| *** epim has joined #openstack-ironic | 00:04 | |
| NobodyCam | devananda: you around to click wip? | 00:15 |
|---|---|---|
| *** matsuhashi has joined #openstack-ironic | 00:26 | |
| devananda | NobodyCam: back | 00:38 |
| NobodyCam | :) | 00:42 |
| * NobodyCam is going to call it a day | 00:42 | |
| NobodyCam | lol | 00:43 |
| *** epim has quit IRC | 00:51 | |
| *** Haomeng|2 has quit IRC | 01:04 | |
| *** Haomeng has joined #openstack-ironic | 01:04 | |
| devananda | NobodyCam: clicked WIP. also - you see hemna's review yet? | 01:07 |
| NobodyCam | devananda: the libvirt stuff | 01:08 |
| devananda | NobodyCam: actually, it's mostly LOG comments -- let me second that. the LOG.info is a bit excessive. I think those are really LOG.debug that you should remove soon, as reviewers are starting to take an interest .... | 01:08 |
| NobodyCam | those are removed | 01:09 |
| devananda | :) | 01:09 |
| NobodyCam | ya | 01:09 |
| devananda | NobodyCam: the libvirt comment is a bit trickier. any thoughts? | 01:09 |
| NobodyCam | other then the fixme :-p | 01:10 |
| devananda | lol | 01:11 |
| devananda | k k. have a good night :) | 01:11 |
| NobodyCam | lol | 01:11 |
| NobodyCam | :-p | 01:11 |
| *** rloo has quit IRC | 01:16 | |
| *** rloo has joined #openstack-ironic | 01:16 | |
| openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Update docstrings in ssh.py https://review.openstack.org/70033 | 01:17 |
| *** epim has joined #openstack-ironic | 01:18 | |
| *** nosnos has joined #openstack-ironic | 01:38 | |
| *** epim has quit IRC | 01:48 | |
| *** rloo has quit IRC | 01:59 | |
| openstackgerrit | A change was merged to openstack/ironic: Add missing parameter in call to _load_one_plugin https://review.openstack.org/69495 | 02:12 |
| *** vkozhukalov has joined #openstack-ironic | 02:20 | |
| *** jcooley_ has quit IRC | 02:47 | |
| NobodyCam | devananda: ^^^^^^^^ woo hoo | 03:04 |
| openstackgerrit | A change was merged to openstack/ironic: Docstrings for ipmitool https://review.openstack.org/68665 | 03:19 |
| *** coolsvap_away has quit IRC | 03:27 | |
| *** jcooley_ has joined #openstack-ironic | 03:27 | |
| *** matsuhashi has quit IRC | 03:29 | |
| *** jcooley_ has quit IRC | 03:34 | |
| *** jcooley_ has joined #openstack-ironic | 03:35 | |
| *** jcooley_ has quit IRC | 03:40 | |
| *** leizhang has joined #openstack-ironic | 03:44 | |
| *** leizhang has quit IRC | 04:01 | |
| *** matsuhashi has joined #openstack-ironic | 04:02 | |
| *** aignatov_ is now known as aignatov | 04:04 | |
| *** jcooley_ has joined #openstack-ironic | 04:10 | |
| *** jcooley_ has quit IRC | 04:16 | |
| *** jcooley_ has joined #openstack-ironic | 04:17 | |
| *** jcooley_ has quit IRC | 04:21 | |
| *** jcooley_ has joined #openstack-ironic | 04:22 | |
| *** aignatov is now known as aignatov_ | 04:50 | |
| *** matsuhashi has quit IRC | 05:14 | |
| *** matsuhas_ has joined #openstack-ironic | 05:17 | |
| *** coolsvap has joined #openstack-ironic | 05:19 | |
| *** rwsu has quit IRC | 05:27 | |
| openstackgerrit | Jenkins proposed a change to openstack/ironic: Imported Translations from Transifex https://review.openstack.org/68024 | 06:07 |
| *** vkozhukalov has quit IRC | 06:18 | |
| *** jcooley_ has quit IRC | 06:59 | |
| *** jcooley_ has joined #openstack-ironic | 07:00 | |
| *** hstimer has quit IRC | 07:06 | |
| *** jcooley_ has quit IRC | 07:16 | |
| *** GheRivero has quit IRC | 07:25 | |
| *** mrda is now known as mrda_away | 07:32 | |
| ifarkas | NobodyCam, sorry, I was out yesterday evening | 07:45 |
| ifarkas | NobodyCam, ping me when you will be online | 07:45 |
| *** jcooley_ has joined #openstack-ironic | 07:51 | |
| *** romcheg has joined #openstack-ironic | 07:53 | |
| *** vkozhukalov has joined #openstack-ironic | 07:57 | |
| *** jistr has joined #openstack-ironic | 08:02 | |
| *** jcooley_ has quit IRC | 08:05 | |
| *** ifarkas_ has joined #openstack-ironic | 08:09 | |
| *** vkozhukalov has quit IRC | 08:10 | |
| *** ifarkas has quit IRC | 08:13 | |
| *** GheRivero has joined #openstack-ironic | 08:16 | |
| *** harlowja is now known as harlowja_away | 08:18 | |
| *** vkozhukalov has joined #openstack-ironic | 08:19 | |
| *** nosnos has quit IRC | 08:39 | |
| *** nosnos has joined #openstack-ironic | 08:39 | |
| *** vkozhukalov has quit IRC | 08:43 | |
| *** mdurnosvistov has joined #openstack-ironic | 08:46 | |
| *** jcooley_ has joined #openstack-ironic | 08:46 | |
| openstackgerrit | Mikhail Durnosvistov proposed a change to openstack/ironic: Get rid of the newline "\" https://review.openstack.org/66793 | 08:52 |
| *** aignatov_ is now known as aignatov | 08:56 | |
| *** vkozhukalov has joined #openstack-ironic | 08:57 | |
| GheRivero | anyone else is seing unique constrains error when running tox? | 08:58 |
| GheRivero | tested in a new clean environment and the same. WTF! | 09:08 |
| openstackgerrit | Yuriy Zveryanskyy proposed a change to openstack/ironic: Add ability to break TaskManager locks via REST API https://review.openstack.org/55549 | 09:10 |
| *** GheRiver1 has joined #openstack-ironic | 09:14 | |
| *** GheRivero has left #openstack-ironic | 09:16 | |
| *** GheRiver1 is now known as GheRivero | 09:16 | |
| *** jcooley_ has quit IRC | 09:16 | |
| yuriyz | morning Ironic | 09:17 |
| yuriyz | GheRivero, I dont see | 09:17 |
| GheRivero | morning | 09:17 |
| yuriyz | Please copy error message | 09:18 |
| GheRivero | :/ | 09:18 |
| GheRivero | http://paste.openstack.org/show/62153/ | 09:19 |
| GheRivero | in Debian Sid + SQLite 3.8.2 | 09:19 |
| yuriyz | Ghe, maybe you need this https://review.openstack.org/#/c/69004/ | 09:21 |
| GheRivero | agh! i now i had seen it someplace but couldn't find it! testing it | 09:22 |
| *** ndipanov has joined #openstack-ironic | 09:23 | |
| mdurnosvistov | Morning all! :) | 09:23 |
| *** derekh has joined #openstack-ironic | 09:26 | |
| *** matsuhas_ has quit IRC | 09:33 | |
| *** matsuhashi has joined #openstack-ironic | 09:39 | |
| *** Alexei_987 has joined #openstack-ironic | 09:44 | |
| *** athomas has joined #openstack-ironic | 09:44 | |
| *** nosnos has quit IRC | 09:46 | |
| *** nosnos has joined #openstack-ironic | 09:46 | |
| *** nosnos has quit IRC | 09:47 | |
| *** martyntaylor has joined #openstack-ironic | 09:48 | |
| *** matsuhashi has quit IRC | 09:50 | |
| *** ifarkas_ has quit IRC | 10:08 | |
| *** jistr has quit IRC | 10:08 | |
| *** jcooley_ has joined #openstack-ironic | 10:09 | |
| *** ifarkas has joined #openstack-ironic | 10:11 | |
| Haomeng | morning all:) | 10:11 |
| GheRivero | morning | 10:12 |
| *** jcooley_ has quit IRC | 10:14 | |
| *** aignatov is now known as aignatov_ | 10:15 | |
| *** aignatov_ is now known as aignatov | 10:15 | |
| *** GheRivero has quit IRC | 10:20 | |
| *** coolsvap has quit IRC | 10:28 | |
| *** aignatov is now known as aignatov_ | 10:31 | |
| *** jistr has joined #openstack-ironic | 10:38 | |
| *** tatyana has joined #openstack-ironic | 10:46 | |
| *** vkozhukalov has quit IRC | 10:48 | |
| *** lucasagomes has joined #openstack-ironic | 10:56 | |
| *** derekh has quit IRC | 10:58 | |
| *** vkozhukalov has joined #openstack-ironic | 11:00 | |
| *** jcooley_ has joined #openstack-ironic | 11:02 | |
| *** aignatov_ is now known as aignatov | 11:04 | |
| *** max_lobur_afk is now known as max_lobur | 11:05 | |
| *** jcooley_ has quit IRC | 11:16 | |
| *** ko__ has joined #openstack-ironic | 11:25 | |
| *** coolsvap has joined #openstack-ironic | 11:31 | |
| openstackgerrit | Max Lobur proposed a change to openstack/ironic: Fix JSONEncodedDict default values https://review.openstack.org/68413 | 12:18 |
| *** coolsvap has quit IRC | 12:20 | |
| *** anyone_else has joined #openstack-ironic | 12:29 | |
| *** anyone_else is now known as romcheg1 | 12:41 | |
| *** jcooley_ has joined #openstack-ironic | 12:51 | |
| lucasagomes | NobodyCam, devananda: w00ts,http://s21.postimg.org/yuxilhft1/Screenshot_from_2014_01_30_12_53_02.png | 12:58 |
| lucasagomes | nova w/ ironic driver + ironic | 12:58 |
| lucasagomes | deloyed :) | 12:58 |
| lucasagomes | ifarkas, ^ | 12:58 |
| ifarkas | lucasagomes, hey, congrats! | 12:59 |
| romcheg1 | lucasagomes: cooooool! | 13:00 |
| lucasagomes | cheers, needs some polish tho | 13:00 |
| ifarkas | lucasagomes, really cool! What image did you use the node? | 13:00 |
| lucasagomes | it's the cirros image | 13:00 |
| romcheg1 | I'm trying to deploy something with Ironic as well | 13:00 |
| lucasagomes | ifarkas, http://s8.postimg.org/gfcjnnf7n/Screenshot_from_2014_01_30_13_01_11.png | 13:01 |
| lucasagomes | now it finished booting | 13:01 |
| romcheg1 | What a beauty :) | 13:02 |
| lucasagomes | romcheg1, great, yea we need to test it as much as possible | 13:02 |
| lucasagomes | I will try to write a guideline | 13:02 |
| romcheg1 | I already found a bug :) | 13:02 |
| lucasagomes | heh yea | 13:02 |
| lucasagomes | in the image you can see that the power state is not being synced | 13:02 |
| ifarkas | +1 for the guide. the previous one was also really helpful! | 13:02 |
| lucasagomes | from ironic to nova heh | 13:02 |
| romcheg1 | We created Ironic and now are trying to guess how to use it :-P | 13:03 |
| lucasagomes | hah yea | 13:03 |
| ifarkas | lol | 13:03 |
| *** vkozhukalov has quit IRC | 13:03 | |
| *** GheRivero has joined #openstack-ironic | 13:09 | |
| *** vkozhukalov has joined #openstack-ironic | 13:18 | |
| *** jcooley_ has quit IRC | 13:25 | |
| ko__ | hi | 13:34 |
| ko__ | I am ne to openstack and I am trying to setup ironic using devstack | 13:34 |
| ko__ | can someone please help me.. | 13:35 |
| lucasagomes | ko__, hi there | 13:35 |
| lucasagomes | please take a look at | 13:35 |
| lucasagomes | https://wiki.openstack.org/wiki/Ironic#Try_it_on_Devstack | 13:35 |
| ko__ | Thanks for responding. I have followed that guide.. | 13:36 |
| ko__ | However, I am not sure if it is necessary to setup a undercloud to see ironic in action.. | 13:36 |
| lucasagomes | ko__, hmm ironic is not 100% ready yet | 13:37 |
| lucasagomes | there's another guide that you can follow using devstack | 13:37 |
| lucasagomes | that you can get ironic to boot a vm | 13:37 |
| ko__ | is there a way to skip setting up undercloud and see ironic in action..? | 13:37 |
| lucasagomes | so u can start doing some experiements with it | 13:37 |
| lucasagomes | ko__, yes | 13:37 |
| lucasagomes | ko__, lemme find the link for ya | 13:37 |
| lucasagomes | ko__, https://etherpad.openstack.org/p/IronicDeployDevstack | 13:37 |
