| *** max_lobur has quit IRC | 00:15 | |
| *** matsuhashi has joined #openstack-ironic | 00:20 | |
| openstackgerrit | Michael Davies proposed a change to openstack/ironic: Return correct HTTP response codes for create ops https://review.openstack.org/71425 | 00:30 |
|---|---|---|
| *** anniec has quit IRC | 00:36 | |
| *** anniec has joined #openstack-ironic | 00:39 | |
| *** nosnos has joined #openstack-ironic | 00:56 | |
| devananda | NobodyCam: will you be available on the 18th to give an update in the project meeting? | 01:00 |
| *** rloo has joined #openstack-ironic | 01:01 | |
| *** anniec has quit IRC | 01:05 | |
| *** igor has joined #openstack-ironic | 01:15 | |
| devananda | NobodyCam: ooh! just saw your patch for nova ... reviewing now | 01:15 |
| *** igor____ has quit IRC | 01:16 | |
| *** anniec has joined #openstack-ironic | 01:23 | |
| openstackgerrit | Michael Davies proposed a change to openstack/ironic: Return correct HTTP response codes for create ops https://review.openstack.org/71425 | 01:24 |
| devananda | NobodyCam: so, it's identical to the baremetal_host_manager module except for doc strings and method name | 01:26 |
| devananda | NobodyCam: could the existing module be updated to support both? | 01:26 |
| *** mrda is now known as mrda_away | 01:27 | |
| devananda | eg, make new_host_state return either a BaremetalHostState or an IronicHostState, depending on thecapabilities | 01:28 |
| NobodyCam | devananda: humm it could. but I made a new file thinking the bm hostmanager was going to revmoved | 01:36 |
| NobodyCam | yes I should be available on the 18th | 01:38 |
| devananda | NobodyCam: great, thanks. I may be on a brief trip and unavailable that day | 01:38 |
| devananda | not sure yet | 01:38 |
| NobodyCam | :) | 01:38 |
| devananda | NobodyCam: so, yes, baremetal is goign to be deprecated ,and then removed -- but taht doesn't necessarily mean we need to duplicate all the files | 01:39 |
| devananda | NobodyCam: I'm also not saying it /doesn't/ mean that :) | 01:39 |
| NobodyCam | I can modify bmhostmanager | 01:39 |
| devananda | NobodyCam: if one way or the other is going to be easier to maintain (less code changes, easier to understand the deprecation path, etc) let's pick that | 01:39 |
| devananda | in this case, it looks to me like we just need a class to differentiate the type of node, but the functionality is the same | 01:40 |
| NobodyCam | I figured new file would allow the most flexibility | 01:40 |
| devananda | true | 01:40 |
| NobodyCam | so as we evoulve we dont have any headache from the old bm hostmanager | 01:41 |
| devananda | so we can have 2 files. or we can have one file with a base class and two inherited classes (one for baremetal, another for ironic) | 01:41 |
| NobodyCam | ya | 01:41 |
| devananda | which approach would be easier for folks to maintain for the next 6 months? which is easier to migrate across? | 01:42 |
| NobodyCam | and mock the bm fakes | 01:42 |
| NobodyCam | for the test | 01:43 |
| NobodyCam | but ya I can do that | 01:43 |
| devananda | meh. just an idea. they both have merit | 01:43 |
| devananda | left some feedback bout the unit tests | 01:44 |
| NobodyCam | :-p | 01:44 |
| NobodyCam | ahh yes | 01:44 |
| NobodyCam | ok :) | 01:45 |
| NobodyCam | will add missing test tomorrow morning | 01:45 |
| NobodyCam | :-p | 01:45 |
| devananda | :) | 01:46 |
| devananda | heading home... bbl or tmw | 02:07 |
| devananda | g'night all! | 02:07 |
| *** epim has quit IRC | 02:09 | |
| *** anniec has quit IRC | 02:11 | |
| NobodyCam | night devananda | 02:13 |
| *** rloo has quit IRC | 02:26 | |
| *** mrda_away is now known as mrda | 02:29 | |
| *** harlowja has quit IRC | 02:44 | |
| *** harlowja has joined #openstack-ironic | 02:55 | |
| *** briancline has quit IRC | 03:10 | |
| *** aignatov_ is now known as aignatov | 03:11 | |
| *** coolsvap_away has quit IRC | 03:13 | |
| openstackgerrit | Jenkins proposed a change to openstack/ironic: Updated from global requirements https://review.openstack.org/71449 | 03:56 |
| openstackgerrit | Michael Davies proposed a change to openstack/ironic: Return correct HTTP response codes for create ops https://review.openstack.org/71425 | 04:14 |
| NobodyCam | devananda: https://review.openstack.org/#/c/71429 now with more tests :-p | 04:17 |
| NobodyCam | mrda: lol Don't think I've tweeted in some time :) | 04:26 |
| mrda | NobodyCam: Just in case :) | 04:28 |
| NobodyCam | hehehe :) | 04:29 |
| mrda | Gotta follow all the cool kids in IronicLand(tm) | 04:29 |
| NobodyCam | lol | 04:29 |
| NobodyCam | :) | 04:29 |
| *** jcooley_ has quit IRC | 04:43 | |
| *** coolsvap has joined #openstack-ironic | 04:44 | |
| *** nosnos_ has joined #openstack-ironic | 05:02 | |
| *** jrist has quit IRC | 05:04 | |
| *** harlowja has quit IRC | 05:05 | |
| *** nosnos has quit IRC | 05:06 | |
| *** harlowja has joined #openstack-ironic | 05:14 | |
| *** blamar has quit IRC | 05:21 | |
| *** aignatov is now known as aignatov_ | 05:27 | |
| *** jrist has joined #openstack-ironic | 05:39 | |
| *** jrist has quit IRC | 05:43 | |
| *** jrist has joined #openstack-ironic | 05:43 | |
| *** jcooley_ has joined #openstack-ironic | 05:49 | |
| *** jrist has quit IRC | 05:52 | |
| *** rsacharya_ has joined #openstack-ironic | 05:59 | |
| *** rsacharya_ has quit IRC | 06:00 | |
| openstackgerrit | Jenkins proposed a change to openstack/ironic: Imported Translations from Transifex https://review.openstack.org/71192 | 06:06 |
| *** matsuhashi has quit IRC | 06:07 | |
| *** matsuhashi has joined #openstack-ironic | 06:07 | |
| *** matsuhas_ has joined #openstack-ironic | 06:09 | |
| *** matsuhashi has quit IRC | 06:09 | |
| *** jcooley_ has quit IRC | 06:11 | |
| *** igor has quit IRC | 06:18 | |
| *** igor has joined #openstack-ironic | 06:18 | |
| openstackgerrit | A change was merged to openstack/ironic: Improve handling of invalid input in HashRing class https://review.openstack.org/70267 | 06:24 |
| *** jcooley_ has joined #openstack-ironic | 06:24 | |
| *** mrda is now known as mrda_away | 06:27 | |
| *** nosnos has joined #openstack-ironic | 06:40 | |
| *** nosnos_ has quit IRC | 06:42 | |
| *** coolsvap_away has joined #openstack-ironic | 06:51 | |
| *** coolsvap has quit IRC | 06:52 | |
| *** coolsvap_away has quit IRC | 07:02 | |
| *** coolsvap has joined #openstack-ironic | 07:07 | |
