complete, stop streaming | 19:25 |
| *** zenfish has joined #openstack-ironic | 19:25 | |
| lifeless | for a second *concurrent* deploy, its exactly the same, but you want to not start a second stream, and stop streaming only when both images are finished | 19:25 |
| lifeless | which suggests to me that you want a hand-off from the per-node deploy code to a multicast stream management thing, which has a refcount and manages the image cache etc | 19:26 |
| lifeless | that could also live in the conductor layers, have its own ring, and be distributed but only one active for a given image at any time | 19:26 |
| devananda | hmm | 19:27 |
| devananda | that's an interesting one, yea | 19:27 |
| lifeless | as a driver writer writing the pxe driver, I'd want that stuff to live in the vendor extension area, so it doesn't need core changes | 19:27 |
| devananda | join existing operation | 19:27 |
| lifeless | note that this example above isn't coalescing *adjacent* requests | 19:28 |
| lifeless | its coalescing over possibly multiple minutes apart | 19:28 |
| devananda | yep | 19:28 |
| lifeless | and if you consider heat rolling updates that might be updating (say) no more than 10% of the cluster at once | 19:28 |
| lifeless | we'll trigger this | 19:28 |
| lifeless | because we'll start a few, wait for finish, start one, wait for finish of any other, start one, repeat | 19:29 |
| lifeless | [*] note that bittorrent or other things along the same line will have the same characteristics | 19:29 |
| devananda | indeed | 19:29 |
| devananda | that kind of optimization will need to live within the driver | 19:30 |
| lifeless | so - personally - I wouldn't do any coalescing work at all at this point, instead I'd look at making the design support this sort of message passing and workers | 19:30 |
| *** datajerk has joined #openstack-ironic | 19:32 | |
| lifeless | because I think you'll get more bang for buck | 19:36 |
| *** datajerk has quit IRC | 19:36 | |
| lifeless | and not have to deal with questions like coalescing requests for different conductors | 19:36 |
| *** datajerk has joined #openstack-ironic | 19:37 | |
| devananda | right, several points in there are interesting | 19:37 |
| devananda | multi-machine APIs doing their own coalescing seems not to fit with their requirements | 19:38 |
| devananda | which is that some layer above the hardware understand the group API that said hardware exposes | 19:38 |
| devananda | this mockup could be a step towards addrsesing that | 19:38 |
| lifeless | 'their' - do you mean vendor features? | 19:39 |
| devananda | yes | 19:39 |
| lifeless | ok | 19:39 |
| lifeless | so I think their driver would do the same basic framework I described for multicast | 19:39 |
| devananda | tuskar folks have been looking at expressing groupings - arbitrary groupings - in the UI | 19:40 |
| lifeless | add in a sleep(1) or something in the 'and now we call our vendor backend library' step, and that would let them let more requests coalesce fairly naturally | 19:40 |
| devananda | and asked hwo to model that in ironic | 19:40 |
| lifeless | sadface | 19:40 |
| devananda | to do that we need something more than just the notion of a chassis | 19:40 |
| lifeless | I think there is some confusion afoot there | 19:40 |
| devananda | possibly | 19:40 |
| devananda | i pushed back :) | 19:40 |
| lifeless | in that I've suggested tags as a useful adhoc thing | 19:40 |
| devananda | saying that we should store tags and allow search-by-tags and such | 19:41 |
| lifeless | and that for anything functional like HA distribution we need to model the specifics | 19:41 |
| lifeless | arbitrary grouping for grouping sake is not something I believe tuskar needs | 19:41 |
| lifeless | or that users need | 19:41 |
| lifeless | devananda: I suspect you and I are entirely on the same page here :) | 19:41 |
| devananda | i think so :) | 19:41 |
