| *** mjturek has quit IRC | 00:02 | |
| *** rh-jelabarre has quit IRC | 00:38 | |
| *** gyee has quit IRC | 00:42 | |
| rpioso | dtantsur, jroll, sambetts, TheJulia, vdrok: I created an RFE for PXE boot configuration of physical ports -- https://storyboard.openstack.org/#!/story/2003580 | 00:57 |
|---|---|---|
| *** rpioso is now known as rpioso|afk | 00:57 | |
| rpioso|afk | Good nite, all | 00:57 |
| *** jcoufal has quit IRC | 01:02 | |
| *** fungi has quit IRC | 01:18 | |
| *** fungi has joined #openstack-ironic | 01:18 | |
| *** mcarden has quit IRC | 01:19 | |
| *** JayF has quit IRC | 01:19 | |
| *** dtantsur|afk has quit IRC | 01:19 | |
| *** dtantsur has joined #openstack-ironic | 01:22 | |
| *** jrcloud has joined #openstack-ironic | 01:23 | |
| *** Bhujay has joined #openstack-ironic | 01:25 | |
| *** pas-ha has quit IRC | 01:49 | |
| *** ianw has quit IRC | 01:49 | |
| *** zigo has quit IRC | 01:49 | |
| *** persia has quit IRC | 01:49 | |
| *** ianw has joined #openstack-ironic | 01:53 | |
| *** persia has joined #openstack-ironic | 01:55 | |
| *** ianw has quit IRC | 02:03 | |
| *** ianw has joined #openstack-ironic | 02:04 | |
| *** spartacloud has joined #openstack-ironic | 02:08 | |
| *** spartacloud has quit IRC | 02:09 | |
| *** jrcloud has quit IRC | 02:10 | |
| *** hamzy has quit IRC | 02:17 | |
| *** hamzy has joined #openstack-ironic | 02:18 | |
| *** rloo has quit IRC | 02:26 | |
| *** spartacloud has joined #openstack-ironic | 02:30 | |
| *** jrcloud has joined #openstack-ironic | 02:35 | |
| *** spartacloud has quit IRC | 02:35 | |
| *** openstack has joined #openstack-ironic | 02:51 | |
| *** zzzeek has joined #openstack-ironic | 02:52 | |
| *** ChanServ sets mode: +o openstack | 02:52 | |
| *** jhesketh has joined #openstack-ironic | 02:57 | |
| *** marlinc has joined #openstack-ironic | 02:59 | |
| *** openstackstatus has joined #openstack-ironic | 03:00 | |
| *** ChanServ sets mode: +v openstackstatus | 03:00 | |
| *** phuongnh has joined #openstack-ironic | 03:02 | |
| *** gkadam has joined #openstack-ironic | 03:39 | |
| *** skazi has joined #openstack-ironic | 04:50 | |
| *** spartacloud has joined #openstack-ironic | 05:05 | |
| *** spartacloud has quit IRC | 05:09 | |
| *** cjloader has joined #openstack-ironic | 05:40 | |
| *** skazi has quit IRC | 06:15 | |
| *** serlex has joined #openstack-ironic | 06:18 | |
| *** adrianc has joined #openstack-ironic | 06:35 | |
| *** skazi has joined #openstack-ironic | 06:45 | |
| etingof | TheJulia, wrt IPA telemetry, I am looking for consensus on design and features. if we have a slot at PTG for that - I think that should boost this work | 06:48 |
| *** pcaruana has joined #openstack-ironic | 06:50 | |
| *** tssurya has joined #openstack-ironic | 06:53 | |
| *** rcernin has quit IRC | 07:00 | |
| *** spartacloud has joined #openstack-ironic | 07:10 | |
| *** jtomasek has joined #openstack-ironic | 07:12 | |
| *** spartacloud has quit IRC | 07:15 | |
| *** jesusaur has quit IRC | 07:23 | |
| *** jesusaur has joined #openstack-ironic | 07:27 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/ironic-ui master: Imported Translations from Zanata https://review.openstack.org/597411 | 07:44 |
| *** threestrands has quit IRC | 07:50 | |
| *** e0ne has joined #openstack-ironic | 07:56 | |
| *** zigo has joined #openstack-ironic | 07:57 | |
| *** e0ne has quit IRC | 08:00 | |
| *** S4ren1 has quit IRC | 08:06 | |
| *** mgoddard has joined #openstack-ironic | 08:26 | |
| *** e0ne has joined #openstack-ironic | 08:38 | |
| dtantsur | morning ironic | 08:42 |
| *** jaganathan has joined #openstack-ironic | 08:44 | |
| *** stendulker has joined #openstack-ironic | 08:44 | |
| etingof | dtantsur, o/ | 08:47 |
| dtantsur | etingof: morning! https://all-systems-go.io/ seems to have a few relevant talks for us, wanna check it out? | 08:49 |
| etingof | dtantsur, some talks sound very interesting indeed! I hope you'll make it to the conference! | 08:58 |
| *** serlex has quit IRC | 09:09 | |