| ko__ | that would be great.. | 13:37 |
| lucasagomes | ko__, it's experimental yet, we are working on the tripleO integration in this cycle as well | 13:38 |
| ko__ | okay.. | 13:39 |
| ko__ | Let me validate my environment with you. I have a ESX host. On one of the VM I have setup devstack and I have another VM that I want to manage using ironic. | 13:40 |
| ko__ | I was not sure what driver I have to use when regisetring the node in ironic so I used fake_ssh driver.. | 13:41 |
| ko__ | And I have a common network shared between devstack vm and the baremetal node. and one uplink on the devstack vm using which I can connect to it.. | 13:42 |
| ko__ | Is there anything spection that I have to take care of before I fire stack.sh..? | 13:43 |
| lucasagomes | ko__, environment seems fine | 13:46 |
| lucasagomes | only the driver used (you can see on the guide) | 13:46 |
| lucasagomes | is pxe_ssh | 13:46 |
| lucasagomes | so we use ssh in the tests as a mock for ipmi | 13:46 |
| ko__ | So you mean to say that I have to use pxe_ssh instead of fake_ssh.. | 13:47 |
| lucasagomes | ko__, yup | 13:47 |
| lucasagomes | fake is just used for our unittests | 13:47 |
| ko__ | do i also have to setup virsh on the esx host..? | 13:48 |
| lucasagomes | ssh = ur power interface (will power on/off the node issuing virsh commands) | 13:48 |
| lucasagomes | ko__, yes | 13:48 |
| ko__ | okay.. thanks a lot.. Let me try it out pretty quick.. | 13:49 |
| lucasagomes | ko__, np :) | 13:49 |
| openstackgerrit | A change was merged to openstack/ironic: API: Add sample() method on Node https://review.openstack.org/65536 | 13:50 |
| *** jdob has joined #openstack-ironic | 13:54 | |
| *** romcheg1 has quit IRC | 14:03 | |
| *** vkozhukalov has quit IRC | 14:06 | |
| ko__ | lucasagomes: I followed the https://etherpad.openstack.org/p/IronicDeployDevstack till line number 65. When I execute line number 68, when I try to start service openvswitch, i get unrecognized service.. | 14:07 |
| ko__ | is this something expected..? | 14:07 |
| ko__ | also, apt-get install openvswitch didnt help either.. | 14:08 |
| lucasagomes | ko__, openvswitch is not installed | 14:09 |
| lucasagomes | I think on ubuntu it;s openvswitch-switch or something like that lemme see | 14:09 |
| lucasagomes | but hmm | 14:09 |
| lucasagomes | you can configure the bridge ip by hand if u want as well | 14:09 |
| ko__ | can you please help me with the commands please.. | 14:09 |
| ko__ | my eth0 on devstack is uplink and eth1 is shared by the baremetal node.. | 14:10 |
| lucasagomes | apt-get install openvswitch-switch | 14:10 |
| lucasagomes | right, yea all you need is a common network between both nodes | 14:11 |
| lucasagomes | so you can pxe boot it | 14:11 |
| ko__ | I checked and it seems openvswitch-switch package is already installed and is latest.. | 14:11 |
| ko__ | but it still errors when I try to start the service.. | 14:12 |
| lucasagomes | ko__, service openvswitch-switch restart ? | 14:12 |
| ko__ | let me try.. | 14:12 |
| ko__ | lucasagomes: that worked. thanks.. | 14:12 |
| lucasagomes | ko__, :) np updated the guide as well for both os | 14:13 |
| lucasagomes | ko__, heh as I told you it's experimental yet (including that guide) | 14:13 |
| lucasagomes | feel free to update it with ur findings | 14:13 |
| lucasagomes | that will be very helpful | 14:13 |
| ko__ | I dont mind trying experimental stuff.. I would be happy to help you in updating the guide.. However i need to know what works before I update it... and I need help from community to do that as i am just a beginner in linux.. | 14:15 |
| *** matty_dubs|gone is now known as matty_dubs | 14:16 | |
| *** jcooley_ has joined #openstack-ironic | 14:18 | |
| lucasagomes | sure :) | 14:18 |
| *** vkozhukalov has joined #openstack-ironic | 14:19 | |
| *** max_lobur is now known as max_lobur_afk | 14:26 | |
| *** ko__ has quit IRC | 14:26 | |
| *** ko__ has joined #openstack-ironic | 14:27 | |
| *** jcooley_ has quit IRC | 14:27 | |
| *** rloo has joined #openstack-ironic | 14:28 | |
| ko__ | lucasagomes: can you please help me with the network creation on ubuntu.. seems like the script is not working for me. | 14:30 |
| lucasagomes | ko__, right, any output? | 14:31 |
| ko__ | one sec.. | 14:31 |
| ko__ | error: command 'net-list' doesn't support option --persistent | 14:33 |
| ko__ | error: command 'net-list' doesn't support option --persistent | 14:33 |
| ko__ | error: Failed to define network from /dev/fd/63 | 14:33 |
| ko__ | error: XML error: unexpected virtualport type -1 | 14:33 |
| ko__ | lucasagomes: these are four lines that get printed when I execute setup-network | 14:34 |
| lucasagomes | right | 14:34 |
| lucasagomes | ko__, I think it's because of ur virsh version | 14:34 |
| lucasagomes | to not support the persistent option | 14:35 |
| ko__ | okay.. | 14:35 |
| lucasagomes | [lucasagomes@lucasagomes nova]$ virsh --version | 14:35 |
| lucasagomes | 1.1.3.3 | 14:35 |
| lucasagomes | is ur virsh 1.1.*+ ? | 14:35 |
| ko__ | so I tried after editing the setup-network script (removed --persistent) and the first two error disappeared.. | 14:36 |
| ko__ | byut the last two are still there.. | 14:36 |
| ko__ | my virsh version is 0.9.8 | 14:36 |
| ko__ | I am using Ubuntu 12.04.3 if that matters in any way.. | 14:37 |
| lucasagomes | right, lemme see if I can find something | 14:38 |
| lucasagomes | (actually I dunno much ubuntu, have been using fedora for a long time since now) | 14:38 |
| ko__ | oh.. | 14:38 |
| lucasagomes | ko__, wondering if it's virsh permission | 14:39 |
| lucasagomes | try to | 14:39 |
| lucasagomes | export LIBVIRT_DEFAULT_URI="qemu:///system" | 14:39 |
| lucasagomes | and run it | 14:39 |
| ko__ | one question.. since I already have one network connected for baremetal VM on devstack, is this step still needed..? | 14:39 |
| lucasagomes | ko__, I don't think so | 14:40 |
| lucasagomes | if the vms can communicate with each other that's what's needed | 14:40 |
| lucasagomes | that script is just to try to make things easier when creating the bridge for both vms | 14:40 |
| lucasagomes | (but it seems it's complicating more than helping) | 14:40 |
| ko__ | So basically, I can skip this step and I can move to line number 82. is that correct.. | 14:41 |
| lucasagomes | (we assumed that people would be familar with tripleO, that's why we are using the tripleO scripts) | 14:41 |
| lucasagomes | ko__, yes, which is also a tripleo script heh | 14:41 |
| lucasagomes | but wait | 14:41 |
| ko__ | I am sorry.. As of now, I am just new to openstack as well as tripleo.. | 14:42 |
| lucasagomes | if you already have a vm you want to manage via ironic you might won't need that as well | 14:42 |
| lucasagomes | cause all that script does is to create/configure a vm to be managed by ironic | 14:42 |
| lucasagomes | you already have one is that correct? | 14:42 |
| lucasagomes | ko__, nothing to be sorry :) | 14:42 |
| ko__ | yes.. I already have a VM on the ESX host pretending as baremetal node. | 14:42 |
| lucasagomes | make sure ur vm is configured to boot via network | 14:43 |
| *** aignatov is now known as aignatov_ | 14:43 | |
| ko__ | it is configured to boot via network. but when I start the VM, it keeps looking for PXE response in a loop.. | 14:43 |
| lucasagomes | ko__, that's correct :) | 14:44 |
| lucasagomes | ko__, in that guideline you will have to setup dnsmasq to answer the pxe dhcp request | 14:44 |
| ko__ | so I should skip to dnsmasq setup.. let me do that.. | 14:45 |
| lucasagomes | (it's just one workaround while the neutron is not complete integrated with ironic yet) | 14:45 |
| lucasagomes | ko__, no, well you'll need the dnsmasq | 14:45 |
| ko__ | ok.. I meant I will skip and start with the lace where dnsmasq config starts.. | 14:46 |
| lucasagomes | ah gotcha | 14:49 |
| *** aignatov_ is now known as aignatov | 14:53 | |
| *** rloo has quit IRC | 14:53 | |
| GheRivero | ~~~p0- | 14:54 |
| GheRivero | ]]]]]]]]]]]]]]]]]]]]] | 14:55 |
| *** max_lobur_afk is now known as max_lobur | 14:56 | |
| ko__ | lucasagomes: I completed the network creation but Now i get error on line 165 | 15:01 |
| ko__ | HTTPForbidden (HTTP 403) | 15:01 |
| ko__ | I mean my baremetal is now able to get pxe IP address... | 15:02 |
| ko__ | and when I execute any ironic command, its givving me this error. | 15:03 |
| lucasagomes | ko__, right so when registering the node with ironic it's returning 403? | 15:03 |
| lucasagomes | any error message? | 15:03 |
| ko__ | yes.. | 15:03 |
| ko__ | even if i execute 'ironic driver-list', it gives me the same error.. | 15:03 |
| ko__ | nope.. | 15:04 |
| lucasagomes | did registering the images with glance worked? | 15:04 |
| ko__ | yes that worked.. | 15:04 |
| lucasagomes | ko__, I think it's a bug in devstack, we have a patch for it at the begginning of the guide | 15:04 |
| lucasagomes | (which is abandoned but is needed) | 15:04 |
| ko__ | I am looking at the ironic api log and see that there is no auth token in header.. | 15:04 |
| lucasagomes | lemme find for u | 15:04 |
| lucasagomes | https://review.openstack.org/#/c/66925/ | 15:04 |
| lucasagomes | check ur auth_url | 15:05 |
| lucasagomes | in the ironic.conf | 15:05 |
| lucasagomes | see if it's incliude the version | 15:05 |
| lucasagomes | ko__, you can add the /v2.0 by hand in the /etc/ironic/ironic.conf | 15:05 |
| lucasagomes | and then you'll have to restart the ironic services | 15:05 |
| ko__ | okay.. let me do that.. | 15:05 |
| NobodyCam | morning ironic says the man as he starts the coffee | 15:06 |
| lucasagomes | NobodyCam, morning | 15:06 |
| lucasagomes | NobodyCam, gotta talk to u | 15:06 |
| NobodyCam | :) | 15:06 |
| lucasagomes | NobodyCam, it's ok if I submit new patch-sets to the driver? | 15:06 |
| lucasagomes | NobodyCam, btw, I managed to deploy a machine with nova+ironic | 15:06 |
| NobodyCam | one minute | 15:06 |
| NobodyCam | awesome and yes please do | 15:07 |
| lucasagomes | NobodyCam, http://i.imgur.com/NvfCFAS.png :) | 15:07 |
| NobodyCam | w00t | 15:08 |
| NobodyCam | brb | 15:08 |
| lucasagomes | NobodyCam, cheers I will work on a patch-set | 15:08 |
| lucasagomes | NobodyCam, and great work at the driver :D devananda as well | 15:08 |
| ifarkas | NobodyCam, Good morning | 15:14 |
| NobodyCam | morning ifarkas :) | 15:14 |