| *** coolsvap_away has joined #openstack-ironic | 07:11 | |
| *** coolsvap has quit IRC | 07:12 | |
| openstackgerrit | A change was merged to openstack/ironic: API tests to check for the return codes https://review.openstack.org/70766 | 07:17 |
| openstackgerrit | A change was merged to openstack/ironic: Handle multiple exceptions raised by jsonpatch https://review.openstack.org/68457 | 07:17 |
| *** harlowja is now known as harlowja_away | 07:23 | |
| GheRivero | morning Ironic | 07:30 |
| *** coolsvap_away is now known as coolsvap | 07:37 | |
| coolsvap | GheRivero: Gm, I waned to try out basics in Ironic and m not getting a document describing the different concepts like chasis,node,port and their relationships | 07:38 |
| coolsvap | any links so that I can try a complete flow might be with a fake driver | 07:38 |
| GheRivero | morning coolsvap | 07:39 |
| openstackgerrit | Ghe Rivero proposed a change to openstack/ironic: Set boot device to PXE when deploying https://review.openstack.org/71332 | 07:40 |
| GheRivero | coolsvap: I not aware of any doc documenting that, but let me check | 07:41 |
| GheRivero | but basically, a chassis (not required) is a group of nodes | 07:43 |
| GheRivero | and every node needs a port for the deployment (PXE for now) | 07:43 |
| coolsvap | okay so to try simple provisioning with Ironic, the flow would be like create node -> create port | 07:45 |
| GheRivero | coolsvap: https://etherpad.openstack.org/p/IronicDeployDevstack | 07:46 |
| coolsvap | but later I am not able to figure out what to do before powering on | 07:46 |
| *** mdurnosvistov_ has joined #openstack-ironic | 07:46 | |
| GheRivero | you should create the images and deploy ramdisk/kernel using diskimage-builder (TripleO) | 07:47 |
| devananda | mrda_away: ping | 07:52 |
| *** killer_prince has quit IRC | 08:03 | |
| *** jistr has joined #openstack-ironic | 08:14 | |
| *** ndipanov_gone is now known as ndipanov | 08:32 | |
| *** athomas has joined #openstack-ironic | 08:33 | |
| *** athomas has quit IRC | 08:34 | |
| *** athomas has joined #openstack-ironic | 08:35 | |
| *** mdurnosvistov_ has quit IRC | 08:40 | |
| *** ifarkas has joined #openstack-ironic | 08:45 | |
| *** killer_prince has joined #openstack-ironic | 08:48 | |
| *** killer_prince is now known as lazy_prince | 08:48 | |
| openstackgerrit | Michael Davies proposed a change to openstack/ironic: Return correct HTTP response codes for create ops https://review.openstack.org/71425 | 08:50 |
| *** romcheg has joined #openstack-ironic | 08:59 | |
| *** lazy_prince is now known as killer_prince | 09:05 | |
| openstackgerrit | Michael Davies proposed a change to openstack/ironic: Return correct HTTP response codes for create ops https://review.openstack.org/71425 | 09:09 |
| *** aignatov_ is now known as aignatov | 09:10 | |
| *** igor_ has joined #openstack-ironic | 09:30 | |
| *** igor has quit IRC | 09:32 | |
| *** jcooley_ has quit IRC | 09:32 | |
| openstackgerrit | Yuriy Zveryanskyy proposed a change to openstack/ironic: Expose 'reservation' field of a node via API https://review.openstack.org/71211 | 09:35 |
| openstackgerrit | Yuriy Zveryanskyy proposed a change to openstack/ironic: Add ability to break TaskManager locks via REST API https://review.openstack.org/71212 | 09:36 |
| *** derekh has joined #openstack-ironic | 09:38 | |
| *** lucasagomes has joined #openstack-ironic | 09:42 | |
| romcheg | lucasagomes: Hi, are you around? | 09:46 |
| lucasagomes | romcheg, hey there | 09:46 |
| lucasagomes | yes | 09:46 |
| romcheg | I'm wondering why we decided to use PUT for breaking node locks https://review.openstack.org/#/c/71212/3/ironic/api/controllers/v1/node.py ? | 09:47 |
| romcheg | I think DELETE is more appropriate for deleting or breaking things | 09:49 |
| yuriyz | hi romcheg | 09:49 |
| lucasagomes | romcheg, yea, actually we didn't think much about it | 09:49 |
| romcheg | Hi yuriyz | 09:49 |
| lucasagomes | romcheg, I took a look yesterday, I saw something similar in the etcd api | 09:49 |
| lucasagomes | and they also use delete | 09:49 |
| lucasagomes | https://coreos.com/docs/distributed-configuration/etcd-api/ | 09:49 |
| lucasagomes | "When the client is finished with the lock, simply send a DELETE request to release the lock" | 09:50 |
| romcheg | yup | 09:50 |
| yuriyz | but 'reservation' resource not deleted when lock breaking, only value changed | 09:52 |
| romcheg | When we unprovision a node we send DELETE as well, don't we? | 09:59 |
| lucasagomes | romcheg, nop | 10:00 |
| lucasagomes | romcheg, in the power/provision states | 10:00 |
| romcheg | Whoops :) | 10:00 |
| lucasagomes | we have a target_*_state for each, so we PUT {"target": "deleted"} | 10:00 |
| romcheg | Ah, exactly | 10:01 |
| lucasagomes | it's close but not a real DELETE | 10:01 |
| romcheg | But I remember we planned to use DELETE for something | 10:01 |
| romcheg | Wasn't that breaking a running provisioning? | 10:01 |
| lucasagomes | romcheg, yea aborting the deploy | 10:01 |
| lucasagomes | but nothing decided yet :/ | 10:02 |
| romcheg | Yes, that's what I remember :) | 10:02 |
| romcheg | So to be consistent with that idea I vote for using DELETE for breaking locks | 10:03 |
| lucasagomes | I like the idea of DELETE as well, although it's possible to argue both sides here | 10:03 |
| romcheg | Let's wait for devananda and NobodyCam then to collect more opinions | 10:05 |
| romcheg | devananda, NobodyCam ^^ | 10:05 |
| yuriyz | yes, let someone else yet look | 10:06 |
| lucasagomes | ack | 10:06 |
| *** aignatov is now known as aignatov_ | 10:22 | |
| *** martyntaylor has joined #openstack-ironic | 10:25 | |
| *** martyntaylor1 has joined #openstack-ironic | 10:25 | |
| *** max_lobur_afk is now known as max_lobur | 10:26 | |
| *** viktors has joined #openstack-ironic | 10:28 | |
| *** martyntaylor has quit IRC | 10:29 | |
| *** mrda_away is now known as mrda | 10:31 | |
| mrda | devananda: pong | 10:31 |
| *** tatyana has joined #openstack-ironic | 10:31 | |
| openstackgerrit | Mikhail Durnosvistov proposed a change to openstack/ironic: Get rid of the newline "\" https://review.openstack.org/66793 | 10:32 |
| openstackgerrit | Mikhail Durnosvistov proposed a change to openstack/ironic: Get rid of the newline "\" https://review.openstack.org/66793 | 10:39 |
