| *** Haomeng has joined #openstack-ironic | 00:04 | |
| devananda | adam_g: think i found the cause of the random test failures -- https://bugs.launchpad.net/ironic/+bug/1321494 | 00:06 |
|---|---|---|
| *** matsuhashi has joined #openstack-ironic | 00:07 | |
| devananda | we're logging ERROR when it's not really an error | 00:07 |
| * devananda really goes afk now | 00:07 | |
| adam_g | devananda, doh! | 00:07 |
| *** romcheg has joined #openstack-ironic | 00:15 | |
| *** rloo has quit IRC | 00:20 | |
| *** romcheg has left #openstack-ironic | 00:24 | |
| *** eguz has quit IRC | 00:40 | |
| *** eghobo has joined #openstack-ironic | 00:41 | |
| *** uberj has quit IRC | 00:51 | |
| *** uberj has joined #openstack-ironic | 00:51 | |
| *** eghobo has quit IRC | 01:15 | |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic: Adding a reference driver for the agent https://review.openstack.org/84795 | 01:18 |
| *** shakamunyi has quit IRC | 01:45 | |
| *** lazy_prince has joined #openstack-ironic | 01:56 | |
| *** lazy_prince is now known as killer_prince | 01:59 | |
| *** shakamunyi has joined #openstack-ironic | 01:59 | |
| *** shakamunyi has quit IRC | 02:05 | |
| *** rwsu has quit IRC | 02:44 | |
| *** jgrimm has joined #openstack-ironic | 02:46 | |
| *** hipster has quit IRC | 02:56 | |
| *** harlowja is now known as harlowja_away | 03:05 | |
| openstackgerrit | A change was merged to openstack/ironic: Fix bad Mock calls to assert_called_once() https://review.openstack.org/93835 | 03:11 |
| *** killer_prince has quit IRC | 03:34 | |
| *** eghobo has joined #openstack-ironic | 03:45 | |
| *** nosnos has joined #openstack-ironic | 03:48 | |
| *** jgrimm has quit IRC | 04:56 | |
| *** igor_ has quit IRC | 05:02 | |
| *** igor has joined #openstack-ironic | 05:06 | |
| *** radsy has quit IRC | 05:08 | |
| *** greghaynes has quit IRC | 05:50 | |
| *** datajerk has joined #openstack-ironic | 06:03 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic: Imported Translations from Transifex https://review.openstack.org/88508 | 06:03 |
| GheRivero | morning all | 06:30 |
| mrda | Hi GheRivero | 06:31 |
| *** jcoufal has joined #openstack-ironic | 07:05 | |
| *** harlowja_away has quit IRC | 07:10 | |
| *** eghobo has quit IRC | 07:12 | |
| *** vkdrao has joined #openstack-ironic | 07:34 | |
| *** vkdrao has quit IRC | 07:36 | |
| *** vkdrao has joined #openstack-ironic | 07:36 | |
| *** lsmola has quit IRC | 07:39 | |
| openstackgerrit | Mikhail Durnosvistov proposed a change to openstack/ironic: Using system call for downloading files https://review.openstack.org/92560 | 07:39 |
| *** lsmola has joined #openstack-ironic | 07:39 | |
| *** ifarkas has joined #openstack-ironic | 07:43 | |
| *** killer_prince has joined #openstack-ironic | 07:47 | |
| *** tatyana has joined #openstack-ironic | 07:48 | |
| *** ifarkas has quit IRC | 07:49 | |
| *** jistr has joined #openstack-ironic | 07:51 | |
| *** tatyana has quit IRC | 07:52 | |
| *** ifarkas has joined #openstack-ironic | 07:53 | |
| *** pelix has joined #openstack-ironic | 07:54 | |
| Mikhail_D_wk | Morning all! :) | 08:08 |
| *** mrda is now known as mrda_away | 08:12 | |
| *** derekh_ has joined #openstack-ironic | 08:14 | |
| *** martyntaylor has joined #openstack-ironic | 08:20 | |
| *** NobodyCam has quit IRC | 08:26 | |
| *** NobodyCam has joined #openstack-ironic | 08:28 | |
| openstackgerrit | Mikhail Durnosvistov proposed a change to openstack/ironic: Cleanup mock patch without `with` part 2 https://review.openstack.org/73256 | 08:29 |
| openstackgerrit | Mikhail Durnosvistov proposed a change to openstack/ironic: Cleanup mock patch without `with` part 3 https://review.openstack.org/86536 | 08:29 |
| openstackgerrit | Mikhail Durnosvistov proposed a change to openstack/ironic: Get rid of the newline "\" https://review.openstack.org/66793 | 08:29 |
| *** athomas has joined #openstack-ironic | 08:29 | |
| *** jistr has quit IRC | 08:34 | |
| *** jistr has joined #openstack-ironic | 08:34 | |
| *** lucasagomes has joined #openstack-ironic | 08:35 | |
| sirushti | lifeless, when you get time, please could you review 73054 and clarify if I've made the changes as you suggested. Thanks. | 08:39 |
| *** rameshg87 has joined #openstack-ironic | 09:03 | |
| *** jcoufal has quit IRC | 09:10 | |
| *** killer_prince has quit IRC | 09:26 | |
| *** killer_prince has joined #openstack-ironic | 09:29 | |
| *** vkdrao has quit IRC | 09:32 | |
| *** vkdrao has joined #openstack-ironic | 09:36 | |
| *** geertj has joined #openstack-ironic | 09:51 | |
| *** matsuhashi has quit IRC | 10:01 | |
| geertj | Good morning Ironic. | 10:10 |
| *** coolsvap|afk is now known as coolsvap | 10:11 | |
| *** vkdrao has quit IRC | 10:11 | |
| geertj | I have a question: how does the final Glance image get deployed to an Ironic bare-metal node? So far I understand that the special deploy kernel and ramdisk are booted via PXE. But How does the deploy initrd get the "real" image onto the server? | 10:12 |
| *** jcoufal has joined #openstack-ironic | 10:15 | |
| lucasagomes | geertj, so the way it works for the default PXE driver _right now_ is... The baremetal metal machine boots the deploy ramdisk and kernel | 10:20 |
| lucasagomes | after the boot the ramdisk will mount the local disk of the machine as an iSCSI driver and send the iqn to the ironic api | 10:20 |
| lucasagomes | so the ironic service will mount that iSCSI target and copy the image onto it (dd'ng) | 10:20 |
| lucasagomes | that's how the image goes to the baremetal disk | 10:21 |
| geertj | lucasgomes: thanks, that makes sense | 10:21 |
| geertj | lucasgomes: you emphasized _right now_. Does this means there's plans to change/ | 10:22 |
| geertj | ? | 10:22 |
| lucasagomes | geertj, yes, the rackspace guys are working on the python-deploy-agent | 10:22 |
| *** athomas has quit IRC | 10:22 | |
| lucasagomes | which fetchs it directly from glance in the baremetal macine | 10:22 |
| lucasagomes | geertj, but there's other people that wants this image to be distribute in diff (more scalable ways) like multicast or bittorrent | 10:23 |
| lucasagomes | in diff ways* | 10:23 |
| geertj | lucasgomes, I'm trying to create a power + deploy driver for Ravello | 10:23 |
| geertj | I'm wondering what the best way is to do it. | 10:23 |
| geertj | I could upload the Glance image into Ravello, and then the bare metal VM can boot directly from it. | 10:24 |
| geertj | Or I could try with the iSCSI like it is now. | 10:24 |
| geertj | any thoughts? | 10:25 |
| lucasagomes | geertj, I personally like the glance approach, you might want to take a peek at what the rackspace guys are doing (https://github.com/openstack/ironic-python-agent) | 10:26 |
| rameshg87 | geertj: what do you mean by bare metal VM here ? | 10:26 |
| lucasagomes | yeah same q ^ | 10:27 |
| geertj | rameshg87: The VMs in Ravello support VT and have bare layer-2 network access. So for all practical purposes they behave like bare metal nodes which can server as an undercloud. | 10:27 |
| *** datajerk has quit IRC | 10:27 | |
| geertj | You can run KVM on it e.g, or use VLANs for Neutron. | 10:27 |
| geertj | see www.ravellosystems.com/how-it-works | 10:28 |
| geertj | and also /technology on how we do it.. | 10:29 |
| lucasagomes | cool will take a look | 10:30 |
| rameshg87 | geertj: so from the above link it looks like ravello provides ability to upload the image and boot a vm from it. is that correct ? | 10:37 |
| *** jcoufal has quit IRC | 10:39 | |
| *** geertj has quit IRC | 10:40 | |
| *** jcoufal has joined #openstack-ironic | 10:40 | |
| rameshg87 | lucasagomes: we were trying to propose a new diskimage-builder element which will download the image directly from glance. any thoughts on that ? | 10:41 |
| *** hipster has joined #openstack-ironic | 10:41 | |
| rameshg87 | lucasagomes: i know the IPA might be the way going forward, but IPA doesn't use diskimage-builder, right ? | 10:42 |
| lucasagomes | rameshg87, right that's what the ironic-python-agent does | 10:42 |