| *** Bhujay has quit IRC | 09:10 | |
| *** S4ren has joined #openstack-ironic | 09:20 | |
| stendulker | Good morning dtantsur, etingof | 09:31 |
| dtantsur | hi stendulker | 09:32 |
| etingof | stendulker, \o | 09:32 |
| stendulker | dtantsur: One query, I created a fedora iso image using kernel/ramdisk from partion image created using DIB and set the command line argument to have 'root=/dev/ram0'. It loads the kernel/ramdisk during vmedia boot and then gets stuck. No error. Any clue on triaging it? | 09:36 |
| stendulker | Basically trying to implement 'ramdisk' deploy interface support for virtual media interface.. | 09:37 |
| stendulker | Tried passing rd.shell and rd.debug as command line args, but no o/p for that either. | 09:39 |
| openstackgerrit | Kaifeng Wang proposed openstack/ironic master: Direct deploy serve HTTP images from conductor https://review.openstack.org/565474 | 09:40 |
| dtantsur | stendulker: hmm, not sure. make sure kernel is not eating anything from the screen, disable KMS, etc | 09:43 |
| stendulker | dtantsur: screen is blank actually. No messag eat all. Just 'Loading /vmlinuz .... ok' and 'Loading /initrd .... ok | 09:44 |
| dtantsur | stendulker: are you sure it's not modesetting right away? | 09:45 |
| stendulker | dtantsur: How do I check that? | 09:46 |
| *** priteau has joined #openstack-ironic | 09:46 | |
| *** spartacloud has joined #openstack-ironic | 09:47 | |
| dtantsur | stendulker: probably the only easy way is to figure out how to disable it (there is a kernel argument, I don't remember it) | 09:48 |
| stendulker | ok, let me try googling it | 09:48 |
| *** sambetts|afk has joined #openstack-ironic | 09:51 | |
| *** spartacloud has quit IRC | 09:52 | |
| stendulker | dtantsur: Its nomodeset | 09:54 |
| dtantsur | yeah, right | 09:54 |
| stendulker | dtantsur: Thank you. Will try with this. | 09:54 |
| *** adrianc has quit IRC | 09:57 | |
| *** adrianc has joined #openstack-ironic | 10:14 | |
| *** mgoddard has quit IRC | 10:28 | |
| dtantsur | rpioso|afk: good idea, commented on the RFE. | 10:33 |
| *** adrianc_ has joined #openstack-ironic | 10:34 | |
| *** adrianc has quit IRC | 10:34 | |
| *** jaganathan has quit IRC | 10:37 | |
| *** hshiina has joined #openstack-ironic | 10:46 | |
| *** serlex has joined #openstack-ironic | 10:51 | |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic master: Adds support for 'ramdisk' deploy with 'ilo-virtual-media' boot https://review.openstack.org/597448 | 10:58 |
| *** d0ugal has quit IRC | 11:03 | |
| *** rpioso|afk is now known as rpioso | 11:07 | |
| rpioso | Good morning, ironicers | 11:07 |
| *** mgoddard has joined #openstack-ironic | 11:07 | |
| rpioso | dtantsur: Thank you. I'll respond on IRC in a few hours. | 11:08 |
| *** rpioso is now known as rpioso|afk | 11:09 | |
| *** bfournie has quit IRC | 11:17 | |
| *** d0ugal has joined #openstack-ironic | 11:17 | |
| *** stendulker has quit IRC | 11:20 | |
| *** jcoufal has joined #openstack-ironic | 11:24 | |
| *** serlex has quit IRC | 11:25 | |
| *** arne_wiebalck_ has joined #openstack-ironic | 11:42 | |
| *** moshele has joined #openstack-ironic | 11:50 | |
| *** bfournie has joined #openstack-ironic | 11:50 | |
| *** nizam037 has joined #openstack-ironic | 11:53 | |
| *** rh-jelabarre has joined #openstack-ironic | 12:03 | |
| *** spartacloud has joined #openstack-ironic | 12:05 | |
| *** phuongnh has quit IRC | 12:05 | |
| *** spartacloud has quit IRC | 12:10 | |
| nizam037 | when the rabbitmq service is down and when i execute the ironic command the ironicclient is not returning to the user. Is there any parameter where i need to set so that ironicclient can return to the user. | 12:13 |
| dtantsur | nizam037: you can look for oslo.messaging configuration options related to timeouts, I guess | 12:15 |
| dtantsur | (note that it happens on ironic side, not ironicclient side) | 12:15 |
| nizam037 | dtantsur : can you tel me which parameter .. i searched i am not getting it. | 12:18 |