| devananda | also, i think there are situations where ironic does need some understanding that nodes x, y, z are discrete but not independent | 19:42 |
| devananda | eg, when it's not possible to perform action A on x without affecting y and z | 19:42 |
| devananda | i dont think we need that soon, though | 19:43 |
| devananda | taking a step back -- this pseudocode came up as a way to clarify how we're handling locks and passing requests to drivers | 19:44 |
| devananda | does our api allow atomically locking >1 node? | 19:44 |
| devananda | *internal api | 19:45 |
| devananda | does that internal api allow passing >1 node to a driver at once? | 19:45 |
| devananda | these are related questions | 19:45 |
| *** vkozhukalov has quit IRC | 19:45 | |
| devananda | if the RESTful API only allows operations on one node at a time -- never operations on multiple nodes OR operations on a single group-of-nodes -- which I think is much clearer and simpler | 19:46 |
| devananda | then how should the code model the coalescing of adjacent (or non-adjascent) requests? | 19:46 |
| devananda | using one TaskManager to contain a grouping of nodes would allow that coalescing to be done at the API or Conductor tiers, rather than inside a driver | 19:47 |
| devananda | is that better? i'm not sure yet :) | 19:47 |
| * devananda gets coffee and breakfast and continues ruminating | 19:48 | |
| *** datajerk has quit IRC | 19:49 | |
| *** rloo has quit IRC | 19:51 | |
| *** rloo has joined #openstack-ironic | 19:52 | |
| *** rloo has quit IRC | 19:53 | |
| *** rloo has joined #openstack-ironic | 19:53 | |
| lifeless | roger | 19:54 |
| lifeless | so, it seems to me that having one-node-lock-at-a-time is going to be much less likely to discover bugs, as its simpler code | 19:55 |
| lifeless | so I'd just do that everywhere | 19:55 |
| *** rloo has quit IRC | 19:58 | |
| *** rloo has joined #openstack-ironic | 19:59 | |
| *** datajerk has joined #openstack-ironic | 20:01 | |
| *** datajerk has quit IRC | 20:08 | |
| NobodyCam | devananda: when your around a keyboard I have a question on your nova-ironic dib element ( https://review.openstack.org/#/c/66461 ) | 20:10 |
| *** rloo has quit IRC | 20:10 | |
| devananda | NobodyCam: i'm sort of here | 20:10 |
| *** rloo has joined #openstack-ironic | 20:10 | |
| NobodyCam | its a easy question | 20:11 |
| *** derekh has joined #openstack-ironic | 20:11 | |
| NobodyCam | I was looking at https://review.openstack.org/#/c/66461/1/elements/nova-ironic/install.d/80-pxelinux | 20:11 |
| NobodyCam | but most of that seems to have already landed: | 20:11 |
| NobodyCam | https://github.com/openstack/tripleo-image-elements/blob/master/elements/ironic-conductor/install.d/68-ironic-tftp-support | 20:11 |
| NobodyCam | did you want to replace that? | 20:12 |
| devananda | replace with? | 20:15 |
| NobodyCam | your review? | 20:15 |
| NobodyCam | I was looking at https://review.openstack.org/#/c/66461/1/elements/nova-ironic/install.d/80-pxelinux | 20:15 |
| devananda | seems totally unrelated | 20:16 |
| devananda | oh! | 20:16 |
| NobodyCam | erally | 20:16 |
| devananda | tht file shouldnt be there | 20:16 |
| NobodyCam | :-p | 20:16 |
| NobodyCam | ok | 20:16 |
| devananda | nova-compute desnt need tftpdir | 20:16 |
| NobodyCam | I'm working thru that now | 20:17 |
| NobodyCam | will remove that file :) | 20:17 |
| devananda | awesome, ty. i havent ad time... | 20:17 |
| NobodyCam | :) | 20:17 |
| *** datajerk has joined #openstack-ironic | 20:20 | |
| *** lucasagomes is now known as lucas-afk | 20:21 | |
| *** rloo has quit IRC | 20:26 | |
| *** rloo has joined #openstack-ironic | 20:26 | |
| *** rloo has quit IRC | 20:26 | |
| *** rloo has joined #openstack-ironic | 20:27 | |
| *** rloo has quit IRC | 20:29 | |
| *** rloo has joined #openstack-ironic | 20:29 | |
| * NobodyCam makes a bagel | 20:30 | |
