| *** rloo has quit IRC | 00:01 | |
| *** jdandrea has quit IRC | 00:07 | |
| *** jdandrea has joined #openstack-ironic | 00:09 | |
| *** jdandrea has joined #openstack-ironic | 00:10 | |
| *** mrda has joined #openstack-ironic | 00:12 | |
| openstackgerrit | Merged openstack/virtualbmc master: Bump hacking to 3.0.0 https://review.opendev.org/715943 | 00:14 |
|---|---|---|
| *** k_mouza has joined #openstack-ironic | 00:21 | |
| kaifeng | TheJulia: if you are around, wrt the https://review.opendev.org/#/c/716845 do we want to backport this? I guess it needs a release note if we do. | 00:46 |
| patchbot | patch 716845 - ironic - Do not autoescape all Jinja2 templates - 2 patch sets | 00:46 |
| openstackgerrit | Merged openstack/ironic master: Add network_type to port local_link_connection https://review.opendev.org/715312 | 00:50 |
| *** rloo has joined #openstack-ironic | 01:01 | |
| openstackgerrit | Merged openstack/ironic master: Additional IP addresses to IPv6 stateful ports https://review.opendev.org/700002 | 01:02 |
| openstackgerrit | Merged openstack/ironic master: Doc - IPv6 Provisioning https://review.opendev.org/715313 | 01:02 |
| openstackgerrit | Verification of a change to openstack/ironic failed: Do not autoescape all Jinja2 templates https://review.opendev.org/716845 | 01:20 |
| *** k_mouza has quit IRC | 01:36 | |
| *** rloo has quit IRC | 01:37 | |
| *** ociuhandu has joined #openstack-ironic | 02:23 | |
| *** mkrai has joined #openstack-ironic | 02:49 | |
| *** stevebaker has quit IRC | 03:21 | |
| *** stevebaker has joined #openstack-ironic | 03:22 | |
| *** stevebaker has quit IRC | 03:26 | |
| *** stevebaker_ has joined #openstack-ironic | 03:26 | |
| *** stevebaker_ has quit IRC | 03:32 | |
| *** stevebaker has joined #openstack-ironic | 03:33 | |
| *** jdandrea has quit IRC | 03:35 | |
| *** k_mouza has joined #openstack-ironic | 03:37 | |
| *** k_mouza has quit IRC | 03:41 | |
| *** Goneri has quit IRC | 03:49 | |
| openstackgerrit | Merged openstack/ironic master: Allow specifying target devices for software RAID https://review.opendev.org/713423 | 04:08 |
| TheJulia | kaifeng: likely on both counts. I was figuring at worst a separate reno can be merged. ... of course it also failed CI :( | 04:17 |
| *** mkrai_ has joined #openstack-ironic | 04:26 | |
| *** mkrai has quit IRC | 04:26 | |
| *** ociuhandu has quit IRC | 04:45 | |
| *** ociuhandu has joined #openstack-ironic | 04:46 | |
| *** ociuhandu has quit IRC | 04:52 | |
| kaifeng | TheJulia: it's currently blocked by CI, so there is a chance to update with a reno, i guess separate reno also works, not an usually path though. | 05:35 |
| TheJulia | I'm kind of on a path of least resistance course at the moment | 05:36 |
| TheJulia | But since it is blocked now it might as well be updated | 05:36 |
| TheJulia | Anyway, something to do tomorrow. I need to get some sleep :) | 05:37 |
| kaifeng | TheJulia: works for me both ways, have some rest! | 06:12 |
| *** igordc has quit IRC | 06:21 | |
| *** uzumaki has joined #openstack-ironic | 06:36 | |
| openstackgerrit | Harald Jensås proposed openstack/networking-generic-switch master: Add netmiko_session_logging https://review.opendev.org/717060 | 06:41 |
| *** ociuhandu has joined #openstack-ironic | 06:46 | |
| *** iurygregory has joined #openstack-ironic | 06:49 | |
| iurygregory | good morning Ironic o/ | 06:50 |
| kaifeng | Good morning iurygregory o/ | 06:51 |
| iurygregory | hey kaifeng o/ | 06:52 |
| uzumaki | o/ ironic | 06:53 |
| *** ociuhandu has quit IRC | 06:55 | |