| dtantsur | nizam037: I don't really know, something in https://docs.openstack.org/oslo.messaging/latest/configuration/opts.html#oslo-messaging-rabbit maybe | 12:19 |
| openstackgerrit | Michal Arbet proposed openstack/ironic master: Fix async keyword for Python 3.7 https://review.openstack.org/597465 | 12:22 |
| *** trown|outtypewww is now known as trown | 12:25 | |
| openstackgerrit | Michal Arbet proposed openstack/ironic master: Fix async keyword for Python 3.7 https://review.openstack.org/597468 | 12:29 |
| TheJulia | good morning | 12:39 |
| dtantsur | morning TheJulia | 12:40 |
| *** gkadam has quit IRC | 12:41 | |
| TheJulia | i see virtual media ramdisk booting seems to be the next thing | 12:43 |
| TheJulia | \o/ | 12:43 |
| dtantsur | yay | 12:44 |
| *** skazi has quit IRC | 12:48 | |
| TheJulia | rpioso left a comment on your rfe, thanks for getting that discussion started | 12:50 |
| *** dtantsur is now known as dtantsur|bbl | 13:04 | |
| *** pcaruana has quit IRC | 13:04 | |
| *** skazi has joined #openstack-ironic | 13:06 | |
| *** skazi has quit IRC | 13:07 | |
| *** skazi has joined #openstack-ironic | 13:08 | |
| jroll | g'morning | 13:27 |
| nizam037 | i asked in channel...but i dint get the answer.. So asking you..please help me out | 13:32 |
| TheJulia | nizam037: if rabbit is down, I think you should be able to list all nodes, but any action upon them would be require rabbitmq to be working so messages could be passed from the api to the conductor | 13:35 |
| TheJulia | nizam037: note: I've not tried anything like that in something like two years | 13:35 |
| openstackgerrit | weizj proposed openstack/ironic stable/ocata: fix bug link in readme https://review.openstack.org/597521 | 13:37 |
| nizam037 | yeah.. But i was trying to validate a node and but the ironicclient was not returning to the user... So is there any parameter where we can set this timeout so that after some time ironicclient can throw error and return to the user | 13:39 |
| *** rloo has joined #openstack-ironic | 13:40 | |
| openstackgerrit | weizj proposed openstack/ironic-tempest-plugin master: fix bug link in readme https://review.openstack.org/597522 | 13:41 |
| jroll | nizam037: did you read the link dmitry gave you? | 13:41 |
| *** rajinir has joined #openstack-ironic | 13:42 | |
| *** cdearborn has joined #openstack-ironic | 13:44 | |
| *** mjturek has joined #openstack-ironic | 13:44 | |
| *** baha has joined #openstack-ironic | 13:55 | |
| *** dtantsur|bbl is now known as dtantsur | 13:58 | |
| dtantsur | morning jroll | 13:58 |
| *** rloo has quit IRC | 14:03 | |
| *** rloo has joined #openstack-ironic | 14:04 | |
| *** cdearborn has quit IRC | 14:05 | |
| *** rloo has quit IRC | 14:10 | |
| *** spartacloud has joined #openstack-ironic | 14:11 | |
| *** rloo has joined #openstack-ironic | 14:11 | |
| *** spartacloud has quit IRC | 14:15 | |
| TheJulia | nizam037: For clarity, if rabbitmq is down, the API functionally cannot request the conductor to do anything and the API is at the mecy of oslo.messaging to return an error. | 14:20 |
| NobodyCam | Good Morning Ironic'ers | 14:22 |
| *** moshele has quit IRC | 14:29 | |
| *** mgoddard has quit IRC | 14:45 | |
| *** pcaruana has joined #openstack-ironic | 14:50 | |
| *** nizam037 has quit IRC | 14:56 | |
| dtantsur | morning NobodyCam | 15:02 |
| NobodyCam | Morning dtantsur :) | 15:03 |
| *** jtomasek has quit IRC | 15:29 | |
| *** rloo has quit IRC | 15:31 | |
| *** rloo has joined #openstack-ironic | 15:31 | |
| *** r-daneel has joined #openstack-ironic | 15:48 | |
| *** rpioso|afk is now known as rpioso | 15:48 | |
| rpioso | TheJulia: Thank you for commenting on the PXE RFE. | 15:49 |
| TheJulia | rpioso: no problem :) | 15:50 |
| * TheJulia hopes she is being helpful | 15:51 | |
| rpioso | dtantsur, TheJulia: Which of your approaches would be more operator friendly when the clean steps are not fleshed out/supported by the h/w type? | 15:53 |