| lucasagomes | rameshg87, downloading from glance require you to have a token to talk to the api... I think the rax guys are avoiding it by using swift as a backend to glance | 10:42 |
| lucasagomes | and having a directly url where the image can be fetched from | 10:42 |
| rameshg87 | lucasagomes, we were also thinking of using swift-temp-url for that purpose | 10:44 |
| rameshg87 | lucasagomes, but we will need to token to inform the completion status to the ironic conductor through a vendor passthru | 10:44 |
| lucasagomes | rameshg87, yeah... so the way it's done today is... ironic puts a token in the tftp and the ramdisk sucks it up | 10:45 |
| lucasagomes | we want to change it by using keystone trust I think | 10:45 |
| lucasagomes | but I don't know if there's someone working on it right now | 10:45 |
| rameshg87 | lucasagomes, the same functionality (of downloading the image and writing the image to disk) done by ironic conductor can be done by an element in diskimage-builder in the baremetal node. | 10:47 |
| lucasagomes | rameshg87, +1 | 10:47 |
| lucasagomes | and I find it more scalable than the current approach | 10:48 |
| lucasagomes | I like that ^ :) | 10:48 |
| rameshg87 | lucasagomes, do you think adding a new diskimage-builder element makes sense ? | 10:48 |
| rameshg87 | deploy-ironic-bm or something like that .. | 10:48 |
| *** hipster has quit IRC | 10:49 | |
| lucasagomes | rameshg87, well I don't think it tons of work, so I would say yes, it makes sense | 10:49 |
| lucasagomes | cause the long term thing I believe is moving towards the IPA | 10:49 |
| lucasagomes | and the guys already have that | 10:49 |
| lucasagomes | IPA = ironic python agent, not beer :) | 10:50 |
| dtantsur | G'afternoon, Ironic | 10:50 |
| rameshg87 | lucasagomes, :-) | 10:50 |
| rameshg87 | lucasagomes, thanks .. yeah, its just retrieving from the url, partitioning the disk, writing it, notifying the completion back to the conductor | 10:50 |
| rameshg87 | dtantsur, good afternoon | 10:50 |
| *** dshulyak has quit IRC | 10:52 | |
| lucasagomes | dtantsur, morning | 10:52 |
| rameshg87 | lucasagomes, i had one more question | 10:54 |
| rameshg87 | dtantsur, you might be able to help too | 10:54 |
| dtantsur | with pleasure | 10:54 |
| rameshg87 | we have ribcl.py module that is currently being reviewed here: https://review.openstack.org/#/c/89500/ | 10:54 |
| rameshg87 | this module interfaces with hp ilo | 10:54 |
| rameshg87 | we wanted to move this out of ironic, something like what has been done by seamicro | 10:55 |
| rameshg87 | is there documented procedure for such a thing ? | 10:55 |
| dtantsur | hmm... I think you start with getting it to stackforge | 10:56 |
| lucasagomes | rameshg87, is it something that you may put into https://github.com/stackforge/pyghmi/ ? | 10:56 |
| dtantsur | then you make it an _optional_ dependency (again like with seamicro) | 10:56 |
| dtantsur | lucasagomes, I doubt it :) | 10:56 |
| dtantsur | I don't think it even similar in protocol | 10:57 |
| lucasagomes | right | 10:58 |
| lucasagomes | rameshg87, so yeah you can use stackforge I guess, or do like seamicro and have it's own github repo | 10:58 |
| lucasagomes | you just need to make it pip-installable | 10:58 |
| dtantsur | yeah, make sure it is on PyPI | 10:58 |
| dtantsur | than the only thing you have to ensure is that unittests for code that stays in Ironic tree can run _without_ this dependency | 10:59 |
| dtantsur | (aka mock everything :) | 10:59 |
| lucasagomes | yup | 10:59 |
| lucasagomes | there's a patch doing that ^ for the seamicro driver in the queu right now | 10:59 |
| lucasagomes | you might want to use that as an example | 10:59 |
| rameshg87 | dtantsur, yeah i saw the unit test code for seamicro | 11:01 |
| rameshg87 | they skip some tests if the library is not available | 11:01 |
| dtantsur | and I guess some tests should still be present and work with mocks | 11:01 |
| dtantsur | so that we know that your code might work :) | 11:01 |
| rameshg87 | okay, :-) | 11:01 |
| rameshg87 | lucasagomes, dtantsur, thanks, so i will pull them out, we are planning to add more to the utils, so we thought it makes sense to pull out of ironic now itself | 11:02 |
| *** vkdrao has joined #openstack-ironic | 11:02 | |
| lucasagomes | ack, good stuff :) | 11:03 |
| rameshg87 | got the code reviewed by ironic reviewers for free though :-) | 11:03 |
| dtantsur | lol | 11:03 |
| lucasagomes | heh | 11:03 |
| rameshg87 | in the next patchset, we will fix up the module in a separate github and remove it from ironic review | 11:04 |
| dtantsur | It would be also much easier for us to review the remaining part | 11:04 |
| rameshg87 | i mean ironic tree | 11:04 |
| lucasagomes | indeed | 11:05 |
| dtantsur | lucasagomes, may I shamelessly ask you to review https://review.openstack.org/#/c/92625/ ? It's around for 2 weeks already | 11:06 |
| lucasagomes | dtantsur, ah yes... actually I was reviewing it | 11:07 |
| *** coolsvap is now known as coolsvap|afk | 11:07 | |
| lucasagomes | dtantsur, i've a quesiton | 11:07 |
| lucasagomes | just a nit | 11:07 |
| dtantsur | sure! | 11:07 |
| lucasagomes | on the pxe.py _fetch_images() | 11:07 |
| lucasagomes | the docstring says """Wrapper around ImageCache, checking for available disk space.""" | 11:07 |
| lucasagomes | it does more than checking for available disk space, it actually fetchs the image | 11:07 |
| lucasagomes | I think that other than that it lgtm | 11:08 |
| dtantsur | I meant `wrapper, also checking` | 11:08 |
| dtantsur | if it's critical, I can fix | 11:08 |
| lucasagomes | it's grand | 11:08 |
| lucasagomes | I will finish the review | 11:08 |
| *** coolsvap|afk is now known as coolsvap | 11:10 | |
| dtantsur | ok, thanks | 11:12 |
| *** killer_prince has quit IRC | 11:16 | |
| *** vkdrao has quit IRC | 11:16 | |
| *** nosnos has quit IRC | 11:16 | |
| *** killer_prince has joined #openstack-ironic | 11:17 | |
| *** datajerk has joined #openstack-ironic | 11:19 | |
| *** coolsvap is now known as coolsvap|afk | 11:23 | |
| openstackgerrit | ChangBo Guo(gcb) proposed a change to openstack/ironic: Fix wrong usage of mock methods https://review.openstack.org/93532 | 11:28 |
| *** killer_prince has quit IRC | 11:29 | |
| *** lazy_prince has joined #openstack-ironic | 11:29 | |
| *** lazy_prince is now known as killer_prince | 11:30 | |
| *** coolsvap|afk is now known as coolsvap | 11:41 | |
| *** dshulyak has joined #openstack-ironic | 11:47 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Fix chassis bookmark link url https://review.openstack.org/94597 | 11:56 |
| dtantsur | lucasagomes, are you planning to take any actions on https://bugs.launchpad.net/ironic/+bug/1306153 ? | 11:57 |
| * dtantsur is looking for some task :) | 11:58 | |
| lucasagomes | dtantsur, not in the moment | 11:58 |
| lucasagomes | dtantsur, go ahead with it :) | 11:58 |
| dtantsur | thnx! | 11:58 |
| *** coolsvap is now known as coolsvap|afk | 12:00 | |
| *** lucasagomes is now known as lucas-hungry | 12:01 | |
| dtantsur | lucas-hungry, could you also (once you have time, I mean) have a look at https://review.openstack.org/#/c/94371/ ? It's a follow-up for previous one | 12:04 |
| *** jistr is now known as jistr|english | 12:05 | |
| NobodyCam | good morning ironic | 12:12 |
| dtantsur | morning NobodyCam! | 12:13 |
| dtantsur | Don't you make coffee? :) | 12:13 |
| NobodyCam | already done | 12:14 |
| NobodyCam | :) | 12:14 |
| * NobodyCam takes big sip of his coffee | 12:14 | |
| NobodyCam | heheheh | 12:15 |
| NobodyCam | and good morning dtantsur | 12:15 |
| *** uberj has quit IRC | 12:22 | |
| *** uberj has joined #openstack-ironic | 12:22 | |
| *** jdob has joined #openstack-ironic | 12:31 | |
| NobodyCam | anyone have a few minutes to review 93133? | 12:36 |
| *** sabah has joined #openstack-ironic | 12:37 | |
| openstackgerrit | A change was merged to openstack/ironic: Don't translate debug level logs https://review.openstack.org/93291 | 12:38 |
| Shrews | So, I'm going to ask what is probably a stupid question(s)... after pxe booting, the node must call back to the ironic api to continue the deploy, correct? | 12:41 |