| *** ociuhandu has joined #openstack-ironic | 06:55 | |
| openstackgerrit | Harald Jensås proposed openstack/ironic master: Fetch netmiko session log https://review.opendev.org/717194 | 06:56 |
| *** gudrutis2 has joined #openstack-ironic | 06:59 | |
| gudrutis2 | Morning o/ | 06:59 |
| iurygregory | gudrutis2, o/ | 07:04 |
| iurygregory | hjensas|afk, hey for me it seems the change to use ipv4/ipv6 will be necessary, I was wondering if it would be better make it configurable | 07:05 |
| *** hjensas|afk is now known as hjensas | 07:05 | |
| *** ociuhandu has quit IRC | 07:05 | |
| arne_wiebalck | Good morning, ironic! | 07:05 |
| iurygregory | good morning arne_wiebalck o/ | 07:06 |
| *** ociuhandu has joined #openstack-ironic | 07:06 | |
| arne_wiebalck | hey iurygregory o/ | 07:07 |
| hjensas | iurygregory: I tried the same in https://review.opendev.org/716893. It did'nt help. I think it's the Netmiko cmd_verify that made the difference? | 07:07 |
| patchbot | patch 716893 - networking-generic-switch - [DNM] Use IPv4 127.0.0.1 ip in ml2 ngs config (ABANDONED) - 1 patch set | 07:07 |
| iurygregory | hjensas, yup | 07:07 |
| iurygregory | only cmd_verify doesn't help | 07:07 |
| iurygregory | but with your change to use ipv4 (I also tested with ::1 for ipv6) it works | 07:07 |
| hjensas | iurygregory: huh, ok. | 07:10 |
| iurygregory | but no idea if using hardcode ipv4 or ipv6 would be the best XD | 07:10 |
| hjensas | I guess we just need to put in place what works. :) | 07:10 |
| hjensas | it's funky that 'localhost' resolving to | 07:11 |
| iurygregory | I will try a test removing localhost from the ipv6 entry in /etc/hosts wdyt? | 07:11 |
| hjensas | ::1 in my local env works. | 07:11 |
| *** ociuhandu has quit IRC | 07:11 | |
| iurygregory | probably something on the cloud provider =( | 07:11 |
| hjensas | iurygregory: yes, I think for CI we can just use 127.0.0.1 hard coded? | 07:13 |
| iurygregory | hjensas, I was thinking this also, and TheJulia asked me to try with ::1 XD | 07:14 |
| hjensas | iurygregory: which also works? | 07:15 |
| iurygregory | hjensas, yup | 07:16 |
| hjensas | iurygregory: the funky thing with 'localhost' is that /etc/hosts have multiple lines with that. So depending on if the file is parsed top down or bottom up results would differ. | 07:17 |
| iurygregory | hjensas, you just made my head explode XD | 07:17 |
| hjensas | iurygregory: :D a crazy theory. | 07:19 |
| iurygregory | I need more coffee now XD | 07:19 |
| *** bnemec has quit IRC | 07:20 | |
| accakks | <TheJulia "accakks: can we help you?"> Yes that would be great | 07:21 |
| *** rachit7 has joined #openstack-ironic | 07:23 | |
| iurygregory | accakks, Julia is probably sleeping atm (US time), feel free to post your question and people will try to answer when possible =) | 07:23 |
| * accakks sent a long message: < https://matrix.org/_matrix/media/r0/download/matrix.org/KpAcXDYJcXBpyiAcgkXAKGTr > | 07:24 | |
| openstackgerrit | Harald Jensås proposed openstack/ironic stable/train: Additional IP addresses to IPv6 stateful ports https://review.opendev.org/717205 | 07:24 |
| *** ociuhandu has joined #openstack-ironic | 07:26 | |
| iurygregory | accakks, well it's a long message in a link... | 07:27 |
| iurygregory | can you split in small messages so we can read in the irc ? | 07:27 |
| accakks | Oh apologies, I'm using matrix client | 07:27 |
| accakks | I'll split it and send again | 07:27 |
| iurygregory | accakks, no worries =) | 07:27 |
| accakks | Hi everyone, I've been trying to run ironic on devstack and finally facing following errors | 07:28 |