| dtantsur | rpioso: I think our standard approach is simply not implementing them | 16:00 |
| *** tssurya has quit IRC | 16:04 | |
| rpioso | dantsur: Does that mean the mixin suggestion would be more consistent with our standard approach? | 16:08 |
| rpioso | dtantsur: ^^^ | 16:09 |
| dtantsur | mixin suggestion? | 16:09 |
| dtantsur | sorry, I may be slightly out of context, I missed the discussion yesterday | 16:09 |
| *** adrianc_ has quit IRC | 16:10 | |
| rpioso | https://storyboard.openstack.org/#!/story/2003580 | 16:10 |
| rpioso | TheJulia's feedback | 16:10 |
| TheJulia | Yeah, I was thinking the interface class could pull in a mix-in that contains the clean step | 16:11 |
| TheJulia | that way the clean step is re-usable, the code is centralized, and it has all the governing logic in case someone did something like... change the management interface to one without methods or something crazy | 16:11 |
| * TheJulia was just kind of tossing the idea out there | 16:12 | |
| * TheJulia is going to go get coffee and allow her brain to unwind for a few minutes | 16:12 | |
| rpioso | Seems like the mixin would provide consistent clean step interfaces that could optionally be implemented by management h/w interface implementations. | 16:13 |
| rpioso | The base class wouldn't implement it. | 16:14 |
| rpioso | ManagementInterface would not list the mixin as a base class. | 16:14 |
| sambetts|afk | personally I think that makes sense, then vendor managment interface can just add the mixin when they decide to support that clean step | 16:15 |
| *** baha has quit IRC | 16:16 | |
| *** dsneddon has quit IRC | 16:18 | |
| *** dsneddon has joined #openstack-ironic | 16:23 | |
| *** dsneddon has quit IRC | 16:27 | |
| *** trown is now known as trown|lunch | 16:33 | |
| *** ptx0 has quit IRC | 16:35 | |
| rpioso | sambetts|afk: Is setting the PXE IP protocol version, IPv4 or IPv6, worth considering? | 16:37 |
| rpioso | What about enable/disable VLAN and VLAN ID? | 16:38 |
| rpioso | Or is that more than what's needed? | 16:38 |
| *** r-daneel has quit IRC | 16:40 | |
| sambetts|afk | i think that sort of thing is dependant on the network interface being used, e.g. when using the neutron network interface the vlan and ip allocation is all handled by neutron and we learn the information we care about from the neutron port that we create, so there is no need for it to be set on the Ironic port, however if someone implemented a network interface that talked directly to an | 16:41 |
| sambetts|afk | SDN or TOR for example you might want something like that | 16:41 |
| sambetts|afk | but I don't think it would be set on the port, I expect it would be set in the driver_info on the node | 16:42 |
| sambetts|afk | as like driver_info/provisioning_vlan for example | 16:42 |
| *** arne_wiebalck_ has quit IRC | 16:43 | |
| TheJulia | I'm in total agreement with sambetts|afk | 16:46 |
| rpioso | What about for stand alone ironic and the flat network interface? | 16:47 |
| *** hshiina has quit IRC | 16:48 | |
| sambetts|afk | standalone ironic is the case for a network interface that talks direct to an SDN, but I don't know of anyone implmenting anything like that ... *cough* yet *cough* | 16:48 |
| *** hshiina has joined #openstack-ironic | 16:48 | |
| TheJulia | flat implies hard wired and static configuration, and standalone is mainly built on flat although it allows a user to do anything they really want to as long as they somehow gain connectivity | 16:48 |
| sambetts|afk | flat is still == neutron too, so all that info comes from neutron | 16:48 |
| sambetts|afk | the flat network interface that is, that really should be called "provider" or "static", because it works fine with neutron provider vlan networks | 16:50 |
| TheJulia | ++ | 16:50 |
| * sambetts|afk considers putting up an RFE to rename that interface | 16:50 | |
| rpioso | Would those additional settings be useful for the tripleo undercloud use case? | 16:50 |
| rpioso | When configuring servers that will be part of the overcloud? | 16:51 |