| NobodyCam | nope | 12:43 |
| *** athomas has joined #openstack-ironic | 12:43 | |
| NobodyCam | conductor pings node to say done | 12:43 |
| NobodyCam | :-p | 12:44 |
| openstackgerrit | Sandhya Balakrishnan proposed a change to openstack/ironic: Update Ironic Guide post landing for 569c0f395a9 https://review.openstack.org/94604 | 12:44 |
| NobodyCam | Shrews: node can not really know when it done with current ramdisk only the conductor knows whe its done writing tot he disk | 12:45 |
| Shrews | NobodyCam: hmm, ok. i think the dib deploy-ironic element confused me | 12:47 |
| Shrews | https://github.com/openstack/diskimage-builder/blob/master/elements/deploy-ironic/init.d/80-deploy-ironic | 12:47 |
| * NobodyCam looks | 12:47 | |
| NobodyCam | https://github.com/openstack/diskimage-builder/blob/master/elements/deploy-ironic/init.d/80-deploy-ironic#L47 | 12:48 |
| NobodyCam | Shrews: ^^^ | 12:48 |
| NobodyCam | its just a simple ping | 12:48 |
| Shrews | NobodyCam: but the curl request above that is what's throwing me off | 12:49 |
| *** linggao has joined #openstack-ironic | 12:50 | |
| NobodyCam | oh I may be confuzlizing it.. node dose ping conductor to START the deploy | 12:51 |
| Shrews | ah ha, ok | 12:52 |
| NobodyCam | https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/deploy_utils.py#L164-L171 | 12:52 |
| NobodyCam | thats the ping to reboot once the dding is done | 12:52 |
| Shrews | NobodyCam: great. so, this obviously requires customized ramdisk/kernel images then. which brings me to the next stupid question... how are the cirros images working for us? we didn't customize those, afaict | 12:53 |
| *** sabah has quit IRC | 12:54 | |
| NobodyCam | cirros is the image we are deploying on the node | 12:55 |
| NobodyCam | not the deploy ramdisk itself | 12:55 |
| *** rloo has joined #openstack-ironic | 12:56 | |
| * Shrews experiences another "duh" moment | 12:56 | |
| * NobodyCam offers Shrews some coffee :) | 12:56 | |
| Shrews | NobodyCam: that may not be enough for me today. :/ | 13:00 |
| *** lucas-hungry is now known as lucasagomes | 13:00 | |
| lucasagomes | morning NobodyCam Shrews | 13:00 |
| NobodyCam | good morning lucasagomes :) | 13:01 |
| NobodyCam | Shrews: LOL | 13:01 |
| Shrews | morning lucasagomes | 13:01 |
| NobodyCam | lucasagomes: gots a second to beet me up on 93133? | 13:01 |
| NobodyCam | hehehe | 13:01 |
| NobodyCam | and good morning lucasagomes :-p | 13:02 |
| lucasagomes | will do in a sec | 13:03 |
| NobodyCam | hehehe :) | 13:03 |
| *** ramineni has joined #openstack-ironic | 13:15 | |
| *** datajerk has quit IRC | 13:24 | |
| *** jistr|english is now known as jistr | 13:28 | |
| openstackgerrit | Rohan Kanade proposed a change to openstack/ironic: Partially remove hard-coded ID's from testdata https://review.openstack.org/76990 | 13:29 |
| *** killer_prince has quit IRC | 13:33 | |
| *** dhellmann_ is now known as dhellmann | 13:35 | |
| *** Penick has joined #openstack-ironic | 13:39 | |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic: Add some real-world testing on DiskPartitioner https://review.openstack.org/94620 | 13:48 |
| *** hipster has joined #openstack-ironic | 13:49 | |
| *** Penick has quit IRC | 13:57 | |
| *** jgrimm has joined #openstack-ironic | 13:57 | |
| NobodyCam | brb | 14:01 |
| *** rameshg87 has left #openstack-ironic | 14:03 | |
| *** datajerk has joined #openstack-ironic | 14:04 | |
| openstackgerrit | Aleksandr Gordeev proposed a change to openstack/ironic-python-agent: Introduce MainlineHardwareManager https://review.openstack.org/92847 | 14:13 |
| openstackgerrit | Aleksandr Gordeev proposed a change to openstack/ironic-python-agent: Add hardware_utils https://review.openstack.org/92399 | 14:13 |
| *** rwsu has joined #openstack-ironic | 14:15 | |
| NobodyCam | dtantsur: blocking your own patches | 14:19 |
| dtantsur | NobodyCam, sorry? | 14:20 |
| NobodyCam | https://review.openstack.org/94620 | 14:20 |
| lucasagomes | NobodyCam, that's the new "WIP" thingy | 14:21 |
| dtantsur | NobodyCam, yes, it's WIP and not finished. What's wrong with it? | 14:21 |
| * dtantsur liked previous WIP button much more | 14:21 | |
| lucasagomes | dtantsur, +1 | 14:21 |
| lucasagomes | although... dtantsur does this new WIP remains | 14:22 |
| lucasagomes | after you upload a new revision ? | 14:22 |
| dtantsur | lucasagomes, no, you have to do it again | 14:22 |
| lucasagomes | dtantsur, damn... yeah the previous one was better | 14:22 |
| lucasagomes | I'd like it to remain and then I could manually say it's ready | 14:22 |
| dtantsur | lucasagomes, actually, I prefer previous approach with +2/+1/-1/-2 separate from WIP and +A | 14:23 |
| dtantsur | new one is overly confusing | 14:23 |
| lucasagomes | heh | 14:23 |
| lucasagomes | dtantsur, have you tried the newest gerrit interface? | 14:23 |
| lucasagomes | you can enable it going to | 14:24 |
| dtantsur | lucasagomes, didn't like it | 14:24 |
| dtantsur | maybe I just didn't give it enough time though | 14:24 |
| lucasagomes | settings->preferences and on the "Change View" dropbox you select "new screen" | 14:24 |
| *** dwalleck has joined #openstack-ironic | 14:24 | |
| lucasagomes | dtantsur, yeah, I'm trying to use it | 14:24 |
| lucasagomes | but def it's overcomplex | 14:24 |
| lucasagomes | the flat buttons kinda annoys me | 14:25 |
| dtantsur | lucasagomes, I hope they're not gonna introduce Metro-style plates in the next release :D | 14:25 |
| lucasagomes | heh | 14:25 |
| lucasagomes | I think ppl are more appleish than windowish | 14:26 |
| lucasagomes | which I consider both pretty bad | 14:26 |
| lucasagomes | but that's my opnion and is biased :) | 14:26 |
| *** ramineni has quit IRC | 14:27 | |
| dtantsur | well, +1 from me :) | 14:27 |
| NobodyCam | nothing just saw the rad x | 14:27 |
| NobodyCam | red even | 14:27 |
| NobodyCam | dtantsur: can you add a test for nodes that already have volume labels | 14:28 |
| NobodyCam | and yea I kinda like the old +-1/2 +A | 14:29 |
| dtantsur | NobodyCam, what I intend to do is to test deploy_utils.make_partitions(). Can I really test it on that level? | 14:29 |
| NobodyCam | humm I'm not sure... but I thought it was worth asking | 14:30 |
| NobodyCam | :-p | 14:30 |
| lucasagomes | NobodyCam, about DiskPartitioner being a wrapper around parted | 14:30 |
| lucasagomes | I think that was the idea | 14:30 |
| lucasagomes | but maybe we should just see it as an class to do disk partitioning operations | 14:31 |
| *** shakamunyi has joined #openstack-ironic | 14:31 | |
| lucasagomes | independent of the tool in the background | 14:31 |
| lucasagomes | like a basic interface | 14:31 |
| NobodyCam | hummm... | 14:31 |
| lucasagomes | cause I kinda see the zap thing as being part of disk operation | 14:31 |
| lucasagomes | NobodyCam, idk what u think about it? | 14:31 |
| lucasagomes | just throwing an idea | 14:32 |
| dtantsur | NobodyCam, actually, I could test work_on_disk, which you change in 93133, it would require more mocking, but will cover more | 14:32 |
| openstackgerrit | Aleksandr Gordeev proposed a change to openstack/ironic-python-agent: Introduce MainlineHardwareManager https://review.openstack.org/92847 | 14:32 |
| openstackgerrit | Aleksandr Gordeev proposed a change to openstack/ironic-python-agent: Add hardware_utils https://review.openstack.org/92399 | 14:32 |
| NobodyCam | lucasagomes: thats in ref to my comment about moving the destroy_mbr_gpt | 14:32 |
| lucasagomes | NobodyCam, yup, but I'm happy with both approachs really | 14:33 |
| lucasagomes | I know it's a blocker | 14:33 |
| NobodyCam | :) | 14:33 |
| lucasagomes | I missed a nit on the previous review :( | 14:33 |
| lucasagomes | I will put a comment there, can you fix it quickly? | 14:33 |
| lucasagomes | other than that it lgtm | 14:33 |
| NobodyCam | yea... I will fix asap. | 14:34 |
| NobodyCam | :) | 14:34 |
| dtantsur | NobodyCam, what do you think about me covering the whole work_on_disk thing (with patching all kinds of is_block_device)? | 14:34 |
| dtantsur | it will even cover dd-ing image to a disk | 14:35 |