| ifarkas | NobodyCam, regarding your question yesterday about my review, does the /mnt/state/boot-stack is created dir exist for you? | 15:15 |
| ifarkas | NobodyCam, My undercloud is missing /mnt/state/boot-stack and /mnt/state/opt/boot-stack | 15:16 |
| NobodyCam | ifarkas: ya.. ws just looking at clints comments, | 15:16 |
| NobodyCam | I will push up a patch that sets it as /mnt/state/opt/boot-stack | 15:17 |
| ifarkas | NobodyCam, so, do you have /mnt/state/opt/boot-stack dir? because I don't | 15:18 |
| NobodyCam | ifarkas: last patch creates it: https://review.openstack.org/#/c/66461/5/elements/nova-ironic/os-refresh-config/configure.d/80-ironic-ssh-power-key | 15:18 |
| ifarkas | NobodyCam, oh, right. cool! | 15:19 |
| ifarkas | NobodyCam, I missed that line somehow :-) | 15:19 |
| NobodyCam | ifarkas: ok new patch up | 15:23 |
| NobodyCam | lucasagomes: fyi : https://review.openstack.org/#/c/66461/ just push new version up | 15:24 |
| lucasagomes | NobodyCam, cheers | 15:24 |
| *** lucasagomes is now known as lucas-hungry | 15:24 | |
| *** rwsu has joined #openstack-ironic | 15:24 | |
| ko__ | lucas-hungry, followed till the end of the guide. but VM wont boot... it is still stuck with the PXE. It is able to get DHCP ip address but its not able to receive boot files. | 15:25 |
| NobodyCam | ko__: is it trying to pxe boot at all? | 15:25 |
| ko__ | yes.. it is... | 15:26 |
| ko__ | there is no change in the /tftproot directory on the devstack box.. | 15:26 |
| ko__ | is there something wrong.. | 15:26 |
| NobodyCam | who is the owner of the /tftpdir | 15:26 |
| ko__ | its root and permission is 777 | 15:27 |
| NobodyCam | ok | 15:27 |
| NobodyCam | that works too | 15:27 |
| NobodyCam | :) | 15:27 |
| NobodyCam | how about /var/lib/ironic | 15:27 |
| ifarkas | NobodyCam, Looks good to me ;-) | 15:28 |
| NobodyCam | :) | 15:28 |
| ko__ | same. root and 777 | 15:28 |
| NobodyCam | ifarkas: thank you for the review :) | 15:29 |
| ifarkas | NobodyCam, thanks for the fix ;-) | 15:29 |
| NobodyCam | ko__: is there anything in the ironic-conductor.log ? | 15:29 |
| ifarkas | NobodyCam, ...and for the whole element | 15:29 |
| openstackgerrit | Ghe Rivero proposed a change to openstack/ironic: Remove deploy kernel and ramdisk global config https://review.openstack.org/70140 | 15:29 |
| openstackgerrit | Ghe Rivero proposed a change to openstack/ironic: Remove net_config_template options https://review.openstack.org/70141 | 15:29 |
| NobodyCam | GheRivero: :) | 15:30 |
| ko__ | NobodyCam, any hint on log file name for ir-cond..? | 15:32 |
| NobodyCam | in dib setup it is /var/log/upstart/ironic-conductor.log | 15:33 |
| NobodyCam | not sure in devstack maybe /var/log/ironic-conductor.log | 15:33 |
| NobodyCam | can you paste the deploy bits to us? | 15:34 |
| ko__ | NobodyCam, I am sorry but i am having trouble finding the logs.. | 15:36 |
| NobodyCam | oh devstack you prob have a screen for that | 15:36 |
| ko__ | yes.. but i thought, its getting logged in some file too.. | 15:37 |
| NobodyCam | gah been awhile for me and devstack | 15:37 |
| NobodyCam | can you cut and paste the screen log to http://paste.openstack.org | 15:38 |
| ko__ | NobodyCam, did that.. Paste #62177 | 15:39 |
| *** rloo has joined #openstack-ironic | 15:40 | |
| ko__ | do you want me to execute curl cmd again and then paste it over there..? | 15:40 |
| NobodyCam | ko__: can you send the entire url | 15:40 |
| ko__ | http://paste.openstack.org/show/62177/ | 15:41 |
| NobodyCam | :) | 15:41 |
| NobodyCam | dont even see it tring the deploy | 15:42 |
| NobodyCam | how about n-compute and n-sch | 15:42 |
| ko__ | one sec.. | 15:45 |
| NobodyCam | sure | 15:46 |
| NobodyCam | GheRivero: why are you removing net_config_template??? | 15:47 |
| ko__ | http://paste.openstack.org/show/62179/ | 15:49 |
| NobodyCam | ko__: second line | 15:50 |
| NobodyCam | paramiko.transport [-] Authentication (password) failed. | 15:50 |
| ko__ | ohh.. is this password for api or for virsh for the host..? | 15:50 |
| *** aignatov is now known as aignatov_ | 15:50 | |
| GheRivero | NobodyCam: we are not doing file injection and there is no template anymore | 15:50 |
| NobodyCam | ko__: that should be virsh host | 15:51 |
| NobodyCam | GheRivero: have you seen the dev email thread re: [openstack-dev] [Ironic] File Injection (and the lack thereof) | 15:52 |
| ko__ | okay.. the fact is that the baremetal VM is on a ESX host which does not have virsh installed. so i used fake_ssh driver for the node. | 15:52 |
| ko__ | is there anyway by which i can avoid installing virsh on the esx host. I dont have the host creds with me and I do not have access to modify the vsphere either.. | 15:54 |
| NobodyCam | ko__: https://bugs.launchpad.net/ironic/+bug/1260414 | 15:55 |
| GheRivero | NobodyCam: true. But we are not doing file injection now, and dhcp config was an ubuntu only option and... | 15:56 |
| NobodyCam | GheRivero: ack, I just wanted some understanding | 15:58 |
| NobodyCam | for the patch | 15:58 |
| GheRivero | and there was some consensus about it: http://paste.openstack.org/show/62181/ | 15:58 |
| GheRivero | but feel free to -2 it! :) | 15:58 |
| GheRivero | maybe I should had open a bug to annotate all this | 15:59 |
| NobodyCam | GheRivero: :) looks good to me | 15:59 |
| NobodyCam | hehe :) | 15:59 |
| NobodyCam | ko__: do you a actual node for testing? | 16:01 |
| NobodyCam | ie a bm-vm? | 16:02 |
| NobodyCam | if so use fake ipmi driver and manually turn on the power | 16:02 |
| *** athomas has quit IRC | 16:02 | |
| NobodyCam | sorry still not enough coffee | 16:03 |
| NobodyCam | fake_pxe | 16:03 |
| *** ifarkas has quit IRC | 16:09 | |
| *** athomas has joined #openstack-ironic | 16:11 | |
| *** jcooley_ has joined #openstack-ironic | 16:12 | |
| ko__ | nope.. i am using bm-vm | 16:14 |
| ko__ | and i use fake_pxe.. but I am getting a different error now.. | 16:14 |
| NobodyCam | foke_ssh is fake for pxe ans real ssh power control | 16:15 |
| NobodyCam | fake_pxe is what you want. as that is real pxe deploy and fake power | 16:15 |
| ko__ | NobodyCam, I am now getting some error in ir-api | 16:18 |
| ko__ | let me paste it and pass the link to you.. | 16:18 |
| NobodyCam | :) | 16:18 |
| NobodyCam | Ty | 16:18 |
| ko__ | http://paste.openstack.org/show/62185/ | 16:18 |
| NobodyCam | while you get that I'm going to grab another cup of coffee ... brb | 16:18 |
| NobodyCam | one sec | 16:18 |
| ko__ | take your time... | 16:19 |
| NobodyCam | ko__: can you paste me the output of ironic node-show <NODE_UUID> | 16:22 |
| ko__ | http://paste.openstack.org/show/62186/ | 16:23 |
| *** lucas-hungry is now known as lucasagomes | 16:24 | |
| ko__ | NobodyCam, I have followed https://etherpad.openstack.org/p/IronicDeployDevstack and I have only skipped line number 180 as I was not sure about it.. | 16:24 |
| ko__ | any luck.. | 16:29 |
| *** zul has quit IRC | 16:29 | |
| NobodyCam | you will need a instance id | 16:30 |
| NobodyCam | but I need to look whats up with the kernel_id | 16:31 |
| ko__ | Can you please help me get one.. | 16:31 |
| lucasagomes | ko__, fake_pxe? | 16:31 |
| ko__ | yes | 16:31 |
| NobodyCam | lucasagomes: WB | 16:31 |
| lucasagomes | :) | 16:31 |
| NobodyCam | lucasagomes: would not instance id on a node cause a kernel_id error in pxe? | 16:31 |
| *** zul has joined #openstack-ironic | 16:32 | |
| ko__ | let me try with line 180 then.. | 16:33 |
| *** martyntaylor has quit IRC | 16:33 | |
| lucasagomes | NobodyCam, yup, if instance doesn't contain a kernel_id or ramdisk_id | 16:33 |
| devananda | morning, all | 16:33 |
| NobodyCam | morning devananda :) | 16:33 |
| lucasagomes | morning devananda | 16:33 |
| *** martyntaylor has joined #openstack-ironic | 16:33 | |
| NobodyCam | lucasagomes: tell him the good news | 16:34 |
| *** lsmola_ has quit IRC | 16:34 | |
| lucasagomes | NobodyCam, yeah | 16:34 |
| lucasagomes | devananda, check it out | 16:34 |
| lucasagomes | devananda, http://i.imgur.com/NvfCFAS.png | 16:34 |
| lucasagomes | nova + ironic | 16:35 |
| NobodyCam | :):):) | 16:35 |
| * devananda tries to read scrollback before coffee finishes brewing | 16:35 | |
| NobodyCam | devananda: look at lucasagomes's link | 16:35 |
| romcheg | Morning devananda, NobodyCam | 16:35 |
| romcheg | How are you? | 16:35 |
| *** jcooley_ has quit IRC | 16:35 | |
| NobodyCam | morning romcheg :) | 16:36 |
| devananda | \o/ | 16:36 |
| NobodyCam | devananda: :) | 16:36 |
| devananda | lucasagomes: full deploy & it got the right IP. Now, could that instance reach out on the network? | 16:36 |
| NobodyCam | devananda: any thoughts about conversation Spamap S and I are having in OoO | 16:37 |
| devananda | fantastic stuff folks!! | 16:37 |
| devananda | NobodyCam: lemme read that scrollback .... | 16:37 |
| lucasagomes | devananda, one thing per time... this is only nova driver + ironic | 16:37 |
| lucasagomes | devananda, we splitted tests to try out the things as atomic parts | 16:37 |
| lucasagomes | devananda, we also got a machine booted with ipmi thank to matty_dubs :) | 16:37 |
| devananda | lucasagomes: awesome! | 16:38 |
| lucasagomes | we will test the neutron integration as well | 16:38 |
| NobodyCam | w00t!!!! | 16:38 |
| lucasagomes | and once we get all the parts in good shape | 16:38 |
| lucasagomes | we are going to test the whole integration | 16:38 |
| ko__ | NobodyCam, lucasagomes: it didnt work with line 180 either.. actually, it gave me an error.. | 16:39 |
| lucasagomes | ko__, so did ironic generate all the config files at the /tftproot folder? | 16:41 |
| ko__ | nope | 16:41 |
| ko__ | it only has pxelinux.0 | 16:41 |
| lucasagomes | right | 16:41 |
| NobodyCam | ko__: are the glance id vaild for your case? | 16:41 |
| ko__ | yes.. i have verified them assuming that I have to use image uuids | 16:42 |
| ko__ | | 8e2d4cb0-3598-4fef-bc97-614c2874f15e | Deploy Kernel | aki | aki | 5631184 | active | | 16:43 |
| ko__ | | 3d3a1233-d2b9-4f96-98e4-971b28a257a5 | Deploy Ramdisk | aki | aki | 80574387 | active | | 16:44 |
| lucasagomes | ko__, the pxe_image_source argument | 16:44 |
| lucasagomes | is the id of the cirros image in glance? | 16:44 |
| ko__ | | 089ab991-8c10-41d5-96a0-2fff8012b1eb | Windwos 2012 R2 | vhd | bare | 10411724800 | active | | 16:44 |
| NobodyCam | oh windows | 16:44 |
| ko__ | nope that a VHD image that I have uploaded and I ma trying to use that.. | 16:44 |
| lucasagomes | ohh windows and pxe boot heh | 16:44 |