| accakks | (I've tried running ./clean.sh and again re running ./stack.sh ) | 07:28 |
| iurygregory | ouch | 07:28 |
| accakks | !. On running ./stack.sh I face error on exit | 07:28 |
| openstack | accakks: Error: "." is not a valid command. | 07:28 |
| iurygregory | devstack after clean is not that good I would say | 07:28 |
| accakks | Oh okay, but I faced same error the first time I ran stack.sh | 07:29 |
| accakks | So before re running next time I sued clean.sh before that | 07:29 |
| iurygregory | when my deployment fails I re-install the OS to try again =( | 07:29 |
| iurygregory | try running clean.sh and unstack.sh | 07:30 |
| accakks | Okay let me try | 07:30 |
| accakks | I'll get back in few mins after trying these | 07:30 |
| iurygregory | accakks, sure | 07:31 |
| arne_wiebalck | ./unstack.sh works surprisingly well for me | 07:31 |
| iurygregory | arne_wiebalck, I know that you have a pact with zuul, now you also have one with devstack? | 07:31 |
| accakks | Awesome, I did not try it before, fingers crossed | 07:32 |
| iurygregory | this is unfair XD | 07:32 |
| *** ociuhandu has quit IRC | 07:32 | |
| arne_wiebalck | the UEFI patch got passed Zuul on the first attempt yesterday | 07:32 |
| hjensas | accakks: ./unstack.sh -> sudo rm -rf /opt/stack/.vbmc/ -> ./stach.sh works for me in devstack. | 07:32 |
| arne_wiebalck | it pays off to talk to Zuul | 07:33 |
| accakks | <hjensas "accakks: ./unstack.sh -> sudo rm"> Trying, thanks | 07:33 |
| arne_wiebalck | hjensas: oh, that's an interesting additional step | 07:33 |
| hjensas | arne_wiebalck: I always get a already exists error if I don't clean out .vbmc directory. | 07:34 |
| arne_wiebalck | accakks: is this with a local.conf file that includes ironic, actually? | 07:34 |
| iurygregory | oh yeah there are some folders that may not be removed with .clean or unstack (vbmc and i think something from nova) | 07:34 |
| arne_wiebalck | is that worth putting somewhere? or fixing somewhere? | 07:35 |
| iurygregory | arne_wiebalck, hehehe you have the power of the words | 07:35 |
| * arne_wiebalck taps on his spell book | 07:35 | |
| iurygregory | worth fixing the clean unstack you mean? | 07:35 |
| hjensas | arne_wiebalck: yes should be trivial to fix. | 07:36 |
| iurygregory | I would say it will be when I can't use VM's to test devstack hehehe | 07:36 |
| arne_wiebalck | I think these are the little things we know (or some of us do), but which make it hard for newcomers | 07:40 |
| iurygregory | arne_wiebalck, yeah, | 07:41 |
| * accakks sent a long message: < https://matrix.org/_matrix/media/r0/download/matrix.org/bNukivmntWevPlpFMLhBlvrd > | 07:44 | |
| openstackgerrit | Iury Gregory Melo Ferreira proposed openstack/networking-baremetal master: [DNM] Testing CI https://review.opendev.org/715411 | 07:44 |
| accakks | I'll try uninstalling this simplejson thing manually maybe and then try again. | 07:45 |
| iurygregory | accakks, if you are pasting logs maybe would be worth putting in paste.openstack.org/ | 07:45 |
| iurygregory | accakks, can you give details of your env? OS you are using, local.conf for devstack | 07:46 |
| accakks | Here : http://paste.openstack.org/show/791575/ | 07:48 |
| accakks | I'm using ubuntu 18.04 | 07:48 |
| accakks | Maybe manual uninstall of simplejson would fix this? | 07:49 |
| iurygregory | this is kinda strange O.o | 07:50 |
| iurygregory | can you share your local.conf? | 07:50 |
| accakks | Here, my local.conf http://paste.openstack.org/show/791576/ | 07:52 |
| iurygregory | ok this is the default example we have in our docs, i will spin a VM with it to see | 07:53 |