| *** rloo has quit IRC | 20:30 | |
| *** rloo has joined #openstack-ironic | 20:31 | |
| *** datajerk has quit IRC | 20:34 | |
| devananda | lifeless: python question -- note this is just a proof-of-concept at this stage to see if I could do it -- is this particularly bad? https://review.openstack.org/#/c/68463/1/ironic/conductor/task_manager.py | 20:34 |
| *** mrda_away is now known as mrda | 20:35 | |
| lifeless | devananda: meep! why? | 20:47 |
| devananda | lifeless: wanted to see if i could remove the "task" parameter from driver methods | 20:47 |
| devananda | while still using the @require_exclusive_lock decorator | 20:47 |
| devananda | in fact, i can - but it seems terrible :) | 20:48 |
| *** datajerk has joined #openstack-ironic | 20:48 | |
| devananda | it seems like nonlocal would be better, but that's not in py2.7 ,and i couldn't find a similar solution | 20:49 |
| devananda | well, nonlocal also isn't quite what i want, i think | 20:49 |
| NobodyCam | brb | 20:49 |
| devananda | lifeless: put another way, is there a pythonic way to pass something to a decorator but NOT to the method it decorates? | 20:50 |
| *** rloo has quit IRC | 20:55 | |
| *** rloo has joined #openstack-ironic | 20:55 | |
| *** datajerk has quit IRC | 20:57 | |
| lifeless | devananda: just don't pass it when you call the decorated thing | 20:58 |
| devananda | lifeless: delete it from *args? | 20:59 |
| lifeless | def wrapper(*args, foo=None, **kwargs): | 20:59 |
| lifeless | ... f(*args, **kwargs) | 21:00 |
| devananda | sometimes the obvious is too simple :) | 21:00 |
| devananda | thanks | 21:00 |
| lifeless | or if task is first | 21:00 |
| lifeless | wrapper(task, *args, **kwargs): | 21:00 |
| lifeless | f(*args, **kwargs) | 21:00 |
| *** mdurnosvistov has joined #openstack-ironic | 21:11 | |
| *** zenfish has quit IRC | 21:18 | |
| *** lucas-afk has quit IRC | 21:25 | |
| *** datajerk has joined #openstack-ironic | 21:26 | |
| *** rloo has quit IRC | 21:27 | |
| *** rloo has joined #openstack-ironic | 21:27 | |
| *** datajerk has quit IRC | 21:31 | |
| *** datajerk has joined #openstack-ironic | 21:35 | |
| *** rloo has quit IRC | 21:35 | |
| *** rloo has joined #openstack-ironic | 21:35 | |
| *** rloo has quit IRC | 21:42 | |
| *** datajerk has quit IRC | 21:44 | |
| *** rloo has joined #openstack-ironic | 21:44 | |
| *** mdurnosvistov has quit IRC | 21:49 | |
| *** derekh has quit IRC | 22:03 | |
| NobodyCam | brb | 22:04 |
| *** datajerk has joined #openstack-ironic | 22:05 | |
| *** linggao has quit IRC | 22:06 | |
| *** ColinTaylor has joined #openstack-ironic | 22:08 | |
| *** datajerk has quit IRC | 22:17 | |
| *** matty_dubs is now known as matty_dubs|gone | 22:17 | |
| *** datajerk has joined #openstack-ironic | 22:19 | |
| *** jdob has quit IRC | 22:27 | |
| *** datajerk has quit IRC | 22:38 | |
| *** datajerk has joined #openstack-ironic | 22:40 | |
| *** datajerk has quit IRC | 22:41 | |
| lifeless | devananda: sorry, ESTUFF | 22:54 |
| lifeless | devananda: I think we sorted that though, no? | 22:55 |
| *** hstimer has joined #openstack-ironic | 23:15 | |
| *** jbjohnso has quit IRC | 23:18 | |
| *** romcheg has quit IRC | 23:27 | |
| devananda | lifeless: yep. been sketching it out for the list | 23:33 |
| *** mdurnosvistov has joined #openstack-ironic | 23:36 | |
| dkehn | NobodyCam: u around | 23:55 |
| NobodyCam | maybe | 23:56 |
| NobodyCam | i'm currently Waiting for seed VM to boot. | 23:57 |
| NobodyCam | :-p | 23:57 |
| dkehn | NobodyCam: I'm seeing some strage things when after running the generate_sample.sh, no errors, toxc -epep8 is giving me a diff error on include_service_catalog=true?/ | 23:58 |
| dkehn | NobodyCam: did a rebase all seems good, one conflict but resolved | 23:58 |
| dkehn | NobodyCam: think I'll hand edit to remove it and see | 23:59 |
| NobodyCam | in pxe.py? | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!