| *** betherly has joined #openstack-ironic | 00:28 | |
| *** betherly has quit IRC | 00:33 | |
| *** tiendc has joined #openstack-ironic | 01:00 | |
| *** betherly has joined #openstack-ironic | 01:01 | |
| *** betherly has quit IRC | 01:05 | |
| *** rloo has quit IRC | 01:08 | |
| *** gyee has quit IRC | 01:16 | |
| *** betherly has joined #openstack-ironic | 01:33 | |
| *** betherly has quit IRC | 01:37 | |
| *** jhesketh has joined #openstack-ironic | 02:12 | |
| openstackgerrit | Kaifeng Wang proposed openstack/ironic master: Workaround for sendfile size limit https://review.openstack.org/622114 | 02:15 |
|---|---|---|
| *** betherly has joined #openstack-ironic | 02:15 | |
| *** betherly has quit IRC | 02:20 | |
| *** dsneddon has quit IRC | 02:22 | |
| *** rh-jelabarre has quit IRC | 02:25 | |
| openstackgerrit | Uros Orozel proposed openstack/ironic master: UEFI iPXE provisioning fails when ipxe_use_swift is enabled https://review.openstack.org/648074 | 02:34 |
| *** dsneddon has joined #openstack-ironic | 02:48 | |
| *** whoami-rajat has joined #openstack-ironic | 02:58 | |
| *** betherly has joined #openstack-ironic | 03:08 | |
| *** betherly has quit IRC | 03:13 | |
| *** betherly has joined #openstack-ironic | 04:02 | |
| *** betherly has quit IRC | 04:07 | |
| *** Lucas_Gray has joined #openstack-ironic | 04:14 | |
| *** Lucas_Gray has quit IRC | 04:43 | |
| *** Lucas_Gray has joined #openstack-ironic | 04:45 | |
| *** rcernin has quit IRC | 04:47 | |
| *** mkrai has joined #openstack-ironic | 04:48 | |
| *** betherly has joined #openstack-ironic | 04:55 | |
| *** betherly has quit IRC | 04:59 | |
| *** rcernin has joined #openstack-ironic | 05:01 | |
| *** whoami-rajat has quit IRC | 05:07 | |
| *** jaypipes has quit IRC | 05:09 | |
| *** jaypipes has joined #openstack-ironic | 05:09 | |
| *** stendulker has joined #openstack-ironic | 05:11 | |
| openstackgerrit | Nisha Agarwal proposed openstack/ironic master: RPC changes for Graphical Console https://review.openstack.org/640827 | 05:11 |
| *** Lucas_Gray has quit IRC | 05:22 | |
| openstackgerrit | Digambar proposed openstack/ironic master: Modify the iDRAC driver to use realtime RAID creation https://review.openstack.org/634903 | 05:31 |
| *** rachit7 has joined #openstack-ironic | 05:34 | |
| *** betherly has joined #openstack-ironic | 05:48 | |
| *** betherly has quit IRC | 05:52 | |
| *** rcernin has quit IRC | 06:03 | |
| *** whoami-rajat has joined #openstack-ironic | 06:31 | |
| *** betherly has joined #openstack-ironic | 06:40 | |
| *** betherly has quit IRC | 06:45 | |
| *** rcernin has joined #openstack-ironic | 06:47 | |
| *** rcernin has quit IRC | 06:47 | |
| *** rcernin has joined #openstack-ironic | 06:48 | |
| openstackgerrit | Kaifeng Wang proposed openstack/ironic master: Workaround for sendfile size limit https://review.openstack.org/622114 | 07:03 |
| *** e0ne has joined #openstack-ironic | 07:17 | |
| *** e0ne has quit IRC | 07:19 | |
| *** pcaruana has joined #openstack-ironic | 07:20 | |
| arne_wiebalck | good morning, ironic! | 07:33 |
| *** rpittau|afk is now known as rpittau | 07:41 | |
| rpittau | good morning ironic! o/ | 07:41 |
| *** rcernin has quit IRC | 07:48 | |
| arne_wiebalck | good morning rpittau o/ | 07:50 |
| rpittau | hey arne_wiebalck :) | 07:50 |
| *** tssurya has joined #openstack-ironic | 08:03 | |
| *** rcernin has joined #openstack-ironic | 08:04 | |
| *** betherly has joined #openstack-ironic | 08:06 | |
| *** iurygregory_ has joined #openstack-ironic | 08:10 | |
| *** betherly has quit IRC | 08:11 | |
| *** iurygregory has quit IRC | 08:12 | |
| *** iurygregory_ is now known as iurygregory | 08:12 | |
| *** Lucas_Gray has joined #openstack-ironic | 08:13 | |
| *** rachit7 has quit IRC | 08:15 | |
| iurygregory | morning ironic o/ | 08:22 |
| rpittau | hey iurygregory :) | 08:22 |
| mkrai | GM Ironic! | 08:54 |
| *** Wryhder has joined #openstack-ironic | 08:54 | |
| *** Lucas_Gray has quit IRC | 08:57 | |
| *** Wryhder is now known as Lucas_Gray | 08:57 | |
| mkrai | Hi, is there any API in Nova that can map to our cleaning API? | 08:58 |
| *** betherly has joined #openstack-ironic | 08:59 | |
| *** betherly has quit IRC | 09:04 | |
| *** S4ren has joined #openstack-ironic | 09:06 | |
| arne_wiebalck | mkrai: What do you mean, trigger ironic cleaning via nova? | 09:06 |
| mkrai | arne_wiebalck: Right | 09:08 |
| arne_wiebalck | mkrai: delete in nova will trigger a clean in ironic, no? | 09:10 |
| *** oanson has quit IRC | 09:10 | |
| mkrai | arne_wiebalck: Explicitly cleaning only | 09:10 |
| arne_wiebalck | mkrai: like you have a node in ironic which does not have an instance and you'd like to clean it via nova? | 09:11 |
| mkrai | arne_wiebalck: Is it possible? | 09:13 |