| *** rpittau|afk is now known as rpittau | 07:56 | |
| rpittau | good morning ironic! o/ | 07:57 |
| iurygregory | rpittau, good morning o/ | 07:57 |
| rpittau | hey iurygregory :) | 07:57 |
| iurygregory | just fyi we have a fix for ngs | 07:58 |
| rpittau | iurygregory: ok, is that also the fix for baremetal-networking ? | 07:58 |
| iurygregory | https://review.opendev.org/#/c/717020/ but maybe would be good get feedback from the cores | 07:58 |
| patchbot | patch 717020 - networking-generic-switch - Testing CI - 3 patch sets | 07:58 |
| iurygregory | rpittau, I pushed with depends-On to see if it will | 07:58 |
| iurygregory | https://review.opendev.org/715411 testing | 07:59 |
| patchbot | patch 715411 - networking-baremetal - [DNM] Testing CI - 3 patch sets | 07:59 |
| rpittau | alright, let's see | 07:59 |
| iurygregory | at least we will be fixing the multinode job hehehe | 08:00 |
| iurygregory | 1 step at time | 08:00 |
| *** tkajinam has quit IRC | 08:08 | |
| *** uzumaki has quit IRC | 08:16 | |
| *** lucasagomes has joined #openstack-ironic | 08:17 | |
| *** dtantsur|afk is now known as dtantsur | 08:22 | |
| dtantsur | morning, ironic. TGIF! | 08:22 |
| rpittau | happy friday dtantsur :) | 08:23 |
| iurygregory | dtantsur, morning \o/ | 08:23 |
| *** ociuhandu has joined #openstack-ironic | 08:27 | |
| rpittau | TheJulia: when you have a minute I think this can move forward now https://review.opendev.org/708827 | 08:31 |
| patchbot | patch 708827 - ironic-python-agent (stable/train) - Rescan after restarting the md device - 2 patch sets | 08:31 |
| openstackgerrit | Mark Goddard proposed openstack/tenks master: Refactor virtualbmc-domain tasks into a module https://review.opendev.org/713043 | 08:34 |
| *** ociuhandu has quit IRC | 08:44 | |
| *** rmart04 has joined #openstack-ironic | 08:49 | |
| *** ociuhandu has joined #openstack-ironic | 08:54 | |
| *** tkajinam has joined #openstack-ironic | 08:54 | |
| *** k_mouza has joined #openstack-ironic | 08:59 | |
| iurygregory | damm I should have tested networking-baremetal with the ipv4 change D: | 09:03 |
| *** alexmcleod has joined #openstack-ironic | 09:08 | |
| accakks | <iurygregory "this is kinda strange O.o"> this fixed it -> pip install --ignore-installed simplejson | 09:14 |
| *** ociuhandu has quit IRC | 09:15 | |
| iurygregory | accakks, awesome! | 09:15 |
| iurygregory | locally I didnt get any problems with simplejson .-. | 09:15 |
| *** ociuhandu has joined #openstack-ironic | 09:15 | |
| accakks | > This error means that this package's metadata doesn't include a list of files that belong to it. Most probably, you have installed this package via your OS' package manager, so you need to use that rather than pip to update or remove it, too. | 09:17 |
| accakks | 09:17 | |
| accakks | It must have been an issue on my system only | 09:17 |
| iurygregory | yeah | 09:17 |
| iurygregory | gald you got it working =) | 09:17 |
| *** ociuhandu has quit IRC | 09:20 | |
| *** ociuhandu has joined #openstack-ironic | 09:43 | |
| *** ociuhandu has quit IRC | 09:52 | |
| openstackgerrit | Dmitry Tantsur proposed openstack/python-ironicclient master: [PoC] A standalone CLI for ironic https://review.opendev.org/717273 | 09:52 |
| dtantsur | TheJulia and anyone curious ^^^ | 09:52 |
| iurygregory | =O | 09:53 |
| dtantsur | hehe | 09:53 |
| * dtantsur is on fire this week | 09:53 | |
| iurygregory | on fire only? | 09:54 |
| iurygregory | I would say it's +1 lvl | 09:54 |
| iurygregory | but i don't have names for this lvl =( | 09:54 |
