| devananda | Shrews: hm. something i'm missing here | 00:08 |
|---|---|---|
| devananda | Shrews: the node has to power down at somepoint anyway -- since it needs to reboot into the deploy ramdisk | 00:08 |
| devananda | Shrews: it sounds like a question of whether that powerdown should be triggered by nova beforehand, or within ironic as part of the rebuild | 00:09 |
| *** pquerna has quit IRC | 00:11 | |
| *** pquerna has joined #openstack-ironic | 00:13 | |
| *** eguz has quit IRC | 00:20 | |
| Shrews | devananda: then it sounds like my forced shutdown is redundant and can be removed | 00:23 |
| Shrews | if the reboot to deploy is a clean shutdown, that is. i suppose i should investigate that code path. | 00:25 |
| openstackgerrit | A change was merged to openstack/ironic-python-agent: Make docker image smaller by using Precise https://review.openstack.org/90845 | 00:26 |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic: Adding swift temp url support https://review.openstack.org/81391 | 00:28 |
| lifeless | Shrews: hi | 00:35 |
| lifeless | devananda: hi, back so rrry | 00:35 |
| lifeless | JayF: how can you do ephemeral partitions in bm with whole disk images? | 00:35 |
| lifeless | Shrews: i think the idea of an agent makes asserting power states from api actions dubious | 00:36 |
| lifeless | Shrews: clearly it will hve to be in the agent to do the disk manipulation (either the current shallow agent or a rich one) | 00:36 |
| lifeless | devananda: partition images allow nova to be responsible for: | 00:37 |
| lifeless | - swap size | 00:37 |
| lifeless | - ephemeral partition | 00:37 |
| devananda | lifeless: ack. was that the reason we moved away from the initially? or was it something else? | 00:37 |
| lifeless | devananda: whole disk images prevent that but allow for legacy boot blocks; theres no reason to ever use whole disk except for windows AFAICT | 00:37 |
| lifeless | devananda: I have no recollection of nova-bm using whole disk ever | 00:38 |
| devananda | lifeless: hmm, i recall us dealing with that in dib's infancy | 00:38 |
| devananda | but haven't tried to grep through ancient logs to figure ou twhy | 00:38 |
| lifeless | devananda: dib has to do both because of test vms | 00:39 |
| devananda | ahh! that's why | 00:39 |
| *** radsy has quit IRC | 00:39 | |
| devananda | ok | 00:39 |
| lifeless | devananda: specifically the seed | 00:39 |
| devananda | artefact of testing | 00:39 |
| lifeless | well, seed is more than test, but broadly yes - the only reason dib originally added whole-disk support was for *non-nova-bm* cases. | 00:39 |
| devananda | thanks | 00:40 |
| devananda | lifeless: and do you see any reason to support /arbitrary/ partitioning (eg, beyond what nova needs) | 00:41 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Sync oslo logging https://review.openstack.org/87346 | 00:43 |
| lifeless | devananda: I can make arguments in favour of that. I'm not sure I'd agree with said arguments :) | 00:45 |
| devananda | :) | 00:46 |
| devananda | s/see any reason/think we should/ | 00:46 |
| devananda | cause, yea, i can see reasons -- but i dont think ironic (or more specifically IPA) should do that | 00:47 |
| lifeless | so as a way to avoid unexpected consequences | 00:49 |
| russell_h | devananda: on this: https://review.openstack.org/#/c/85742/ | 00:49 |
| russell_h | shouldn't a standard interface be able to assume inputs are valid? | 00:50 |
| lifeless | I think openstack (e.g. Ironic // cinder) should either own the partition table for a given physical disk, or not touch the partition table for that disk. | 00:50 |
| lifeless | devananda: so consider the software raid 5 scenario - ironic would own the partition table for the backing disks | 00:50 |
| russell_h | like, a passthru interface I understand the implementation needing to do validation | 00:50 |
| lifeless | devananda: regardless of whether its deploying a whole disk image or a partition image to the upper layer | 00:50 |
| devananda | russell_h: ah - you're assuming teh API would have validated the input. and you're correct. | 00:50 |
| devananda | russell_h: though I dont believe I've seen a REST API patch that corresponds to this driver API yet | 00:51 |
| russell_h | right | 00:51 |
| lifeless | devananda: so starting from that desire, the question about arbitrary partition layouts becomes more 'are there users that can't achieve what they need if we don't support arbitrary layouts'? and 'is it worth supporting them' | 00:52 |
| lifeless | devananda: so far I haven't seen any need to do the crazy 7-fs layouts you might do on a precious snowflake server with a cloud based approach | 00:52 |
| lifeless | devananda: so, I'm skeptical that there are such users. IMBW though! | 00:53 |
| devananda | lifeless: and, is ironic the right tool to support those needs, or does that expand the scope too much | 00:53 |
| *** datajerk has joined #openstack-ironic | 00:53 | |
| lifeless | devananda: so ironic seems to need whole-disk and partition modes | 00:54 |
| lifeless | devananda: in the former, I'd like to not touch (read *or* write) the table at all. | 00:54 |
| devananda | lifeless: as an aside, every time i chat with cinder folks, i seem to hear that it doesn't do anything even resembling what we want -- in that it doesn't have a way to define raid or partition layout | 00:55 |
| lifeless | devananda: in the latter, we have three basic primitives: lay down an image, make a partition and leave it untouched, make a partition and make a filesystem | 00:55 |
| devananda | lifeless: the cinder admin can create a pool from a raid and describe it as-such to users, who then select which pool to carve off a chunk from | 00:55 |
| lifeless | devananda: yes; when we spoke face to face in portland they were open to the idea of such a thing, which we'd define | 00:55 |
| lifeless | devananda: doing a complex layout would seem to require being able to specify an arbitrary mix of those primitives, and then the user *building* images has a consistency problem to deal with | 00:57 |
| lifeless | devananda: OTOH android and similar devices seem to get by without having lots of partitions | 00:57 |
| devananda | lifeless: yep. for a cloud-like environment (even on bare metal) i don't see a compelling reason for complex partition layouts | 00:59 |
| *** datajerk has quit IRC | 01:05 | |
| *** lazy_prince has quit IRC | 01:09 | |
| devananda | heading out for dinner shortly... bbtmw | 01:18 |
| jroll | \o see ya | 01:20 |
| openstackgerrit | David Shrewsbury proposed a change to openstack/ironic: Implement instance rebuild in nova.virt.driver https://review.openstack.org/90429 | 01:24 |
| *** lazy_prince has joined #openstack-ironic | 01:24 | |
| *** lazy_prince is now known as killer_prince | 01:25 | |
| openstackgerrit | David Shrewsbury proposed a change to openstack/ironic: Implement instance rebuild in nova.virt.driver https://review.openstack.org/90429 | 01:26 |
| *** datajerk has joined #openstack-ironic | 01:31 | |
| lifeless | devananda: whats the url for the agent review? | 01:46 |
| *** newell_ has quit IRC | 01:50 | |
| *** zdin0bot has quit IRC | 01:53 | |
| lifeless | NobodyCam: adam_g hi - can I pass ownership of my race condition patch to one of you two ? | 01:55 |
| *** zdin0bot has joined #openstack-ironic | 02:02 | |
| *** Haomeng has quit IRC | 02:02 | |
| *** Haomeng has joined #openstack-ironic | 02:03 | |
| *** rameshg87 has joined #openstack-ironic | 02:06 | |
| *** shakamunyi has quit IRC | 02:11 | |
| *** rloo has quit IRC | 02:13 | |
| *** Haomeng has quit IRC | 02:28 | |
| *** coolsvap|afk is now known as coolsvap | 02:31 | |
| *** Haomeng has joined #openstack-ironic | 02:32 | |
| *** dwalleck_ has joined #openstack-ironic | 02:42 | |
| lifeless | https://review.openstack.org/#/c/84361/ specifically | 02:43 |
| lifeless | NobodyCam: adam_g ^ | 02:43 |
| *** datajerk1 has joined #openstack-ironic | 02:44 | |
| *** datajerk has quit IRC | 02:45 | |
| *** dwalleck__ has quit IRC | 02:46 | |
| *** Haomeng has quit IRC | 02:50 | |
| *** Haomeng has joined #openstack-ironic | 02:51 | |
| *** datajerk1 has quit IRC | 02:52 | |
| adam_g | lifeless, yeah, the workaround seems simple enough. the proper fix will be more work but i would love to drive that. will push a new patchset there tomorrow | 03:30 |
| *** eghobo has joined #openstack-ironic | 03:33 | |
| *** zdin0bot has quit IRC | 03:48 | |