| mkrai | Nova would not know about the Ironic node in that case | 09:13 |
| arne_wiebalck | mkrai: that would have been my next question: how do you refer to an ironic node from nova? | 09:14 |
| mkrai | arne_wiebalck: Actually I am working on this feature which is manual deployment step | 09:14 |
| mkrai | Which will be similar to clean API specifying deploy_steps | 09:14 |
| *** diga has joined #openstack-ironic | 09:14 | |
| mkrai | So to rephrase my question, I am thinking how would we expose the manual deploy API from Nova | 09:15 |
| mkrai | arne_wiebalck: Sorry for the confusion | 09:15 |
| diga | Hello Ironic | 09:16 |
| diga | arne_wiebalck rpittau iurygregory: Hello | 09:17 |
| iurygregory | o/ | 09:17 |
| arne_wiebalck | mkrai: mgoddard would be the one to talk to about this | 09:18 |
| mkrai | arne_wiebalck: Ok thank you for your response :) | 09:18 |
| mkrai | mgoddard: Hi! | 09:18 |
| arne_wiebalck | mkrai: FWIU, the deploy step framework will be more or less in analogy how the clean steps are done | 09:19 |
| openstackgerrit | Mark Goddard proposed openstack/bifrost stable/rocky: Use a virtualenv in openSUSE CI testing https://review.openstack.org/648221 | 09:19 |
| rpittau | hi diga :) | 09:19 |
| openstackgerrit | Mark Goddard proposed openstack/bifrost stable/rocky: Use a virtualenv in openSUSE CI testing https://review.openstack.org/648221 | 09:19 |
| arne_wiebalck | mkrai: but users will have the possibility to execute certain deploy steps when deploying an instance | 09:19 |
| * arne_wiebalck better leaves the floor to mgoddard | 09:20 | |
| arne_wiebalck | hi diga | 09:20 |
| diga | arne_wiebalck: Hey | 09:20 |
| mkrai | arne_wiebalck: The use case is allow deploy steps after the node is provisioned | 09:20 |
| mgoddard | hi mkrai arne_wiebalck | 09:20 |
| mkrai | And the deploy template works while provisioning only | 09:20 |
| mkrai | Hi mgoddard | 09:21 |
| arne_wiebalck | mkrai: yes, that's also a use case, I have a story for it ... | 09:21 |
| diga | arne_wiebalck: I updated my patch, not sure reason behind why zuul is failing - https://review.openstack.org/#/c/634903/ | 09:21 |
| patchbot | patch 634903 - ironic - Modify the iDRAC driver to use realtime RAID creation - 11 patch sets | 09:21 |
| * arne_wiebalck is looking for the active node deploy step story ... | 09:21 | |
| mkrai | mgoddard: I was discussing about the manual deploy step feature with arne_wiebalck | 09:21 |
| diga | do I need recheck ? | 09:22 |
| arne_wiebalck | mkrai: https://storyboard.openstack.org/#!/story/2005129 | 09:22 |
| mgoddard | mkrai: just catching up | 09:22 |
| mkrai | arne_wiebalck: Here it is https://storyboard.openstack.org/#!/story/2005129 | 09:22 |
| mgoddard | mkrai: we don't really expose anything too baremetal-specific in nova | 09:22 |
| mkrai | mgoddard: Ok I agree | 09:23 |
| mgoddard | mkrai: instead we try to abstract things to use the same interfaces as VMs | 09:23 |
| mkrai | mgoddard: Yes so I am looking for similar interface in Nova | 09:23 |
| *** betherly has joined #openstack-ironic | 09:23 | |
| mgoddard | mkrai: this is for during deployment? | 09:23 |
| mkrai | mgoddard: post deployment | 09:24 |
| mgoddard | mkrai: what is the use case? | 09:24 |
| arne_wiebalck | mkrai: coud you provide a use case example? | 09:24 |
| mkrai | arne_wiebalck: mgoddard Yes | 09:24 |
| rpittau | diga: you can recheck, that error unfortunately sometimes happen in that job | 09:24 |
| mkrai | Run a deploy_step "do_nothing" post deployment via Nova | 09:24 |
| *** lucasagomes has joined #openstack-ironic | 09:25 | |
| mgoddard | mkrai: that doesn't sound very useful :p | 09:25 |
| mkrai | the do_nothing can be changing a boot parameter post deployment which might need a power recycle as well | 09:25 |
| mkrai | mgoddard: Trying to elaborate :) Sorry | 09:25 |
| mgoddard | ok, I think we have discussed this before | 09:25 |
| arne_wiebalck | mgoddard: mkrai : I think my story ^ has an example | 09:26 |
| mkrai | mgoddard: Yes. | 09:26 |
| mkrai | Copying from the story, A real-life example for such a required post-deploy configuration change is the change of BIOS settings to disable hyper-threading in order to address a security vulnerability. | 09:26 |
| arne_wiebalck | mgoddard: mkrai : changing BIOS settings post deployment | 09:26 |
| diga | rpittau: Sure. Thanks man | 09:27 |
| mgoddard | there is no such API currently. After ironic deploys an instance it does not really do much other than monitor and set the power state | 09:27 |
| mkrai | mgoddard: Yes I know. I am talking about the post deploy step feature implementation :) | 09:27 |
| *** betherly has quit IRC | 09:28 | |
| mgoddard | I'm struggling to see how we would fit this into nova | 09:28 |