| openstackgerrit | A change was merged to openstack/ironic: Fix 'run_as_root' parameter check in utils https://review.openstack.org/70324 | 10:41 |
| romcheg | mrda: Hi, I guess it's a late night for him | 10:42 |
| *** jistr has quit IRC | 10:44 | |
| openstackgerrit | Michael Davies proposed a change to openstack/ironic: Make the parameter order of assertEqual consistent https://review.openstack.org/71486 | 10:44 |
| *** aignatov_ is now known as aignatov | 10:58 | |
| mrda | thanks romcheg, I'll catch him tomorrow | 11:02 |
| mrda | it's a late night for me too :) | 11:02 |
| *** mrda is now known as mrda_away | 11:02 | |
| *** jistr has joined #openstack-ironic | 11:10 | |
| *** coolsvap_away has joined #openstack-ironic | 11:29 | |
| *** coolsvap has quit IRC | 11:30 | |
| *** martyntaylor1 has quit IRC | 11:33 | |
| *** martyntaylor has joined #openstack-ironic | 11:44 | |
| coolsvap_away | GheRivero: ping | 11:44 |
| *** coolsvap_away is now known as coolsvap | 11:46 | |
| coolsvap | can anybody tell me what pxe_image_source would be in https://etherpad.openstack.org/p/IronicDeployDevstack #173 | 11:46 |
| *** athomas has quit IRC | 11:50 | |
| Haomeng | coolsvap: pxe_image_source should be an image uuid which used be 'copy' to pxe node local disk I think | 11:51 |
| *** athomas has joined #openstack-ironic | 11:51 | |
| Haomeng | coolsvap: it is different one with pxe_deploy_ramdisk and pxe_deploy_kernel | 11:52 |
| coolsvap | Haomeng: currently I have 3 images in glance, default cirros and two images for kernel and ramdisk i created after dib run | 11:52 |
| Haomeng | coolsvap: both pxe_deploy_ramdisk and pxe_deploy_kernel are used on deploy time, they are deployment util image | 11:52 |
| Haomeng | coolsvap: yes, we need 3 images | 11:52 |
| coolsvap | so I think cirros would not be the image for pxe_source, correct? | 11:53 |
| Haomeng | coolsvap: but not sure if cirros image can be used for your hardware, it should include the hardware network adaptor driver | 11:54 |
| coolsvap | okay so I would need to add a ubuntu/fedora cloud image which can be used for source | 11:55 |
| coolsvap | is it correct? | 11:55 |
| Haomeng | coolsvap: in our ironic deployment steps, we will 'dd' the pxe_image_source image into the hardware server's local disk, and restart sys from disk once it is ready | 11:55 |
| Haomeng | coolsvap: depens on your test case I think | 11:56 |
| coolsvap | Haomeng: I am giving first try to Ironic | 11:56 |
| Haomeng | coolsvap: suggest you have a try with your cirros image as pxe_image_source first, maybe it works with your hardware | 11:56 |
| coolsvap | Haomeng: sure let me try that | 11:57 |
| Haomeng | coolsvap: welcome | 11:57 |
| openstackgerrit | Andrey Kurilin proposed a change to openstack/python-ironicclient: Reuse exceptions from common code https://review.openstack.org/71500 | 11:59 |
| coolsvap | Haomeng: just one more query, the ssh_address mentioned here is 192.122.x.x, but with my devstack installation I have default network 172.24.4.x, I can use address from available network, right? | 11:59 |
| Haomeng | coolsvap: ssh_address | 12:07 |
| Haomeng | coolsvap: is used for pxe_ssh driver with vm to simulate a physical node | 12:07 |
| Haomeng | coolsvap: so make sure the ssh_address is your ip which host vms | 12:08 |
| coolsvap | oh..k got it | 12:08 |
| *** pradipta` has joined #openstack-ironic | 12:10 | |
| Haomeng | coolsvap: I guss you use pxe_ssh as your driver | 12:10 |
| coolsvap | Haomeng: I am trying to use the steps mentioned in etherpad I mentioned above | 12:10 |
| coolsvap | Haomeng: I did not get any MAC from node-create | 12:12 |
| *** pradipta has quit IRC | 12:12 | |
| openstackgerrit | Andrey Kurilin proposed a change to openstack/python-ironicclient: Reuse exceptions from common code https://review.openstack.org/71500 | 12:13 |
| Haomeng | coolsvap: that is fine, ssh_virt_type=virsh -i ssh_address=192.168.122.1, these mean we use linux kvm as vm host, and 192.168.122.1 is the host ip | 12:13 |
| Haomeng | coolsvap: mac should be your vm mac which used to simulate a physical node | 12:14 |
| coolsvap | okay | 12:15 |
| Haomeng | coolsvap: you can check node with 'ironic port-list' to check the port with your mac, if there is not existing as well, have to create with command 'ironic node-create' | 12:16 |
| Haomeng | coolsvap: good luck | 12:16 |
| coolsvap | Haomeng: just to get the steps clear to test with VM, Create a VM -> then Create a Node -> Associate VM IP as for ssh_address -> Associate VMs mac to port -> Associate instance to node -> Deploy | 12:17 |
| Haomeng | coolsvap: " ./tripleo-incubator/scripts/create-nodes 1 512 10 amd64 1" | 12:19 |
| Haomeng | you did not get mac returned by this command? | 12:19 |
| Haomeng | coolsvap: yes, you are correct I think | 12:19 |
| Haomeng | coolsvap: we can follow the guide - https://etherpad.openstack.org/p/IronicDeployDevstack , the steps are verified, should work | 12:20 |
| coolsvap | hmm | 12:20 |
| Haomeng | coolsvap: so which step you run out what issue which we can discuss | 12:20 |
| coolsvap | Haomeng: let me complete that once | 12:21 |
| Haomeng | coolsvap: good luck:) | 12:22 |
| coolsvap | Haomeng: thx! | 12:22 |
| Haomeng | coolsvap: welcome your any questions:) | 12:22 |
| *** jcooley_ has joined #openstack-ironic | 12:26 | |
| *** jcooley_ has quit IRC | 12:31 | |
| *** coolsvap has quit IRC | 12:34 | |
| *** coolsvap has joined #openstack-ironic | 12:34 | |
| *** r-mibu has quit IRC | 12:41 | |
| *** romcheg has quit IRC | 12:43 | |
| *** igor has joined #openstack-ironic | 12:45 | |
| *** igor_ has quit IRC | 12:46 | |
| *** coolsvap has quit IRC | 12:55 | |
| *** datajerk has joined #openstack-ironic | 13:07 | |
| *** romcheg has joined #openstack-ironic | 13:12 | |
| *** romcheg has quit IRC | 13:15 | |
| *** romcheg has joined #openstack-ironic | 13:15 | |
| *** max_lobur is now known as max_lobur_afk | 13:17 | |
| *** datajerk has quit IRC | 13:26 | |
| *** max_lobur_afk is now known as max_lobur | 13:29 | |
| *** jdob has joined #openstack-ironic | 13:49 | |
| *** jbjohnso has joined #openstack-ironic | 13:56 | |
| *** jcooley_ has joined #openstack-ironic | 13:57 | |
| *** jcooley_ has quit IRC | 14:03 | |
| openstackgerrit | Andrey Kurilin proposed a change to openstack/python-ironicclient: Reuse exceptions from common code https://review.openstack.org/71500 | 14:03 |