| *** killer_prince is now known as lazy_prince | 04:06 | |
| *** dwalleck_ has quit IRC | 04:12 | |
| *** zdin0bot has joined #openstack-ironic | 04:13 | |
| *** pradipta_away has quit IRC | 04:13 | |
| *** pradipta_away has joined #openstack-ironic | 04:13 | |
| lifeless | adam_g: thanks | 04:16 |
| rameshg87 | Hi all | 04:23 |
| rameshg87 | i am getting "oslo.config.cfg.DuplicateOptError: duplicate option: rpc_backend" error when devstack runs /usr/local/bin/ironic-dbsync with the latest ironic code | 04:24 |
| rameshg87 | i am not able to run the tests as well | 04:24 |
| rameshg87 | this seems to be related to https://review.openstack.org/#/c/88307 | 04:25 |
| rameshg87 | anyone has any idea already regarding this ? | 04:25 |
| *** eghobo has quit IRC | 04:25 | |
| sirushti | rameshg87, did you try with RECLONE=True in your localrc? | 04:26 |
| rameshg87 | sorry my mistake | 04:27 |
| rameshg87 | it is due to leftover .pyc files | 04:27 |
| rameshg87 | :-) | 04:27 |
| sirushti | rameshg87, :-) | 04:27 |
| rameshg87 | thanks sirushti, it was due to .pyc files :-) | 04:27 |
| *** eghobo has joined #openstack-ironic | 04:46 | |
| *** vkozhukalov has joined #openstack-ironic | 04:54 | |
| *** killer_p- has joined #openstack-ironic | 04:55 | |
| *** killer_p- is now known as killer_prince | 04:55 | |
| *** lazy_prince has quit IRC | 04:56 | |
| *** harlowja is now known as harlowja_away | 04:56 | |
| *** sabah has joined #openstack-ironic | 04:59 | |
| *** zdin0bot has quit IRC | 05:15 | |
| *** sabah has quit IRC | 05:40 | |
| *** Alexei_987 has quit IRC | 05:49 | |
| *** vkozhukalov has quit IRC | 05:50 | |
| *** pradipta_away is now known as pradipta | 05:51 | |
| *** Alexei_987 has joined #openstack-ironic | 05:52 | |
| *** sabah has joined #openstack-ironic | 05:54 | |
| *** max_lobur has joined #openstack-ironic | 06:02 | |
| *** sabah has quit IRC | 06:05 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic: Imported Translations from Transifex https://review.openstack.org/88508 | 06:07 |
| *** rameshg87 has quit IRC | 06:16 | |
| *** rameshg87 has joined #openstack-ironic | 06:17 | |
| *** rameshg87 has quit IRC | 06:25 | |
| Mikhail_D_wk | Morning Ironic! :) | 06:25 |
| *** rameshg87 has joined #openstack-ironic | 06:27 | |
| *** viktors|afk is now known as viktors | 06:29 | |
| *** rameshg87 has quit IRC | 06:31 | |
| *** rameshg87 has joined #openstack-ironic | 06:33 | |
| *** rameshg87 has quit IRC | 06:38 | |
| *** rameshg87 has joined #openstack-ironic | 06:38 | |
| *** sabah has joined #openstack-ironic | 06:38 | |
| *** rameshg87 has quit IRC | 06:44 | |
| *** rameshg87 has joined #openstack-ironic | 06:45 | |
| openstackgerrit | Ramakrishnan G proposed a change to openstack/ironic: Add IloDriver and its utils https://review.openstack.org/89500 | 06:47 |
| *** rameshg87 has quit IRC | 06:48 | |
| *** rameshg87 has joined #openstack-ironic | 06:48 | |
| *** rameshg87 has left #openstack-ironic | 06:55 | |
| *** datajerk has joined #openstack-ironic | 07:01 | |
| *** yuriyz has joined #openstack-ironic | 07:11 | |
| *** ifarkas has joined #openstack-ironic | 07:12 | |
| *** Manishanker has joined #openstack-ironic | 07:13 | |
| Haomeng | Mikhail_D_wk: morning:) | 07:18 |
| *** mrda is now known as mrda_away | 07:20 | |
| *** eghobo has quit IRC | 07:20 | |
| *** max_lobur1 has joined #openstack-ironic | 07:21 | |
| *** max_lobur has quit IRC | 07:24 | |
| *** sabah has quit IRC | 07:30 | |
| *** vkozhukalov has joined #openstack-ironic | 07:40 | |
| *** jistr has joined #openstack-ironic | 07:51 | |
| *** foexle has joined #openstack-ironic | 07:53 | |
| *** ndipanov_gone is now known as ndipanov | 07:59 | |
| *** datajerk has quit IRC | 08:01 | |
| *** datajerk has joined #openstack-ironic | 08:10 | |
| *** pradipta is now known as pradipta_away | 08:13 | |
| *** pradipta_away is now known as pradipta | 08:14 | |
| *** pradipta is now known as pradipta_away | 08:14 | |
| *** derekh has joined #openstack-ironic | 08:17 | |
| *** athomas has joined #openstack-ironic | 08:22 | |
| *** lucasagomes has joined #openstack-ironic | 08:26 | |
| dtantsur | Morning Ironic | 08:27 |
| *** sabah has joined #openstack-ironic | 08:43 | |
| *** max_lobur1 has quit IRC | 08:51 | |
| *** derekh has quit IRC | 09:03 | |
| *** athomas has quit IRC | 09:07 | |
| romcheg | Morning dtantsur | 09:08 |
| romcheg | lucasagomes: Morning, around? | 09:09 |
| *** martyntaylor has joined #openstack-ironic | 09:11 | |
| *** athomas has joined #openstack-ironic | 09:12 | |
| *** anusha has joined #openstack-ironic | 09:12 | |
| Mikhail_D_wk | dtantsur, romcheg g'morning :) | 09:15 |
| romcheg | I noticed that it's not possible to specify region_name to the client | 09:17 |
| agordeev | morning :) | 09:28 |
| agordeev | morning dtantsur Haomeng romcheg Mikhail_D_wk | 09:28 |
| dtantsur | morning agordeev, Mikhail_D_wk, romcheg :) | 09:29 |
| *** anusha has quit IRC | 09:31 | |
| romcheg | Ok, it is possible but somehow it's not clear for me yet... | 09:35 |
| *** sabah has quit IRC | 09:36 | |
| *** max_lobur has joined #openstack-ironic | 09:41 | |
| openstackgerrit | A change was merged to openstack/ironic: Document ClusteredComputeManager https://review.openstack.org/89533 | 09:47 |
| romcheg | So I definitely need to discuss that with lucasagomes. Please poke me when you're available | 09:49 |
| *** radsy has joined #openstack-ironic | 09:54 | |
| *** radsy has joined #openstack-ironic | 09:54 | |
| *** datajerk has quit IRC | 09:59 | |
| lifeless | lucasagomes: morning | 10:16 |
| *** mdenny has quit IRC | 10:17 | |
| lifeless | lucasgomes nvm | 10:17 |
| lucasagomes | romcheg, hey | 10:30 |
| lucasagomes | lifeless, morning | 10:30 |
| lucasagomes | anyone able to push changes to gerrit?! I'm getting a "fatal: Could not read from remote repository." | 10:33 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Fix chassis-node relationship https://review.openstack.org/90964 | 10:36 |
| lifeless | lucasagomes: have you refreshed the ssh key ? | 10:37 |
| lucasagomes | lifeless, yeah fixed following https://wiki.openstack.org/wiki/GerritUpgrade | 10:37 |
| lucasagomes | first time I got this error :) lifeless thanks! | 10:37 |
| *** coolsvap is now known as coolsvap|afk | 10:40 | |
| *** datajerk has joined #openstack-ironic | 10:45 | |
| openstackgerrit | Imre Farkas proposed a change to openstack/ironic: Fix workaround for the "device is busy" problem https://review.openstack.org/90126 | 10:54 |
| *** sabah has joined #openstack-ironic | 10:56 | |
| *** pradipta_away is now known as pradipta | 11:02 | |
| *** sabah has quit IRC | 11:09 | |
| openstackgerrit | A change was merged to openstack/ironic: Overwrite instance_exists in Nova Ironic Driver https://review.openstack.org/88611 | 11:17 |
| openstackgerrit | A change was merged to openstack/ironic: Sync oslo logging https://review.openstack.org/87346 | 11:22 |
| romcheg | lucasagomes: Hi, sorry, was afk for a lunch | 11:25 |
| vkozhukalov | guys, i have a question | 11:28 |
| vkozhukalov | oslo.config by default tries to find config file in /etc/{project}/{project}.conf | 11:29 |
| vkozhukalov | the question is | 11:29 |
| vkozhukalov | ironic-python-agent or ironic_python_agent? | 11:29 |
| vkozhukalov | which is correct? | 11:29 |
| vkozhukalov | or it is set manually somewhere? | 11:30 |
| romcheg | vkozhukalov: in devstack config is explicitly passed | 11:46 |
| vkozhukalov | romcheg: thanks, but it is not very helpful for my case | 11:47 |
| romcheg | I see | 11:47 |
| romcheg | vkozhukalov: I've checked oslo.config's code | 11:50 |
| romcheg | you can explicitly pass project=project when calling the config, | 11:50 |
| romcheg | %s/calling/parsing/ | 11:51 |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic: Consider Glance checksum when caching master images https://review.openstack.org/90390 | 11:55 |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic: Implement more robust caching for master images https://review.openstack.org/85387 | 11:55 |
| vkozhukalov | romcheg: thanks, now i see that every project sets it's project name explicitely | 11:58 |
| romcheg | vkozhukalov: yup, however, I could not find that in IPA | 11:58 |
| vkozhukalov | romcheg: for example ironic sets it in ironic/common/config.py:parse_args | 11:59 |
| vkozhukalov | romcheg: yep | 11:59 |
| vkozhukalov | romcheg: IPA does not set it at all | 11:59 |