| openstackgerrit | Digambar proposed openstack/ironic master: Modify the iDRAC driver to use realtime RAID creation https://review.openstack.org/634903 | 09:28 |
| mkrai | mgoddard: If we implement this feature in Ironic. How do we expose via Nova? | 09:28 |
| arne_wiebalck | the story does not suggest to expose this to nova | 09:28 |
| mgoddard | we have deploy steps in ironic, could easily provide a list of steps to run | 09:28 |
| mgoddard | but I don't see how we would map that to a concept in nova | 09:28 |
| mkrai | mgoddard: I tried looking for similar interface in Nova and think rebuild might be a possible option | 09:29 |
| arne_wiebalck | mkrai: mgoddard: I don't think this should be exposed via nova. | 09:29 |
| mgoddard | mkrai: you could rebuild, but it would overwrite the image on the disk | 09:30 |
| arne_wiebalck | I am not sure the owner of an instance should have the rights to reconfigure the BIOS. | 09:30 |
| mgoddard | arne_wiebalck: with deploy templates they implicitly could do | 09:31 |
| arne_wiebalck | mgoddard: right, but in a controlled way | 09:31 |
| mgoddard | yes | 09:31 |
| arne_wiebalck | it's probably a grey area | 09:31 |
| arne_wiebalck | in our cloud we basically hand the instance over to the user | 09:31 |
| arne_wiebalck | the user can do anything with it | 09:32 |
| mgoddard | even the BMC? | 09:32 |
| arne_wiebalck | yes | 09:32 |
| arne_wiebalck | mostly for "backwards compatibility" | 09:32 |
| mgoddard | do you reimage the BMC firmware afterwards? | 09:32 |
| arne_wiebalck | we should :D | 09:32 |
| arne_wiebalck | we plan to do that, yes | 09:33 |
| mgoddard | or could check checksums | 09:33 |
| arne_wiebalck | atm there are other DBs keeping track of credentials as well, so we need to make sure these are updated once we reset things | 09:33 |
| arne_wiebalck | we plan to count the number of admins accounts and reset the pw | 09:34 |
| arne_wiebalck | so, in principle users can already now change BIOS settings (in our cloud) | 09:35 |
| arne_wiebalck | the story I created was more for fleet-wide campaigns | 09:35 |
| arne_wiebalck | change the BIOS on 1000 nodes | 09:35 |
| mkrai | arne_wiebalck: mgoddard Thank you for the discussion :) | 09:36 |
| arne_wiebalck | anyway, I think it'll be hard to do that via nova | 09:36 |
| *** derekh has joined #openstack-ironic | 09:37 | |
| *** oanson has joined #openstack-ironic | 09:41 | |
| *** e0ne has joined #openstack-ironic | 09:47 | |
| diga | dtantsur|afk: Hi | 09:54 |
| diga | ohh. looks like he is away | 09:54 |
| *** priteau has joined #openstack-ironic | 09:55 | |
| openstackgerrit | Digambar proposed openstack/ironic stable/rocky: [WIP] Remove nic method dependencies from dracclient==1.5.0 from stable releases https://review.openstack.org/648360 | 09:55 |
| *** dtantsur|afk is now known as dtantsur | 09:59 | |
| dtantsur | hi diga | 10:00 |
| dtantsur | morning ironic | 10:00 |
| rpittau | hi dtantsur :) | 10:01 |
| dtantsur | eandersson: hi! the reasoning was that each hardware type has someone responsible for it (community for ipmi, <vendor name> for <vendor type>), etc. and we wanted these people to define what is compatible with what hardware type. | 10:01 |
| dtantsur | I do agree that networking interfaces are weird citizens here | 10:01 |
| diga | dtantsur: Hi | 10:02 |
| dtantsur | I don't mind opening this topic again, but back then some people felt very strongly | 10:02 |
| diga | dtantsur: I just pushed patch for removing dracclient 1.5.0 dependency from ironic stable version - https://review.openstack.org/#/c/648360/ | 10:02 |
| patchbot | patch 648360 - ironic (stable/rocky) - [WIP] Remove nic method dependencies from dracclie... - 1 patch set | 10:02 |
| diga | dtantsur: Its in WIP now, working on unittests | 10:03 |
| dtantsur | great! I'll check it soon | 10:03 |
| diga | :) | 10:03 |
| iurygregory | morning dtantsur | 10:17 |
| dtantsur | mgoddard: hi, did you have a chance to check my questions on https://review.openstack.org/#/c/647485/ ? | 10:18 |
| patchbot | patch 647485 - ironic - Check for deploy.deploy deploy step in heartbeat - 1 patch set | 10:18 |
| mgoddard | sorry dtantsur, got lost in the noise. Looking now | 10:19 |
| mgoddard | dtantsur: I'm not sure about touching provisioning. I don't think it will hurt, but I also don't think it affects timeout, does it? | 10:23 |
| dtantsur | mgoddard: I think we use this to determine if IPA is still alive for a node. I may be horribly wrong though. | 10:23 |
| mgoddard | dtantsur: ok, you are right - deploy timeout is CONF.conductor.deploy_callback_timeout seconds since last provision_updated_at | 10:26 |
| *** melwitt has quit IRC | 10:26 | |
| mgoddard | dtantsur: so the missing case is when in DEPLOYWAIT and not self.in_deploy_deploy_step(task)? | 10:27 |