| dtantsur | :D | 09:55 |
| openstackgerrit | Dmitry Tantsur proposed openstack/python-ironicclient master: [PoC] A standalone CLI for ironic https://review.opendev.org/717273 | 09:55 |
| *** ociuhandu has joined #openstack-ironic | 09:55 | |
| dtantsur | I need your help folks: what could be a name for ironic-inspector standalone CLI? | 09:57 |
| dtantsur | baremetal-introspection is way too long | 09:57 |
| dtantsur | bm-introspection? also a bit long(ish) | 09:57 |
| iurygregory | bm-inspect? | 09:57 |
| dtantsur | maybe | 09:57 |
| dtantsur | or maybe we should make the ironic's standalone CLI support ironic-inspector? | 09:58 |
| dtantsur | so that we keep 'baremetal introspection'? | 09:58 |
| * dtantsur needs to think about it | 09:58 | |
| *** dtantsur is now known as dtantsur|brb | 09:58 | |
| iurygregory | Im wondering here, I think we should have the networking-baremetal-multitenant-vlans running in networking-generic-switch ... thoughts? | 09:58 |
| iurygregory | since changes in ngs may affect netwroking-baremetal... | 09:59 |
| iurygregory | at least in check | 10:00 |
| *** ociuhandu has quit IRC | 10:09 | |
| openstackgerrit | Iury Gregory Melo Ferreira proposed openstack/networking-baremetal master: [DNM] Testing CI https://review.opendev.org/715411 | 10:11 |
| *** tkajinam has quit IRC | 10:11 | |
| *** rachit7 has quit IRC | 10:16 | |
| hjensas | iurygregory: or make ironic-tempest-ipa-wholedisk-direct-tinyipa-multinode boting in n-g-s ? | 10:16 |
| iurygregory | hjensas, boting? | 10:17 |
| hjensas | iurygregory: s/boting/voting/ | 10:17 |
| iurygregory | hjensas, well we solved ironic-temepest-ipa.... | 10:17 |
| iurygregory | but the job in networking-baremetal is not working yet.. | 10:17 |
| iurygregory | Im debugging to see if it can only be ipv4 or maybe some config | 10:17 |
| hjensas | oh, still same netmiko connect/no data issue? | 10:19 |
| iurygregory | Im looking now at the fresh logs =) | 10:19 |
| iurygregory | Apr 03 08:32:06.511427 ubuntu-bionic-openedge-us-east-0015685205 neutron-dhcp-agent[1898]: ERROR neutron.agent.linux.ip_lib [None req-987d7880-c3cf-4c8b-b1bf-b1ec31bd5bed None None] Device tap56ac1674-ad cannot be used as it has no MAC address | 10:20 |
| iurygregory | the vms are starting to boot bu they get stuck https://storage.gra.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_cf5/715411/3/check/networking-baremetal-multitenant-vlans/cf50b75/controller/logs/ironic-bm-logs/node-3_no_ansi_2020-04-03-09%3A58%3A41_log.txt | 10:21 |
| iurygregory | I was expecting to se more in the vm logs .-. | 10:21 |
| *** ociuhandu has joined #openstack-ironic | 10:22 | |
| openstackgerrit | Merged openstack/ironic-python-agent master: Change ipa partition jobs to netboot https://review.opendev.org/716624 | 10:25 |
| openstackgerrit | Merged openstack/ironic-lib master: Add netboot to all partition jobs https://review.opendev.org/716605 | 10:25 |
| hjensas | ok, so next issue behind the n-g-s issue. My fetch of the netmiko_session.log does'nt seem to work - https://review.opendev.org/#/c/717194/1/zuul.d/ironic-jobs.yaml | 10:25 |
| patchbot | patch 717194 - ironic - Fetch netmiko session log - 1 patch set | 10:25 |
| iurygregory | =( | 10:26 |
| openstackgerrit | Harald Jensås proposed openstack/ironic master: Fetch netmiko session log https://review.opendev.org/717194 | 10:27 |
| *** rpittau is now known as rpittau|bbl | 10:29 | |
| *** rcernin has quit IRC | 10:29 | |
| *** ociuhandu has quit IRC | 10:33 | |