| vkozhukalov | romcheg: need to set it | 11:59 |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic: Consider Glance checksum when caching master images https://review.openstack.org/90390 | 12:01 |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic: Implement more robust caching for master images https://review.openstack.org/85387 | 12:01 |
| *** radsy has quit IRC | 12:03 | |
| openstackgerrit | Aleksandr Gordeev proposed a change to openstack/ironic-python-agent: Make encoding.serialize() more programmatical https://review.openstack.org/90989 | 12:04 |
| *** datajerk has quit IRC | 12:08 | |
| NobodyCam | good mroning from Loisiana | 12:13 |
| openstackgerrit | Vladimir Kozhukalov proposed a change to openstack/ironic-python-agent: Fixed circular imports https://review.openstack.org/90991 | 12:14 |
| dtantsur | morning, NobodyCam! | 12:14 |
| NobodyCam | morning dtantsur | 12:15 |
| openstackgerrit | Vladimir Kozhukalov proposed a change to openstack/ironic-python-agent: Fixed circular imports https://review.openstack.org/90991 | 12:15 |
| dtantsur | Maybe someone knows: is it possible to update contents of an image in Glance? I my experiments glance image-update succeeds, but content is not updated | 12:16 |
| dtantsur | (it is to decide, whether we need https://review.openstack.org/#/c/90390 ) | 12:16 |
| agordeev | NobodyCam: morning | 12:18 |
| NobodyCam | morning agordeev | 12:19 |
| romcheg | Morning NobodyCam! | 12:19 |
| NobodyCam | morning romcheg | 12:20 |
| NobodyCam | dtantsur: humm good question.. I think you have remove and reload the image but I am not sure | 12:20 |
| NobodyCam | brb quick mornig walkies :-p | 12:21 |
| agordeev | romcheg: hey hey, glad to know you have become the core reviewer in IPA too. Congrats! | 12:28 |
| *** jdob has joined #openstack-ironic | 12:28 | |
| romcheg | agordeev: I think that ironic-core was merged to ipa-core | 12:28 |
| openstackgerrit | Aleksandr Gordeev proposed a change to openstack/ironic-python-agent: Make encoding.serialize() more programmatical https://review.openstack.org/90989 | 12:32 |
| NobodyCam | nice:http://onlineathens.com/breaking-news/2014-04-28/tornado-watch-covers-17-georgia-counties | 12:37 |
| NobodyCam | I hope this weather is cleared up for the meeting | 12:37 |
| NobodyCam | s/meeting/summit/ | 12:37 |
| lucasagomes | NobodyCam, :( be careful on ur way to atlanta | 12:38 |
| NobodyCam | morning lucasagomes ... I am trying... :) we in western Louisiana now.. | 12:39 |
| lucasagomes | NobodyCam, :) | 12:43 |
| *** jbjohnso has joined #openstack-ironic | 12:44 | |
| NobodyCam | with really bad bandwidth :( | 12:45 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Remove 'node' parameter from the VendorPassthru interface https://review.openstack.org/90363 | 12:45 |
| *** linggao has joined #openstack-ironic | 12:47 | |
| NobodyCam | gah I cann't even get on the vpn :( | 12:51 |
| Shrews | NobodyCam: the weather is supposed to be really nice friday, sat, sun. so just take your time and ride behind the weather for a few days :) | 12:51 |
| NobodyCam | Shrews: yep :) thats the plan :-p | 12:52 |
| NobodyCam | come in behind it all | 12:52 |
| *** datajerk has joined #openstack-ironic | 12:55 | |
| *** sseago has quit IRC | 13:00 | |
| *** sseago has joined #openstack-ironic | 13:01 | |
| *** rloo has joined #openstack-ironic | 13:02 | |
| GheRivero | morning all | 13:10 |
| dtantsur | morning, GheRivero | 13:10 |
| NobodyCam | morning gh:) | 13:10 |
| NobodyCam | humm | 13:11 |
| NobodyCam | moning GheRivero | 13:11 |
| NobodyCam | thats better | 13:11 |
| GheRivero | :) | 13:11 |
| NobodyCam | any one here used suse studio to create SuSe cloud images? | 13:12 |
| dtantsur | NobodyCam, GheRivero may I ask you guys to look at https://blueprints.launchpad.net/ironic/+spec/pxe-master-images-caching and https://review.openstack.org/#/c/85387/ ? | 13:12 |
| dtantsur | I'm also a bit stuck with question, whether glance actually supports updating image content | 13:12 |
| NobodyCam | dtantsur: let get my dev laptop up and going | 13:13 |
| dtantsur | sure, thank you :) | 13:13 |
| *** zdin0bot has joined #openstack-ironic | 13:13 | |
| *** datajerk has quit IRC | 13:15 | |
| *** sseago has quit IRC | 13:21 | |
| romcheg | Uploaded basic scenario to Rally | 13:22 |
| romcheg | Discussing it with Boris | 13:22 |
| NobodyCam | dtantsur: http://docs.openstack.org/user-guide/content/cli_manage_images.html <- just fyi | 13:23 |
| *** jgrimm has joined #openstack-ironic | 13:24 | |
| dtantsur | NobodyCam, yes, I've seen it, but the docs are misleading. Judging by source code of v1 API you can only update image content if it does not have one, i.e. is in queued state | 13:25 |
| NobodyCam | ahh | 13:26 |
| NobodyCam | I have not looked at the code :-p | 13:27 |
| dtantsur | I started to suspect something only after manually checking. Call to image-update succeeds, but content does not get updated | 13:28 |
| NobodyCam | GheRivero: do we use v1 or v2 for glance api? with out looking to hard it looks like we support both | 13:29 |
| dtantsur | NobodyCam, that is on of the reasons I want feedback on a blueprint: it's not clear for me if we should deal with changing images in Glance when caching | 13:29 |
| GheRivero | NobodyCam: I was just looking at that. We use both, jsut need to indicate which version are we using. | 13:29 |
| GheRivero | There was some attempt in glance to autodiscover the version supported by the glance server | 13:30 |
| GheRivero | but hasn't landed yet | 13:30 |
| NobodyCam | dtantsur: the end goal I have is for IPA to pull the images directly witch will affect cacheing | 13:30 |
| NobodyCam | dtantsur: what about v2 of the glance api? | 13:30 |
| GheRivero | https://review.openstack.org/#/c/53189/ | 13:31 |
| dtantsur | NobodyCam, not sure, code for v2 is much harder to read :( | 13:31 |
| NobodyCam | GheRivero: Abandoned :( | 13:32 |
| *** zdin0bot has quit IRC | 13:32 | |
| *** datajerk has joined #openstack-ironic | 13:35 | |
| *** zdin0bot has joined #openstack-ironic | 13:36 | |
| *** matty_dubs|gone is now known as matty_dubs | 13:37 | |
| *** sseago has joined #openstack-ironic | 13:37 | |
| NobodyCam | morning matty_dubs :) | 13:38 |
| openstackgerrit | David Shrewsbury proposed a change to openstack/ironic: Implement instance rebuild in nova.virt.driver https://review.openstack.org/90429 | 13:41 |
| GheRivero | NobodyCam: actually, re-reading the code, we support both glance versions, but it's hardcoded to v1 | 13:48 |
| NobodyCam | ahh | 13:49 |
| *** zdin0bot has quit IRC | 13:50 | |
| dtantsur | NobodyCam, I've asked on #openstack-glance, but got no answer. So I'm inclining to abandon patch about checksums and remove this from blueprint.. | 13:51 |
| *** zdin0bot has joined #openstack-ironic | 13:52 | |
| *** pradipta is now known as pradipta_away | 13:52 | |
| NobodyCam | :) dtantsur we should ask the other *CORES* but I can see that as a logical step | 13:53 |
| jroll | good morning ironic | 13:54 |
| NobodyCam | lucasagomes: rloo: romcheg: | 13:55 |
| *** zdin0bot has quit IRC | 13:55 | |
| NobodyCam | morning jroll :) | 13:55 |
| dtantsur | morning, jroll | 13:55 |
| NobodyCam | jroll: thoughts on ^^^^ | 13:55 |
| lucasagomes | dtantsur, maybe asking eglynn? he's glance-core for red hat | 13:55 |
| lucasagomes | morning jroll | 13:56 |
| jroll | NobodyCam: thoughts on this? 06:30:27 NobodyCam | dtantsur: the end goal I have is for IPA to pull the images directly witch will affect cacheing | 13:56 |
| dtantsur | will try | 13:56 |
| jroll | is your question? | 13:56 |
| jroll | NobodyCam: or the caching things in general? | 13:56 |
| NobodyCam | jroll: on this: 13:51 | dtantsur > NobodyCam, I've asked on #openstack-glance, but got no answer. So I'm inclining to abandon patch about checksums and remove this from blueprint.. | 13:57 |
| jroll | mmm | 13:57 |
| openstackgerrit | Mikhail Durnosvistov proposed a change to openstack/ironic: Unified object model https://review.openstack.org/85065 | 13:57 |
| openstackgerrit | Mikhail Durnosvistov proposed a change to openstack/ironic: Get rid object model `dict` methods part 3 https://review.openstack.org/64108 | 13:57 |
| * jroll looks for blueprint | 13:57 | |
| openstackgerrit | Mikhail Durnosvistov proposed a change to openstack/ironic: Get rid object model `dict` methods part 1 https://review.openstack.org/60025 | 13:57 |