| dtantsur | mgoddard: yep, I think so | 10:29 |
| mgoddard | dtantsur: ack | 10:29 |
| *** e0ne has quit IRC | 10:31 | |
| *** melwitt has joined #openstack-ironic | 10:32 | |
| *** e0ne has joined #openstack-ironic | 10:36 | |
| *** hoonetorg has quit IRC | 10:39 | |
| openstackgerrit | Mark Goddard proposed openstack/ironic master: Check for deploy.deploy deploy step in heartbeat https://review.openstack.org/647485 | 10:41 |
| *** mkrai has quit IRC | 10:42 | |
| openstackgerrit | Merged openstack/tenks master: Fix Zuul galaxy install task https://review.openstack.org/648185 | 10:52 |
| *** hoonetorg has joined #openstack-ironic | 10:52 | |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic master: [WIP] Allow backfilling/assigning allocations https://review.openstack.org/648178 | 10:54 |
| *** tiendc has quit IRC | 10:54 | |
| dtantsur | mgoddard: sorry, another question | 10:58 |
| *** e0ne has quit IRC | 10:58 | |
| mgoddard | dtantsur: fire away | 10:59 |
| *** e0ne has joined #openstack-ironic | 11:05 | |
| mgoddard | dtantsur: good catch | 11:06 |
| openstackgerrit | Mark Goddard proposed openstack/ironic master: Check for deploy.deploy deploy step in heartbeat https://review.openstack.org/647485 | 11:06 |
| mgoddard | So many values for a bool, it's sometimes hard to know which one to pick | 11:07 |
| openstackgerrit | Merged openstack/tenks master: Ignore COMPUTE_NET_ATTACH_INTERFACE trait when waiting for placement https://review.openstack.org/648144 | 11:08 |
| *** stendulker has quit IRC | 11:11 | |
| *** lennyb has quit IRC | 11:12 | |
| *** lennyb has joined #openstack-ironic | 11:14 | |
| dtantsur | lol | 11:17 |
| *** diga has quit IRC | 11:29 | |
| openstackgerrit | Merged openstack/sushy stable/stein: Change sushy devstack job to python3 https://review.openstack.org/648045 | 11:30 |
| *** Lucas_Gray has quit IRC | 11:37 | |
| *** w14161_1_ has quit IRC | 11:40 | |
| *** Lucas_Gray has joined #openstack-ironic | 11:49 | |
| *** jtomasek has joined #openstack-ironic | 11:51 | |
| *** cdearborn has quit IRC | 11:54 | |
| openstackgerrit | Arne Wiebalck proposed openstack/ironic-specs master: Add details on target_raid_config for Software RAID support https://review.openstack.org/643100 | 11:56 |
| *** jtomasek has quit IRC | 12:00 | |
| *** rh-jelabarre has joined #openstack-ironic | 12:01 | |
| *** Lucas_Gray has quit IRC | 12:09 | |
| *** Wryhder has joined #openstack-ironic | 12:09 | |
| *** whoami-rajat has quit IRC | 12:10 | |
| *** Wryhder is now known as Lucas_Gray | 12:10 | |
| *** trown|outtypewww is now known as trown | 12:11 | |
| openstackgerrit | Ilya Etingof proposed openstack/ironic master: Add UEFI firmware to Redfish emulator config https://review.openstack.org/648382 | 12:17 |
| *** whoami-rajat has joined #openstack-ironic | 12:20 | |
| *** w14161_1 has joined #openstack-ironic | 12:27 | |
| *** jtomasek has joined #openstack-ironic | 12:32 | |
| *** e0ne has quit IRC | 12:41 | |
| *** rloo has joined #openstack-ironic | 12:42 | |
| *** e0ne has joined #openstack-ironic | 12:43 | |
| TheJulia | dtantsur|afk: any thoughts? | 12:45 |
| TheJulia | err, thoguhts on the discussion from last night re driver interfaces and overrides | 12:45 |
| iurygregory | dtantsur, the problem was that devstack-minimal was branched yesterday so they didnt have stable/stein https://review.openstack.org/#/c/647774 https://zuul.openstack.org/status now shows ironicclient-functional and ironicclient-tempest \o/ | 12:46 |
| patchbot | patch 647774 - python-ironicclient (stable/stein) - Move to zuulv3 - 2 patch sets | 12:46 |
| * TheJulia reads | 12:46 | |
| TheJulia | I suspect people feel less strongly at this poitn, fwiw | 12:46 |
| *** jtomasek has quit IRC | 12:46 | |
| dtantsur | TheJulia: oh, I posted something about it this morning. I feel less strongly as well, maybe we should relax restrictions at least around network/storage interfaces | 12:50 |
| TheJulia | I think that is really reasonable | 12:53 |
| *** jcoufal has joined #openstack-ironic | 12:53 | |
| dtantsur | It may be a bit funky from the coding point of view though :) | 12:57 |
| jroll | I agree allowing additional network/storage interfaces should be fine | 12:58 |
| iurygregory | morning jroll o/ | 13:05 |
| iurygregory | morning TheJulia o/ | 13:05 |
| jroll | morning :) | 13:06 |
| *** jtomasek has joined #openstack-ironic | 13:07 | |
| dtantsur | morning jroll | 13:08 |
| jroll | heya dtantsur \o | 13:08 |
| *** jtomasek has quit IRC | 13:08 | |
| openstackgerrit | Merged openstack/ironic master: Workaround for sendfile size limit https://review.openstack.org/622114 | 13:09 |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic master: Allocation API: backfilling allocations https://review.openstack.org/648178 | 13:10 |