| NobodyCam | dtantsur: I like the sound of it... but would need to see how it ends up getting reworked :) | 14:36 |
| dtantsur | ok, let's try. Covering the whole partitioning sounds fruitful | 14:37 |
| NobodyCam | :) | 14:38 |
| lucasagomes | NobodyCam, commented, after that it lgtm I will +2... sorry i didn't see it in the previous review | 14:38 |
| NobodyCam | hehehe all good TY lucasagomes :) | 14:39 |
| lucasagomes | NobodyCam, cooleo! | 14:39 |
| NobodyCam | ahh I yeppers I should have added a doc string there :) | 14:40 |
| openstackgerrit | Chris Krelle proposed a change to openstack/ironic: Workaround for conflicting volume labels https://review.openstack.org/93133 | 14:43 |
| *** ifarkas has quit IRC | 14:44 | |
| lucasagomes | :) | 14:47 |
| lucasagomes | just waiting for gerrit | 14:47 |
| NobodyCam | :) | 14:47 |
| *** dwalleck has quit IRC | 14:48 | |
| NobodyCam | yeppers | 14:49 |
| lucasagomes | jroll, you around? re instance_info | 14:50 |
| *** dwalleck has joined #openstack-ironic | 14:52 | |
| *** coolsvap|afk is now known as coolsvap | 14:54 | |
| *** datajerk has quit IRC | 14:58 | |
| JoshNang | lucasagomes: jroll's going to be intermittent for the next week or so..he's heading home for a wedding. what's up? | 15:03 |
| lucasagomes | JoshNang, oh hey... just wanted to know if he's working on moving the parameters from the pxe driver to the instance info | 15:03 |
| lucasagomes | if not I will probably jump on it, cause we need this refactored before adding the driver to nova | 15:04 |
| lucasagomes | so want to get it done soonish | 15:04 |
| lucasagomes | adding to nova I mean... sending it to review | 15:04 |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic: Add some real-world testing on DiskPartitioner https://review.openstack.org/94620 | 15:05 |
| *** geertj has joined #openstack-ironic | 15:05 | |
| dtantsur | NobodyCam, that's how it will look like ^^^ (some tests are to be written) | 15:06 |
| openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Implement API to get driver properties https://review.openstack.org/73005 | 15:06 |
| JoshNang | he was planning on doing all of them, but if you want to get some of them moved, i doubt he'll be upset. i'm not sure how far he got | 15:06 |
| openstackgerrit | A change was merged to openstack/ironic: Remove explicit dependency on amqplib https://review.openstack.org/92153 | 15:07 |
| dtantsur | lucasagomes, I would really like to also see this merged before we submit the driver: https://review.openstack.org/#/c/86536/ | 15:07 |
| dtantsur | Mikhail_D_wk, ^^^ | 15:07 |
| lucasagomes | JoshNang, right, I just want to move the parameters from pxe because it's going to affect the driver | 15:07 |
| lucasagomes | dtantsur, +1 | 15:08 |
| lucasagomes | gotta review it | 15:08 |
| lucasagomes | JoshNang, I will do a quick patch fixing that and tag he created on it | 15:09 |
| lucasagomes | NobodyCam, quick one for uhttps://review.openstack.org/#/c/94202/ | 15:10 |
| lucasagomes | NobodyCam, this one as well https://review.openstack.org/#/c/92844 | 15:10 |
| JoshNang | sounds good! | 15:17 |
| NobodyCam | lucasagomes: both +A'd :) | 15:27 |
| NobodyCam | morning rloo | 15:27 |
| rloo | morning NobodyCam! Where are you these days? | 15:28 |
| NobodyCam | rloo: nebraska | 15:29 |
| NobodyCam | rloo: got a free minute to take a look at https://review.openstack.org/#/c/93133 | 15:30 |
| matty_dubs | Wow, you made good time! | 15:30 |
| *** coolsvap is now known as coolsvap|afk | 15:30 | |
| rloo | NobodyCam: hope you're enjoying Nebraska. | 15:30 |
| rloo | NobodyCam: sure, I'll take a look. | 15:30 |
| NobodyCam | matty_dubs: had too. rloo here because my dad had a pritty major stroke :( | 15:31 |
| lucasagomes | NobodyCam, o wow thanks! | 15:31 |
| rloo | NobodyCam: bad news. Hope he's doing 'ok'... | 15:31 |
| matty_dubs | NobodyCam: Oh geez, sorry to hear that :'( | 15:31 |
| lucasagomes | rloo, morning :) | 15:31 |
| NobodyCam | :( I have a meeting with the doctors today.. Will know more after that :-/ | 15:32 |
| lucasagomes | NobodyCam, :/ good lucky, hope he's fine | 15:32 |
| rloo | Hi lucasagomes. Morning matty_dubs. | 15:32 |
| lucasagomes | luck* | 15:32 |
| NobodyCam | thank you all :) | 15:32 |
| NobodyCam | sorry I've been distracted the last week.. but that is why :( | 15:33 |
| JoshNang | NobodyCam: no need to apologize. hope he's ok! | 15:34 |
| NobodyCam | :) Thank you.. | 15:35 |
| rloo | lucasagomes: if you have a minute, you might be able to answer my question: https://review.openstack.org/#/c/91366/ | 15:36 |
| *** krtaylor has quit IRC | 15:38 | |
| lucasagomes | rloo, ah | 15:40 |
| lucasagomes | rloo, so that 10 there I kinda came up with because it sounded reasonable | 15:41 |
| lucasagomes | but that parameter that the david added is being used on all the others API calls | 15:41 |
| lucasagomes | till now 2 seconds pooling seems to be working... idk if there's an ideal value | 15:41 |
| lucasagomes | but I would say it's 2 diff problems | 15:42 |
| *** ramineni has joined #openstack-ironic | 15:42 | |
| lucasagomes | the first step is to have the api to use the same config value for the pooling, which is what david's patch is doing | 15:42 |
| lucasagomes | so it's more consistent | 15:42 |
| rloo | lucasagomes: ok. well it sort of is 2 diff problems. i don't mind landing it, but it could break if 2 is not large enough. | 15:42 |
| lucasagomes | then the value itself if changed for something more resonable should be done in another patch or something | 15:42 |
| lucasagomes | rloo, right too seems grand, the tripleo tests passed and all | 15:43 |
| lucasagomes | so it's not breaking anything | 15:43 |
| rloo | lucasagomes: ok... | 15:43 |
| dtantsur | morning, rloo! | 15:49 |
| *** igor has quit IRC | 15:50 | |
| *** igor__ has joined #openstack-ironic | 15:51 | |
| rloo | hi dtantsur! | 15:51 |
| *** coolsvap|afk is now known as coolsvap | 15:55 | |
| *** eghobo has joined #openstack-ironic | 15:55 | |
| dtantsur | Folks, do you think we actually have `parted` on gate machines? I'm debugging http://logs.openstack.org/20/94620/2/check/gate-ironic-python26/fb841ed/testr_results.html.gz | 15:56 |
| *** derekh_ has quit IRC | 15:56 | |
| dtantsur | Test for Py27 passed though... | 15:56 |
| NobodyCam | dtantsur: I would think parted would be there | 16:00 |
| geertj | question: when using Nova with Ironic, is it Nova that is responsible for setting pxe_image_source, pxe_root_gb, etc? Or does the user need to do this when enrolling the node? | 16:01 |
| dtantsur | well... not sure what went wrong than... | 16:01 |
| dtantsur | geertj, user is not setting these, afaik | 16:01 |
| NobodyCam | geertj: yes, source should be set by the --image tag and size info should be from the --flavor | 16:02 |
| *** ramineni has quit IRC | 16:03 | |
| geertj | dtantsur, NobodyCam thx! However, how does does "nova boot" know to set "pxe_image_source" vs just "image_source". Does Nova know about Ironic drivers? | 16:04 |
| dtantsur | geertj, I think it's done by Ironic Nova driver | 16:06 |
| *** rloo has quit IRC | 16:07 | |
| *** rloo has joined #openstack-ironic | 16:08 | |
| dtantsur | geertj, yeah, see ironic/nova/virt/ironic/patcher.py | 16:08 |
| NobodyCam | yep was just about to paste that link ^^^ | 16:08 |
| *** jistr is now known as jistr|gone | 16:09 | |
| geertj | OK it all makes sense now :) I didn't realize the Ironic driver for Nova was part of Ironic itself. For some reason I had assume it was part of Nova. | 16:09 |
| NobodyCam | geertj: it will be moved to nova later | 16:10 |
| *** dwalleck has quit IRC | 16:16 | |
| NobodyCam | rloo lucasagomes not pushing, just checking on 93133 :-p | 16:21 |
| *** igordcard_ has joined #openstack-ironic | 16:22 | |
| rloo | NobodyCam: am looking at it now. I'm slow... | 16:22 |
| openstackgerrit | Dan Prince proposed a change to openstack/ironic: Port iBoot PDU driver from Nova https://review.openstack.org/50977 | 16:22 |
| dtantsur | lucasagomes, NobodyCam: while I'm writing this tests, I see that we can't set ephemeral + swap + disk + 1 MiB padding == disk size | 16:22 |