| lucasagomes | right, ko__ that won't work that easily, actually we lack in support for windows in the moment | 16:45 |
| ko__ | yup.. but atleast it should boot to PXE for deployment.. right.. | 16:45 |
| NobodyCam | ko__: https://blueprints.launchpad.net/ironic/+spec/windows-pxe-localboot0 | 16:45 |
| ko__ | and then should fail at a later stage.. | 16:45 |
| lucasagomes | ko__, hmm no really, cause the pxe driver will try to cache the final image (to dd it to the disk) | 16:45 |
| lucasagomes | and also the ramdisk+kernel associated with that image | 16:45 |
| lucasagomes | in that case ur windows image has no kernel and ramdisk | 16:46 |
| ko__ | k.. let me try with the cirrus then | 16:46 |
| lucasagomes | in order to pxe windows, we will need some mechanism in ironic to tell that it's an windows image and generate a diff pxe config file | 16:47 |
| lucasagomes | ^ not implemented yet (take a look at the link NobodyCam gave u( | 16:47 |
| *** hemna has quit IRC | 16:47 | |
| ko__ | okay.. but that part comes into picture once bm-vm has booted into deployment mode. | 16:48 |
| ko__ | in my case its not pxe booting into deployment mode... | 16:48 |
| lucasagomes | ko__, yes but the pxe driver will cache the final image before | 16:48 |
| ko__ | okay.. so is there a logic in pxe driver which would inspect the image to check if its windows or linux and takes appropriate action.. Sorry.. I am not clear and I just want to understand better.. | 16:49 |
| lucasagomes | ko__, yup, but don't apologize :) | 16:50 |
| lucasagomes | we lack in documentation in the moment | 16:50 |
| lucasagomes | it's our fault heh | 16:50 |
| NobodyCam | :-p | 16:51 |
| ko__ | can you please be specific as in what inspection logic exists.. any line numbers from the code would be great.. | 16:52 |
| lucasagomes | ko__, in the moment there's no logic for that yet | 16:52 |
| lucasagomes | ko__, https://blueprints.launchpad.net/ironic/+spec/windows-pxe-localboot0 | 16:53 |
| lucasagomes | that would implement that ^ | 16:53 |
| devananda | so, no, we shouldn't be inspecting the image _contents_ to determine what kind of OS it is | 16:54 |
| devananda | there should be image meta-data which specifies | 16:54 |
| lucasagomes | devananda, +1 | 16:55 |
| lucasagomes | sorry I think I missed what he said, yea it won't mount the image to inspect it | 16:56 |
| lucasagomes | even because by mounting the filesystem you already modify an image (fs flags for checksum etc) | 16:56 |
| lucasagomes | it needs to be something external | 16:56 |
| devananda | yep | 16:57 |
| max_lobur | morning / afternoon Everyone | 16:58 |
| NobodyCam | morning max_lobur :) | 16:59 |
| ko__ | I get that.. with cirrus image pxe config is getting created under /tftproot. but bm-vm still does not boot. | 17:00 |
| NobodyCam | lucasagomes: note: keyname change in https://review.openstack.org/#/c/66461 just pushed up | 17:01 |
| ko__ | meaning its not getting response from TFTP server.. | 17:01 |
| NobodyCam | ko__: dnsmasq running? | 17:02 |
| dkehn | devananda: the ironic sprint, 3/3, I saw the email to book, but to where? | 17:02 |
| NobodyCam | dkehn: I think seattle but not sure? | 17:03 |
| devananda | nope | 17:03 |
| devananda | lemme check before i say for sure | 17:03 |
| ko__ | NobodyCam: bm-vm is able to get IP address. so I can safly assume that its running.. | 17:04 |
| NobodyCam | humm... ko__ are you running neutron? | 17:04 |
| lucasagomes | NobodyCam, cheers | 17:04 |
| NobodyCam | that would give ip with out pxe booting | 17:04 |
| devananda | dkehn: which email are you referring to? | 17:05 |
| ko__ | I guess so.. I can see the logs in screen.. | 17:05 |
| dkehn | devananda: HP | 17:05 |
| NobodyCam | brb | 17:05 |
| *** blamar has joined #openstack-ironic | 17:06 | |
| dkehn | devananda: Thu 1/30/2014 3:36 AM | 17:06 |
| dkehn | devananda: maybe jumping the gun, read proposed dates | 17:06 |
| *** martyntaylor has quit IRC | 17:07 | |
| dkehn | devananda: don't waste anytime looking it up will bug SpamapS or lifeless | 17:07 |
| NobodyCam | bbt...brb | 17:07 |
| dkehn | was so hoping for Seattle | 17:07 |
| devananda | dkehn: location is almost definitely the bay area. looks like a few things aren't confirmed yet as no official announcement has gone out | 17:08 |
| dkehn | devananda: thx, at least its not Northern part of the world | 17:09 |
| lucasagomes | devananda, that's the mid-cycle meetup? | 17:09 |
| devananda | dkehn: i haven't seen anything yet indicate we should start booking | 17:09 |
| devananda | lucasagomes: yea | 17:09 |
| devananda | lucasagomes: lifeless is arranging the budget/space/etc. I'd really like you & other redhat folks to be there, and some of the mirantis folks | 17:10 |
| devananda | rloo - will you be around the bay area march 3 - 7? | 17:10 |
| devananda | romcheg, max_lobur - any chance you guys will be able to fly to california in early march for a code sprint? | 17:11 |
| *** matty_dubs is now known as matty_dubs|lunch | 17:12 | |
| rloo | devananda: I wasn't planning on it. I suppose I could. Umm, maybe not. I think I am single-parenting that week til the Tues. Are the dates confirmed? | 17:12 |
| *** Haomeng has quit IRC | 17:12 | |
| * rloo need to read emails | 17:13 | |
| devananda | rloo: dates are not yet confirmed | 17:13 |
| lucasagomes | devananda, I will talk to my manager about the budget etc, haven't heard anything yet. But yea the last mid-cycle was very useful | 17:13 |
| devananda | rloo: i thought you lived near SF ? | 17:13 |
| rloo | devananda. If Toronto Canada is near, then I do ;) | 17:13 |
| NobodyCam | lol | 17:14 |
| devananda | rloo: hah! um.... | 17:14 |
| romcheg | devananda: I don't know yet | 17:14 |
| romcheg | I do not have a US visa yet | 17:14 |
| devananda | romcheg: ah. then it's not likely | 17:14 |
| romcheg | I will ask my mgr when he's back from his vacation | 17:14 |
| NobodyCam | romcheg: may be tuff for you to get one now. | 17:15 |
| devananda | romcheg: know if anyone else @mirantis, working on ironic, has US visa? | 17:15 |
| rloo | has the US blocked visas for Ukraine? | 17:15 |
| romcheg | devananda: AFAIK no one | 17:15 |
| romcheg | rloo: Only for certain officials | 17:15 |
| max_lobur | devananda, Idk too | 17:16 |
| max_lobur | btw my visa interview will happen next Monday :) wish me a luck :) | 17:16 |
| devananda | brb | 17:16 |
| rloo | good luck max_lobur! | 17:17 |
| romcheg | Getting a visa is not a big deal. | 17:17 |
| max_lobur | and point two - even if I have vise I'm not sure if they approve me to fly | 17:17 |
| max_lobur | rloo, they blocked only for our government guys | 17:17 |
| romcheg | I can do it without Mirantis. However I need sponsorship to get to the summit | 17:17 |
| max_lobur | not for all | 17:17 |
| lucasagomes | max_lobur, g luck :) | 17:18 |
| rloo | max_lobur. Good to hear,just the bad guys are blocked ;) | 17:18 |
| max_lobur | lucasagomes, thx :) | 17:18 |
| max_lobur | rloo yep! :) | 17:18 |
| lucasagomes | visas r always a pain in the neck :/ | 17:20 |
| max_lobur | true | 17:20 |
| NobodyCam | post bbt walkies... bbiafm | 17:20 |
| *** ko__ has quit IRC | 17:23 | |
| romcheg | The reason I will never stop using Google is NobodyCam's bbts, brbs, bbiafms and other wthits :D | 17:24 |
| max_lobur | lol | 17:24 |
| *** Alexei_987 has quit IRC | 17:25 | |
| rloo | romcheg: ha ha. and I thought it was just me! ;) | 17:25 |
| *** jcooley_ has joined #openstack-ironic | 17:25 | |
| devananda | back | 17:26 |
| max_lobur | I was always wondering what a hell is bbt :D | 17:26 |
| devananda | bubby breakfast time -- probably wont find that in google search | 17:26 |
| devananda | :) | 17:27 |
| devananda | (i think) | 17:27 |
| max_lobur | ahh :) | 17:27 |
| max_lobur | so sweet :) | 17:28 |
| lucasagomes | hah | 17:28 |
| *** jcooley_ has quit IRC | 17:28 | |
| lucasagomes | urban dictionary says something complete different :P | 17:29 |
| max_lobur | lucasagomes, pls say you didn't know too :) | 17:29 |
| lucasagomes | max_lobur, I didn't hah | 17:29 |
| lucasagomes | I only know bbq = barbecue | 17:29 |
| lucasagomes | :P | 17:29 |
| max_lobur | :D | 17:29 |
| *** aignatov_ is now known as aignatov | 17:31 | |
| openstackgerrit | Max Lobur proposed a change to openstack/ironic: Allow concurrent image downloads in pxe logic https://review.openstack.org/63904 | 17:32 |
| lucasagomes | devananda, quick q | 17:32 |
| lucasagomes | when issuing nova boot, we have to specify the --image parameter | 17:32 |
| lucasagomes | should we use that instead of having to add pxe_image_source directly in the driver_info in ironic? | 17:33 |
| lucasagomes | by use that I mean, should the nova driver inject the pxe_image_source information in ironic | 17:33 |
| *** jcooley_ has joined #openstack-ironic | 17:34 | |
| NobodyCam | lol | 17:34 |
| NobodyCam | yep bubbie breakfast time | 17:34 |
| NobodyCam | bubbie are the great danes! | 17:34 |
| lucasagomes | NobodyCam, ^ same q | 17:35 |
| NobodyCam | :) | 17:35 |
| max_lobur | :) | 17:35 |
| NobodyCam | lucasagomes: I would yes to the use --image | 17:36 |
| lucasagomes | NobodyCam, cool :D | 17:36 |
| lucasagomes | thanks | 17:37 |
| NobodyCam | we're getting that from https://review.openstack.org/#/c/51328/13/nova/virt/ironic/ironic_driver_fields.py | 17:37 |
| *** coolsvap has joined #openstack-ironic | 17:38 | |
| NobodyCam | and driver/py lines 359 - 370 | 17:38 |
| devananda | lucasagomes: no | 17:38 |
| lucasagomes | NobodyCam, a-ha! right so it's already done, when I tried I added it directly in the ironic driver_info | 17:39 |
| devananda | NobodyCam: um, why do you say "yes" ? | 17:39 |
| lucasagomes | devananda, no? | 17:39 |
| NobodyCam | devananda: no to use --image | 17:39 |
| devananda | so the Nova driver doesn't know which deploy driver that Ironic node is configured with | 17:39 |
| devananda | Nova should be able to issue a deploy to nodes that use different drivers | 17:40 |
| NobodyCam | devananda: yes it dose | 17:40 |
| devananda | Nova knows the user image -- it doesnt know the underlying mechanism by which Ironic will deploy that image | 17:40 |
| NobodyCam | line 359 of driver.py | 17:40 |
| devananda | NobodyCam: ok - it can know it. but then we need to change Nova code for every ironic vendor driver? | 17:40 |
| NobodyCam | I think we should! | 17:41 |
| devananda | hmm | 17:41 |
| lucasagomes | right, cause I noticed one thing | 17:41 |