| hjensas | iurygregory: wonder if the IPA is simply stuck at EDD probing for too long. I rechecked https://review.opendev.org/714421, maby tinyipa does not have that issue? | 10:37 |
| patchbot | patch 714421 - networking-baremetal - Explicitly set ramdisk type - 3 patch sets | 10:37 |
| hjensas | iurygregory: we could also add edd=off to [pxe]/pxe_append_params ? | 10:38 |
| *** derekh has joined #openstack-ironic | 10:40 | |
| iurygregory | hjensas, gotcha I will try that | 10:48 |
| openstackgerrit | Ilya Etingof proposed openstack/ironic master: Support burning configdrive into boot ISO https://review.opendev.org/687250 | 11:05 |
| openstackgerrit | Merged openstack/ironic-tempest-plugin master: Upddate tox config https://review.opendev.org/714185 | 11:07 |
| openstackgerrit | Merged openstack/tenks master: Refactor virtualbmc-domain tasks into a module https://review.opendev.org/713043 | 11:07 |
| *** ociuhandu has joined #openstack-ironic | 11:17 | |
| iurygregory | hjensas, do you think I should give a try running the ngs with 127.0.0.1 to see how networking-baremetal will respond? | 11:24 |
| hjensas | iurygregory: since DHCP iPXE is fetching the deploy kernel+ramdisk the n-g-s port plugging must have succeded already? | 11:27 |
| iurygregory | to change the /pxe_append_params I would need to push a patch in ironic first and then test with networking-baremetal right? | 11:30 |
| *** ociuhandu has quit IRC | 11:32 | |
| hjensas | iurygregory: I think you can set IRONIC_EXTRA_PXE_PARAMS in the networking-baremetal job? | 11:33 |
| iurygregory | oh | 11:34 |
| hjensas | iurygregory: I can push a change with that. | 11:36 |
| iurygregory | hjensas, just did =) | 11:37 |
| iurygregory | not sure if it's correct... | 11:37 |
| openstackgerrit | Ilya Etingof proposed openstack/ironic master: Support burning configdrive into boot ISO https://review.opendev.org/687250 | 11:37 |
| openstackgerrit | Iury Gregory Melo Ferreira proposed openstack/networking-baremetal master: [DNM] Testing CI https://review.opendev.org/715411 | 11:37 |
| iurygregory | gerrit could be less slow | 11:37 |
| hjensas | iurygregory: :D looks good. Now we wait and see ... | 11:38 |
| openstackgerrit | Ilya Etingof proposed openstack/ironic master: Support burning configdrive into boot ISO https://review.opendev.org/687250 | 11:39 |
| openstackgerrit | Ilya Etingof proposed openstack/ironic master: Add `network_data` field to ironic node object https://review.opendev.org/687910 | 11:40 |
| openstackgerrit | Ilya Etingof proposed openstack/ironic master: Add `network_data` field to ironic node object https://review.opendev.org/687910 | 11:41 |
| *** ociuhandu has joined #openstack-ironic | 11:43 | |
| *** ociuhandu has quit IRC | 11:49 | |
| *** dtantsur|brb is now known as dtantsur | 11:57 | |
| openstackgerrit | Dmitry Tantsur proposed openstack/python-ironicclient master: [PoC] A standalone CLI for ironic https://review.opendev.org/717273 | 12:07 |
| *** rpittau|bbl is now known as rpitau | 12:08 | |
| *** rpitau is now known as rpittau | 12:09 | |
| *** mkrai_ has quit IRC | 12:20 | |
| openstackgerrit | Dmitry Tantsur proposed openstack/python-ironicclient master: [PoC] A standalone CLI for ironic https://review.opendev.org/717273 | 12:21 |
| rpittau | dtantsur: just a thought, why not having a separate project of a standalone CLI client that integrates also support for inspector? Instead of having that in python-ironicclient | 12:23 |
| dtantsur | rpittau: it's a bit of high maintenance costs for 200 lines of simple code | 12:23 |
| dtantsur | also there is a desire for people to be able to `dnf install python3-ironicclient` and just use it | 12:24 |