| dtantsur | huh, this ^^ proved bigger than I hoped for | 13:10 |
| *** jtomasek has joined #openstack-ironic | 13:13 | |
| *** bfournie has joined #openstack-ironic | 13:18 | |
| *** jhesketh has quit IRC | 13:18 | |
| *** e0ne has quit IRC | 13:20 | |
| *** munimeha1 has joined #openstack-ironic | 13:20 | |
| *** derekh has quit IRC | 13:24 | |
| dtantsur | TheJulia, jroll, please check https://review.openstack.org/#/c/645523/ | 13:51 |
| patchbot | patch 645523 - ironic (stable/stein) - Prepare stable/stein testing - 1 patch set | 13:51 |
| *** baha has joined #openstack-ironic | 13:56 | |
| jroll | dtantsur: +2 | 13:57 |
| jroll | thanks for doing that. | 13:57 |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic-tempest-plugin master: Allocation API: add tests on backfilling allocations https://review.openstack.org/648403 | 13:59 |
| *** derekh has joined #openstack-ironic | 14:04 | |
| TheJulia | dtantsur: thanks | 14:12 |
| TheJulia | too many meetings this week :( | 14:13 |
| bdodd | dtantsur I added a reply to your comments in settings.py of my sushy patch set. If you and/or etingof could take a look and give me feedback it would be appreciated. https://review.openstack.org/#/c/643656/1/sushy/resources/settings.py | 14:13 |
| patchbot | patch 643656 - sushy - Update model to support ApplyTime annotations - 1 patch set | 14:13 |
| dtantsur | will do | 14:13 |
| bdodd | o/ | 14:14 |
| openstackgerrit | Dmitry Tantsur proposed openstack/python-ironicclient master: Allocation API: support allocation backfilling https://review.openstack.org/648406 | 14:16 |
| openstackgerrit | Ilya Etingof proposed openstack/sushy-tools master: Fix Flask config load under WSGI https://review.openstack.org/648407 | 14:16 |
| iurygregory | People, ironic can create nodes with windows images or nodes with dual boot (Linux/Windows)? my friend asked me this question and i told him would do some search XD | 14:19 |
| dtantsur | iurygregory: if it's a whole disk image, you can do anything (in theory) | 14:19 |
| iurygregory | dtantsur, ty! | 14:20 |
| dtantsur | I'd be curious to hear if they succeed or not :) | 14:20 |
| *** jtomasek has quit IRC | 14:20 | |
| etingof | bdodd's patch reminds me that sushy models are mostly suitable for GET'ing not for POST'ing... | 14:23 |
| etingof | for example, ActionField object does not capture action parameters | 14:24 |
| iurygregory | dtantsur, I will let you know if he decides to use ironic o/ | 14:25 |
| *** jtomasek has joined #openstack-ironic | 14:26 | |
| etingof | ironic should take over their windows shop by reinstalling linux | 14:26 |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic master: Allocation API: backfilling allocations https://review.openstack.org/648178 | 14:27 |
| *** whoami-rajat has quit IRC | 14:30 | |
| dtantsur | folks, a quick poll: how stupid is it to accept names into a field called node_uuid? | 14:36 |
| dtantsur | if the alternative is to have two fields: node and node_uuid with just slightly different semantics? | 14:36 |
| *** mbeierl has joined #openstack-ironic | 14:36 | |
| dtantsur | and another alternative is to have one more place in our API where names are not accepted | 14:37 |
| dtantsur | context: https://storyboard.openstack.org/#!/story/2005014 | 14:37 |
| *** e0ne has joined #openstack-ironic | 14:38 | |
| rpittau | not stupid, just not "correct" | 14:39 |
| dtantsur | yeah, I have 3 variants that I do not like :) | 14:39 |
| rpittau | xD | 14:39 |
| dtantsur | rpittau: which one would you prefer? | 14:40 |
| rpittau | hehhhhhh | 14:40 |
| rpittau | to be honest, I would keep them separated | 14:40 |
| *** mbeierl has quit IRC | 14:41 | |
| dtantsur | so, having both node_uuid and node, even though node will not be in the result of GET? | 14:41 |
| rpittau | avoid confusion, misunderstanding, headaches | 14:41 |
| dtantsur | or just having node in POST and node_uuid in GET? | 14:41 |
| dtantsur | it's not very restful, but probably the most intuitive | 14:42 |
| rpittau | it's a hard choice :/ | 14:42 |
| dtantsur | right? | 14:43 |
| rpittau | I'd go for the most intuitive, but I'm being very practical here | 14:43 |
| * dtantsur wonders if TheJulia has a minute for ^^^ | 14:43 | |
| * TheJulia blinks | 14:43 | |
| TheJulia | sorry meetings | 14:43 |
| *** sthussey has joined #openstack-ironic | 14:44 | |
| TheJulia | ugh, because of users expecting to be able to use the name as well, I think we kind of have to support it | 14:45 |
| dtantsur | yeah, we already have POST /v1/ports that only supports UUIDs, and it is pretty inconvenient | 14:45 |
| TheJulia | yeah, we've had a few people demand/ask it be fixed | 14:45 |