| openstackgerrit | Mikhail Durnosvistov proposed a change to openstack/ironic: Get rid object model `dict` methods part 2 https://review.openstack.org/62331 | 13:57 |
| * romcheg is looking | 13:57 | |
| openstackgerrit | Mikhail Durnosvistov proposed a change to openstack/ironic: Get rid object model `dict` methods part 5 https://review.openstack.org/64278 | 13:57 |
| Mikhail_D_wk | Folks please look at this chain of patches, if you have time :) In general, I added in the tests `mock.patch.object (self.dbapi, [appropriate method]) in order to avoid the return of the resource object, and return Dict-Object :) | 13:57 |
| jroll | NobodyCam: what's the problem with getting a checksum? | 13:57 |
| NobodyCam | jroll: 13:25 | dtantsur > NobodyCam, yes, I've seen it, but the docs are misleading. Judging by source code of v1 API you can only update image content if it does not have one, i.e. is in queued state | 13:58 |
| jroll | oh | 13:58 |
| jroll | what about v2 api? | 13:58 |
| jroll | I feel like someone on our team has said similar | 13:59 |
| NobodyCam | jroll: 13:48 | GheRivero > NobodyCam: actually, re-reading the code, we support both glance versions, but it's hardcoded to v1 | 13:59 |
| jroll | that if the image content needs to change, it will be a new image and thus new uuid | 13:59 |
| jroll | right | 13:59 |
| jroll | but that could change, no? | 13:59 |
| jroll | I don't want to make an assumption based on using v1, if we might use v2 later | 13:59 |
| jroll | i.e. JoshNang's temp urls patch uses glance v2 | 14:00 |
| GheRivero | jroll: yes. it would change, if something like https://review.openstack.org/#/c/53189/ lands someday | 14:00 |
| jroll | https://review.openstack.org/#/c/81391/ | 14:00 |
| jroll | GheRivero: right | 14:00 |
| jroll | does v2 behave the same way as far as updating image content? | 14:01 |
| GheRivero | can't remember. TEsting | 14:01 |
| romcheg | dtantsur, NobodyCam: so the question is basically regarding to usage of the checksum, isn't it? | 14:01 |
| jroll | so if v2 behaves the same way, then not using checksum is probably fine | 14:02 |
| jroll | if v2 allows updates, then we should use checksum | 14:02 |
| jroll | it can't hurt to use checksum anyway, right? | 14:02 |
| dtantsur | romcheg, essentially, question is: when caching, should we expect image in Glance to ever change? | 14:02 |
| romcheg | dtantsur: that's what I mean :) | 14:02 |
| romcheg | To be honest, I never thought that glance allows updating images (sounds weird for me). | 14:03 |
| jroll | you might be right, romcheg | 14:03 |
| *** derekh has joined #openstack-ironic | 14:03 | |
| jroll | I'm pretty sure at rackspace that we never update image content, but not sure if that is a glance limitation or operational thing | 14:03 |
| romcheg | But if that is a concern, why not ask Glance folks? | 14:03 |
| dtantsur | romcheg, I got no answer yet | 14:04 |
| jroll | is the patch currently pulling all glance data for the image? if so, why not use the checksum? | 14:04 |
| romcheg | Hmm, then we should check the API specs while waiting for the response | 14:04 |
| dtantsur | jroll, I'm working on enhanced caching for master images. You can see details here: https://blueprints.launchpad.net/ironic/+spec/pxe-master-images-caching | 14:05 |
| jroll | http://docs.openstack.org/developer/glance/glanceapi.html#update-an-image | 14:05 |
| dtantsur | jroll, my doubts are about the 3rd option | 14:05 |
| jroll | dtantsur: right | 14:05 |
| jroll | I think v2 behaves the same based on that doc I linked | 14:05 |
| dtantsur | jroll, thanks for this link, could not google for it :) that means, I'm dropping suggestion about checksums, right? | 14:06 |
| romcheg | dtantsur: That should update the metadata, not images themself | 14:06 |
| jroll | https://github.com/openstack/glance/blob/master/glance/api/v2/images.py#L112 | 14:07 |
| jroll | https://github.com/openstack/glance/blob/master/glance/api/v2/images.py#L159 | 14:07 |
| jroll | I think v2 behaves the same | 14:07 |
| dtantsur | ok, so we don't expect image body to change and can cache it as long as we want, right? | 14:08 |
| jroll | dtantsur: I think so | 14:08 |
| jroll | dtantsur: I would like to hear that from the glance team, but I think you are right | 14:08 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Fix chassis-node relationship https://review.openstack.org/90964 | 14:08 |
| romcheg | jroll: changes to metadata should not change our cache I think | 14:09 |
| dtantsur | Great, thank you jroll, romcheg, NobodyCam, GheRivero! Now I wonder whether the blueprint itself is necessary, as the other 2 suggestions there are based on existing bugs... | 14:10 |
| jroll | romcheg: right, but I think the same path is taken for image content | 14:10 |
| * jroll pokes around a bit more | 14:10 | |
| jroll | dtantsur: probably not :) | 14:10 |
| romcheg | merge-and-pray :) | 14:10 |
| jroll | :P | 14:12 |
| dtantsur | I would also incredibly thankful, if you guys review the previous patch (one that introduces caching): https://review.openstack.org/#/c/85387/ | 14:14 |
| jroll | how is this for an answer? https://github.com/openstack/glance/blob/3db3b67388bb86af93184c1836414838cf0b449a/glance/tests/unit/v2/test_image_data_resource.py#L201 | 14:16 |
| jroll | :) | 14:16 |
| jroll | they have a v2 api test to make sure an error happens when data already exists | 14:16 |
| dtantsur | jroll, perfect! :) | 14:17 |
| jroll | dtantsur: to be sure, this caching is for the deploy ramdisk, correct? or for other images? | 14:17 |
| jroll | in the patch you mentioned | 14:18 |
| *** dwalleck_ has joined #openstack-ironic | 14:18 | |
| dtantsur | jroll, everything PXE has. | 14:18 |
| jroll | a gigabyte seems small if you are caching images that will be deployed | 14:19 |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic: Implement more robust caching for master images https://review.openstack.org/85387 | 14:19 |
| jroll | but it's probably fine | 14:20 |
| dtantsur | jroll, these number are completely random, I'll change it to whatever seems more reasonable. Anyway, if you _need_ images that don't fit, they will be stored, but deleted right after they're no longer needed | 14:21 |
| dtantsur | does it make sense for you? | 14:21 |
| jroll | right, I just read how the invalidation works | 14:22 |
| jroll | I think it's fine :) | 14:22 |
| *** mdenny has joined #openstack-ironic | 14:25 | |
| *** blamar has joined #openstack-ironic | 14:26 | |
| agordeev | jroll: morning! | 14:33 |
| NobodyCam | any one played with booting from nfs volumes? | 14:35 |
| *** ndipanov has quit IRC | 14:36 | |
| NobodyCam | brb | 14:36 |
| openstackgerrit | Ihar Hrachyshka proposed a change to openstack/ironic: Synced jsonutils from oslo-incubator https://review.openstack.org/91055 | 14:37 |
| jroll | dtantsur: reviewed | 14:39 |
| jroll | agordeev: morning :) | 14:39 |
| agordeev | jroll: it seems in order to get more information about hardware i'm in a need of using `dmidecode`. It has python binding, but hasn't pypi package yet. Do you know if python-dmidecode package is available for coreos? or just could you tell me how check that | 14:40 |
| jroll | coreos... doesn't have packages or a package manager | 14:41 |
| jroll | they expect you to run everything in docker/systemd-nspawn | 14:41 |
| jroll | so we use pip for all packages | 14:41 |
| dtantsur | jroll, many thanks, will fix. Regarding writing to actual files in unit tests: I don't know what the common practice is, but I don't think it hurts to have more real-world test :) | 14:42 |
| jroll | agordeev: if there is no pypi package for dmidecode bindings, I'd prefer to shell out to dmidecode | 14:42 |
| jroll | dtantsur: yeah, I'm just surprised infra lets you write to disk at all :) | 14:42 |
| *** dwalleck__ has joined #openstack-ironic | 14:43 | |
| jroll | dtantsur: I didn't notice what master_dir you're using when writing, but we should probably ensure that is some sort of temp directory | 14:43 |
| *** datajerk has quit IRC | 14:43 | |
| dtantsur | jroll, I still suspect I'm not the first one to write to $TEMP in one of the tests :) | 14:43 |
| dtantsur | yeah, it's from mkdtemp | 14:43 |
| jroll | I guess that should be fine, then | 14:44 |
| *** ndipanov has joined #openstack-ironic | 14:44 | |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic: Implement more robust caching for master images https://review.openstack.org/85387 | 14:45 |