| NobodyCam | --image is part of the normal nova command | 17:41 |
| devananda | --image ==> the user image. NOT the deploy image | 17:42 |
| devananda | pxe_deploy_kernel ==> the deploy image. | 17:42 |
| lucasagomes | after I deleted the instance, nova removed all (actually almost all) the pxe_ attributes from my ironic node | 17:42 |
| devananda | i'm goign to side track for a moment | 17:42 |
| NobodyCam | ack | 17:42 |
| lucasagomes | right | 17:43 |
| devananda | when we add the various utility functionality to ironic that we've been discussing (eg, firmware update, wipe disk, etc) | 17:43 |
| devananda | we'll really be adding that t othe PXE driver | 17:43 |
| devananda | and we'll need some persistent "use this ramdisk & kernel for these operations" information | 17:43 |
| devananda | in the node.driver_info['pxe_*'] area | 17:43 |
| devananda | or in some CONF.pxe.$options arae | 17:44 |
| *** vkozhukalov has quit IRC | 17:44 | |
| devananda | using a CONF option for that is going to be harder to manage in a heterogeneous, long-running environment --- so sticking the utility ramdisk IDs in node.driver_info['pxe_*'] gives deployers more flexibility and better support for different hardware | 17:45 |
| devananda | and thus, IMO, is better | 17:45 |
| devananda | so | 17:45 |
| devananda | thoughts? | 17:45 |
| * lucasagomes thinking | 17:45 | |
| * NobodyCam thinking | 17:46 | |
| lucasagomes | right, because in the moment the deploy ramdisk, kernel id | 17:46 |
| lucasagomes | is inject by nova | 17:47 |
| lucasagomes | and should be part of the nova flavor | 17:47 |
| NobodyCam | devananda: vers say on the flavor like current deploy info... | 17:47 |
| NobodyCam | flavor could also have extra specs recovery_k&r | 17:47 |
| lucasagomes | so should we ignore the kernel/ramdisk id from flavor? (I mean ignore, but we won't even need to set it) | 17:47 |
| lucasagomes | NobodyCam, one problem about having it in the flavor | 17:47 |
| lucasagomes | is that our | 17:48 |
| lucasagomes | node-validate would fail | 17:48 |
| lucasagomes | cause the informations is not present in the driver_info before booting | 17:48 |
| NobodyCam | lucasagomes: nova could set all of that like it does now with deploy K&r | 17:48 |
| NobodyCam | before it calll validate for spawn | 17:49 |
| lucasagomes | NobodyCam, it could, but it only does after boot | 17:49 |
| lucasagomes | NobodyCam, you won't be able to let's say | 17:49 |
| lucasagomes | enroll the nodes, and issue a "ironic node-validate <uuid>" | 17:49 |
| lucasagomes | cause that would fail | 17:49 |
| lucasagomes | (it will complain about missing parameters in the pxe configuration) | 17:49 |
| *** ndipanov has quit IRC | 17:50 | |
| *** athomas has quit IRC | 17:50 | |
| *** matty_dubs|lunch is now known as matty_dubs | 17:50 | |
| lucasagomes | NobodyCam, s/after boot/after(during) deploy/g | 17:50 |
| NobodyCam | true, but the that means for a user to boot they need to have access to ironic interface | 17:51 |
| NobodyCam | and that the whole nova --image support is lost | 17:52 |
| lucasagomes | NobodyCam, well... they need to enroll the node anyway | 17:52 |
| lucasagomes | that would happen before | 17:52 |
| NobodyCam | no an admn can setup a pool of nodes for users to boot (under current scheme) | 17:52 |
| NobodyCam | and user can just issue nova boot | 17:52 |
| lucasagomes | right | 17:52 |
| lucasagomes | hmmm | 17:53 |
| *** jistr has quit IRC | 17:53 | |
| NobodyCam | what (if i under stand) you are saying is that the admin would have to set the image on the node | 17:53 |
| NobodyCam | and users would have no control of that | 17:53 |
| lucasagomes | NobodyCam, yea I forgot the tenants part | 17:54 |
| lucasagomes | but yea that's basically what i was saying | 17:54 |
| lucasagomes | not only image, but deploy kernel+ramdisk | 17:54 |
| lucasagomes | as well | 17:54 |
| lucasagomes | but I can see ur point now | 17:54 |
| lucasagomes | in have a non-admin tenant issuing nova boot | 17:54 |
| lucasagomes | to provision a baremetal machine | 17:54 |
| NobodyCam | i think that would break nova (or at least make it so users can not pick there own images | 17:55 |
| NobodyCam | ) | 17:55 |
| devananda | right, so there's another point | 17:55 |
| devananda | the deploy ramdisk runs outside of the tenant network, ya? | 17:55 |
| lucasagomes | yea indeed it will be a bit controversial in nova's pov | 17:55 |
| devananda | so the user shouldn't be able to specify what k&r gets used for that | 17:55 |
| NobodyCam | thats why I stuck with getting the image from nova the way i did | 17:55 |
| devananda | same for rescue, etc | 17:56 |
| devananda | end user (non-admin) shouldn't be able to influence that. and, granted, in most public clouds, users can't upload their own images | 17:56 |
| openstackgerrit | Max Lobur proposed a change to openstack/ironic: Allow concurrent image downloads in pxe logic https://review.openstack.org/63904 | 17:56 |
| devananda | i think it's more a question of, which service is authoritative regarding the deploy,rescue,etc mechanism for node X | 17:57 |
| devananda | is it Nova or Ironic? | 17:57 |
| NobodyCam | by having it on the flavor as it is now. users can pick --flavor and --image that match thier needs | 17:57 |
| devananda | which ever service it is needs to know things like machine architecture, what driver to use, etc | 17:57 |
| NobodyCam | I see deploy / rescue / etc as extra specs of the nova side | 17:57 |
| devananda | NobodyCam: what if user picks a --flavor with x86 deploy ramdisk and --image that is ARM ? | 17:58 |
| NobodyCam | they get a node stuck in deploying | 17:58 |
| devananda | heh | 17:58 |
| NobodyCam | :-p | 17:58 |
| lucasagomes | lol | 17:58 |
| lucasagomes | true | 17:58 |
| devananda | i'll call them utility ramdisks -- whether deploy, rescue, firmware, etc | 17:59 |
| * NobodyCam need to refill coffe for this | 17:59 | |
| devananda | so if utility ramdisks are a property of the ironic node.driver_info['pxe*'], | 17:59 |
| devananda | and users requests an image of some $ARCH | 17:59 |
| devananda | nova-scheduler will always pick a node of the same $ARCH (if one is available) and that node will use the appropriate utility ramdisks, regardless of --flavor | 18:00 |
| openstackgerrit | Max Lobur proposed a change to openstack/ironic: Add futures to requirements https://review.openstack.org/70177 | 18:00 |
| *** jcooley_ has quit IRC | 18:00 | |
| devananda | (arguign the other side) we could probably add functionality to Nova to check that flavor and image are the same $ARCH | 18:00 |
| *** jcooley_ has joined #openstack-ironic | 18:01 | |
| devananda | ok - here's another perspective | 18:01 |
| devananda | what if the admin wants to get a recovery console on a node that doesn't have a nova instance on it | 18:02 |
| lucasagomes | yea about the dpeloy k&r I think it's fine to have it in the driver_info or flavor | 18:02 |
| lucasagomes | but the image, the image should come from the nova boot | 18:02 |
| devananda | lucasagomes: the user image -- yes, that must come from the nova boot request | 18:02 |
| max_lobur | https://review.openstack.org/#/c/70177/1 guys, could you please review/approve Add futures to requirements ? | 18:02 |
| lucasagomes | yea user image | 18:02 |
| devananda | lucasagomes: oh! I thought you were asking abotu the deploy image this whole time | 18:03 |
| lucasagomes | devananda, no that --image is the user image | 18:03 |
| devananda | lucasagomes: right | 18:03 |
| lucasagomes | devananda, but it's an interesting point about whether we should have the deploy ramdisk/kernel in the flavor or not | 18:03 |
| lucasagomes | alias image="user image" | 18:03 |
| *** tatyana has quit IRC | 18:04 | |
| lucasagomes | devananda, yes, I see more benefits in have the deploy ramdisk/kernel in the driver_info directly instead of flavor | 18:05 |
| lucasagomes | because in the future we would want to add more capabilities to that ramdisk | 18:05 |
| lucasagomes | such as hw burning, etc | 18:05 |
| lucasagomes | so it can't be only set when a nova boot is issued | 18:06 |
| devananda | right - actions which exist outside of Nova's purvue | 18:06 |
| lucasagomes | we would = we will | 18:06 |
| lucasagomes | exactly | 18:06 |
| devananda | max_lobur: waiting for jenkins, then will approve | 18:07 |
| max_lobur | devananda, thx! | 18:07 |
| max_lobur | I'm going home, will join irc later from there | 18:07 |
| *** max_lobur is now known as max_lobur_afk | 18:08 | |
| NobodyCam | night max_lobur :) | 18:08 |
| lucasagomes | max_lobur, safe drive back home | 18:08 |
| max_lobur_afk | thank :) | 18:08 |
| lucasagomes | NobodyCam, if you think about the deploy r&k to be used for more operations apart from deploy (and perhaps we might even want to change this name) | 18:08 |
| lucasagomes | I think it makes sense to live in the driver_info instead of the flavor | 18:09 |
| lucasagomes | and be enrolled by the admin in the moment he's enrolling the node in ironic | 18:09 |
| NobodyCam | ture | 18:09 |
| devananda | guys, if you could take a moment to re-review https://review.openstack.org/#/c/66995/3 that'd be great | 18:10 |
| devananda | it's got a dependent patch already approved, too | 18:10 |
| lucasagomes | will do | 18:10 |
| lucasagomes | this one as well: https://review.openstack.org/#/c/69004/ | 18:11 |
| * NobodyCam looking | 18:11 | |
| devananda | NobodyCam: and https://review.openstack.org/#/c/68182/ | 18:11 |
| lucasagomes | without it the tests are broken in my machine | 18:11 |
| devananda | lucasagomes: really? unit tests? | 18:12 |
| lucasagomes | devananda, yup | 18:12 |
| devananda | hmm | 18:12 |
| lucasagomes | devananda, it's has to do with the version of the sqlite | 18:12 |
| * devananda runs unit tests | 18:13 | |
| lucasagomes | the newer sqlite has new exceptions, so the code in oslo which whould wrap the exceptions and issue a DuplicatedEntry exception can't do that anymore | 18:13 |
| lucasagomes | it raises DbError and the test fails | 18:13 |
| devananda | ahhh | 18:13 |
| devananda | yuck | 18:14 |
| NobodyCam | devananda: why the _ in line 75 of doc/source/dev/architecture.rst | 18:14 |
| lucasagomes | devananda, https://review.openstack.org/#/c/61405 | 18:14 |
| lucasagomes | oslo patch fixing it | 18:14 |
| devananda | lucasagomes: ah, i see | 18:15 |
| devananda | lucasagomes: fwiw, passes locally w/o that patch | 18:15 |