| *** mariojv has joined #openstack-ironic | 16:51 | |
| sambetts|afk | the undercloud uses neutron flat networking for provisioning at the moment, and even if it did use dynmaic vlans I expect that would be configured through neutron | 16:52 |
| *** S4ren has quit IRC | 16:52 | |
| *** dtantsur is now known as dtantsur|afk | 16:54 | |
| rpioso | sambetts|afk: What's the mechanism for applying the settings that come from neutron to the physical ports? | 16:56 |
| *** e0ne has quit IRC | 16:56 | |
| sambetts|afk | rpioso: no sure what settings you mean, but vlans are applied at the TOR, and IP addresses get applied by DHCP or similar for IPv6 | 16:57 |
| sambetts|afk | unless you are running a fancy network card like a Cisco VIC you can't apply VLANs directly to the network interface | 16:58 |
| sambetts|afk | on the server | 16:58 |
| sambetts|afk | and for that case you would need to write a custom Ironic network interface to support that feature | 16:58 |
| *** gyee has joined #openstack-ironic | 16:59 | |
| *** baha has joined #openstack-ironic | 16:59 | |
| rpioso | sambetts|afk: I ask, because the iDRAC can apply VLAN config directly to the network interface on the server. | 16:59 |
| rpioso | Same with the IP proto version. | 17:00 |
| rpioso | It sounds like those fancy settings, VLAN and IP proto version, aren't generally needed. | 17:01 |
| rpioso | Do you agree? | 17:01 |
| sambetts|afk | correct, we have similar features on our Cisco gear and haven't found a reason to use them for the ironic case because the TOR is programmed for us via neutron | 17:02 |
| *** ptx0 has joined #openstack-ironic | 17:02 | |
| sambetts|afk | we did experiment with using them, and to do it we created a vendor specific network interface which replaced the neutron network interface for our gear, and added in steps which programmed the server side at the right point in the process | 17:03 |
| sambetts|afk | but its extremely vendor and feature specific | 17:03 |
| rpioso | Gotcha | 17:05 |
| rpioso | And I prefer simple, as long as that meets the need. | 17:05 |
| *** zul has quit IRC | 17:06 | |
| sambetts|afk | yeah thats basically what we landed on | 17:06 |
| TheJulia | operators also prefer relatively simple :) | 17:06 |
| rpioso | Another question I have is how to identify the port that is being configured. There are a few identifiers that come to mind, port object UUID, MAC address, and perhaps a vendor specific ID. | 17:07 |
| rpioso | Thoughts? | 17:07 |
| TheJulia | is being configured? | 17:07 |
| rpioso | By the clean step. | 17:08 |
| TheJulia | oh | 17:08 |
| TheJulia | well, I think it would enumerate through all clean steps | 17:08 |
| sambetts|afk | I think that needs to be one of the inputs to the clean step right? | 17:08 |
| rpioso | Exactly | 17:08 |
| TheJulia | if it was automatic though running on pxe_enabled? | 17:09 |
| rpioso | A list of ports to configure and how to configure them. | 17:09 |
| sambetts|afk | depends if we want to make it use pxe_enabled and set the value to that, or take an input, set the value then update pxe_enabled | 17:09 |
| rpioso | I'm under the impression that pxe_enabled is informative, rather than declarative. | 17:09 |
| TheJulia | I feel like I've had discussions with operators who thought it was declaritive, and in a sense it does declare intent when used a particular way in a configuration | 17:10 |
| sambetts|afk | there are lots of things in ironic that people things are declaritive but are actually informative :( | 17:10 |
| TheJulia | this is likely a good operator topic to discuss in that operators may have varying needs and I suspect they are going to want to keep custom inputs to a minimum beyond established defaults that they put in at a point in time. | 17:10 |
| TheJulia | or that are declaritive in one case and informative in another | 17:11 |
| rpioso | I'm now very confused. I recall discussing this in channel yesterday. | 17:11 |
| * rpioso is looking through the channel logs | 17:11 | |