| *** viktors is now known as viktors|afk | 14:45 | |
| dtantsur | jroll, updated ^^^mind looking again? | 14:46 |
| *** dwalleck_ has quit IRC | 14:47 | |
| jroll | looking | 14:47 |
| jroll | and +1'd | 14:48 |
| dtantsur | jroll, thanks :) | 14:49 |
| *** zdin0bot has joined #openstack-ironic | 14:52 | |
| NobodyCam | oh, here a question what if the node had two disks and the user wanted to set one as data vol and the other as log vol.. | 14:53 |
| *** jistr has quit IRC | 14:54 | |
| *** jistr has joined #openstack-ironic | 14:55 | |
| *** derekh has quit IRC | 15:00 | |
| openstackgerrit | Mikhail Durnosvistov proposed a change to openstack/ironic: Old value 'updated_at' field returned after update https://review.openstack.org/75430 | 15:07 |
| *** jbjohnso has quit IRC | 15:07 | |
| *** vkozhukalov has quit IRC | 15:11 | |
| linggao | Hi ironic. seems all the test cases are broken in the master | 15:12 |
| linggao | I got DuplicateOptError: duplicate option: rpc_backend for every test cases. | 15:12 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Remove 'node' parameter from the VendorPassthru interface https://review.openstack.org/90363 | 15:12 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Remove 'node' parameter from the Deploy interface https://review.openstack.org/91062 | 15:12 |
| openstackgerrit | linggao proposed a change to openstack/ironic: Support serial console access https://review.openstack.org/64100 | 15:14 |
| dtantsur | linggao, ifarkas discovered that it's enough to delete all .pyc files | 15:17 |
| ifarkas | that was actually lucasagomes ;-) | 15:18 |
| linggao | dtantsur, ifarkas, thanks. let me try... | 15:18 |
| *** Alexei_987 has quit IRC | 15:21 | |
| *** shakamunyi has joined #openstack-ironic | 15:21 | |
| lucasagomes | :) | 15:22 |
| *** jbjohnso has joined #openstack-ironic | 15:22 | |
| NobodyCam | lucasagomes: got a sec for me to pick your brain? | 15:22 |
| lucasagomes | yeah the oslo.message patch deleted the openstack/common/rpc/ stuff, but we need to clean up the repo to get rid of the pyc | 15:23 |
| lucasagomes | NobodyCam, hey sure | 15:23 |
| NobodyCam | we have pyc in the repo? | 15:23 |
| lucasagomes | NobodyCam, local repo* | 15:23 |
| NobodyCam | ahh | 15:23 |
| NobodyCam | lucasagomes: so my question is on https://github.com/openstack/ironic/blob/master/ironic/common/disk_partitioner.py | 15:24 |
| lucasagomes | NobodyCam, right | 15:24 |
| NobodyCam | can you think of a nice clean way to add multi disk support? ie one hdd for data, onr for logs, one for OS (note os will evolve to nfs vol) | 15:25 |
| lucasagomes | hmm | 15:26 |
| lucasagomes | NobodyCam, I think I need some context... we are going to create partition on these 3 disks right? | 15:27 |
| lucasagomes | cause I like the way DiskPartitioner works with 1 disk, you can create an instance of each for each disk | 15:28 |
| NobodyCam | lucasagomes: yep.. systems have >1 hdd and users are currently setting (example only) sda = OS, sdb = data vol, sdc = log vol | 15:28 |
| lucasagomes | NobodyCam, right... and the deploy ramdisk would expose all 3 via iscsi? | 15:29 |
| NobodyCam | I'm not sure.. but it culd | 15:30 |
| NobodyCam | *could | 15:30 |
| *** jistr has quit IRC | 15:30 | |
| NobodyCam | just working thru some ideas for a conf call later today | 15:30 |
| *** jistr has joined #openstack-ironic | 15:30 | |
| lucasagomes | right, yeah, I don't think that DiskPartitioner should support multiple disk (I dunno any disk partitioning tool that does)... | 15:31 |
| lucasagomes | we better use one insteand of DiskPartitioner for each disk | 15:31 |
| lucasagomes | even why they might want to have different sizes and different partition layout | 15:31 |
| lucasagomes | they might have* | 15:32 |
| NobodyCam | right we we would need to cycle thru th disks setting each one up | 15:32 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Remove 'node' parameter from the Deploy interface https://review.openstack.org/91062 | 15:32 |
| lucasagomes | NobodyCam, yup, I think it's more sane... because even if they have the same disk layout | 15:33 |
| lucasagomes | but they have different sizes | 15:33 |
| lucasagomes | how would the partition create looks like, you specify size but it doesn't fit in one of the disks | 15:34 |
| agordeev | NobodyCam: at least there is a problem with recognizing disks as sda/sdb/etc. Different udev configuration/version, kernel version could totally reassign them. | 15:34 |
| NobodyCam | lucasagomes: I am thinking it would be something like sda: use whole disk as ext4, sdb use whole disk as zfs, sdc whole disk as gfs | 15:35 |
| *** hemna_ has quit IRC | 15:35 | |
| NobodyCam | agordeev: yes | 15:35 |
| *** hemna has joined #openstack-ironic | 15:35 | |
| lucasagomes | NobodyCam, ah right... just 1 parition for the whole disk? | 15:35 |
| NobodyCam | lucasagomes: yep... as I see it now | 15:36 |
| *** foexle has quit IRC | 15:38 | |
| *** jbjohnso has quit IRC | 15:38 | |
| jroll | NobodyCam: I would argue that ironic should just install the OS, and the user should do the rest | 15:39 |
| jroll | partitioning and all | 15:39 |
| agordeev | NobodyCam: so that's the one of the reason to provide more detailed hw info about disks especially. And i'm working on that now | 15:39 |
| *** ndipanov has quit IRC | 15:42 | |
| NobodyCam | jroll: I wold find it tuff to defent that argument | 15:42 |
| openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Implement API to get driver properties https://review.openstack.org/73005 | 15:42 |
| *** ndipanov has joined #openstack-ironic | 15:43 | |
| jroll | NobodyCam: why? ironic is in the business of provisioning a box with an operating system, not defining where data/logs/etc are stored, right? | 15:43 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Remove 'node' parameter from the Deploy interface https://review.openstack.org/91062 | 15:43 |
| romcheg | dtantsur: Regarding to the caching | 15:47 |
| dtantsur | ? | 15:47 |
| romcheg | dtantsur: does clean_up work without locks? | 15:48 |
| romcheg | or I just missed that? | 15:48 |
| dtantsur | romcheg, @lockutils.synchronized('master_image', 'ironic-') | 15:49 |
| romcheg | dtantsur: Oh, sorry.... | 15:50 |
| dtantsur | np :) | 15:50 |
| romcheg | dtantsur: I was looking for more specific locks and missed the big one :) | 15:50 |
| dtantsur | unfortunately, clean_up (and one more small peace inside fetch_image) require more or less global locks | 15:51 |
| dtantsur | have to go for now, my Czech lesson is waiting for me :) see you | 15:53 |
| *** matty_dubs is now known as matty_dubs|lunch | 15:54 | |
| romcheg | Have a nice time dtantsur! | 15:54 |
| *** Manishanker has quit IRC | 15:56 | |
| *** coolsvap|afk is now known as coolsvap | 15:56 | |
| *** eghobo has joined #openstack-ironic | 15:58 | |
| NobodyCam | brb | 16:04 |
| *** derekh has joined #openstack-ironic | 16:08 | |
| *** max_lobur has quit IRC | 16:10 | |
| *** BadCub has joined #openstack-ironic | 16:24 | |
| *** athomas_ has joined #openstack-ironic | 16:29 | |
| *** athomas has quit IRC | 16:30 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Remove 'node' parameter from the Power interface https://review.openstack.org/91093 | 16:31 |
| *** GheRivero has quit IRC | 16:31 | |
| *** jbjohnso has joined #openstack-ironic | 16:33 | |
| lucasagomes | NobodyCam, when u get some time: https://review.openstack.org/#/c/86063/ | 16:34 |
| *** athomas_ has quit IRC | 16:35 | |
| NobodyCam | lucasagomes: ack on conf call now | 16:35 |
| *** matty_dubs|lunch is now known as matty_dubs | 16:36 | |
| lucasagomes | NobodyCam, no hurry :) | 16:36 |
| *** ndipanov has quit IRC | 16:38 | |
| *** romcheg1 has joined #openstack-ironic | 16:39 | |
| *** athomas has joined #openstack-ironic | 16:40 | |
| *** martyntaylor has left #openstack-ironic | 16:40 | |
| *** zdin0bot has quit IRC | 16:43 | |
| *** ndipanov has joined #openstack-ironic | 16:44 | |
| *** jbjohnso has quit IRC | 16:44 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Remove 'node' parameter from the Console and Rescue interface https://review.openstack.org/91097 | 16:48 |
| *** GheRivero has joined #openstack-ironic | 16:48 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Remove 'node' parameter from the Power interface https://review.openstack.org/91093 | 16:49 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Remove 'node' parameter from the Console and Rescue interface https://review.openstack.org/91097 | 16:49 |