| dtantsur | TheJulia: so, would you have node_uuid accept names, or rather have 'node' in POST, while returning 'node_uuid' in GET? | 14:46 |
| TheJulia | node in post I think | 14:46 |
| dtantsur | okay, I guess I'll go with it | 14:47 |
| *** mbeierl has joined #openstack-ironic | 14:57 | |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic master: Allocation API: backfilling allocations https://review.openstack.org/648178 | 15:00 |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic-tempest-plugin master: Allocation API: add tests on backfilling allocations https://review.openstack.org/648403 | 15:00 |
| *** Lucas_Gray has quit IRC | 15:01 | |
| openstackgerrit | Dmitry Tantsur proposed openstack/python-ironicclient master: Allocation API: support allocation backfilling https://review.openstack.org/648406 | 15:02 |
| openstackgerrit | Dmitry Tantsur proposed openstack/python-ironicclient master: Allocation API: support allocation backfilling https://review.openstack.org/648406 | 15:04 |
| *** hamzy has quit IRC | 15:10 | |
| *** e0ne has quit IRC | 15:14 | |
| *** baha has quit IRC | 15:18 | |
| *** jtomasek has quit IRC | 15:19 | |
| *** e0ne has joined #openstack-ironic | 15:20 | |
| *** jtomasek has joined #openstack-ironic | 15:20 | |
| openstackgerrit | Ilya Etingof proposed openstack/ironic master: WIP: Add Redfish Virtual Media support https://review.openstack.org/638453 | 15:27 |
| *** jhesketh has joined #openstack-ironic | 15:28 | |
| *** jhesketh has quit IRC | 15:34 | |
| *** hamzy has joined #openstack-ironic | 15:35 | |
| *** e0ne has quit IRC | 15:40 | |
| openstackgerrit | Merged openstack/ironic stable/stein: Prepare stable/stein testing https://review.openstack.org/645523 | 15:42 |
| *** gyee has joined #openstack-ironic | 16:12 | |
| *** jtomasek has quit IRC | 16:13 | |
| *** tssurya has quit IRC | 16:27 | |
| *** e0ne has joined #openstack-ironic | 16:27 | |
| *** whoami-rajat has joined #openstack-ironic | 16:32 | |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic-tempest-plugin master: Allocation API: add tests on backfilling allocations https://review.openstack.org/648403 | 16:46 |
| *** jtomasek has joined #openstack-ironic | 16:49 | |
| *** lucasagomes has quit IRC | 17:00 | |
| arne_wiebalck | TheJulia: dtantsur : do we have a reference architecture document for larger deployments anywhere? Quickly looking I only found one for "small clouds" ... | 17:06 |
| *** trown is now known as trown|lunch | 17:07 | |
| *** e0ne has quit IRC | 17:07 | |
| *** e0ne has joined #openstack-ironic | 17:08 | |
| arne_wiebalck | Background is that we plan to add quite some nodes over the next couple of months, starting in 2 weeks or so with an additional 600. | 17:08 |
| arne_wiebalck | And in addition to unknown issues, we already see that the resource tracker (one our 1 compute node) is getting to its limits ... | 17:09 |
| arne_wiebalck | ... with the 1700 nodes we have currently in ironic. | 17:10 |
| arne_wiebalck | So, one option we're discussing is to shard and scale out nova compute (which we had in the past, but then removed again). | 17:11 |
| arne_wiebalck | In general, I'd be interested in feedback from deployers that are one order of magnitude bigger than we are, so ~10k nodes, to prepare for the next months :) | 17:14 |
| *** e0ne has quit IRC | 17:17 | |
| dtantsur | arne_wiebalck: we need input from someone like you to write such a document :) | 17:18 |
| arne_wiebalck | dtantsur: I was afraid of that answer ;) | 17:18 |
| *** e0ne has joined #openstack-ironic | 17:21 | |
| rpittau | good night! o/ | 17:24 |
| *** rpittau is now known as rpittau|afk | 17:24 | |
| arne_wiebalck | bye rpittau|afk o/ | 17:24 |
| eandersson | dtantsur, for sure makes sense | 17:44 |
| *** e0ne has quit IRC | 17:45 | |
| jroll | heh, good answer dtantsur | 17:53 |
| jroll | arne_wiebalck: from my experience, I can totally recommend scaling out nova-compute, it should "just work" to spin more up | 17:53 |
| jroll | we have a hash ring in there, they will figure out how to distribute management of nodes | 17:54 |
| arne_wiebalck | IIRC, James mentioned you were running with one nova-compute only, no? | 17:55 |
| arne_wiebalck | Maybe my memory does not serve me well here ... | 17:56 |
| jroll | yes, we're also on juno :) | 17:56 |
| arne_wiebalck | :-D | 17:56 |
| jroll | this is from previous job experience, as well as writing the code to make nova-compute work that way :P | 17:56 |
| arne_wiebalck | jroll: juno is the reason you have only 1 nova-compute? | 17:56 |
| jroll | arne_wiebalck: part of it, yes | 17:57 |
| jroll | (we actually have two in active/passive) | 17:57 |
| jroll | (or maybe more) | 17:57 |
| arne_wiebalck | jroll: I see | 17:57 |
| arne_wiebalck | jroll: the thing getting to its limits now is the resource tracker | 17:57 |