| *** rlooh has joined #openstack-ironic | 14:24 | |
| *** rlooh has quit IRC | 14:25 | |
| *** jrist has joined #openstack-ironic | 14:32 | |
| *** rloo has joined #openstack-ironic | 14:35 | |
| *** matty_dubs|gone is now known as matty_dubs | 14:36 | |
| *** rloo has left #openstack-ironic | 14:37 | |
| *** rloo has joined #openstack-ironic | 14:38 | |
| openstackgerrit | Yuriy Zveryanskyy proposed a change to openstack/ironic: Improve NodeLocked exception message https://review.openstack.org/71541 | 14:39 |
| *** rloo_ has joined #openstack-ironic | 14:40 | |
| openstackgerrit | Yuriy Zveryanskyy proposed a change to openstack/ironic: Improve NodeLocked exception message https://review.openstack.org/71541 | 14:42 |
| *** rloo has left #openstack-ironic | 14:49 | |
| *** blamar has joined #openstack-ironic | 14:55 | |
| openstackgerrit | Jarrod Johnson proposed a change to stackforge/pyghmi: Fix session reuse stall for async usage https://review.openstack.org/71549 | 15:01 |
| openstackgerrit | A change was merged to stackforge/pyghmi: Fix session reuse stall for async usage https://review.openstack.org/71549 | 15:02 |
| openstackgerrit | Andrey Kurilin proposed a change to openstack/python-ironicclient: Reuse exceptions from common code https://review.openstack.org/71500 | 15:08 |
| openstackgerrit | Jarrod Johnson proposed a change to stackforge/pyghmi: Move previous fix out to command and console https://review.openstack.org/71554 | 15:12 |
| openstackgerrit | Yuriy Zveryanskyy proposed a change to openstack/ironic: Improve NodeLocked exception message https://review.openstack.org/71541 | 15:13 |
| openstackgerrit | A change was merged to stackforge/pyghmi: Move previous fix out to command and console https://review.openstack.org/71554 | 15:17 |
| *** matsuhas_ has quit IRC | 15:17 | |
| *** nosnos has quit IRC | 15:21 | |
| *** shausy has joined #openstack-ironic | 15:24 | |
| *** romcheg has quit IRC | 15:27 | |
| *** romcheg has joined #openstack-ironic | 15:30 | |
| openstackgerrit | Jenkins proposed a change to openstack/ironic: Updated from global requirements https://review.openstack.org/71449 | 15:38 |
| *** ifarkas_ has joined #openstack-ironic | 15:40 | |
| *** ifarkas__ has joined #openstack-ironic | 15:41 | |
| *** ifarkas has quit IRC | 15:43 | |
| *** ifarkas has joined #openstack-ironic | 15:43 | |
| *** romcheg has quit IRC | 15:44 | |
| *** jcooley_ has joined #openstack-ironic | 15:46 | |
| *** aignatov is now known as aignatov_ | 15:48 | |
| *** jistr has quit IRC | 15:48 | |
| devananda | morning, all | 15:49 |
| *** ifarkas has quit IRC | 15:49 | |
| *** ifarkas__ has quit IRC | 15:49 | |
| lucasagomes | devananda, morning | 15:50 |
| max_lobur | morning Ironic | 15:50 |
| GheRivero | morning all | 15:51 |
| *** romcheg has joined #openstack-ironic | 15:51 | |
| lucasagomes | morning max_lobur GheRivero | 15:51 |
| mdurnosvistov | morning all! :) | 15:51 |
| rloo_ | morning/aft/evening everyone ;) | 15:51 |
| max_lobur | :) | 15:51 |
| romcheg | Morning folks! | 15:52 |
| *** jcooley_ has quit IRC | 15:53 | |
| *** jistr has joined #openstack-ironic | 15:54 | |
| yuriyz | morning all | 15:56 |
| *** shausy has quit IRC | 15:57 | |
| max_lobur | bbiaw | 16:02 |
| *** max_lobur is now known as max_lobur_afk | 16:02 | |
| openstackgerrit | Andrey Kurilin proposed a change to openstack/python-ironicclient: Reuse exceptions from common code https://review.openstack.org/71500 | 16:06 |
| NobodyCam | good morning Ironic | 16:14 |
| matty_dubs | Morning NobodyCam | 16:14 |
| matty_dubs | Are you guys seeing a flake8 failure on master? | 16:14 |
| matty_dubs | ./ironic/tests/test_neutron.py:23:1: H302 import only modules.'from ironic.common import neutron' does not import a module | 16:14 |
| matty_dubs | Or do I have something messed up locally? | 16:15 |
| NobodyCam | mornig matty_dubs | 16:15 |
| NobodyCam | hummm I have not seen that. yet | 16:16 |
| NobodyCam | thou the day is young | 16:16 |
| matty_dubs | Odds are also fairly decent that I'm doing something dumb, hence my asking. I'd have expected Jenkins to not allow such a thing to be merged. | 16:18 |
| NobodyCam | ya | 16:18 |
| NobodyCam | I don't think Jenkins would allow that | 16:18 |
| *** tatyana has left #openstack-ironic | 16:19 | |
| NobodyCam | thou I have see the flake/pep 8 rules change after somehtign lands | 16:19 |
| lucasagomes | morning NobodyCam | 16:20 |
| NobodyCam | morning lucasagomes | 16:20 |
| NobodyCam | lucasagomes: did you see ?https://review.openstack.org/#/c/71429 | 16:21 |
| lucasagomes | NobodyCam, nice! didn't see | 16:21 |
| lucasagomes | NobodyCam, we prolly need to remove that files from the main patch-set | 16:22 |
| lucasagomes | want me to do that? | 16:22 |
| lucasagomes | I added a couple of more tests (about 30 now) | 16:22 |
| NobodyCam | ea, Sure :) | 16:22 |
| lucasagomes | ack | 16:22 |
| NobodyCam | Yea | 16:22 |
| lucasagomes | will submit a new patch-set soon | 16:22 |
| lucasagomes | and then I rmeove it | 16:22 |
| lucasagomes | remove* | 16:22 |
| NobodyCam | :) | 16:22 |
| lucasagomes | NobodyCam, good stuff :D | 16:22 |
| NobodyCam | :-p | 16:23 |
| NobodyCam | brb getting coffee | 16:23 |
| *** jcooley_ has joined #openstack-ironic | 16:30 | |
| NobodyCam | ok /me need to address the test issues on the hostmanager patch | 16:34 |
| NobodyCam | s/need/needs/ | 16:34 |
| NobodyCam | lucasagomes: devananda: have you seen this client patch? https://review.openstack.org/#/c/71500/ | 16:37 |
| *** killer_prince has quit IRC | 16:38 | |
| *** max_lobur_afk is now known as max_lobur | 16:38 | |
| max_lobur | back | 16:38 |
| NobodyCam | Wb max_lobur | 16:39 |
| max_lobur | morning NobodyCam :) | 16:39 |
| NobodyCam | Morning max_lobur :) and romcheg too! | 16:39 |
| lucasagomes | NobodyCam, uuuu didn't see that yet | 16:39 |
| lucasagomes | but that would mess up with the driver heh | 16:39 |
| romcheg | Morning NobodyCam, devananda! | 16:40 |
| lucasagomes | I mean, we would need to refactor part of it | 16:40 |
| NobodyCam | I have to check to see if we trap any errors | 16:40 |
| lucasagomes | ack | 16:40 |
| NobodyCam | hey hey romcheg saw your first +2 lastnight | 16:40 |
| romcheg | NobodyCam: Honestly it was a bit scaring to put the first +2 :) | 16:42 |
| NobodyCam | for me too :) | 16:42 |
| * max_lobur still afraid :D | 16:43 | |