| lucasagomes | devananda, it's because ur sqlite is old | 18:15 |
| lucasagomes | [lucasagomes@lucasagomes ironic]$ sudo rpm -q sqlite | 18:16 |
| lucasagomes | sqlite-3.8.2-1.fc20.x86_64 | 18:16 |
| devananda | 3.7.17 | 18:16 |
| devananda | ah | 18:16 |
| lucasagomes | after my upgrade to fedora 20 it started failing | 18:16 |
| devananda | "SQLite has new error messages since 3.7.16. | 18:16 |
| lucasagomes | works on f19 tho | 18:16 |
| devananda | is what the oslo patch says | 18:16 |
| devananda | and i'm newer than that :p | 18:16 |
| devananda | k | 18:16 |
| lucasagomes | yea I think that's wrong | 18:16 |
| lucasagomes | but that patch solves the problems in my localmachine | 18:16 |
| devananda | cool | 18:17 |
| lucasagomes | I tested it | 18:17 |
| devananda | d'oh. needs to be rebased | 18:17 |
| devananda | breaks _load_one_plugin | 18:17 |
| lucasagomes | devananda, ah, lemme rebase | 18:17 |
| lucasagomes | urgh will have to rebase the patch it's dependent on as well | 18:18 |
| devananda | i have - want me to push the rebase? | 18:18 |
| devananda | they both pass locally now | 18:18 |
| lucasagomes | devananda, if u have it handy there feel free to do | 18:19 |
| lucasagomes | if not I can rebase it here quickly | 18:19 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Add lockfile>=0.8 to requirements.txt https://review.openstack.org/69358 | 18:19 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Sync db.sqlalchemy code from Oslo https://review.openstack.org/69004 | 18:19 |
| lucasagomes | cheers | 18:19 |
| lucasagomes | fyi the errors when I run the unittests on master: http://paste.openstack.org/show/62194/ | 18:20 |
| devananda | rebasing one more dep | 18:21 |
| devananda | lucasagomes: cheers, thanks. taht's what i'd expect :) | 18:22 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Removes use of timeutils.set_time_override https://review.openstack.org/67432 | 18:22 |
| NobodyCam | lucasagomes: not to sidetrack you, bit did you end having to do major changes to get the nova driver working? or just the execption | 18:24 |
| NobodyCam | s/bit/but/ | 18:25 |
| devananda | NobodyCam: https://review.openstack.org/#/c/67877/ also needs another review | 18:26 |
| *** rloo has quit IRC | 18:27 | |
| devananda | as a reminder to reviewers, http://status.openstack.org/reviews/ gives us some hints about prioritizing our reviewing | 18:27 |
| *** harlowja_away is now known as harlowja | 18:28 | |
| * NobodyCam *(blushes)* | 18:28 | |
| lucasagomes | NobodyCam, I've exceptions and some changes on the destroy() | 18:34 |
| NobodyCam | lucasagomes: got a quick second for: https://review.openstack.org/#/c/66995 | 18:37 |
| lucasagomes | NobodyCam, doing it now | 18:37 |
| lucasagomes | building the doc | 18:37 |
| NobodyCam | :-p | 18:38 |
| lucasagomes | cooleo +a'd | 18:40 |
| NobodyCam | :) | 18:40 |
| NobodyCam | bagel make'n time | 18:42 |
| lucasagomes | much better now to have a message routing explanation in the docs | 18:42 |
| * lucasagomes always got confused when tried to explain it to others | 18:42 | |
| devananda | dkehn: reviewing your neutron patch now | 18:50 |
| devananda | dkehn: the code looks good. i'm leaving a bunch of suggestions on the unit tests | 18:50 |
| devananda | dkehn: almost there! | 18:50 |
| openstackgerrit | A change was merged to openstack/ironic: Remove deploy kernel and ramdisk global config https://review.openstack.org/70140 | 18:50 |
| dkehn | devananda: thanks | 18:50 |
| openstackgerrit | A change was merged to openstack/ironic: Bump version of sphinxcontrib-pecanwsme https://review.openstack.org/66078 | 18:51 |
| openstackgerrit | A change was merged to openstack/ironic: Removed unused exceptions https://review.openstack.org/68852 | 18:53 |
| *** jcooley_ has quit IRC | 18:56 | |
| devananda | dkehn: oh, i take taht back -- one suggestion on the class too. but it's small :) | 18:57 |
| *** jcooley_ has joined #openstack-ironic | 18:59 | |
| NobodyCam | devananda: want to over ride my -1 on https://review.openstack.org/#/c/60528 | 18:59 |
| openstackgerrit | A change was merged to openstack/ironic: Run mkfs as root https://review.openstack.org/68182 | 18:59 |
| NobodyCam | the only one I think worth the fix the doc strion issue | 18:59 |
| NobodyCam | string even | 18:59 |
| devananda | dkehn: posted | 19:00 |
| devananda | lucasagomes: have you seen that ^ client patch? | 19:00 |
| devananda | lucasagomes: looks like you +2'd the dependent patch | 19:01 |
| NobodyCam | ya I will too | 19:01 |
| lucasagomes | devananda, 60528? I think I reviewed that before | 19:01 |
| lucasagomes | lemme re-review | 19:01 |
| openstackgerrit | A change was merged to openstack/ironic: Delete the iscsi target https://review.openstack.org/67877 | 19:03 |
| lucasagomes | devananda, NobodyCam so... I'm always a bit confused with sync patches and i18n problems | 19:06 |
| lucasagomes | because it's kinda out of our control | 19:06 |
| NobodyCam | ya again the only one i think needs fixing is hte doc string | 19:07 |
| NobodyCam | :-p | 19:07 |
| lucasagomes | yea | 19:07 |
| NobodyCam | doc the wrong error being raised | 19:07 |
| lucasagomes | yea, that needs to be fixed in oslo first tho | 19:08 |
| *** hstimer has joined #openstack-ironic | 19:08 | |
| NobodyCam | so do we land this as is? and wait / patch oslo to fix it? | 19:08 |
| devananda | NobodyCam: wrong error? or just not i18n? | 19:08 |
| lucasagomes | devananda, there's a docstring error | 19:09 |
| devananda | lucasagomes: we don't have to merge things from oslo -- but it's nice if we do | 19:09 |
| *** hstimer has quit IRC | 19:09 | |
| lucasagomes | it says raises the wrong exception | 19:09 |
| *** marty_ has joined #openstack-ironic | 19:09 | |
| devananda | oh | 19:09 |
| *** marty_ is now known as Guest10145 | 19:09 | |
| NobodyCam | see comment on https://review.openstack.org/#/c/60528/14/ironicclient/openstack/common/apiclient/auth.py | 19:09 |
| devananda | indeed.... | 19:10 |
| devananda | looks like that is not fixed upsteram yet either | 19:11 |
| devananda | https://github.com/openstack/oslo-incubator/blob/master/openstack/common/apiclient/auth.py#L79 | 19:11 |
| lucasagomes | yup | 19:11 |
| lucasagomes | apart from that error I'm fine with the rest | 19:11 |
| NobodyCam | ya I can live with i18n stuff | 19:11 |
| devananda | i'll file bug and fix it | 19:13 |
| lucasagomes | devananda, right, so can I go and +2/a it? | 19:13 |
| openstackgerrit | A change was merged to openstack/ironic: Minor documentation update https://review.openstack.org/66995 | 19:14 |
| openstackgerrit | A change was merged to openstack/ironic: Update method doc strings in pxe.py https://review.openstack.org/66996 | 19:14 |
| devananda | sure | 19:14 |
| lucasagomes | ok | 19:14 |
| NobodyCam | :) | 19:14 |
| *** romcheg has quit IRC | 19:17 | |
| NobodyCam | nice to the gate working again :) | 19:19 |
| *** vkozhukalov has joined #openstack-ironic | 19:19 | |
| devananda | indeed! | 19:20 |
| devananda | dkehn: pls let me know if you have any questions about my feedback on 66071 | 19:20 |
| lucasagomes | NobodyCam, r u working on another review for the driver? | 19:22 |
| lucasagomes | I think I will have to make some other changes around the code to make it retry the request if a 409 is returned | 19:23 |
| devananda | oh, how do you guys want to start splitting up the nova driver patch? | 19:23 |
| devananda | it's already too big to review... and we need unit tests, too | 19:23 |
| lucasagomes | devananda, +2 | 19:23 |
| NobodyCam | lucasagomes: I have been holding off waiting to see what you changed | 19:23 |
| lucasagomes | NobodyCam, ouch hmm | 19:23 |
| NobodyCam | I can move to some tests | 19:24 |
| lucasagomes | NobodyCam, I think it would be easier to you to go ahead and submit it | 19:24 |
| devananda | all 3 of us working on N different patch sets in a sort of "lets play twister" mode is going to be terrible | 19:24 |
| lucasagomes | cause I don't think I will finish it today to submit a new patch-set | 19:24 |
| devananda | until they land | 19:24 |
| NobodyCam | lol yes | 19:24 |
| lucasagomes | I wanna test it first | 19:24 |
| lucasagomes | yea :( it's hard to have a patch-set with 3 people working on it | 19:24 |
| devananda | yep | 19:24 |
| devananda | even 2 | 19:25 |
| NobodyCam | lucasagomes: do you want to paste me the changes and i'll put them up? | 19:25 |
| lucasagomes | NobodyCam, I can do that, but I'm kinda in the middle of it | 19:26 |
| NobodyCam | ahh | 19:26 |
| lucasagomes | Maybe it's better to me to fix the exception and the destroy() first | 19:26 |
| lucasagomes | and then do the retry in another patch-set | 19:26 |
| * lucasagomes saves a copy of the file and start removing some code | 19:26 | |
| lucasagomes | NobodyCam, or u can go ahead ur changes and I can sync/rebase mines on top it | 19:27 |
| lucasagomes | devananda, do you have any idea in which parts we could start breaking that code? | 19:27 |
| NobodyCam | sure.. I'll start looking tests and volume_driver stuff | 19:28 |
| devananda | break on common function boundaaries | 19:28 |
| devananda | ? | 19:28 |
| openstackgerrit | A change was merged to openstack/python-ironicclient: Sync apiclient and strutils from Oslo https://review.openstack.org/60528 | 19:28 |
| openstackgerrit | A change was merged to openstack/python-ironicclient: Reuse Resource from oslo https://review.openstack.org/62921 | 19:28 |
| lucasagomes | break on different patches | 19:28 |
| lucasagomes | split* | 19:28 |
| devananda | i mean, split the code into patches where each patch adds N functions and corresponding unit tests | 19:29 |
| devananda | that should reduce merge conflicts a bit | 19:29 |
| devananda | and when al the patches are assembled (the branch HEAD) we have a working driver | 19:29 |
| lucasagomes | I see, yea that works | 19:30 |
| devananda | the rebasign stuff can be tricky, see https://wiki.openstack.org/wiki/GerritWorkflow#Add_dependency | 19:31 |
| lucasagomes | I was looking at breaking it in some atomic parts like: e.g one patch for the host scheduler part, another for volume | 19:31 |
| lucasagomes | etc | 19:31 |
| *** rloo has joined #openstack-ironic | 19:31 | |
| devananda | ah | 19:31 |
| devananda | so split by file in this case may not work so well | 19:31 |
| devananda | but maybe there are some files taht it'll work well for | 19:32 |
| devananda | we still need to split the driver, though. > 1k lines .... | 19:32 |