| arne_wiebalck | it takes ~25 mins or so for one cycle | 17:57 |
| arne_wiebalck | and during that time it does not necessarily like to do other things such as scheduling instances :-S | 17:58 |
| arne_wiebalck | but the resource tracker is serial | 17:58 |
| jroll | arne_wiebalck: right, so since newton(?) if you have multiple computes, each will only manage a subset of the ironic nodes. so e.g. 10k nodes and 5 n-cpu means each manages 2k nodes. the time to run the resource tracker is roughly linear with respect to the number of nodes. | 17:58 |
| arne_wiebalck | which is fine for "normal" compute nodes | 17:58 |
| arne_wiebalck | I was wondering if it would make sense to parallelise the RT, similar to what etingof has done for ironic's power sync | 17:59 |
| jroll | it might, that's a conversation for nova folks :) | 18:00 |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic-tempest-plugin master: Allocation API: add tests on backfilling allocations https://review.openstack.org/648403 | 18:00 |
| arne_wiebalck | yeah ... probably such a parallelisation is not sensible in 99% of the deployments | 18:00 |
| jroll | I mean, it doesn't hurt in a VM deployment either | 18:01 |
| arne_wiebalck | I'm just worried that multiple n-cpu will bring some other "surprises" | 18:02 |
| arne_wiebalck | jroll: right | 18:02 |
| arne_wiebalck | jroll: or you simply configure the number of workers to 1 | 18:02 |
| jroll | arne_wiebalck: there should be minimal surprises, lemme find the commit message which has some notes | 18:02 |
| jroll | any surprises should be eventually consistent types of things | 18:02 |
| *** derekh has quit IRC | 18:02 | |
| arne_wiebalck | jroll: well, I think we had some surprises initially ... which is why we dropped to 1 n-cpu | 18:03 |
| jroll | interesting, I'd be curious to hear more about those | 18:03 |
| arne_wiebalck | jroll: b/c we started with several n-cpu | 18:03 |
| jroll | this has some notes: https://review.openstack.org/#/c/348443/18//COMMIT_MSG | 18:03 |
| patchbot | patch 348443 - nova - Ironic: allow multiple compute services (MERGED) - 18 patch sets | 18:03 |
| arne_wiebalck | awesome, thx, I'll go through it | 18:04 |
| arne_wiebalck | jroll: I think some of the issues were related with the concurrent move to cells v2 | 18:04 |
| arne_wiebalck | jroll: and with the fact that we were early adopters | 18:04 |
| arne_wiebalck | jroll: many if not all issues have meanwhile been addressed I believe | 18:05 |
| jroll | ah, cool | 18:05 |
| arne_wiebalck | jroll: our nova experts are actually pushing for scale-out :) | 18:05 |
| *** trown|lunch is now known as trown | 18:06 | |
| arne_wiebalck | jroll: thx for your input! | 18:06 |
| jroll | arne_wiebalck: no problem :) | 18:07 |
| *** dtantsur is now known as dtantsur|afk | 18:08 | |
| dtantsur|afk | see you | 18:08 |
| arne_wiebalck | dtantsur|afk: o/ | 18:09 |
| openstackgerrit | Zane Bitter proposed openstack/ironic master: Docs: correct expected host format for drac_address https://review.openstack.org/648493 | 18:11 |
| openstackgerrit | Ilya Etingof proposed openstack/ironic master: Add systemd unit for sushy emulator in devstack https://review.openstack.org/644822 | 18:12 |
| openstackgerrit | Ilya Etingof proposed openstack/ironic master: Add UEFI firmware to Redfish emulator config https://review.openstack.org/648382 | 18:12 |
| *** S4ren has quit IRC | 18:16 | |
| openstackgerrit | Ilya Etingof proposed openstack/ironic master: Add systemd unit for sushy emulator in devstack https://review.openstack.org/644822 | 18:18 |
| openstackgerrit | Ilya Etingof proposed openstack/ironic master: Add UEFI firmware to Redfish emulator config https://review.openstack.org/648382 | 18:18 |
| openstackgerrit | Ilya Etingof proposed openstack/ironic master: Run IPMI and Redfish emulators as stack https://review.openstack.org/648496 | 18:18 |
| openstackgerrit | Ilya Etingof proposed openstack/ironic master: Inhibit automatic vbmcd start in devstack https://review.openstack.org/648498 | 18:26 |
| openstackgerrit | Ilya Etingof proposed openstack/ironic master: WIP: Add Redfish Virtual Media support https://review.openstack.org/638453 | 18:27 |
| arne_wiebalck | good night everyone o/ | 18:31 |
| jroll | \o | 18:33 |
| *** jtomasek has quit IRC | 18:33 | |
| *** jtomasek has joined #openstack-ironic | 18:34 | |
| *** jtomasek has quit IRC | 18:35 | |
| *** MattMan_ has quit IRC | 19:08 | |
| *** MattMan_ has joined #openstack-ironic | 19:09 | |
| openstackgerrit | Ilya Etingof proposed openstack/ironic master: WIP: Add Redfish Virtual Media support https://review.openstack.org/638453 | 19:10 |
| openstackgerrit | Ilya Etingof proposed openstack/ironic master: Improve VirtualBMC use in Devstack https://review.openstack.org/648510 | 19:10 |