| max_lobur | devananda, lucasagomes please take a look https://github.com/openstack/ironic/blob/master/ironic/api/controllers/v1/node.py#L173-L180 | 16:43 |
| NobodyCam | there is also -2 which is very scarry | 16:43 |
| max_lobur | devananda, according to our yesterday discussion, where it was decided to perform node power action in conductor fully async (including async validation) | 16:45 |
| max_lobur | this race will remain | 16:45 |
| max_lobur | maybe to remove this pre-validation from api code? | 16:46 |
| lucasagomes | I think I missed the discussion | 16:47 |
| lucasagomes | so it will be fully async now? | 16:47 |
| devananda | max_lobur: that is awkward, yea.. .but what about the space between booting the deploy ramdisk and when it calls back? ie the DEPLOYWAIT period in patch https://review.openstack.org/#/c/71297/ | 16:47 |
| *** athomas has quit IRC | 16:48 | |
| * max_lobur looks | 16:48 | |
| lucasagomes | NobodyCam, btw, that structure in the driver mapping the required fields for each driver | 16:48 |
| *** athomas has joined #openstack-ironic | 16:48 | |
| max_lobur | lucasagomes, yea, according to this mockup https://review.openstack.org/#/c/71281/1/ironic/conductor/manager.py it was decided to make it fully async for the sake of simplicity | 16:48 |
| lucasagomes | NobodyCam, maybe it wouldn't be better to use something like dicts instead of a class? I find it a bit obscure to import classes in the middle of the code (using importutils) | 16:49 |
| NobodyCam | lucasagomes: this one? https://review.openstack.org/#/c/51328/17/nova/virt/ironic/ironic_driver_fields.py | 16:49 |
| devananda | lucasagomes: for actions that require a TaskManager lock, instead of call-then-cast, they should all be call-and-use-greenthread | 16:49 |
| *** jistr has quit IRC | 16:49 | |
| lucasagomes | NobodyCam, yup | 16:49 |
| *** athomas has quit IRC | 16:49 | |
| NobodyCam | there is a comment to not import it like that | 16:49 |
| lucasagomes | devananda, max_lobur right | 16:49 |
| lucasagomes | devananda, but call-and-use-greenthread would still return a validation no? | 16:50 |
| max_lobur | lucasagomes, validation will be inside greenthread | 16:50 |
| max_lobur | so no | 16:50 |
| lucasagomes | call->spawn the greenthread -> return | 16:50 |
| lucasagomes | so why use greenthread then? and just use cast() ? | 16:50 |
| devananda | max_lobur: so we /could/ pu tthe validation outside the greenthread | 16:50 |
| lucasagomes | just use cast()* | 16:50 |
| NobodyCam | lucasagomes: something more like this: https://github.com/openstack/nova/blob/master/nova/virt/baremetal/virtual_power_driver.py#L86-L91 | 16:50 |
| NobodyCam | is what we need | 16:51 |
| max_lobur | lucasagomes, devananda call -> get log -> spawn thread -> return ** -> release lock in the end of greenthread | 16:51 |
| max_lobur | *get lock | 16:51 |
| devananda | right | 16:51 |
| devananda | or | 16:51 |
| lucasagomes | NobodyCam, ah will take a look, yea, I'm writing the tests for spawn/destroy | 16:51 |
| lucasagomes | but that's kinda obscure to have those importutils there | 16:51 |
| max_lobur | lucasagomes, to see if lock was acquired | 16:51 |
| devananda | call -> get lock -> do validation -> spawn thread -> return ** greenthread updates state && release lock at end | 16:51 |
| max_lobur | devananda, right :) | 16:51 |
| devananda | that way user will get immediate feedback if either get-lock or do-validation fail | 16:51 |
| max_lobur | this is what I was trying to say yesterday :) | 16:52 |
| lucasagomes | devananda, +1 | 16:52 |
| lucasagomes | so it's async but not really | 16:52 |
| lucasagomes | the main job is async the validation is sync | 16:52 |
| devananda | right | 16:52 |
| lucasagomes | which is good for the user feedback | 16:52 |
| max_lobur | Therefore my comment to https://review.openstack.org/#/c/71281/2/ironic/conductor/manager.py | 16:52 |
| devananda | this way the user knows at least if the work /started/. and by the time that call returns, they should be able to enter a GET status loop | 16:53 |
| NobodyCam | lucasagomes: we also need to get rid if the " if 'pxe' in node.driver:" line and do something better then string match | 16:53 |
| devananda | to observe when it ends | 16:53 |
| max_lobur | should be fully applicable, except futures part right? | 16:53 |
| lucasagomes | NobodyCam, hmm | 16:53 |
| *** yuriyz has quit IRC | 16:53 | |
| *** lsmola has quit IRC | 16:53 | |
| max_lobur | devananda, lucasagomes ok, got it | 16:54 |
| lucasagomes | NobodyCam, can't think about another way to findout if the node is using the pxe driver | 16:54 |
| devananda | there is still a race -- see what I posted at the bottom of https://etherpad.openstack.org/p/action-started | 16:54 |
| devananda | but it is a different race taht we can't solve today | 16:54 |
| *** ifarkas_ has quit IRC | 16:55 | |
| max_lobur | yea, all the N finished jobs get N-1th result | 16:55 |
| max_lobur | I agree it's different | 16:55 |
| max_lobur | so | 16:55 |
| max_lobur | call -> get lock -> do validation -> spawn thread -> return ** greenthread updates state && release lock at end | 16:55 |
| max_lobur | our target right? | 16:55 |
| *** hemna has joined #openstack-ironic | 16:55 | |
| devananda | ++ | 16:55 |
| lucasagomes | devananda, max_lobur which line on that etherpad is about the race we can't solve? | 16:56 |
| devananda | lucasagomes: the very bottom | 16:56 |
| NobodyCam | lucasagomes: dicts would also work too! | 16:56 |
| max_lobur | 297 | 16:56 |
| lucasagomes | ack | 16:56 |
| *** jcooley_ has quit IRC | 16:56 | |
| max_lobur | devananda, ack | 16:56 |
| *** jcooley_ has joined #openstack-ironic | 16:57 | |
| lucasagomes | hmmm | 16:59 |
| lucasagomes | max_lobur, devananda maybe one way to solve that problem | 16:59 |
| lucasagomes | is to have a factory-like api for the state changes | 16:59 |
| lucasagomes | e.g | 16:59 |
| *** matty_dubs is now known as matty_dubs|lunch | 16:59 | |
| lucasagomes | POST {target: active} /nodes/<uuid>/states/power | 16:59 |
| lucasagomes | returns | 16:59 |
| devananda | lucasagomes: aside from a scheduling service that generates tokens for each request | 16:59 |
| lucasagomes | /nodes/<uuid>/states/power/<task uuid> | 17:00 |
| lucasagomes | to track the state of that task you would GET /nodes/<uuid>/states/power/<task uuid> | 17:00 |