| rloo | if anyone has some time to do a review, I'd really like to get this merged. For selfish reasons. I don't think I can look at another revision of it: https://review.openstack.org/#/c/60025/ | 16:50 |
| *** Mikhail_D_ltp has joined #openstack-ironic | 16:52 | |
| lucasagomes | rloo, this also https://review.openstack.org/#/c/73223/ | 16:52 |
| *** zdin0bot has joined #openstack-ironic | 16:52 | |
| rloo | lucasagomes: thx, i'll take a look at that in a few minutes. | 16:52 |
| *** martyntaylor has joined #openstack-ironic | 16:56 | |
| *** krtaylor has quit IRC | 16:56 | |
| *** vkozhukalov has joined #openstack-ironic | 16:58 | |
| *** harlowja_away is now known as harlowja | 16:59 | |
| *** romcheg1 is now known as romcheg_ltp | 16:59 | |
| *** derekh has quit IRC | 17:02 | |
| rloo | lucasagomes, Mikhail_D_ltp: wrt https://review.openstack.org/#/c/73223/, why do we need to get rid of the 'with' when using mocks? | 17:08 |
| lucasagomes | rloo, hmm cosmetic change | 17:09 |
| *** newell_ has joined #openstack-ironic | 17:09 | |
| rloo | lucasagomes: ok, the commit should be clearer about that. It says 'need'. I'll look at it anyway ;) | 17:09 |
| lucasagomes | rloo, heh... right I don't think it's something we _need_ | 17:10 |
| lucasagomes | looks better | 17:10 |
| rloo | lucasagomes: yeah, looks better and the fewer indented lines of code, the better! :-) | 17:10 |
| rloo | levels of indentation I mean ;) | 17:10 |
| lucasagomes | yup | 17:10 |
| *** newell_ has quit IRC | 17:10 | |
| *** martyntaylor has quit IRC | 17:14 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Remove 'node' parameter from the Deploy interface https://review.openstack.org/91062 | 17:15 |
| *** eghobo has quit IRC | 17:15 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Remove 'node' parameter from the Console and Rescue interfaces https://review.openstack.org/91097 | 17:16 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Remove 'node' parameter from the Power interface https://review.openstack.org/91093 | 17:16 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Remove 'node' parameter from the validate() methods https://review.openstack.org/91104 | 17:16 |
| lucasagomes | uu done heh | 17:17 |
| *** jistr has quit IRC | 17:21 | |
| *** notq has joined #openstack-ironic | 17:22 | |
| *** eghobo has joined #openstack-ironic | 17:22 | |
| lucasagomes | /nick lucas-dinner | 17:30 |
| *** lucasagomes is now known as lucas-dinner | 17:30 | |
| lucas-dinner | ops | 17:30 |
| lucas-dinner | heh have a good night everybody! | 17:30 |
| *** openstackgerrit has quit IRC | 17:32 | |
| *** openstackgerrit has joined #openstack-ironic | 17:33 | |
| *** martyntaylor has joined #openstack-ironic | 17:37 | |
| rloo | 'night lucas-dinner! | 17:37 |
| *** zdin0bot has quit IRC | 17:38 | |
| *** max_lobur has joined #openstack-ironic | 17:39 | |
| NobodyCam | brb | 17:39 |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic: Adding swift temp url support https://review.openstack.org/81391 | 17:41 |
| *** zdin0bot has joined #openstack-ironic | 17:43 | |
| *** zdin0bot has quit IRC | 17:49 | |
| * NobodyCam attempts to eat a bagel before the next conf call | 17:51 | |
| *** martyntaylor has quit IRC | 17:52 | |
| *** martyntaylor1 has joined #openstack-ironic | 17:52 | |
| *** jgrimm has quit IRC | 18:06 | |
| *** BadCub has quit IRC | 18:06 | |
| *** BadCub01 has joined #openstack-ironic | 18:13 | |
| *** Nisha has joined #openstack-ironic | 18:22 | |
| *** newell_ has joined #openstack-ironic | 18:23 | |
| *** jgrimm has joined #openstack-ironic | 18:24 | |
| Nisha | @Deva, Is there any plan for Gui remote console for end user in ironic | 18:25 |
| *** coolsvap is now known as coolsvap|afk | 18:30 | |
| *** Nisha_ has joined #openstack-ironic | 18:35 | |
| matty_dubs | Nisha_: I'm not sure that anyone is working on that presently; right now it's just serial console (via IPMI serial-over-LAN) | 18:40 |
| matty_dubs | I think it'd have to be a vendor-specific thing | 18:41 |
| devananda | Nisha: "end user" -- no. Ironic is (today) admin-only. | 18:42 |
| devananda | Nisha: exporting console log to provide via the Nova API -- i think that'd be good, but no one is working on it yet | 18:42 |
| *** Mikhail_D_ltp has left #openstack-ironic | 18:43 | |
| Nisha | Deva: Ok, So if we say the console is for admin only, then admin has the privilege for all the operations on that server | 18:44 |
| devananda | Nisha: admin has all privileges in ironic today already | 18:44 |
| Nisha | Deva: Then do we want to display GUI remote console via ironic? | 18:44 |
| devananda | Nisha: what do you mean GUI remote console? | 18:45 |
| devananda | Nisha: SOL? | 18:45 |
| Nisha | Deva: Like ILO provides extra capabilities in GUI remote console | 18:45 |
| devananda | no | 18:45 |
| Nisha | Deva: No, SOL will be text console only correct | 18:45 |
| devananda | ironic isn't a proxy for all the mgmt extra goodies -- it's a provisioning tool | 18:45 |
| devananda | sometimes provisioning fails and you need to collect the kernel log | 18:46 |
| devananda | hence we need the SOL access | 18:46 |
| devananda | but that's where it ends IMHO | 18:46 |
| Nisha | Sorry i didnt understand last statement | 18:46 |
| devananda | proxying the OOB MGMT GUI is not relevant | 18:46 |
| devananda | to a cloud deployment on baremetal | 18:47 |
| *** iron1 has joined #openstack-ironic | 18:47 | |
| Nisha | Ok. | 18:47 |
| Nisha | And for Windows deployed servers? | 18:47 |
| devananda | what about them? | 18:48 |
| Nisha | where you can see the the graphics only in GUI console | 18:48 |
| Nisha | For example | 18:48 |
| devananda | wait - that's got nothing to do with mgmt | 18:48 |
| devananda | once you've deployed a windows instance, remote-desktop is just the end-user connecting to their instance | 18:49 |
| devananda | assuming networking is set up properly (eg by neutron) that has nothing to do with ironic, the OOB management card, iLO, etc | 18:49 |
| Nisha | yes correct, but you can still want to login to the console for some reason? | 18:49 |
| *** Haomeng has quit IRC | 18:51 | |
| NobodyCam | brb | 18:52 |
| devananda | Nisha: who would log in? and why? | 18:53 |
| lifeless | Nisha: so what I think you are saying is that ILO is for more than just DC ops and cloud infra ops folk. | 18:55 |
| lifeless | Nisha: that it is for the end user as well ? | 18:56 |
| Nisha | Admin? say if the OS stops responding for some reason or some other issue on Windows based systems? If there is a scenario admin needs the GUI console of the deployed server(mostly it would be required for windows,if required) , then how does ironic handles it | 18:56 |
| Nisha | DC ops? | 18:57 |
| lifeless | Nisha: which admin - the instance admin or the cloud infrastructure admin or the dc ops (datacentre operators) admin | 18:57 |
| Nisha | the cloud infrastructure admin and may be even dc admin | 18:58 |
| lifeless | Nisha: with a baremetal cloud we have the concept of three different 'admins' - the folk that put the hardware in the rack and fix physical issues, the folk that manage the cloud and allocate resources to regions, deal with nova/ironic issues and the like, and then the user who requested the instance via Nova, whose data is on the machine. | 18:58 |
| Nisha | Yes, so end user shall not have admin privilege in the remote console except for logs | 18:58 |
| devananda | sorry to run ... will catch up when i get back in an hour ... | 18:59 |
| Nisha | the cloud infrastructure admin and may be even dc admin? | 19:00 |
| *** derekh has joined #openstack-ironic | 19:00 | |
| Nisha | provisioning is done by which user here? | 19:00 |
| notq | nisha, in vmware with windows. you have a major error, you go to the console and fix. you're asking, how do you do that with baremetal? you go to the imm and check the console. | 19:01 |
| Nisha | imm? | 19:01 |
| Nisha | check the console means sol? | 19:02 |
| notq | yes | 19:02 |
| *** dwalleck_ has joined #openstack-ironic | 19:03 | |
| *** dwalleck__ has quit IRC | 19:03 | |
| Nisha | could you expand imm, sorry am not aware of it | 19:04 |
| notq | sorry, i should have said ipmi | 19:05 |
| Nisha | oh | 19:05 |
| notq | imm is ibm's management for servers. i'm using ibm a lot lately | 19:05 |