| mgoddard | now that's more like it - custom RAID config via deploy templates in under 11 minutes | 19:10 |
| openstackgerrit | Merged openstack/ironic master: Docs: correct expected host format for drac_address https://review.openstack.org/648493 | 19:10 |
| *** priteau has quit IRC | 19:27 | |
| *** e0ne has joined #openstack-ironic | 19:33 | |
| *** rcernin has quit IRC | 19:36 | |
| openstackgerrit | Mark Goddard proposed openstack/ironic master: WIP: Add iDRAC RAID deploy steps https://review.openstack.org/641731 | 19:37 |
| *** dsneddon has quit IRC | 19:38 | |
| *** jaypipes_ has joined #openstack-ironic | 19:54 | |
| *** jaypipes has quit IRC | 19:54 | |
| *** jtomasek has joined #openstack-ironic | 19:57 | |
| *** e0ne has quit IRC | 20:00 | |
| *** pcaruana has quit IRC | 20:08 | |
| *** brandor5 has joined #openstack-ironic | 20:21 | |
| *** efried has quit IRC | 20:22 | |
| brandor5 | hello everyone: I was wondering if ironic in newton can provision ppc64le systems, i tried googling but wasn't very successful | 20:22 |
| *** dsneddon has joined #openstack-ironic | 20:23 | |
| *** jtomasek has quit IRC | 20:24 | |
| *** efried has joined #openstack-ironic | 20:26 | |
| *** dsneddon has quit IRC | 20:30 | |
| TheJulia | brandor5: I think back in newton some peopel may ahve been but I'm not sure. I know we more formalized support and addded capabilities more recently to help enable that more officially, but that was more like Rocky time frame. | 20:30 |
| brandor5 | TheJulia: ok, thanks! | 20:30 |
| TheJulia | brandor5: You may want to try early tomorrow and chat with mjturek | 20:30 |
| TheJulia | Or maybe inquire with... | 20:31 |
| TheJulia | trying to think of the name | 20:31 |
| TheJulia | tonyb: You may be able to help brandor5 | 20:31 |
| brandor5 | cool, thanks for the pointers! | 20:31 |
| TheJulia | happy to help | 20:32 |
| tonyb | say waht now? | 20:33 |
| tonyb | brandor5: I didn't try newton, but thinking about the chnages I landed in queens I'd say 'not without fiddling' | 20:34 |
| tonyb | brandor5: You'd need to use 'whole disk' images, You'll probably need DIB from a newer release than newton | 20:35 |
| brandor5 | tonyb: ok, thanks for the heads up... we use whole disk by default so we're good with that | 20:37 |
| tonyb | brandor5: cool | 20:37 |
| tonyb | I can't think of anything that'd stop it | 20:37 |
| brandor5 | just trying to do some research now, i'll probably be back in a couple weeks when we're actually trying to make it happen :D | 20:38 |
| tonyb | brandor5: but I will do the thing where I point out that newton is well passed End-Of-Life so using a newer ironic would be good if you can | 20:38 |
| tonyb | brandor5: cool beans | 20:38 |
| tonyb | brandor5: ping me if you need help | 20:38 |
| brandor5 | yep, we've got the upgrade on the map and will be jumping up to rocky this summer | 20:38 |
| tonyb | brandor5: I'm in Australia (UTC+1000) sodepending where you are overlap could be tricky | 20:39 |
| brandor5 | unfortunately the ppc project is happening before that | 20:39 |
| tonyb | brandor5: Well rocky shoudl be good *and* you can use partition images \o/ | 20:39 |
| brandor5 | no worries | 20:39 |
| tonyb | brandor5: and it wil need to be ppc64le not ppc64(be) | 20:39 |
| brandor5 | yep, that's what we have :) | 20:40 |
| tonyb | cool | 20:40 |
| *** efried has quit IRC | 20:44 | |
| *** hamzy has quit IRC | 20:54 | |
| *** openstackgerrit has quit IRC | 21:07 | |
| *** trown is now known as trown|outtypewww | 21:08 | |
| *** brandor5 has quit IRC | 21:31 | |
| *** jcoufal has quit IRC | 21:31 | |
| *** andrein has joined #openstack-ironic | 21:41 | |
| *** rcernin has joined #openstack-ironic | 21:44 | |
| *** rcernin has quit IRC | 21:55 | |
| *** rcernin has joined #openstack-ironic | 22:04 | |
| *** munimeha1 has quit IRC | 22:20 | |
| *** whoami-rajat has quit IRC | 22:52 | |
| *** S4ren has joined #openstack-ironic | 23:05 | |
| *** S4ren has quit IRC | 23:10 | |
| *** S4ren has joined #openstack-ironic | 23:11 | |
| *** S4ren has quit IRC | 23:16 | |
| eandersson | TheJulia, the yahoo way of doing it seem fine. We are just doing that. | 23:33 |
| eandersson | I assumed it would make upgrade more difficult, but we have to change the driver anyway. | 23:33 |
| eandersson | Since we are upgrading from Newton to Rocky. | 23:33 |
| eandersson | I still think we should allow ops to override these constants somehow though (at least for network), but not sure how that would work. | 23:34 |
| *** dsneddon has joined #openstack-ironic | 23:35 | |
| *** dsneddon has quit IRC | 23:40 | |
| *** gyee has quit IRC | 23:57 | |
| *** jhesketh has joined #openstack-ironic | 23:58 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!