| lucasagomes | once the task is finished that <task uuid> is removed | 17:00 |
| max_lobur | lucasagomes, you just exactly described the scheduling service :) | 17:01 |
| lucasagomes | no more than 1 <task uuid> can live inside states/power/ | 17:01 |
| max_lobur | ah | 17:01 |
| lucasagomes | cause only 1 power operation can be issued per node only | 17:01 |
| devananda | lucasagomes: i was with you up till "no more than 1 <task uuid> can live inside states/power/" | 17:01 |
| max_lobur | so, what to do after the second task already completed | 17:01 |
| max_lobur | we're doing /nodes/<uuid>/states/power/<first task uuid> | 17:02 |
| devananda | lucasagomes: then what is the point of giving out a request token if that token will be invalidated // not GETable when a second request comes in | 17:02 |
| max_lobur | but there's already <second task uuid> right? | 17:02 |
| devananda | lucasagomes: tokens should exist independently of how many more tasks come in. at scale things will happen at different times | 17:02 |
| lucasagomes | right hmm | 17:03 |
| devananda | lucasagomes: eg, network congestion. assume that client 1's GET may be delayed | 17:03 |
| devananda | so | 17:03 |
| lucasagomes | max_lobur, in my case the second task uuid wouldn't be created | 17:03 |
| devananda | long term, something like that is possible | 17:03 |
| lucasagomes | unless the first is ended | 17:03 |
| devananda | a) definitely not going in Icehouse | 17:03 |
| max_lobur | lucasagomes, right, and what if we do /nodes/<uuid>/states/power/<first task uuid> when the second is already started | 17:04 |
| lucasagomes | max_lobur, 404 | 17:04 |
| max_lobur | there's no <first task uuid> already right | 17:04 |
| devananda | yea, that's wrong | 17:04 |
| lucasagomes | max_lobur, once the task is created the Location to be GET'ed | 17:04 |
| lucasagomes | and tracked | 17:04 |
| lucasagomes | is returned in the HTTP header | 17:04 |
| max_lobur | so we will be able to get results only until the next one is started | 17:04 |
| max_lobur | well, this may work. But I think it's better to queue such things. So /nodes/<uuid>/states/power/<first ask uuid> will always be available | 17:05 |
| max_lobur | even if there were 10 tasks after the first | 17:05 |
| devananda | so, guys - we're going waaay off into the weeds :) | 17:06 |
| lucasagomes | right | 17:06 |
| lucasagomes | max_lobur, like a history? | 17:06 |
| lucasagomes | devananda, heh yea sorry | 17:06 |
| max_lobur | yea | 17:06 |
| max_lobur | :) | 17:06 |
| lucasagomes | max_lobur, that could be useful yea | 17:06 |
| lucasagomes | max_lobur, so you would know when each state change happened | 17:06 |
| lucasagomes | and when it finished, if it failed or not | 17:06 |
| lucasagomes | that would be useful I think | 17:06 |
| devananda | backgrounding long running operations solves a few problems which are causing deploys to fail _today_. and we can solve it fairly easily | 17:07 |
| devananda | completely changing our API (in a way that is unlike any other openstack service, and non-REST-ful) isn't going to solve those problems (we still will need backgrounded operations) and there's no time to get that into Icehouse | 17:08 |
| lucasagomes | +1 | 17:08 |
| devananda | and, with backgrounded operations, there is other work that needs to be done to take advantage of the synchronous feedback | 17:08 |
| devananda | eg the nova driver | 17:08 |
| lucasagomes | just one note, that would be backgrouded | 17:08 |
| lucasagomes | just diff way to trigger and track that | 17:08 |
| devananda | needs to understand NodeAlreadyLocked and retry | 17:08 |
| devananda | lucasagomes: right -- my point is, we still need to do backgrounding | 17:09 |
| lucasagomes | making each request to have it's own resource | 17:09 |
| lucasagomes | devananda, +1 | 17:09 |
| *** GheRivero has quit IRC | 17:09 | |
| devananda | lucasagomes: so lets actually solve that :) | 17:09 |
| max_lobur | devananda, you asked RE: .but what about the space between booting the deploy ramdisk and when it calls back? ie the DEPLOYWAIT period in patch https://review.openstack.org/#/c/71297/ | 17:09 |
| max_lobur | sorry, I can't get how this relates to current discussion | 17:09 |
| devananda | ahh | 17:09 |
| devananda | max_lobur: so, taskmanager lock is released between those two steps | 17:10 |
| max_lobur | yup | 17:10 |
| max_lobur | how to prevent races? | 17:10 |
| max_lobur | between those two calls | 17:10 |
| devananda | max_lobur: ah, sorry. i was looking at https://github.com/openstack/ironic/blob/master/ironic/api/controllers/v1/node.py#L168 | 17:11 |
| devananda | max_lobur: but you were looking at L173 | 17:11 |
| * lucasagomes a bit confused heh | 17:11 | |
| devananda | max_lobur: i think that's where the confusion is. I thought you meant, move L168 into conductor (which you didn't, right?) | 17:12 |
| max_lobur | ahh I got now | 17:12 |
| max_lobur | yea | 17:12 |
| max_lobur | https://github.com/openstack/ironic/blob/master/ironic/api/controllers/v1/node.py#L173-L180 | 17:12 |
| max_lobur | I meant another | 17:12 |
| devananda | k k | 17:12 |
| devananda | so yes, L173 should be removed | 17:12 |
| max_lobur | k, thanks | 17:12 |
| max_lobur | that's all I guess | 17:13 |
| lucasagomes | devananda, deva, I will submit a new patch-set to the driver (including tests, few ones missing I plan to finish tomorrow) | 17:26 |
| lucasagomes | devananda, can u mark it as WIP please? | 17:27 |
| devananda | lucasagomes: ack. thanks! | 17:27 |
| *** GheRivero has joined #openstack-ironic | 17:27 | |
| lucasagomes | devananda, sent | 17:28 |
| lucasagomes | https://review.openstack.org/#/c/51328/18 | 17:28 |
| NobodyCam | devananda: not 100% sure thats what you were looking for but: https://review.openstack.org/#/c/71429 has mo tests? got a second for a quick poke | 17:32 |
| devananda | NobodyCam: looking | 17:33 |
| devananda | NobodyCam: whitespace error, pep's going t ocomplain | 17:33 |
| NobodyCam | gah | 17:33 |
| NobodyCam | :-p will fix | 17:34 |
| devananda | NobodyCam: yep, the new tests look ilke they cover what i felt was missing. also, please add a description here https://review.openstack.org/#/c/71429/5/nova/scheduler/ironic_host_manager.py | 17:35 |
| *** martyntaylor has quit IRC | 17:35 | |
| devananda | NobodyCam: those tests (larger|smaller_instance) are probably goign to raise eyebrows | 17:35 |