| dtantsur | it dies with e.g. 'Error: The location 24 is outside of the device /tmp/tmpW8Yjp_/tmp6tXBTN' | 16:23 |
| dtantsur | oh, sorry, disregard this ^^^ | 16:23 |
| dtantsur | I need to recheck my math :) | 16:23 |
| NobodyCam | hehehe | 16:24 |
| NobodyCam | ty rloo :) | 16:24 |
| *** dwalleck has joined #openstack-ironic | 16:25 | |
| *** hemna_ is now known as hemna | 16:26 | |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic: Add some real-world testing on DiskPartitioner https://review.openstack.org/94620 | 16:27 |
| dtantsur | NobodyCam, lucasagomes: my math seems ok to me :) could you please look at test_whole_disk here https://review.openstack.org/#/c/94620/3/ironic/tests/drivers/test_deploy_utils.py | 16:28 |
| dtantsur | it fails for me with Error: The location 20 is outside of the device. Should it work? | 16:28 |
| lucasagomes | dtantsur, maybe that's the 512 bytes used to allocate the MBR? | 16:29 |
| dtantsur | lucasagomes, at least with 1 more reserved MiB it succeeds :) | 16:31 |
| dtantsur | should I make this test also account for it? | 16:31 |
| NobodyCam | ya thats what I was thinking .. dtantsur dose it work if you set ephemeral to 5mib | 16:31 |
| lucasagomes | dtantsur, :) yeah because the 512 bytes is not included in the math there | 16:31 |
| dtantsur | NobodyCam, yeah, works | 16:32 |
| lucasagomes | things can get a bit more complex if you use GPT where the end of the disk is also used to allocate the partitions structure | 16:32 |
| *** dwalleck has quit IRC | 16:32 | |
| lucasagomes | 4k if I can recall correct | 16:32 |
| dtantsur | wait, guys! Aren't we already accounting for MBR in this + 1MiB? | 16:33 |
| dtantsur | # NOTE(lucasagomes): Lead in with 1MiB to allow room for the partition table itself. | 16:34 |
| dtantsur | so, I added this 1MiB to my calculations already | 16:34 |
| dtantsur | for me it seems like some difficult rounding by parted itself | 16:34 |
| *** athomas has quit IRC | 16:34 | |
| lucasagomes | hmm yeah, I thought it was but I'm not 100% sure now | 16:35 |
| dtantsur | so that we can't just `mkpart SMTH END` where END == size of the disk | 16:35 |
| *** matty_dubs is now known as matty_dubs|lunch | 16:35 | |
| NobodyCam | could we be looking at MiB vs MB issue? | 16:36 |
| dtantsur | parted -a optimal -s /tmp/tmpOv5DQB/tmpouq22z -- unit MiB mklabel msdos mkpart primary 1 7 mkpart primary linux-swap 7 10 mkpart primary 10 20 | 16:37 |
| dtantsur | I've created a file with dd ... bs=1024 count=20 * 1024 | 16:37 |
| dtantsur | Ok, guys: mkpart expects BEGIN END as arguments. File size is beyond END. So we do need END == file size - 1 | 16:38 |
| dtantsur | Am I right? | 16:38 |
| openstackgerrit | A change was merged to openstack/ironic: Speed up Nova Ironic driver tests https://review.openstack.org/94202 | 16:39 |
| lucasagomes | dtantsur, you can do bs=1MiB count=20 | 16:39 |
| NobodyCam | dtantsur: lucasagomes: could it be the -a optimal moving things slightly | 16:39 |
| *** rloo has quit IRC | 16:40 | |
| *** rloo has joined #openstack-ironic | 16:40 | |
| * dtantsur starts to conduct manual experiments | 16:41 | |
| lucasagomes | NobodyCam, hmm i'm not 100% sure but I think it might | 16:42 |
| lucasagomes | dtantsur, ok it seems to be the MBR | 16:43 |
| lucasagomes | if you create the file with | 16:43 |
| dtantsur | NobodyCam, even without optimal it does not allow me to mkpart SMTH END | 16:43 |
| lucasagomes | bs=1024 count=20 * 1024 + 512 | 16:44 |
| lucasagomes | it works | 16:44 |
| dtantsur | lucasagomes, we already accounted for MBR | 16:44 |
| openstackgerrit | A change was merged to openstack/ironic: Fix 'pxe_preserve_ephemeral' parameter leakage https://review.openstack.org/92844 | 16:44 |
| *** martyntaylor1 has joined #openstack-ironic | 16:45 | |
| lucasagomes | dtantsur, we accounted it within that 1mb padding | 16:45 |
| lucasagomes | but that's what I'm investigating now | 16:45 |
| dtantsur | lucasagomes, yes, and I deducted 1MiB from total size | 16:45 |
| lucasagomes | if the mbr is really included in that 1mib | 16:45 |
| lucasagomes | or if it's 1mib after the mbr | 16:45 |
| dtantsur | but now I need more padding from the end | 16:45 |
| dtantsur | lucasagomes, oh, that's a good point! | 16:46 |
| openstackgerrit | A change was merged to openstack/ironic: Remove non-existent 'pxe_default_format' parameter from patcher https://review.openstack.org/93012 | 16:47 |
| *** martyntaylor has quit IRC | 16:48 | |
| NobodyCam | gah I'm out of ink for my printer. :( | 16:49 |
| *** ndipanov has quit IRC | 16:50 | |
| *** martyntaylor1 has quit IRC | 16:50 | |
| dtantsur | Warning: You requested a partition from 0.00MiB to 10.0MiB (sectors 0..20479). The closest location we can manage is 0.00MiB to 10.0MiB (sectors 1..20479). | 16:50 |
| dtantsur | lucasagomes, it still seems to me that it includes MBR ^^^ | 16:51 |
| dtantsur | I suspect it's alignment issue, not mbr | 16:51 |
| openstackgerrit | A change was merged to openstack/ironic: Virt driver change to use API retry config value https://review.openstack.org/91366 | 16:51 |
| lucasagomes | dtantsur, yeah, I def have to read a bit more about it to know exactly how it works | 16:54 |
| *** dwalleck has joined #openstack-ironic | 16:56 | |
| *** datajerk has joined #openstack-ironic | 16:57 | |
| *** vkdrao has joined #openstack-ironic | 16:57 | |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic: Add some real-world testing on DiskPartitioner https://review.openstack.org/94620 | 16:58 |
| dtantsur | will leave it with additional 1MiB for now ^^^ | 16:59 |
| *** igordcard_ has quit IRC | 16:59 | |
| *** igordcard has joined #openstack-ironic | 16:59 | |
| NobodyCam | dtantsur: lucasagomes in googling there seems to be some issues with optimal and md devices, I haven't found anything about regular disks/files, but I just did a quick look | 17:00 |
| rloo | NobodyCam: have some questions about 93133 (didn't look at the tests yet). Going to grab lunch, back in a few minutes. | 17:00 |
| NobodyCam | rloo: sure ...thank you :) | 17:01 |
| * NobodyCam needs to run to grab some ink so he can print a pdf sign it then scan it back in email it :-p ... happy happy joy joy | 17:02 | |
| NobodyCam | s/in email/in AND email/ | 17:03 |
| *** dwalleck has quit IRC | 17:04 | |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic: Consider free disk space before downloading images into cache https://review.openstack.org/92625 | 17:04 |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic: PXE to pass hints to ImageCache on how much space to reclaim https://review.openstack.org/94371 | 17:04 |
| *** tatyana has joined #openstack-ironic | 17:04 | |
| dtantsur | lucasagomes, I've fixed merge conflicts for these 2, could you look again ^^^ | 17:05 |
| dtantsur | (no real code changes) | 17:05 |
| *** martyntaylor has joined #openstack-ironic | 17:05 | |
| *** rloo has quit IRC | 17:05 | |
| *** rloo has joined #openstack-ironic | 17:06 | |
| *** harlowja has joined #openstack-ironic | 17:06 | |
| *** tatyana_ has joined #openstack-ironic | 17:09 | |
| *** ndipanov has joined #openstack-ironic | 17:11 | |
| *** tatyana has quit IRC | 17:11 | |
| *** tatyana_ is now known as tatyana | 17:11 | |
| lucasagomes | dtantsur, cool will do | 17:13 |
| *** matty_dubs|lunch is now known as matty_dubs | 17:13 | |
| *** rloo has quit IRC | 17:16 | |
| *** rloo has joined #openstack-ironic | 17:17 | |
| *** rloo has quit IRC | 17:17 | |
| *** rloo has joined #openstack-ironic | 17:17 | |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic: Factoring out PXE and TFTP functions https://review.openstack.org/90233 | 17:17 |
| NobodyCam | ok bbiab.. prob 1/2 an hour | 17:18 |
| lucasagomes | I'm also done for today | 17:18 |
| lucasagomes | have a great night everybody | 17:18 |
| Shrews | night lucasagomes | 17:19 |
| NobodyCam | lucasagomes: 93133? | 17:19 |
| NobodyCam | hehehe | 17:19 |
| NobodyCam | and have a good night lucasagomes :) | 17:19 |
| devananda | morning, all | 17:19 |
| NobodyCam | good morning devananda :) | 17:19 |