| Nisha | yes i am aware of ipmi sol | 19:05 |
| *** dwalleck__ has joined #openstack-ironic | 19:06 | |
| *** jistr has joined #openstack-ironic | 19:06 | |
| Nisha | but my question is when we have a baremetal, say proliant servers then the text console(SOL) after the OS is deployed gives the statement like "in graphical mode or unsupported text mode" | 19:07 |
| Nisha | at the same time if i open the GUI console on the ilo, i can see the graphical display of OS running there | 19:07 |
| Nisha | this is not seen in text mode or SOL | 19:08 |
| *** dwalleck_ has quit IRC | 19:09 | |
| Nisha | as Deva said we dont want extra managemnet goodies by console, those can be avoided to be given to the admin when he tries to open GUI console | 19:10 |
| NobodyCam | brb again | 19:13 |
| *** ifarkas has quit IRC | 19:23 | |
| *** max_lobur has quit IRC | 19:25 | |
| Shrews | JoshNang: _wait_for_active *is* factored out | 19:32 |
| JoshNang | Shrews: hmm, didn't look like it. must have missed it. sorry!! | 19:38 |
| *** BadCub01 has quit IRC | 19:38 | |
| Shrews | no worries | 19:38 |
| *** vkozhukalov has quit IRC | 19:50 | |
| Nisha_ | Deva, lifeless: IMHO, GUI console will be useful for admin and easy to use | 19:50 |
| *** ndipanov is now known as ndipanov_gone\ | 19:51 | |
| *** jbjohnso has joined #openstack-ironic | 19:51 | |
| *** ndipanov_gone\ is now known as ndipanov_gone | 19:51 | |
| *** krtaylor has joined #openstack-ironic | 19:58 | |
| *** dwalleck__ has quit IRC | 20:00 | |
| openstackgerrit | Jarrod Johnson proposed a change to stackforge/pyghmi: Correct mistakes in constants https://review.openstack.org/91152 | 20:00 |
| *** jistr has quit IRC | 20:00 | |
| openstackgerrit | A change was merged to stackforge/pyghmi: Correct sensor offset for byte 5 state values https://review.openstack.org/90778 | 20:02 |
| *** derekh has quit IRC | 20:05 | |
| openstackgerrit | A change was merged to stackforge/pyghmi: Correct mistakes in constants https://review.openstack.org/91152 | 20:10 |
| *** Nisha_ has quit IRC | 20:14 | |
| *** Nisha has quit IRC | 20:14 | |
| *** wendar has quit IRC | 20:24 | |
| *** wendar has joined #openstack-ironic | 20:24 | |
| *** zdin0bot has joined #openstack-ironic | 20:25 | |
| Shrews | wow. deploying a vm with ironic+devstack is MUCH faster on my spare laptop compared to a cloud vm. just under 2 minutes compared to the 15 or so in the vm | 20:26 |
| *** zdin0bot has quit IRC | 20:26 | |
| devananda | Shrews: nested virt is terrible. | 20:26 |
| *** zdin0bot has joined #openstack-ironic | 20:26 | |
| Shrews | devananda: indeed. very glad i took the time to get this working | 20:27 |
| devananda | Shrews: also, it takes only a few min for me with nested virt. (not counting time to start devstack) | 20:27 |
| Shrews | devananda: hrm, perhaps your vm has more juice than mine | 20:28 |
| devananda | Shrews: 2 core 4GB RAM | 20:28 |
| NobodyCam | Shrews: ++ on spare laptop | 20:29 |
| Shrews | devananda: hrm, mine is 4 core, 8G. *shrug* | 20:30 |
| *** zdin0bot has quit IRC | 20:30 | |
| Shrews | NobodyCam: and I don't even have to run mine upside down! :) | 20:30 |
| openstackgerrit | Jarrod Johnson proposed a change to stackforge/pyghmi: Clean up command.py https://review.openstack.org/91168 | 20:32 |
| NobodyCam | :-p sure rub it in | 20:32 |
| devananda | adam_g: any thoughts on how the bump from 512 to 1024 RAM will affect the devstack-gate environment? | 20:32 |
| adam_g | devananda, we should be fine. those get run on 8GB nodes | 20:32 |
| devananda | right | 20:32 |
| devananda | adam_g: and when we need to run >1 instance at a time? | 20:32 |
| devananda | i'm assuming we'll need tempest to do that at some point | 20:32 |
| devananda | but IMBW | 20:33 |
| adam_g | devananda, we should be okay to scale to 1 or 2 more, but anything beyond that is probably stretching it | 20:33 |
| devananda | k, if we can do 3, i think that's fine | 20:33 |
| adam_g | devananda, not sure we have a choice tho | 20:33 |
| *** zdin0bot has joined #openstack-ironic | 20:34 | |
| devananda | k | 20:34 |
| devananda | adam_g: "DIB produced ramdisks tend to be ~250MB" -- huh? | 20:35 |
| devananda | adam_g: last i saw they were ~50 | 20:35 |
| adam_g | devananda, uncompressed | 20:35 |
| devananda | ah | 20:35 |
| adam_g | devananda, if we really need to scale devstack gate to many VMs we may be able to get creative with the initrams + kernel parameters | 20:36 |
| adam_g | *initramfs | 20:36 |
| devananda | or we request larger instances for this job | 20:36 |
| adam_g | or that, yeah | 20:37 |
| devananda | k | 20:37 |
| openstackgerrit | linggao proposed a change to openstack/ironic: Support serial console access https://review.openstack.org/64100 | 20:37 |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic: Adding swift temp url support https://review.openstack.org/81391 | 20:42 |
| openstackgerrit | A change was merged to stackforge/pyghmi: Clean up command.py https://review.openstack.org/91168 | 20:47 |
| *** zdin0bot has quit IRC | 20:49 | |
| *** zdin0bot has joined #openstack-ironic | 20:52 | |
| *** zdin0bot1 has joined #openstack-ironic | 20:54 | |
| *** zdin0bot1 has quit IRC | 20:56 | |
| *** zdin0bot1 has joined #openstack-ironic | 20:56 | |
| *** zdin0bot has quit IRC | 20:56 | |
| *** jdob has quit IRC | 21:00 | |
| *** linggao has quit IRC | 21:09 | |
| openstackgerrit | Adam Gandelman proposed a change to openstack/ironic: Wait for Neutron port updates when using SSHPower https://review.openstack.org/91174 | 21:10 |
| *** romcheg_ltp has quit IRC | 21:10 | |
| *** jbjohnso has quit IRC | 21:10 | |
| *** romcheg1 has joined #openstack-ironic | 21:10 | |
| *** newell_ has quit IRC | 21:11 | |
| adam_g | lifeless, ^ | 21:11 |
| lifeless | cool | 21:11 |
| adam_g | devananda, should i file a BP to get discussion started on a proper fix for that? eg, API callbacks similar to os-external-events in nova/neutron currently | 21:14 |
| lifeless | webhooks FTW | 21:14 |
| devananda | adam_g: that'd be great | 21:15 |
| devananda | adam_g: also, i think you want to make those changes in conductor.manager, not drivers.modules.pxe | 21:15 |
| adam_g | hmm. ill take a look | 21:17 |
| devananda | adam_g: well. i dont know if you /can/ right now -- the sleep has to happen in the middle of pxe.deploy | 21:17 |
| devananda | adam_g: but conceptually, one driver shouldn't be depending on another like this | 21:17 |
| adam_g | devananda, yeah, and the relevant neutron updates are only relevant to the pxe driver, tho IMBW | 21:18 |
| lifeless | well deploy does need to know if power has completed its thing | 21:18 |
| *** krtaylor has quit IRC | 21:18 | |
| devananda | adam_g: perhaps factoring the _update_neutron out of pxe is the right thing, since the agent driver will need it too | 21:18 |
| devananda | JayF: you were working on factoring the tftp / file manipulation stuff out of pxe, right? | 21:19 |
| adam_g | devananda, oh right. i seem to remember a patch up that does similar | 21:19 |
| devananda | JayF: were you also looking at the _update_neutron method? | 21:19 |
| lifeless | actually no | 21:19 |
| lifeless | the thing here isn't knowing the power has completed | 21:19 |
| devananda | lifeless: right | 21:19 |
| lifeless | its knowing the minimum latency power has | 21:19 |
| devananda | lifeless: deploy driver knowswhen power has completed | 21:19 |
| lifeless | and the maximum latency neutron has | 21:19 |
| devananda | lifeless: deploy needs to know when network update has completed (or guess how long it takes to propagate until we have a callback hook) | 21:20 |
| lifeless | right | 21:20 |
| JoshNang | devananda: that's me working on the factoring | 21:21 |
| devananda | adam_g: i think moving update_neutron from drivers.modules.pxe to conductor.utils would solve it | 21:21 |
| devananda | JoshNang: ah! sorry :) | 21:21 |
| JoshNang | heh no worries. :) | 21:21 |
| devananda | JoshNang: so ... ^ ? | 21:21 |
| JoshNang | so i did move update_neutron | 21:22 |
| lifeless | slemax(0, power latency - neutron latency) | 21:23 |
| JoshNang | (but to a new tftp.py, rather than conductor utils) | 21:23 |
| *** newell_ has joined #openstack-ironic | 21:23 | |
| JoshNang | as for needing to know when update_neutron is complete, we need that too in the agent driver, but for switching from a provisioning network to the open network | 21:24 |