| lucasagomes | I see | 19:32 |
| lucasagomes | maybe it would also be easy to work on github first? | 19:33 |
| NobodyCam | lucasagomes: if you are workinng on driver I will not change it untill your done! | 19:33 |
| lucasagomes | like submitting pull requests to one of our repos | 19:33 |
| lucasagomes | so one of us will take care of merge the different changes | 19:34 |
| lucasagomes | NobodyCam, devananda ^ | 19:34 |
| NobodyCam | we could split out scheduler in to it's own patch | 19:34 |
| lucasagomes | NobodyCam, ok, I will have a patch-set done for tomorrow (me wants to give it a try first) | 19:35 |
| devananda | russellb: question about the nova-ironic driver and how we're going to split it up for reviews. got a minute? | 19:35 |
| NobodyCam | lucasagomes: ack | 19:35 |
| russellb | sure | 19:35 |
| NobodyCam | just to driver.py | 19:35 |
| NobodyCam | lucasagomes: ^^ | 19:35 |
| devananda | russellb: hi! so two aspects to this | 19:35 |
| lucasagomes | NobodyCam, yup, driver.py | 19:36 |
| devananda | 1 - multiple people are making changes to the driver as we're testing it, fleshing out some functionality, etc | 19:36 |
| devananda | 2 - it's already too big to review sanely | 19:36 |
| NobodyCam | lucasagomes: I'll work else where today :) | 19:36 |
| lucasagomes | NobodyCam, ack :) | 19:36 |
| devananda | russellb: if we had a complete-and-finished patchset, sure, it could be split into multiple smaller ones, with a little work | 19:36 |
| devananda | russellb: but maintaining a patch-chain while multiple people refine each patchset is, well, really messy... we've all tried that before, though | 19:37 |
| russellb | devananda: certainly split out any mods needed outside of the new driver | 19:37 |
| russellb | for just an isolated driver and its tests ... i dunno, just 1 patch seems fine | 19:37 |
| *** rloo has quit IRC | 19:37 | |
| devananda | russellb: even if driver is ~2k lines and unit tests are similar? | 19:37 |
| devananda | it's a bit hard to review that much, IMO anyway | 19:38 |
| russellb | it's certainly not ideal | 19:38 |
| *** rloo has joined #openstack-ironic | 19:38 | |
| russellb | how is it coming along | 19:38 |
| devananda | if you guys are OK with that, that's fine with me | 19:38 |
| devananda | good! lucasagomes and crew have it work on real hardware now, apparently | 19:39 |
| russellb | honestly, i'd be relying pretty hard on you to tell me it's sane | 19:39 |
| russellb | :) | 19:39 |
| NobodyCam | russellb: first deploy success with nova + ironic this morning :) | 19:39 |
| russellb | nice | 19:39 |
| devananda | russellb: if you're ok with landing partial stuff and letting us iteratively improve it, we could probably do that now | 19:40 |
| devananda | russellb: or wait another few weeks while we improve it out-of-trunk. up to you guys | 19:40 |
| russellb | would rather have it worked out of trunk until you think it's ready for others to start using | 19:40 |
| devananda | ack | 19:41 |
| russellb | you should be able to work faster that way anyway | 19:41 |
| devananda | there's already a solid base | 19:41 |
| devananda | not bug free and probably missing some bits | 19:41 |
| devananda | but yea, we can work faster out of trunk | 19:42 |
| devananda | russellb: thanks! | 19:42 |
| NobodyCam | devananda: russellb: what about splitting nova/scheduler/ironic_host_manager.py it to its own patch? | 19:43 |
| russellb | np | 19:43 |
| NobodyCam | can that land with out the (working) driver | 19:43 |
| russellb | NobodyCam: yeah, anything outside of the driver is a good candidate to split out ... though that may make sense to submit with the driver itself | 19:44 |
| russellb | to make it clear how it's used | 19:44 |
| russellb | i don't like merging unused code | 19:44 |
| russellb | without something using it right behind it | 19:44 |
| devananda | russellb: right. so there are several dependent-but-external bits, such as a scheduler filter, a volume driver, etc | 19:44 |
| NobodyCam | russellb: ack! (i agree) | 19:44 |
| devananda | russellb: by "submit with" - i assume you mean as a dependent patchset. we can do that easily enough | 19:45 |
| russellb | yes | 19:45 |
| devananda | NobodyCam: do you want to handle the patch splitting? | 19:46 |
| NobodyCam | I can pull sch stuff now and put up as wip | 19:47 |
| NobodyCam | vol driver I a=really need to take a look at and see what we are really useing there | 19:47 |
| NobodyCam | s/a=// | 19:47 |
| NobodyCam | using even | 19:48 |
| NobodyCam | lol | 19:48 |
| lucasagomes | :) | 19:49 |
| NobodyCam | are there currently no tests for baremetal_host_manager.py?? | 19:51 |
| devananda | :( | 19:52 |
| devananda | lucasagomes: another one we should land soon: https://review.openstack.org/#/c/63937/7 | 19:52 |
| NobodyCam | :-p | 19:52 |
| lucasagomes | devananda, oh haven't reviewed that yet | 19:53 |
| lucasagomes | going to add here on my todo list | 19:53 |
| NobodyCam | quick afk then I'll stat in on ironic scheduler patch and tests | 19:54 |
| devananda | max_lobur_afk: 69135 is looking much better than the first shot at this, but i'm going to need more time to dig into it | 19:57 |
| lucasagomes | right guys I'm done for today | 19:59 |
| lucasagomes | NobodyCam, devananda g'night | 19:59 |
| dkehn | devananda: looking over the review, are we sure about cacheing the neutron client, see https://review.openstack.org/#/c/66071/8/ironic/common/neutron.py , where it was taken out | 20:03 |
| NobodyCam | night lucasagomes | 20:03 |
| dkehn | devananda: line 92 | 20:03 |
| *** aignatov is now known as aignatov_ | 20:04 | |
| devananda | dkehn: that was only for admin=True, and used local.strongstore | 20:05 |
| devananda | dkehn: what i'm suggesting is that the PXE driver may instantiate a NeutronAPI object and re-use that object | 20:06 |
| dkehn | devananda: totally understand, it makes sense to me | 20:06 |
| devananda | :) | 20:06 |
| devananda | what's the cost, in network round=trips, of calling get_client()? | 20:07 |
| dkehn | devananda: once the tokens and what not there, about the same as anything else, pretty much doing an db updates | 20:08 |
| dkehn | devananda: or insert | 20:08 |
| devananda | erm... you maen, "not doing any .." ? | 20:08 |
| *** mdurnosvistov has quit IRC | 20:08 | |
| dkehn | erm um correct | 20:08 |
| devananda | dkehn: does clientv20.Client() make any network calls? or just instantiate an object locally? | 20:09 |
| *** vkozhukalov has quit IRC | 20:10 | |
| NobodyCam | woo hoo https://review.openstack.org/#/c/66461/ merged | 20:10 |
| *** aignatov_ is now known as aignatov | 20:10 | |
| devananda | NobodyCam: yep! :) | 20:10 |
| devananda | NobodyCam: time to update our wiki instructions too, hehe | 20:10 |
| dkehn | devananda: well with keystone to get propoer info is the only cost, the get_client, once it has token does not talk to neutron only the update does | 20:10 |
| NobodyCam | ya :) | 20:10 |
| devananda | ok, so if there's no auth_token in the context, then 2 network calls | 20:11 |
| *** mdurnosvistov has joined #openstack-ironic | 20:11 | |
| devananda | keystone.get_service_url('neutron') and then getting the token | 20:11 |
| dkehn | pretty much | 20:12 |
| devananda | *1 or 2, depending on config | 20:12 |
| devananda | cool | 20:12 |
| dkehn | devananda: I'm fine with the change, actually I like it better, that was my point | 20:12 |
| devananda | so it makes sense to cache that object it we don't already have a token -- otherwise it'll get re-generated _every_ time | 20:12 |
| devananda | dkehn: oh! cool :) | 20:12 |
| dkehn | didn't like taking it out in the first place | 20:12 |
| devananda | yea, though i think this way is better than using local.strongstore. but i can see why nova did the other way | 20:14 |
| *** lucasagomes has quit IRC | 20:14 | |
| devananda | dkehn: just caght myself -- s/assertCalledOnceWith/assert_called_once_with/ in all my comments please | 20:22 |
| devananda | dkehn: http://www.voidspace.org.uk/python/mock/mock.html#mock.Mock.assert_called_once_with | 20:22 |
| dkehn | devananda: will look at it, torn away on something else | 20:23 |
| devananda | k | 20:23 |
| dkehn | at the moment | 20:23 |
| devananda | NobodyCam: another one for your queue :) https://review.openstack.org/#/c/66990/ | 20:26 |
| * NobodyCam looks | 20:27 | |
| *** jcooley_ has quit IRC | 20:31 | |
| *** jcooley_ has joined #openstack-ironic | 20:32 | |
| NobodyCam | devananda: it and dep +a'd (with out pulling down and testing) | 20:34 |
| NobodyCam | did we mode ssh execute out of ironic in to process utils some time ago? | 20:35 |
| NobodyCam | just a lol point | 20:35 |
| *** max_lobur has joined #openstack-ironic | 20:38 | |
| NobodyCam | ahh I was thinking of this one:https://review.openstack.org/#/c/61709/ | 20:46 |
| *** aignatov is now known as aignatov_ | 20:58 | |
| devananda | NobodyCam: processutils is in oslo now | 21:04 |
| NobodyCam | ya | 21:04 |
| NobodyCam | the _ssh_execute (https://review.openstack.org/#/c/66990/6/ironic/drivers/modules/ssh.py) made me think od the trycmd we moved out | 21:06 |
| NobodyCam | https://review.openstack.org/#/c/61709/4/ironic/common/utils.py | 21:07 |
| NobodyCam | just lol on my part | 21:08 |
| *** coolsvap is now known as coolsvap_away | 21:10 | |
| openstackgerrit | A change was merged to openstack/ironic: mock's return value for processutils.ssh_execute https://review.openstack.org/69479 | 21:11 |
| *** mdurnosvistov has quit IRC | 21:18 | |
| openstackgerrit | A change was merged to openstack/ironic: SSHPower driver raises IronicExceptions https://review.openstack.org/66990 | 21:19 |
| openstackgerrit | Jarrod Johnson proposed a change to stackforge/pyghmi: Detect broken sessions https://review.openstack.org/70219 | 21:24 |
| *** epim has joined #openstack-ironic | 21:26 | |
| NobodyCam | brb | 21:31 |
| *** jcooley_ has quit IRC | 21:32 | |
| *** jcooley_ has joined #openstack-ironic | 21:32 | |
| openstackgerrit | A change was merged to stackforge/pyghmi: Detect broken sessions https://review.openstack.org/70219 | 21:34 |
| devananda | lifeless: in your experience with APIs, should a multi-part PATCH request be accepted if the net result is valid, even if one or more individual parts are invalid? | 21:58 |
| *** jcooley_ has quit IRC | 21:58 | |