| TheJulia | rpioso: this is likely best discussed at the ptg when we've kidnapped some operators | 17:11 |
| TheJulia | err, I mean... operatornapped | 17:11 |
| * TheJulia thinks that still doesn't sound right | 17:11 | |
| TheJulia | Captured? | 17:11 |
| rpioso | The discussion begins at http://eavesdrop.openstack.org/irclogs/%23openstack-ironic/%23openstack-ironic.2018-08-28.log.html#t2018-08-28T10:59:01 | 17:13 |
| *** gkadam has joined #openstack-ironic | 17:14 | |
| rpioso | Specifically, http://eavesdrop.openstack.org/irclogs/%23openstack-ironic/%23openstack-ironic.2018-08-28.log.html#t2018-08-28T12:25:57 | 17:15 |
| *** gkadam has quit IRC | 17:15 | |
| *** gkadam has joined #openstack-ironic | 17:15 | |
| *** ElCoyote_ has quit IRC | 17:21 | |
| * rpioso goes to refuel | 17:21 | |
| *** rpioso is now known as rpioso|eat | 17:21 | |
| TheJulia | refueling ++ | 17:27 |
| TheJulia | rpioso|eat: yeah, the initial thing was running on what is in the db and I see it as declaritive in how we interact with with neutron as to what we expect to occur. If pxe_enabled is set to false, we won't try to use the port for PXE boot operations.. I guess part of hte following thought is reconciling impossible configuration where what is set does not match what the hardware can do? but there may also be no way | 17:31 |
| TheJulia | to accurately determine that if the BMC can't understand a card's capabilities and bios/uefi enablement or not. | 17:31 |
| *** dsneddon has joined #openstack-ironic | 17:31 | |
| *** jtomasek has joined #openstack-ironic | 17:45 | |
| sambetts|afk | there are a few things that we've changed from informative to declaritive too as new features have been introduced, for example the portgroups bond mode initially was informative, and now the port group info is passed to the neutron ml2 driver so it can take action to change it | 17:50 |
| * sambetts|afk --> home | 17:50 | |
| sambetts|afk | night all o/ | 17:50 |
| *** trown|lunch is now known as trown|brb | 17:50 | |
| *** trown|brb is now known as trown | 17:50 | |
| *** S4ren has joined #openstack-ironic | 18:05 | |
| openstackgerrit | Merged openstack/networking-generic-switch master: Convert error detection to use a decorator https://review.openstack.org/592009 | 18:09 |
| openstackgerrit | Merged openstack/networking-generic-switch master: Detection of errors for Dell PowerConnect devices https://review.openstack.org/589961 | 18:15 |
| *** e0ne has joined #openstack-ironic | 18:19 | |
| *** rh-jelabarre has quit IRC | 18:22 | |
| *** e0ne_ has joined #openstack-ironic | 18:23 | |
| *** e0ne has quit IRC | 18:25 | |
| *** rh-jelabarre has joined #openstack-ironic | 18:29 | |
| *** S4ren has quit IRC | 18:38 | |
| *** e0ne_ has quit IRC | 18:39 | |
| *** e0ne has joined #openstack-ironic | 18:41 | |
| *** e0ne has quit IRC | 18:44 | |
| *** gkadam has quit IRC | 18:56 | |
| *** e0ne has joined #openstack-ironic | 18:56 | |
| TheJulia | Oh land of ironic! I've created a forum ideas etherpad: https://etherpad.openstack.org/p/BER-stein-forum-ironic-brainstorming | 18:59 |
| *** rpioso|eat is now known as rpioso | 19:04 | |
| rpioso | TheJulia: It seems that declarative v. informative can be seen through a couple of different lenses, a) its affect on OpenStack components such as neutron and ironic and b) its affect on hardware configuration. | 19:06 |
| rpioso | That may be the cause of my confusion. | 19:07 |
| *** pcaruana has quit IRC | 19:09 | |
| rpioso | jroll: Thank you for reviewing the PXE config RFE. Regarding its prospective use by automatic cleaning, does your question, "Or does it take the current values from the database and just apply those?", refer to the infamous pxe_enabled setting on the ports? | 19:10 |
| jroll | rpioso: yes | 19:11 |
| jroll | rpioso: my question is "for auto cleaning, does it use the port.pxe_enabled setting from the database to determine how to configure the hardware? or something else?" | 19:11 |