| JoshNang | so its on my "need to figure this out very soon" list | 21:24 |
| *** matty_dubs is now known as matty_dubs|gone | 21:26 | |
| adam_g | hmm. sounds like we're in the same boat | 21:28 |
| *** romcheg1 has quit IRC | 21:28 | |
| adam_g | devananda, i'm curious why you think conductor.utils should meddling with neutron details? | 21:29 |
| *** BadCub has joined #openstack-ironic | 21:30 | |
| JoshNang | we also need to do a reboot give or take the same time we're switching networks, right before the box is ready to go | 21:31 |
| JoshNang | adam_g: what are you looking to do? | 21:31 |
| adam_g | JoshNang, we need to know that neutron updates have completed on the agents before booting | 21:33 |
| devananda | adam_g: first thought - because one driver interface shouldnt be importing another driver interface | 21:33 |
| adam_g | devananda, yeah, that seems awkward | 21:33 |
| devananda | adam_g: and the conductor is, broadly speaking, making the decisions about what steps to do when. ... but on second thought, the conductor isn't the only common ground there. | 21:34 |
| devananda | adam_g: lemme add a few comments to the review, see what else you come up with | 21:34 |
| adam_g | JoshNang, nova had to solve almost an exact issue during icehouse.. they added os-external-events API and client support, neutron just uses that to callback when certain events complete | 21:34 |
| adam_g | devananda, cool, thanks | 21:35 |
| JoshNang | adam_g: that would be fantastic | 21:35 |
| adam_g | JoshNang, https://wiki.openstack.org/wiki/Nova/ExternalEventAPI | 21:35 |
| JoshNang | adam_g: ok that makes sense. how many changes on the neutron side would need to be made to make this work? (realize that's a big question) | 21:38 |
| JoshNang | or is everything needed already there for nova? | 21:38 |
| adam_g | JoshNang, https://git.openstack.org/cgit/openstack/neutron/tree/neutron/notifiers looks relatively extensible | 21:40 |
| NobodyCam | lucas-dinner: sorry it took me so long to look at 86063, I gave it a -1 only for the log line format other wise LGTM! | 21:40 |
| adam_g | JoshNang, the nova changes were more, one sec let me find that blueprint. i wonder what we can re-use | 21:41 |
| NobodyCam | adam_g: awesome Ty :) | 21:41 |
| adam_g | JoshNang, https://blueprints.launchpad.net/nova/+spec/admin-event-callback-api | 21:42 |
| adam_g | JoshNang, i need to step out for a bit but id like to start collecting use cases for something similar in ironic and get a BP filed today/tomrrw | 21:42 |
| lucas-dinner | NobodyCam, hey will take a look | 21:42 |
| JoshNang | adam_g: awesome. i'd love to discuss this more | 21:43 |
| lucas-dinner | NobodyCam, hmm I can't do that, using ',' is only for the logging module | 21:43 |
| lucas-dinner | lemme exemplify | 21:43 |
| rloo | NobodyCam: was just going to mention -- it doesn't matter for the exception since you want it to interpolate. | 21:43 |
| rloo | NobodyCam: the diff with logging, is that you only want it done if it is going to be logged, and that depends on the log level first. | 21:44 |
| lucas-dinner | NobodyCam, http://paste.openstack.org/show/77813/ | 21:44 |
| lucas-dinner | see [3] | 21:44 |
| rloo | NobodyCam: which is why you don't always want it interpolated, cuz it might not be needed. | 21:44 |
| rloo | NobodyCam: I also question whether that is worth a -1, or a good-to-go and please fix later ;) | 21:45 |
| lucas-dinner | rloo, using ', ' just works for the logging module | 21:45 |
| lucas-dinner | so can't fix later :) | 21:46 |
| rloo | lucas-dinner: yes, that too ;) | 21:46 |
| devananda | comstud: does the object code have a means to get the old value for a changed field? | 21:46 |
| rloo | lucas-dinner: but if it could be fixed later -- should we -1 if there is eg 1 minor thing, or just approve and fix later? | 21:46 |
| lucas-dinner | rloo, hmm I depends really, if it's really minor I would -1, I would say something like "in case you have to submit a new patch-set please use this insteand of that" | 21:48 |
| rloo | lucas-dinner: but when you were doing those review jams before, it seems like you (cores) approved some, with a 'please fix xyz later'? | 21:48 |
| lucas-dinner | rloo, oh yeah... on the review jam we usually approve and then submit a following patch | 21:48 |
| lucas-dinner | that's correct | 21:49 |
| lucas-dinner | I wouldn't -1! | 21:49 |
| lucas-dinner | on my comment there | 21:49 |
| lucas-dinner | not would | 21:49 |
| lucas-dinner | s/minor I would/minor I wouldn't/g | 21:49 |
| NobodyCam | I only -1'd as I have been i il | 21:50 |
| NobodyCam | grrrr | 21:50 |
| NobodyCam | sorry have been unable to reply | 21:50 |
| rloo | lucas-dinner: just want to make sure I am consistent with the rest of the cores ;) | 21:50 |
| NobodyCam | lucas-dinner: I -1 as I have been called out several times on that issue | 21:50 |
| NobodyCam | With you example I happy to +2 | 21:51 |
| lucas-dinner | NobodyCam, ah :( np... cheers | 21:51 |
| NobodyCam | (and agreed -1 was to harsh) :( | 21:51 |
| lucas-dinner | NobodyCam, heh it's grand really | 21:51 |
| lucas-dinner | rloo, :) | 21:51 |
| * rloo thinks NobodyCam is really a nice guy. | 21:52 | |
| * NobodyCam *(BLUSH)* | 21:52 | |
| NobodyCam | lucas-dinner: on its way down to trunk land | 21:54 |
| lucas-dinner | NobodyCam, yay!!! | 21:54 |
| lucas-dinner | NobodyCam, thank u! | 21:54 |
| NobodyCam | no no thank you | 21:54 |
| lucas-dinner | :) | 21:55 |
| comstud | devananda: Not a built-in way | 21:55 |
| devananda | comstud: k | 21:55 |
| devananda | comstud: i think i can work around that | 21:55 |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic-python-agent: Fix broken path to shell scripts https://review.openstack.org/91180 | 21:56 |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic-python-agent: Fix broken path to shell scripts https://review.openstack.org/91180 | 21:56 |
| comstud | devananda: We did add something for system_metadata or something in Instance in nova | 21:58 |
| comstud | devananda: but basically so we could tell when it changed... compared old dict with new one | 21:58 |
| comstud | but yeah | 21:58 |
| NobodyCam | bbiab | 22:04 |
| *** datajerk has joined #openstack-ironic | 22:12 | |
| openstackgerrit | A change was merged to openstack/ironic-python-agent: Fix broken path to shell scripts https://review.openstack.org/91180 | 22:13 |
| *** mrda_away is now known as mrda | 22:14 | |
| *** jbjohnso has joined #openstack-ironic | 22:20 | |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic: Adding swift temp url support https://review.openstack.org/81391 | 22:28 |
| *** zdin0bot1 has quit IRC | 22:29 | |
| *** zigo_ has joined #openstack-ironic | 22:35 | |
| *** zigo has quit IRC | 22:37 | |
| *** greghaynes has quit IRC | 22:37 | |
| *** jgrimm has quit IRC | 22:37 | |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Don't update neutron VIF if MAC is duplicated https://review.openstack.org/90436 | 22:37 |
| *** _dguerri has quit IRC | 22:37 | |
| *** zdin0bot has joined #openstack-ironic | 22:37 | |
| *** greghaynes_ has joined #openstack-ironic | 22:37 | |
| *** Shrews has quit IRC | 22:37 | |
| *** zdin0bot1 has joined #openstack-ironic | 22:39 | |
| *** lucas-dinner has quit IRC | 22:39 | |
| *** BadCub has quit IRC | 22:39 | |
| *** zdin0bot has quit IRC | 22:40 | |
| jroll | what's up with the firewall driver calls in the virt driver? | 22:45 |
| *** datajerk has quit IRC | 22:47 | |
| *** greghaynes_ is now known as greghaynes | 22:59 | |
| *** radsy has joined #openstack-ironic | 23:01 | |
| *** radsy has joined #openstack-ironic | 23:01 | |
| *** zdin0bot1 has quit IRC | 23:03 | |
| *** zdin0bot has joined #openstack-ironic | 23:03 | |
| *** jbjohnso has quit IRC | 23:06 | |
| *** derekh has joined #openstack-ironic | 23:10 | |
| *** shakamunyi has quit IRC | 23:14 | |
| *** shakamunyi has joined #openstack-ironic | 23:19 | |
| *** matsuhashi has joined #openstack-ironic | 23:38 | |
| *** rloo has quit IRC | 23:39 | |
| *** zdin0bot has quit IRC | 23:41 | |
| *** zdin0bot has joined #openstack-ironic | 23:42 | |
| *** eguz has joined #openstack-ironic | 23:46 | |
| *** iron1 has quit IRC | 23:47 | |
| *** eghobo has quit IRC | 23:50 | |
| *** dguerri has joined #openstack-ironic | 23:57 | |
| *** datajerk has joined #openstack-ironic | 23:58 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!