| devananda | which is fine -- the point of this HostManager is it just uses 100% of the node | 17:35 |
| devananda | but it should document that :) | 17:36 |
| lucasagomes | devananda, btw, did you try to setup ironic+devstack ? did it work? | 17:38 |
| devananda | lucasagomes: the instructions seemed to be missing some steps, i didn't get it to finish | 17:38 |
| lucasagomes | devananda, hmm | 17:39 |
| lucasagomes | devananda, u trying the guide 2? (nova + ironic) | 17:39 |
| devananda | or rather, some of the steps for devstack+ironic seem to not apply to gude 2 | 17:39 |
| devananda | guide | 17:39 |
| lucasagomes | ack | 17:40 |
| lucasagomes | yea I still using the dnsmasq for pxe | 17:41 |
| lucasagomes | I gotta give neutron a go with the new patch u put up | 17:41 |
| *** jcooley_ has quit IRC | 17:41 | |
| *** jcooley_ has joined #openstack-ironic | 17:42 | |
| devananda | yea | 17:42 |
| devananda | that was a chunk of the steps I skipped :) | 17:42 |
| lucasagomes | heh | 17:43 |
| *** viktors has quit IRC | 17:47 | |
| *** shausy has joined #openstack-ironic | 17:50 | |
| *** aignatov_ is now known as aignatov | 17:53 | |
| *** john-n-seattle has quit IRC | 17:55 | |
| *** john-n-seattle has joined #openstack-ironic | 17:55 | |
| *** harlowja_away is now known as harlowja | 17:59 | |
| *** derekh has quit IRC | 18:02 | |
| lucasagomes | ok gotta run (I'm in the office) | 18:04 |
| lucasagomes | devananda, NobodyCam max_lobur night! | 18:04 |
| NobodyCam | have a good night lucasagomes | 18:04 |
| *** lucasagomes has quit IRC | 18:04 | |
| *** matty_dubs|lunch is now known as matty_dubs | 18:10 | |
| NobodyCam | brb | 18:18 |
| *** jcooley_ has quit IRC | 18:21 | |
| *** GheRivero has quit IRC | 18:23 | |
| *** GheRivero has joined #openstack-ironic | 18:23 | |
| openstackgerrit | Ghe Rivero proposed a change to openstack/ironic: Allow to tear-down a node waiting to be deployed https://review.openstack.org/71297 | 18:27 |
| *** mdurnosvistov_ has joined #openstack-ironic | 18:33 | |
| NobodyCam | wow TripleO queue hit 90 hours :( | 18:35 |
| *** shausy has quit IRC | 18:35 | |
| GheRivero | ugh. that's not good :/ | 18:37 |
| *** ndipanov is now known as ndipanov_gone | 18:37 | |
| NobodyCam | :( | 18:39 |
| *** igor has quit IRC | 18:40 | |
| *** igor has joined #openstack-ironic | 18:44 | |
| *** jcooley_ has joined #openstack-ironic | 18:58 | |
| matty_dubs | devananda: Are you working on https://bugs.launchpad.net/ironic/+bug/1260337 ? I think it got auto-assigned to you when you rebased a patch. Should I take and finish this up? | 19:01 |
| devananda | matty_dubs: ah, nope, not working on it | 19:01 |
| * matty_dubs will claim and finish | 19:02 | |
| NobodyCam | brb ... /me makes a bagel! | 19:04 |
| *** romcheg1 has joined #openstack-ironic | 19:04 | |
| *** romcheg1 is now known as romcheg_ltp | 19:04 | |
| *** epim has joined #openstack-ironic | 19:08 | |
| devananda | hm, some doc links are broken | 19:17 |
| devananda | probably my fault, lol | 19:17 |
| NobodyCam | :-p | 19:17 |
| openstackgerrit | Max Lobur proposed a change to openstack/ironic: Fix race condition when changing node states https://review.openstack.org/69135 | 19:20 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Fix broken doc links on the index page https://review.openstack.org/71632 | 19:23 |
| NobodyCam | Open question: the ironic nova volume driver imports the libvirt utils lib, this is bad, we are useing just one function ( https://github.com/openstack/nova/blob/master/nova/virt/libvirt/utils.py#L55-L66 ) better to move that one function into a shared util file or jost copy it into our volume_driver file? | 19:24 |
| max_lobur | g'night folks | 19:28 |
| NobodyCam | night max_lobur ! | 19:28 |
| *** max_lobur is now known as max_lobur_afk | 19:30 | |
| *** russell_h has joined #openstack-ironic | 19:33 | |
| russell_h | the ironic docs make some mention of a deployment agent, is that implemented yet? | 19:35 |
| *** jcooley_ has quit IRC | 19:35 | |
| *** jcooley_ has joined #openstack-ironic | 19:36 | |
| NobodyCam | russell_h: Like baremetal deploy helper? | 19:38 |
| NobodyCam | can you point me to the doc? | 19:38 |
| russell_h | NobodyCam: http://docs.openstack.org/developer/ironic/dev/architecture.html | 19:39 |
| russell_h | "A Deployment Ramdisk or Deployment Agent, which provide control over the hardware which is not available remotely to the Conductor." | 19:39 |
| NobodyCam | ahh | 19:39 |
| NobodyCam | I know there is a dib element for such | 19:40 |
| NobodyCam | https://github.com/openstack/diskimage-builder/tree/master/elements/deploy-ironic | 19:40 |
| *** jcooley_ has quit IRC | 19:40 | |
| *** jcooley_ has joined #openstack-ironic | 19:41 | |
| russell_h | oh, interseting | 19:41 |
| NobodyCam | thats basicly like the current baremetal deploy ramdisk | 19:41 |
| russell_h | so Ironic basically causes a box to PXE boot that ramdisk, which exports the root volume over iscsi | 19:41 |
| russell_h | then Ironic mounts that volume to write out an image? | 19:41 |
| NobodyCam | just like current nova bm | 19:41 |
| *** wusWWW has joined #openstack-ironic | 19:41 | |
| NobodyCam | we also have other options that nova bm does not | 19:42 |
| NobodyCam | however | 19:42 |
| NobodyCam | the pxe work is done here https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/pxe.py | 19:43 |
| *** jcooley_ has quit IRC | 19:45 | |
| russell_h | whoa, wild | 19:47 |
| russell_h | does the vendor_passthru thing return a shell script to be evaluated on the host? | 19:47 |
| NobodyCam | it currently dose not, what would be evaluated in such a script | 19:48 |
| russell_h | https://github.com/openstack/diskimage-builder/blob/master/elements/deploy-ironic/init.d/80-deploy-ironic#L39 | 19:48 |
| russell_h | I'm reading that as a way for a vendor to cause some code to run on a box, probably the actual vendor deploy thing happens | 19:49 |
| russell_h | the 'eval' bit | 19:50 |
| russell_h | s/probably the/probably after the/ | 19:50 |
| *** jcooley_ has joined #openstack-ironic | 19:50 | |
| NobodyCam | ahh that becomes part of over are init.d script that does the actual exposing of the disk and such | 19:50 |
| NobodyCam | ahh ya drivers other then pxe COULD do such things | 19:52 |
| russell_h | NobodyCam: ok, thats kind of neat, gotcha | 19:54 |