| lucasagomes | NobodyCam, oh true | 17:19 |
| Shrews | hey devananda | 17:19 |
| lucasagomes | devananda, morning | 17:19 |
| lucasagomes | devananda, btw, I have a q about the hashring and rpc versioning | 17:20 |
| NobodyCam | ok be back in about 1/2 an hour :) | 17:20 |
| rloo | night lucasagomes. morning devananda. | 17:20 |
| lucasagomes | devananda, should we take account the version of the conductor when choosing the conductor that manage a specific node | 17:20 |
| lucasagomes | ? I think it makes sense | 17:20 |
| russell_h | lucasagomes: how would that work? | 17:20 |
| russell_h | lucasagomes: only use the newest conductor versions found anywhere in the ring? | 17:20 |
| lucasagomes | the conductor might have the driver loaded, but the rpc version might be incompatible with the rpcapi | 17:21 |
| devananda | lucasagomes: version of conductor? | 17:22 |
| lucasagomes | https://review.openstack.org/#/c/94379/ | 17:22 |
| lucasagomes | russell_h, devananda ^ | 17:22 |
| *** diga has joined #openstack-ironic | 17:22 | |
| lucasagomes | perhaps I should elaborate it a bit more | 17:22 |
| devananda | ah | 17:22 |
| devananda | i see the issue | 17:22 |
| lucasagomes | but... right now | 17:23 |
| lucasagomes | we always pick the first conductor that has that driver | 17:23 |
| lucasagomes | from the list | 17:23 |
| devananda | lucasagomes: we should recommend // require that all services of a certain type be upgraded before the other services | 17:23 |
| lucasagomes | we don't take in account the RPC_API_VERSION that the conductor supports | 17:23 |
| devananda | eg, upgrade all conductors first, then upgrade all api services | 17:23 |
| russell_h | ^ that | 17:23 |
| lucasagomes | devananda, yes | 17:23 |
| devananda | we can't assume that all conductors upgrade at once, but as long as the API services are all the same version WHILE the conductors are upgrading, it should be fine | 17:23 |
| *** notq has quit IRC | 17:24 | |
| devananda | since no messages with higher RPC version will be sent | 17:24 |
| devananda | and the conductor side of the RPC bus has to maintain backwards compat | 17:24 |
| russell_h | right | 17:24 |
| lucasagomes | yeah makes sense | 17:25 |
| russell_h | it also means you don't necessarily need to migrate every node between conductors during a rolling upgrade | 17:25 |
| devananda | well. if you want the management plane to continue operating, you'd need to do a lot of rebalancing | 17:25 |
| devananda | if you degrade service briefly (ie, set everything into maintenance mode briefly) then an upgrade wouldn't need to rebalance | 17:26 |
| devananda | we haven't talked about that yet :) | 17:26 |
| russell_h | eh, as long as you can do a graceful restart of the conductor, some builds might get delayed a little, but nothing should actually break | 17:26 |
| russell_h | and if you can't do a graceful restart of the conductor, you've got problems anyway | 17:26 |
| *** igordcard has quit IRC | 17:26 | |
| devananda | right | 17:27 |
| devananda | and a graceful restart should be well under the heartbeat timeout | 17:27 |
| russell_h | right | 17:27 |
| russell_h | in fact as long as the conductor doesn't do a lot of really-long-running operations, it could be very very fast | 17:27 |
| devananda | it would need to stop accepting new RPC requests while it finishes anything in flight, then restart | 17:28 |
| russell_h | right | 17:28 |
| russell_h | or | 17:28 |
| russell_h | well, it sucks | 17:28 |
| russell_h | but just kill -9 | 17:28 |
| lucasagomes | heh | 17:28 |
| russell_h | and hope for the best | 17:28 |
| russell_h | a lot of problems would eventually sort themselves out | 17:28 |
| russell_h | not, like a firmware upgrade in progress :) | 17:29 |
| devananda | sure | 17:29 |
| russell_h | my point is mostly about lost RPC messages | 17:29 |
| russell_h | we should eventually rectify the situation | 17:29 |
| russell_h | I was thinking some about locking. Maybe this is already what you had in mind, but the "reserved" field today is used in two ways: 1) to tell other conductors "I'm working on this node, don't touch", and 2) to tell the API not to try to kick off additional changes | 17:31 |
| russell_h | what if the API ignored that field entirely | 17:32 |
| russell_h | and instead looked at whether any target_something_state != something_state | 17:32 |
| *** eguz has joined #openstack-ironic | 17:33 | |
| russell_h | so for example, if you set power_state to OFF, we'd do a transactional "set target power state to off, if all actual states match their target states" | 17:33 |
| russell_h | but not actually mark the node reserved at all | 17:34 |
| russell_h | then cast to the conductor and return | 17:34 |
| russell_h | then conductors reserve the node and make the states match | 17:34 |
| russell_h | it'd actually be a very small change to the API I think | 17:34 |
| *** eghobo has quit IRC | 17:37 | |
| *** ndipanov has quit IRC | 17:42 | |
| *** vkdrao has quit IRC | 17:43 | |
| lucasagomes | russell_h, right, but if it's just mark the target_state and not lock the node we could have a race condition no? | 17:44 |
| lucasagomes | 2 requests changing the same state | 17:44 |
| * lucasagomes brb, gotta go out buy something to eat | 17:46 | |
| *** lucasagomes is now known as lucas-afk | 17:46 | |
| *** tatyana has quit IRC | 17:47 | |
| *** tatyana has joined #openstack-ironic | 17:51 | |
| *** Mikhail_D_ltp has joined #openstack-ironic | 17:51 | |
| *** datajerk has quit IRC | 17:51 | |
| *** jistr|gone has quit IRC | 17:55 | |
| *** tatyana has quit IRC | 17:57 | |
| dtantsur | devananda, hi! Could you answer my questions here: https://review.openstack.org/#/c/79194/ This probably supersedes some of my work | 18:04 |
| devananda | dtantsur: looking | 18:05 |
| devananda | dtantsur: full user image isn't needed, but the user kernel&ramdisk are | 18:06 |
| dtantsur | devananda, look, code you're introducing will call prepare on rebalance, covering actually the same case | 18:07 |
| dtantsur | (I'm about superseding https://review.openstack.org/#/c/92115/ aka https://bugs.launchpad.net/ironic/+bug/1279331) | 18:08 |
| *** coolsvap is now known as coolsvap|afk | 18:08 | |
| devananda | dtantsur: that's an old patch. I didn't notice jenkins abandoned it until yesterday | 18:08 |
| devananda | which is why I marked it WIP -- i haven't reviewed it yet to see how current it is | 18:09 |
| devananda | I agree that prepare, when used for rebalancing, shouldn't fetch the whole user image again. | 18:09 |
| dtantsur | re instance image, I'm suggesting moving downloading it from prepare() to deploy(), anyway we're going to have it deleted in the end of deploy(). Are you ok with it? | 18:09 |
| dtantsur | patch about deleting by ifarkas: https://review.openstack.org/#/c/92340/ | 18:10 |
| *** diga has quit IRC | 18:10 | |
| devananda | move that to deploy && delete at the end of deploy -- yep, sounds good to me | 18:11 |
| dtantsur | ack, either ifarkas or I will do | 18:11 |
| Shrews | rloo: hi! great comment on 92105. it could very well lead us to a more in-depth discussion about how to document those variables, but my main reason for including IRONIC_VM_EPHEMERAL_DISK in our quickstart guide is that it makes it easier for a dev to test ephemeral partitions if they know that variable exists. | 18:11 |
| dtantsur | devananda, now re my patches: should a just postpone https://review.openstack.org/#/c/92115/ (and parent) until you're decided about your patch? | 18:12 |
| Shrews | rloo: the other variables are more "tuning" type things, which is probably why they haven't been documented anywhere | 18:12 |
| dtantsur | btw, core folks, easy target for approving (two +2's already): https://review.openstack.org/#/c/92340/ :) | 18:15 |
| devananda | dtantsur: quick note on use of INFO logging: https://wiki.openstack.org/wiki/LoggingStandards | 18:28 |
| devananda | also, we need to clean up our log usage overall.... | 18:29 |
| *** pelix has quit IRC | 18:30 | |
| rloo | Shrews: ok, so you want me to land it? :-) | 18:31 |