| *** ijw has joined #openstack-ironic | 19:11 | |
| jroll | rpioso: but, that question depends on "do we allow using this in automated clean" :) | 19:11 |
| *** ijw has quit IRC | 19:13 | |
| *** ijw has joined #openstack-ironic | 19:13 | |
| *** ijw has quit IRC | 19:14 | |
| rpioso | jroll: Kk. If it did, that would make port.pxe_enabled declarative ;-) If not, I'm not clear on how the desired settings would be specified by the operator. Must manual cleaning execute the clean step before it can be used automatically? If so, its args could be saved off on the node somewhere. | 19:16 |
| * rpioso has no experience with automatic cleaning | 19:16 | |
| jroll | rpioso: nope, I have never once run a manual clean step | 19:17 |
| jroll | rpioso: "something else" is an option. could have node.driver_info[pxe_ports] to configure it, for instance. | 19:18 |
| jroll | just things to think about, I don't have much interest in this other than making it useful :) | 19:18 |
| *** arne_wiebalck_ has joined #openstack-ironic | 19:20 | |
| rpioso | jroll: Is it acceptable practice for a clean step to use args, if provided; otherwise look at "something else" for its inputs? | 19:20 |
| rpioso | I appreciate that :) | 19:21 |
| jroll | rpioso: that seems fine to me - have a place to look for defaults, allow override via arguments | 19:21 |
| rpioso | jroll: Regarding "do we allow using this in automated clean?", I don't understand how that would be useful. Changing the ironic or h/w PXE config while a node is active sounds verboten. Would it be to tidy up after the mess created by doing that? | 19:25 |
| jroll | rpioso: I'm not sure if it's useful, but it's a question I always ask about clean steps. but yes, I imagine it would to be to clean up any hardware config touched while it's active | 19:26 |
| rpioso | jroll: Gotcha. 'clean step' suggests it would address that sort of thing. | 19:29 |
| jroll | yep | 19:29 |
| rpioso | jroll: Could those inputs be tweaked via automated actions? I'm thinking about the possibility of a baremetal server's network configuration being changed between deployments and that these new steps could become deploy steps, too. | 19:32 |
| jroll | rpioso: possibly, I don't think I would think that far ahead yet. we can refactor the code if it comes to that | 19:34 |
| rpioso | jroll: Ack. Could allowing these to run by default in automated cleaning be done as a follow-up effort? | 19:36 |
| jroll | rpioso: totally! I was just asking if it makes sense to do so, and if so, how do we do it :) | 19:36 |
| jroll | rpioso: I'm fine with manual-only | 19:36 |
| rpioso | jroll: And finally, are you still inclined to view pxe_enabled as informative? | 19:38 |
| jroll | rpioso: thinking yes, because not all drivers can be declarative about it. | 19:40 |
| jroll | ironic currently uses that to decide what interfaces to use for DHCP, it will still need to do that | 19:41 |
| rpioso | jroll: I see. And yeah, that seems like a different lens than the hardware config one. It informs ironic of the h/w config and affects its DHCP behavior, but can't be used to affect the PXE config of all of the h/w in the wild. | 19:47 |
| jroll | rpioso: yep. above all, I want consistent behavior, and having some drivers view it as informative and some as declarative is inconsistent | 19:48 |
| * jroll feels like he channeled rloo well there | 19:48 | |
| rloo | huh? | 19:49 |
| rpioso | jroll: Thank you | 19:49 |
| rpioso | LOL | 19:49 |
| rloo | :D | 19:49 |
| jroll | rloo: just feeling like I'm on your page :) | 19:49 |
| rpioso | rloo: I thought I heard you ;-) | 19:50 |
| *** arne_wiebalck_ has quit IRC | 19:50 | |
| rloo | heh. i still need to make consistency ingrained in all of us ;) | 19:50 |
| TheJulia | rloo: possibly a genetically engineered virus? | 19:54 |
| TheJulia | Of course, at that point... nanotech might be better | 19:54 |
| rloo | TheJulia: nah, we're dealing with reasonable people here :) Or at least, not desparate enuf ;) | 19:54 |
| *** e0ne has quit IRC | 19:56 | |