| *** aignatov is now known as aignatov_ | 19:56 | |
| *** jcooley_ has quit IRC | 19:56 | |
| *** jcooley_ has joined #openstack-ironic | 19:57 | |
| NobodyCam | russell_h: see https://github.com/openstack/diskimage-builder/blob/master/bin/disk-image-create to see how DIB creates its deploy ram disk | 19:57 |
| * NobodyCam steps afk for a few minutes | 19:58 | |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Add notes to start ironic services in a local venv https://review.openstack.org/71638 | 20:01 |
| devananda | davidlenwell: taht ^ is what you are looking for | 20:01 |
| devananda | russell_h: so the ability for the driver to run arbitrary code in the ramdisk is, well, interesting, but risky. | 20:02 |
| devananda | russell_h: something along those lines is needed to support firmware udpates etc. there is at least one open blueprint to that effect | 20:03 |
| devananda | russell_h: but no support for it today | 20:03 |
| davidlenwell | devananda: might be .. i review in a bit .. have refstack meeting in 2 hours im prepping for | 20:03 |
| NobodyCam | morning davidlenwell :) | 20:09 |
| NobodyCam | devananda: did you see my open question above? | 20:11 |
| *** jcooley_ has quit IRC | 20:12 | |
| *** jcooley_ has joined #openstack-ironic | 20:12 | |
| *** mrda_away is now known as mrda | 20:22 | |
| mrda | morning all | 20:23 |
| NobodyCam | morning mrda | 20:28 |
| mrda | hey NobodyCam | 20:28 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Add testing and doc sections to docs/dev-quickstart https://review.openstack.org/71638 | 20:29 |
| devananda | NobodyCam: mm,which one? | 20:29 |
| devananda | NobodyCam: oh, about get_iscsi_initiator? | 20:31 |
| devananda | NobodyCam: if that's the only thing we need libvirt/utils for -- then just copy it to ironic/utils or something | 20:31 |
| NobodyCam | :) cool hehehe | 20:31 |
| devananda | NobodyCam: I thought the objection was w.r.t. the volume driver | 20:31 |
| devananda | and iirc we use a lot of code from that | 20:31 |
| NobodyCam | we have our onw VD file | 20:32 |
| NobodyCam | so was down to the libvirt import | 20:32 |
| NobodyCam | devananda: see Daniel Berrange's comment on https://review.openstack.org/#/c/71026 | 20:37 |
| devananda | yuck | 20:39 |
| devananda | i get his point, but ... jsut copying the libvirt vol driver code? | 20:39 |
| devananda | we'll probably have to push on nova to refactor it out into a shared lib, it seems | 20:40 |
| devananda | how do other hypervisors deal with this? | 20:40 |
| NobodyCam | nova bm imported the libvirt module | 20:40 |
| NobodyCam | :-p | 20:40 |
| NobodyCam | I can just copy the function and then start a patch to refactor libvirt/nova bm and Ironic to use a shared lib? | 20:42 |
| NobodyCam | brb quick walkies :-p | 20:44 |
| openstackgerrit | Michael Davies proposed a change to openstack/ironic: Return correct HTTP response codes for create ops https://review.openstack.org/71425 | 20:48 |
| openstackgerrit | Michael Davies proposed a change to openstack/ironic: Make the parameter order of assertEqual consistent https://review.openstack.org/71486 | 20:49 |
| devananda | NobodyCam: see openstack-nova -- we should refactor to a common lib, eg. nova/virt/volume, or such | 20:50 |
| devananda | NobodyCam: core agrees with daniel berrange -- no importing from /within/ another virt driver | 20:50 |
| NobodyCam | devananda: one patch to do all refactor work or two (1) add common lib & (2) refactor virt drivers (with out looking) libvirt/nova bm / and Ironic | 20:54 |
| devananda | mrda: looks like a rebasing issue with 71486 has brought a bunch of app.AppError tests back in? | 20:55 |
| devananda | NobodyCam: one patch to: add the method in common location, remove implementation from libvirt and nova/bm, and invoke the common lib in those two places | 20:57 |
| devananda | NobodyCam: and one patch to add it for ironic | 20:57 |
| NobodyCam | devananda: ack :) | 20:57 |
| devananda | NobodyCam: mention in the refactor-existing-code patch's commit message why you're doing it, though :) | 20:57 |
| NobodyCam | :) | 20:58 |
| NobodyCam | yes | 20:58 |
| devananda | GheRivero: 71297 is a draft. can you un-draft it? we can't vote | 21:07 |
| devananda | fwiw, there's almost never a good time to use a draft | 21:07 |
| mrda | devananda: yeah I saw that re: review 71486. Once I can work out how to fix I'll upload new patches | 21:08 |
| mrda | I'd rather not do it by hand :) | 21:09 |
| devananda | mrda: indeed! | 21:09 |
| devananda | mrda: fwiw the first one is probably going to merge soon | 21:09 |
| *** athomas has joined #openstack-ironic | 21:09 | |
| devananda | dunno if that helps or makes it worse | 21:10 |
| mrda | devananda: cool, thnx | 21:10 |
| devananda | afk for a lil while | 21:24 |
| *** jcooley_ has quit IRC | 21:28 | |
| openstackgerrit | A change was merged to openstack/ironic: Return correct HTTP response codes for create ops https://review.openstack.org/71425 | 21:29 |
| *** jcooley_ has joined #openstack-ironic | 21:29 | |
| GheRivero | devananda: ups. I didn't intend to made it a draft. Done | 21:41 |
| *** epim has quit IRC | 22:01 | |
| *** epim has joined #openstack-ironic | 22:03 | |
| *** jbjohnso has quit IRC | 22:17 | |
| *** jdob has quit IRC | 22:23 | |
| *** epim has quit IRC | 22:26 | |
| *** epim has joined #openstack-ironic | 22:27 | |
| NobodyCam | brb | 22:46 |
| *** mdurnosvistov_ has quit IRC | 22:49 | |
| openstackgerrit | Michael Davies proposed a change to openstack/ironic: Make the parameter order of assertEqual consistent https://review.openstack.org/71486 | 22:55 |
| openstackgerrit | Michael Davies proposed a change to openstack/ironic: Make the parameter order of assertEqual consistent https://review.openstack.org/71486 | 22:56 |
| openstackgerrit | A change was merged to openstack/ironic: Fix broken doc links on the index page https://review.openstack.org/71632 | 23:00 |
| *** epim has quit IRC | 23:04 | |
| *** matty_dubs is now known as matty_dubs|gone | 23:08 | |
| *** epim has joined #openstack-ironic | 23:09 | |
| *** romcheg_ltp has quit IRC | 23:26 | |
| openstackgerrit | A change was merged to openstack/ironic: Fix log and test for NeutronAPI.update_port_dhcp_opts https://review.openstack.org/71094 | 23:32 |
| NobodyCam | brb... walkies | 23:54 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!