| devananda | lifeless: eg, if the PATCH contains 3 operations, and by itself, op 2 would be rejected, but when applied 1-2-3, the net is still allowed -- should they be rejected or accepted? | 21:59 |
| *** mdurnosvistov has joined #openstack-ironic | 22:03 | |
| *** jdob has quit IRC | 22:04 | |
| NobodyCam | gah quick walkies brb | 22:04 |
| max_lobur | devananda: I guess you're about breaking resource locks | 22:10 |
| devananda | max_lobur: yes | 22:10 |
| devananda | i'm reading http://tools.ietf.org/html/draft-ietf-appsawg-json-patch-10 | 22:10 |
| devananda | ugh... | 22:10 |
| devananda | reading IETF drafts before lunch .... | 22:10 |
| devananda | er, bad paste. this one: http://tools.ietf.org/html/rfc6902 | 22:11 |
| max_lobur | I'll take a look now | 22:12 |
| devananda | max_lobur: look at top of page 3 | 22:12 |
| max_lobur | in general I can't imagine if someone will replace the lock with something, and then replace back | 22:12 |
| devananda | Evaluation continues until all operations are successfully applied or until an error condition is encountered. | 22:12 |
| max_lobur | in the same patch | 22:12 |
| max_lobur | I see | 22:13 |
| *** mrda_away is now known as mrda | 22:16 | |
| *** epim has quit IRC | 22:17 | |
| mrda | morning all | 22:17 |
| devananda | max_lobur: comment posted to review | 22:18 |
| devananda | mornin, mrda | 22:18 |
| devananda | NobodyCam: another one for your queue! https://review.openstack.org/#/c/55549/ | 22:18 |
| devananda | NobodyCam: if you are happy with it, I'm ok reversign my -1 and landing it, then tossing up the doc string quickly -- jus tso we can land this | 22:19 |
| * mrda is reading the long scrollback | 22:19 | |
| * NobodyCam looks | 22:19 | |
| max_lobur | devananda: ack | 22:19 |
| NobodyCam | morning mrda | 22:20 |
| NobodyCam | devananda: reading your's and max_loburs comments on that patch | 22:31 |
| NobodyCam | skimming rfc to :-p | 22:32 |
| max_lobur | I also found this | 22:34 |
| max_lobur | If a normative requirement is violated by a JSON Patch document, or | 22:34 |
| max_lobur | if an operation is not successful, evaluation of the JSON Patch | 22:34 |
| max_lobur | document SHOULD terminate and application of the entire patch | 22:34 |
| max_lobur | document SHALL NOT be deemed successful. | 22:34 |
| max_lobur | http://tools.ietf.org/html/rfc6902#section-5 | 22:34 |
| devananda | max_lobur: yes - a "normative rquiremetn" | 22:35 |
| devananda | gah, cant type | 22:35 |
| devananda | max_lobur: AIUI, normative requirement in this context means that the JSON PATCH document adheres to the syntax in that RFC | 22:36 |
| devananda | max_lobur: not that our server knows what to dow ith the ontents | 22:36 |
| devananda | max_lobur: "or if an operation is not successful" | 22:37 |
| devananda | in this case, we have multi-operation PATCH document | 22:38 |
| devananda | but today, we are not evaluating each operation individually at all | 22:38 |
| devananda | and we really can't | 22:38 |
| max_lobur | devananda: yea I was going to say that | 22:38 |
| max_lobur | if we will evaluate each one individually | 22:38 |
| max_lobur | the quote above will mean that we should reject the whole patch if at least one is invalid | 22:39 |
| devananda | we are applying all the operations to the node's JSON representation, then passing the new node object to the conductor to apply the changes | 22:39 |
| max_lobur | but since we applying a net weight | 22:39 |
| devananda | if we tried to apply each op individually, we couldn't (today, at least) roll back the changes | 22:39 |
| max_lobur | yea, I understand | 22:40 |
| devananda | and things would be really slow if we tried to | 22:40 |
| max_lobur | yes | 22:40 |
| NobodyCam | devananda: would this be a valid example.. {power_off, remove instance_uuid} where insance uuid can be removed unless the node is first powered off | 22:40 |
| devananda | eg, get a PATCH with 10 {op: add, path: /driver_info/xxxx, ....} requests | 22:40 |
| devananda | NobodyCam: there's a separate API for power control | 22:41 |
| devananda | NobodyCam: and power is controlled with PUT, not PATCH, IIRC | 22:41 |
| * NobodyCam notes poor examle | 22:41 | |
| devananda | eg, [{op: replace, path: driver_name}, {op: add, path: driver_info/foo}] | 22:42 |
| devananda | we're not calling validate just to PATCH the data, so that should work even if evaluated individually | 22:43 |
| devananda | NobodyCam: if power was controlled via PATCH, then yes, your example would make sense :) | 22:44 |
| devananda | ok, i really need to step away and make lunch .... | 22:45 |
| devananda | bbiah | 22:45 |
| NobodyCam | enjoy | 22:46 |
| NobodyCam | :) | 22:46 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: API validates driver name for both POST and PATCH https://review.openstack.org/68018 | 22:48 |
| NobodyCam | max_lobur: do you have a example for #1 (&@) on your last comment on https://review.openstack.org/#/c/55549/20/ironic/api/controllers/v1/node.py | 22:48 |
| NobodyCam | s/\@/\&/ | 22:48 |
| NobodyCam | gah | 22:49 |
| NobodyCam | s/\@/2/ | 22:49 |
| NobodyCam | this is throwing me "2nd and 4th are not supported by our service" | 22:49 |
| max_lobur | do you mean do I have an example of what we don't support? | 22:51 |
| max_lobur | for example patch with 2 ops | 22:53 |
| max_lobur | 1st is | 22:53 |
| max_lobur | {'op': 'replace', 'path': '/reservation', 'value': 'test'} | 22:53 |
| NobodyCam | an example of what would fail thrn be fixed by th e next op | 22:53 |
| max_lobur | 2nd is | 22:53 |
| max_lobur | {'op': 'remove', 'path': '/reservation'} | 22:54 |
| max_lobur | 1st is not allowed | 22:54 |
| max_lobur | but if we apply 1-2 | 22:54 |
| max_lobur | it won't be noticable | 22:54 |
| max_lobur | and net result will be acceptable by Ironic | 22:55 |
| max_lobur | From my point of view it doesn't have its use in the service, it just may confuse the users | 22:56 |
| max_lobur | also I don't think someone will try my example =) it does'n makes sense =) | 22:57 |
| max_lobur | 1. If the patch has some restricted action we should reject it | 22:58 |
| max_lobur | 2. If net weight of the patch is acceptable we should not - I agree | 22:58 |
| max_lobur | but if 1 and 2 together - this means the patch has two opposite actions | 22:59 |
| max_lobur | or 2 mutually exclusive actions | 22:59 |
| *** thedodd has joined #openstack-ironic | 22:59 | |
| max_lobur | that may be replaced with one | 22:59 |
| max_lobur | therefore I don't think such patch makes sense. It's unlikely that someone will do so | 23:00 |
| max_lobur | but to support this behaviour we need either have a very complex patch-analyzing tool and clear code | 23:01 |
| max_lobur | or have a comlex code :) | 23:01 |
| NobodyCam | I can see the point | 23:02 |
| max_lobur | if we decide to not support it - we will have simple jsonpatch-analyzing tool + simple code :) | 23:02 |
| *** max_lobur has quit IRC | 23:07 | |
| *** max_lobur has joined #openstack-ironic | 23:08 | |
| NobodyCam | i see this in section 5: http://paste.openstack.org/show/XYrGtumWBLfv85XkfGRL/ | 23:10 |
| *** max_lobur has quit IRC | 23:10 | |
| *** max_lobur has joined #openstack-ironic | 23:10 | |
| max_lobur | {power_off, remove instance_uuid} the second op of this patch is not acceptable without 1st | 23:10 |
| *** mdurnosvistov has quit IRC | 23:10 | |
| NobodyCam | ya | 23:10 |
| NobodyCam | I tend to want to reject just based off we wont know what really failed and are dumping the (possiable) clean up on the admin | 23:12 |
| NobodyCam | but I'm thinking | 23:13 |
| *** max_lobur has quit IRC | 23:16 | |
| *** matty_dubs is now known as matty_dubs|gone | 23:18 | |
| NobodyCam | devananda: when you get back, how would this impact our ablity to "plan" for a conductor outage.. ie. as a admin i want to take a conductor down to add some ram, buy I can update any reseravations untill i take the node down? | 23:22 |
| NobodyCam | s/buy/but/ | 23:22 |
| NobodyCam | gah bad typing s/can/cann't/ | 23:23 |
| *** max_lobur_cell has joined #openstack-ironic | 23:23 | |
| max_lobur_cell | sorry. I have some connection problems :-\ | 23:24 |
| NobodyCam | hehe | 23:24 |
| NobodyCam | max you see my question? | 23:24 |
| NobodyCam | is something like that one of our concerns? | 23:24 |
| NobodyCam | our your :-p | 23:25 |
| max_lobur_cell | just wanted to say that having that json util wont stop us from analyzing the resulting operation. and most of the time it will be enough to check for particular operations | 23:25 |
| max_lobur_cell | sorry, I can't get the question:-) | 23:27 |
| NobodyCam | how would this impact our ablity to "plan" for a conductor outage.. ie. as a admin i wanted to take a conductor down to add some ram, but I cann't update any reseravations untill i take the node down? | 23:28 |
| max_lobur_cell | hmm | 23:30 |
| devananda | NobodyCam: reservation == taskManager lock | 23:30 |
| max_lobur_cell | well in this case we'll need analyze the resulting node | 23:30 |
| devananda | NobodyCam: so, a) we need to implement a gentle shutdown for ConductorManager | 23:30 |
| devananda | NobodyCam: and b) we should implement a way for a ConductorManager to stop sending its hearbeat | 23:32 |
| devananda | when CM drops out of hte hash ring, all nodes will be redistributed away from it | 23:32 |
| devananda | but any in-progrses operations will continue until they are finished | 23:32 |
| devananda | that is the period when an admin could use this API to "break" a lock -- but they shouldn't. because CM is still working | 23:33 |
| devananda | only in the case wher CM has crashed and left a dangling lock open, then this API is needed | 23:33 |
| devananda | NobodyCam: that make sense? | 23:34 |
| NobodyCam | yes... | 23:35 |
| openstackgerrit | dekehn proposed a change to openstack/ironic: Adds Neutron support to Ironic https://review.openstack.org/66071 | 23:36 |
| devananda | woot - our tempest API tests found a real bug in my code! | 23:37 |
| NobodyCam | nice :) | 23:37 |
| devananda | http://logs.openstack.org/18/68018/4/check/check-tempest-dsvm-ironic/094a973/console.html#_2014-01-30_23_23_02_345 | 23:37 |
| NobodyCam | nice | 23:38 |
| max_lobur_cell | :) | 23:39 |
| max_lobur_cell | ok guys, have a nice day :) seems my internet provider trying to say that it's time for rest:) | 23:41 |
| NobodyCam | hehehe ok max_lobur_cell have a good night | 23:42 |
| *** max_lobur_cell has quit IRC | 23:44 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