| devananda | dtantsur: so I'd like a spec for handling the rebalancing | 18:32 |
| devananda | dtantsur: and I would rather not land either of our patches taht implement rebalancing until then | 18:33 |
| Shrews | rloo: i think it would be helpful, so... yes please :) | 18:33 |
| devananda | dtantsur: also, mrda_away volunteered to work on that at the summit | 18:33 |
| *** igor__ has quit IRC | 18:34 | |
| *** igor_ has joined #openstack-ironic | 18:35 | |
| rloo | Shrews: done! | 18:36 |
| Shrews | rloo: muchas gracias! | 18:38 |
| *** geertj has quit IRC | 18:41 | |
| dtantsur | devananda, cool, will follow, thanks | 18:53 |
| dtantsur | g'night | 18:55 |
| *** krtaylor has joined #openstack-ironic | 18:56 | |
| *** jcoufal has quit IRC | 18:56 | |
| *** jcoufal has joined #openstack-ironic | 18:56 | |
| *** shakamunyi has quit IRC | 18:57 | |
| * NobodyCam is back | 18:57 | |
| NobodyCam | rloo: would you like me to reply to the questions on 93133 or here? | 18:59 |
| rloo | NobodyCam: probably best to reply so there is a 'record' there. Unless you think we should discuss? | 19:00 |
| NobodyCam | :) | 19:00 |
| *** harlowja is now known as harlowja_away | 19:04 | |
| *** datajerk has joined #openstack-ironic | 19:06 | |
| openstackgerrit | Chris Krelle proposed a change to openstack/ironic: Workaround for conflicting volume labels https://review.openstack.org/93133 | 19:10 |
| NobodyCam | rloo: replyed and pushed!!! | 19:10 |
| rloo | NobodyCam: thx. | 19:10 |
| NobodyCam | let me know if there are other questions | 19:10 |
| rloo | NobodyCam: so it still isn't clear: line 208 https://review.openstack.org/#/c/93133/25..26/ironic/drivers/modules/deploy_utils.py | 19:13 |
| rloo | NobodyCam: if utils.execute raises an exception, what is block_sz? | 19:14 |
| rloo | NobodyCam: and for line 230 - do you have the link from the google search? | 19:14 |
| NobodyCam | rloo: 208 should be zero. see lucas's comment on rev 24.. https://review.openstack.org/#/c/93133/24/ironic/drivers/modules/deploy_utils.py | 19:25 |
| NobodyCam | rloo: link: http://www.noah.org/wiki/Dd_-_Destroyer_of_Disks#Erase_GPT_.28GUID_Partition_Table.29 | 19:26 |
| rloo | NobodyCam: my python isn't that great. Are you sure that if utils.execute() raises an exception, block_sz = '0'? | 19:26 |
| *** harlowja_away is now known as harlowja | 19:27 | |
| rloo | NobodyCam: I suppose I should figure it out myself. I'll test it out in a bit... | 19:28 |
| NobodyCam | I wanted to set it to zero. like rev 24, but lucas says not to. | 19:29 |
| rloo | NobodyCam: yeah, I just don't see how it can be set to '0' if an exception is raised. even if lucas says so. I'll have to prove it to myself ;) | 19:30 |
| NobodyCam | :) | 19:30 |
| NobodyCam | anyone else have time to take a look at 93133? | 19:31 |
| *** lucas-afk has quit IRC | 19:32 | |
| *** datajerk has quit IRC | 19:32 | |
| NobodyCam | rloo Now that I think about it what iff I set to zero in the except clause of get_dev_block_size | 19:33 |
| rloo | NobodyCam: yes, that will work. I just can't see how it will work as-is. | 19:33 |
| NobodyCam | ack | 19:33 |
| NobodyCam | one sec | 19:33 |
| rloo | well, no need to push it out now. cuz i'm going to ask for another change. just don't know what it is yet. need to read the link first... | 19:34 |
| NobodyCam | lol.. ack... I'm ready to push/change | 19:35 |
| rloo | NobodyCam: it may take me awhile. i want to understand where 35 comes from... | 19:35 |
| rloo | NobodyCam: it seems related to the dd command, the count=X variable... | 19:36 |
| rloo | NobodyCam: my first guess is that you should be using 1, not 35. but that's just by pattern matching; i don't understand it yet. | 19:36 |
| NobodyCam | rloo ack. but count 35 with bs of 4096 is 143360 | 19:40 |
| NobodyCam | (thats why I left the 35) that and it works in my testing | 19:41 |
| NobodyCam | :-p | 19:41 |
| *** lucas-afk has joined #openstack-ironic | 19:44 | |
| *** eguz has quit IRC | 19:47 | |
| *** eghobo has joined #openstack-ironic | 19:47 | |
| Shrews | NobodyCam, rloo: yeah, that formula makes *no* sense to me | 19:54 |
| Shrews | if this table is accurate (https://wiki.archlinux.org/index.php/GUID_Partition_Table#GUID_Partition_Table), then that formula would be dd'ing more than necessary | 19:55 |
| NobodyCam | Shrews: so -1 vs the -35 | 19:56 |
| devananda | lunching, bbiab | 19:56 |
| NobodyCam | enjoy lunch devananda | 19:56 |
| Shrews | NobodyCam: did you read the "Note 1" part of that website you got that from? | 19:58 |
| Shrews | "Note 1, you may have to zero out a larger range of blocks for the secondary GPT. This is because I am not certain of my math." .. yada yada yada | 19:59 |
| NobodyCam | :) | 20:01 |
| NobodyCam | ok all, I have to run to a meeting with doctors. I will be back... Please leave comments on 93133 and I will address as soon as I get back. | 20:02 |
| NobodyCam | Thank you all for the help / assistence with this :) | 20:02 |
| Shrews | my concern would be that, since i *think* the ephemeral partion is the first partition created, you might be accidentally killing data there. the part at the end probably isn't as big a deal since the root partition should be the last one | 20:03 |
| *** tatyana has joined #openstack-ironic | 20:06 | |
| rloo | Shrews, so the question is what is the minimal amount that needs to be wiped from the beginning and the end? | 20:07 |
| Shrews | rloo: right | 20:07 |
| Shrews | rloo: but not being an expert with gpt, not sure what the correct math is | 20:07 |
| Shrews | leaving my concerns on the review | 20:08 |
| rloo | Shrews: ha ha. Neither am I. Just reading http://en.wikipedia.org/wiki/GUID_Partition_Table made my eyes glaze over. | 20:08 |
| rloo | Shrews: based on that, it seems like if we wipe out LBA0&1, we're OK. | 20:09 |
| rloo | Shrews: from the beginning. | 20:09 |
| Shrews | rloo: i would think so. just don't want to go *beyond* those, which i think this is doing | 20:13 |
| rloo | Shrews: that also says that LBA 34 is the first usable sector. if we want to erase the partitiion entry array too. but do we know that GPT is always there, vs MBR? | 20:14 |
| Shrews | rloo: not sure. i'd need to research that | 20:15 |
| Shrews | anyway, i left a note on the review for NobodyCam | 20:16 |
| *** jrist_ has joined #openstack-ironic | 20:19 | |
| *** jrist has quit IRC | 20:21 | |
| *** jrist_ is now known as jrist | 20:21 | |
| Shrews | rloo, NobodyCam: so, if an LBA sector is 512 bytes (not sure how to determine that), then the first dd covers the MBR and the GPT and then some. I think there is some "free space" between the GPT and the first physical partition, but again, not sure how to verify that. | 20:29 |
| openstackgerrit | A change was merged to openstack/ironic: Update dev quickstart guide for ephemeral testing https://review.openstack.org/92105 | 20:29 |
| Shrews | rloo, NobodyCam: if all of that is true, then the change is probably ok, but it doesn't make me feel very confident | 20:29 |
| rloo | Shrews: the first DD covers 4096 bytes, right? | 20:31 |
| Shrews | yeah | 20:31 |
| rloo | The MBR is only 512 bytes. A GPT is at least 18 KiB. | 20:31 |
| rloo | Shrews: according to https://wiki.archlinux.org/index.php/GUID_Partition_Table#GUID_Partition_Table | 20:32 |
| *** r-daneel has joined #openstack-ironic | 20:33 | |
| Shrews | rloo: yeah, but i'm supposing you only need to erase the gpt header, which is the first 512 of that. i'm probably wrong | 20:33 |
| rloo | Do we know if we're always going to deal with GPT, or if we might have MBR only? | 20:34 |
| * Shrews is probably only adding more confusion and not really helping | 20:34 | |
| Shrews | rloo: i have no idea :/ | 20:34 |
| rloo | if we have MBR only, do we know that the first partition or whever the ephemeral part starts, is beyond 4096 bytes from the start? | 20:34 |
| *** Mikhail_D_ltp has quit IRC | 20:35 | |
| rloo | Shrews: hehe. I don't know either. And I haven't even looked at the wiping at the end of the disk yet! | 20:35 |