| *** r-daneel has joined #openstack-ironic | 20:01 | |
| *** MattMan has quit IRC | 20:14 | |
| *** MattMan has joined #openstack-ironic | 20:14 | |
| *** rh-jelabarre has quit IRC | 20:20 | |
| *** mgagne has joined #openstack-ironic | 20:21 | |
| *** priteau has quit IRC | 20:23 | |
| *** rh-jelabarre has joined #openstack-ironic | 20:26 | |
| *** jcoufal has quit IRC | 20:38 | |
| *** hwoarang has quit IRC | 20:39 | |
| * TheJulia is starting to wonder if the clean step just needs a quick spec because it feels like we've rehashed the discussion twice-ish | 20:47 | |
| jroll | TheJulia: eh, mostly clarifying questions I asked | 20:55 |
| jroll | I'm ready to approve it as a manual clean step now, if we add the name and arguments to the RFE | 20:55 |
| rpioso | Have we reached consensus on whether the clean step should use port.pxe_enabled as the source of truth or not? | 21:01 |
| jroll | rpioso: I had assumed (maybe wrongly?) that the manual clean step would take an argument that specifies which NICs should be enabled for PXE, and the clean step would configure the hardware and update the database | 21:04 |
| *** trown is now known as trown|outtypewww | 21:05 | |
| rpioso | jroll: That's what I intended when I wrote the RFE. | 21:05 |
| jroll | rpioso: ok cool, I guess I wasn't aware there was anybody advocating for something else | 21:06 |
| rpioso | jroll: My impression is that TheJulia was, but I may have misunderstood. | 21:06 |
| jroll | I think I glossed over that bit of scrollback | 21:07 |
| jroll | hm, maybe it's beer o'clock, struggling to make sense of that scrollback at the moment | 21:08 |
| rpioso | See the backscroll, starting at 1:07 PM EDT. | 21:08 |
| jroll | right, I just re-read | 21:08 |
| *** ijw has joined #openstack-ironic | 21:11 | |
| *** ijw has quit IRC | 21:14 | |
| TheJulia | my impression was more towards automated, because manual is an additional step that must be progressed through otherwise when the pxe_enabled data is already used for provisioning dhcp assignments. | 21:22 |
| jroll | my thought was that a manual step is part of the "prepare a node to be enrolled into ironic" | 21:23 |
| jroll | instead of whoever racks the machine going through the bios, they call ironic to do it | 21:23 |
| jroll | sounds like a PTG topic to me, but again, I really don't care about this feature, I only care about it being user-friendly if we accept it | 21:24 |
| * jroll cares about walking away from the computer more at the moment, have a good night everyone :) | 21:25 | |
| TheJulia | good plan :) | 21:30 |
| TheJulia | goodnight jroll! | 21:30 |
| *** bfournie has quit IRC | 21:34 | |
| *** mcarden has joined #openstack-ironic | 21:37 | |
| openstackgerrit | Thomas Goirand proposed openstack/ironic master: Fix async keyword for Python 3.7 https://review.openstack.org/597468 | 21:44 |
| *** rcernin has joined #openstack-ironic | 21:49 | |
| *** r-daneel_ has joined #openstack-ironic | 22:05 | |
| *** r-daneel has quit IRC | 22:05 | |
| *** r-daneel_ is now known as r-daneel | 22:05 | |
| *** baha has quit IRC | 22:08 | |
| *** ijw has joined #openstack-ironic | 22:17 | |
| *** threestrands has joined #openstack-ironic | 22:19 | |
| *** threestrands has quit IRC | 22:19 | |
| *** threestrands has joined #openstack-ironic | 22:22 | |
| *** rloo has quit IRC | 22:33 | |
| *** rloo has joined #openstack-ironic | 22:34 | |
| TheJulia | well, I was hoping we could finally tackle ceph rados, and it is looking more unlikely than ever | 22:54 |
| *** r-daneel has quit IRC | 22:56 | |
| *** ijw has quit IRC | 23:06 | |
| *** bfournie has joined #openstack-ironic | 23:43 | |
| rpioso | TheJulia: I created a new story for your reading pleasure -- https://storyboard.openstack.org/#!/story/2003594 | 23:52 |
| rpioso | I look forward to your feedback. | 23:52 |
| TheJulia | Thanks! | 23:53 |
| rpioso | TheJulia: yw | 23:54 |
| *** mjturek has quit IRC | 23:55 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!