| Shrews | rloo: "This typically causes the first primary partition to start at LBA 63 on disks accessed via LBA, leaving a gap of 62 sectors with MBR-based disks, sometimes called "MBR gap", "boot track", or "embedding area"." | 20:35 |
| Shrews | how "typical" that is, i have no clue | 20:36 |
| rloo | Shrews: yeah, I saw that. But the 'typically' worried me. | 20:36 |
| Shrews | lol ++ | 20:36 |
| rloo | ha ha. | 20:36 |
| rloo | As long as we say 'we assume that ...' :-) | 20:36 |
| rloo | ok, maybe i'll look at the end part now ;) | 20:36 |
| *** tatyana has quit IRC | 20:43 | |
| *** pradipta_away is now known as pradipta | 20:43 | |
| *** matty_dubs is now known as matty_dubs|gone | 21:03 | |
| *** jdob has quit IRC | 21:07 | |
| devananda | back | 21:14 |
| devananda | rloo: Shrews: we are definitely going to deal with both MBR and GPT | 21:15 |
| rloo | devananda: ah, ok thx. | 21:15 |
| devananda | and we need to support a) detecting which one is present, eg. during rebuild. b) allowing the user // image // flavor to indicate which one we should use. c) accepting an override (eg, some property of the node) that indicates the capabilities of the hardware itself | 21:16 |
| devananda | anyone want to write a spec for that? :) | 21:16 |
| devananda | I'm happy to brain dump the reasoning behind it | 21:16 |
| rloo | devananda: are you in a hurry for that spec? | 21:18 |
| devananda | rloo: not terribly, but the sooner something gets started the better | 21:20 |
| devananda | (isn't taht always true?) | 21:20 |
| rloo | devananda: ha ha. I could take a stab at it (the spec anyway), but don't want to before next week. | 21:21 |
| devananda | rloo: let's see what happens by then | 21:21 |
| devananda | i'm working on specs today | 21:21 |
| *** jcoufal has quit IRC | 21:21 | |
| rloo | devananda: ok, always good to procrastinate. I'll check back with you next week ;) | 21:22 |
| *** Penick has joined #openstack-ironic | 21:28 | |
| adam_g | is there a way i can force a node out of 'wait call-back' state so that it can be deleted by nova? | 21:31 |
| devananda | adam_g: waitcallback should be deletable | 21:35 |
| devananda | adam_g: if it's in an active deploying state, it can't be interrupted. but wait callback, last I tried it, was interruptible | 21:36 |
| *** Penick has quit IRC | 21:36 | |
| *** mrda_away is now known as mrda | 21:43 | |
| mrda | Morning ironic! | 21:43 |
| devananda | g'morning, mrda ! | 21:45 |
| mrda | hey devananda | 21:50 |
| *** hipster has quit IRC | 21:55 | |
| *** krtaylor has quit IRC | 22:05 | |
| adam_g | devananda, thoughts on making the virutal ironic job voting in the ironic gate? i was into the idea a few weeks ago, but then a change slipped into tempest that broke the job for a few days everywhere | 22:15 |
| adam_g | it could have been caught in the tempest gate if the virtual-ironic job was voting, but AIUI that can't happen until graduation/integration | 22:15 |
| adam_g | but maybe there was talk about this last week at summit | 22:15 |
| devananda | adam_g: strongly in favor of it. though we may want to fix the bug I posted yesterday, or at least post an elastic recheck for it | 22:15 |
| devananda | and yea, we can't vote on other projects yet | 22:16 |
| devananda | but that should be voting on ironic and python-ironic-client | 22:16 |
| devananda | (we cant vote on integrated projects until Ironic gradautes) | 22:16 |
| adam_g | devananda, right, im just worried more things slipping thru tempest or similar and it blocking ironic code from landing till its fixed | 22:17 |
| devananda | adam_g: as am I. however, we still need to do it | 22:18 |
| adam_g | devananda, cool | 22:19 |
| devananda | two things on this from the summit | 22:20 |
| *** uberj has quit IRC | 22:20 | |
| devananda | * folks on the QA team should start paying attention to our jobs, even though they're not voting yet. sdague is aware of this, but not everyone else is yet | 22:21 |
| *** uberj has joined #openstack-ironic | 22:21 | |
| devananda | * we should be reporting to nova in the same way that other third-party drivers are reporting | 22:21 |
| devananda | because folks actually look at those | 22:21 |
| devananda | (as opposed to the jenkins-run nonvoting jobs, which the nova-core team mostly ignores) | 22:21 |
| adam_g | devananda, as in a separate 3rd party CI system? | 22:22 |
| devananda | adam_g: want to work with infra on getting our virtual-ironic job, which is already run on every nova reviwe, to *report* as though it's a third-party CI ? | 22:22 |
| devananda | it still needs to run in the same way it runs today. it just needs to report differently | 22:22 |
| adam_g | ah okay | 22:22 |
| adam_g | yea. sounds doable | 22:22 |
| adam_g | wonder if there are any others that do similar | 22:22 |
| mrda | adam_g: yay! | 22:23 |
| devananda | jeblair had some ideas as to how we can do that, whcih I didn't really understand | 22:23 |
| devananda | adam_g: afaik, nope. it was quite an odd discussion :) | 22:23 |
| devananda | it's also temporary | 22:23 |
| devananda | once ironic graduates, that'll no longer be necessary | 22:23 |
| adam_g | right | 22:24 |
| devananda | I believe it was mikal's suggestion, as a means to increase awareness of nova changes that break ironic prior to graduation | 22:24 |
| mrda | devananda: it's all about cultural change - teaching nova reviewers to care about what we're doing, and gaining their trust, so that there will be no blockers to voting going on eventually | 22:25 |
| devananda | mrda: right | 22:25 |
| mrda | I think it's a great thing! | 22:26 |
| mrda | devananda: in other news, one of my peers has been doing work on the ironic horizon tab. Here's his screenshots to date: http://neill.id.au/horizon-ironic/screenshots | 22:29 |
| devananda | mrda: see the ML thread on this? jcoufal is working on this as well, and the two of them should collaborate :) | 22:30 |
| mrda | no reviews have taken place so far, but he's made some progress and probably isn't far from seeking thoughts | 22:30 |
| *** radsy has joined #openstack-ironic | 22:31 | |
| *** sanek11 has joined #openstack-ironic | 22:31 | |
| mrda | devananda: I haven't seen the thread from jcoufal. I'll take a look | 22:31 |
| devananda | good stuff though :) | 22:31 |
| mrda | :) | 22:31 |
| *** dividebin has joined #openstack-ironic | 22:31 | |
| *** dividehex has quit IRC | 22:33 | |
| *** igor__ has joined #openstack-ironic | 22:34 | |
| *** morgabra_ has joined #openstack-ironic | 22:36 | |
| *** morgabra has quit IRC | 22:38 | |
| *** agordeev has quit IRC | 22:38 | |
| *** anteaya has quit IRC | 22:38 | |
| *** igor_ has quit IRC | 22:38 | |
| *** uberj has quit IRC | 22:38 | |
| *** jgrimm has quit IRC | 22:39 | |
| *** r-daneel has quit IRC | 22:41 | |
| *** uberj has joined #openstack-ironic | 22:42 | |
| *** Hefeweiz1n has joined #openstack-ironic | 22:47 | |
| *** rwsu has quit IRC | 22:49 | |
| *** linggao has quit IRC | 22:49 | |
| *** sseago_ has quit IRC | 22:49 | |
| *** morgabra_ is now known as morgabra | 22:50 | |
| *** morgabra has quit IRC | 22:50 | |
| *** morgabra has joined #openstack-ironic | 22:50 | |
| *** Hefeweizen has quit IRC | 22:55 | |
| *** anteaya has joined #openstack-ironic | 22:56 | |
| *** zdin0bot has joined #openstack-ironic | 22:57 | |
| *** eguz has joined #openstack-ironic | 22:59 | |
| *** eghobo has quit IRC | 23:02 | |
| *** sseago_ has joined #openstack-ironic | 23:09 | |
| *** nosnos has joined #openstack-ironic | 23:10 | |
| *** linggao has joined #openstack-ironic | 23:10 | |
| *** rwsu has joined #openstack-ironic | 23:12 | |
| *** lucas-afk has quit IRC | 23:13 | |
| *** zdin0bot has quit IRC | 23:29 | |
| *** pradipta is now known as pradipta_away | 23:29 | |
| *** harlowja has quit IRC | 23:33 | |
| *** harlowja_ has joined #openstack-ironic | 23:33 | |
| openstackgerrit | Dan Prince proposed a change to openstack/ironic: Port iBoot PDU driver from Nova https://review.openstack.org/50977 | 23:34 |
| *** zdin0bot has joined #openstack-ironic | 23:40 | |
| *** openstackgerrit has quit IRC | 23:49 | |
| *** openstackgerrit has joined #openstack-ironic | 23:51 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!