| *** Lucas_Gray has joined #openstack-ironic | 12:30 | |
| openstackgerrit | Dmitry Tantsur proposed openstack/python-ironicclient master: [PoC] A standalone CLI for ironic https://review.opendev.org/717273 | 12:33 |
| *** Lucas_Gray has quit IRC | 12:35 | |
| openstackgerrit | Dmitry Tantsur proposed openstack/python-ironicclient master: [PoC] A standalone CLI for ironic https://review.opendev.org/717273 | 12:36 |
| *** Lucas_Gray has joined #openstack-ironic | 12:37 | |
| *** rh-jelabarre has joined #openstack-ironic | 12:38 | |
| *** Lucas_Gray has quit IRC | 12:40 | |
| iurygregory | tests still failling =( I wish I could check the configs while the job is running hehe | 12:45 |
| *** Lucas_Gray has joined #openstack-ironic | 12:46 | |
| *** Lucas_Gray has quit IRC | 12:47 | |
| *** ociuhandu has joined #openstack-ironic | 12:47 | |
| *** zigo has quit IRC | 12:48 | |
| openstackgerrit | Riccardo Pittau proposed openstack/ironic-lib master: Convert jobs to dib https://review.opendev.org/716998 | 12:50 |
| *** rloo has joined #openstack-ironic | 12:51 | |
| *** zigo has joined #openstack-ironic | 12:51 | |
| *** ameya49 has joined #openstack-ironic | 13:11 | |
| openstackgerrit | Dmitry Tantsur proposed openstack/python-ironicclient master: A standalone CLI for ironic https://review.opendev.org/717273 | 13:13 |
| dtantsur | ready for review and testing ^^^ | 13:13 |
| dtantsur | TheJulia: I've got a Friday present for you ^^ | 13:13 |
| *** rachit7 has joined #openstack-ironic | 13:13 | |
| TheJulia | <3 | 13:13 |
| dtantsur | and good morning | 13:17 |
| openstackgerrit | Merged openstack/ironic master: Make oslo.reports an optional dependency https://review.opendev.org/716904 | 13:17 |
| iurygregory | good morning TheJulia o/ | 13:17 |
| hjensas | iurygregory: I wonder if enabling cleaning on the networking-baremetal job would shorten the feedback loop. | 13:18 |
| openstackgerrit | Dmitry Tantsur proposed openstack/python-ironicclient master: A standalone CLI for ironic https://review.opendev.org/717273 | 13:19 |
| dtantsur | a small issue in the docs ^^^ | 13:19 |
| iurygregory | hjensas, we can try | 13:20 |
| hjensas | iurygregory: or if we can lower the timeout when waiting-for-callback ... | 13:20 |
| * TheJulia tries to wake up | 13:21 | |
| dtantsur | give up, it's Friday | 13:21 |
| TheJulia | I at least have to write some words in some status update doc and then try and fumble through those words | 13:22 |
| iurygregory | dtantsur, I think after your gift TheJulia will try to wake up anyway =) | 13:22 |
| dtantsur | heh, hopefully | 13:22 |
| TheJulia | I noticed the bulk of priorities for the week merged | 13:24 |
| TheJulia | People should add more things | 13:24 |
| dtantsur | MOAR THINGS | 13:24 |
| TheJulia | What about ALLTHETHINGS? | 13:24 |
| iurygregory | ++ ALLTHETHINGS | 13:24 |
| TheJulia | MOAR ALLTHETHINGS! | 13:25 |
| *** k_mouza has quit IRC | 13:26 | |
| TheJulia | etingof: looks like you have merge conflicts :( | 13:26 |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic-python-agent master: Add an ability to run in-band deploy steps https://review.opendev.org/698770 | 13:27 |
| * arne_wiebalck has solved the UEFI RAID patch merge conflict (hint) | 13:27 | |
| TheJulia | arne_wiebalck: linkity link link? | 13:28 |
| * arne_wiebalck should probably check with another browser ... | 13:28 | |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic-python-agent master: Add an ability to run in-band deploy steps https://review.opendev.org/698770 | 13:28 |
| * arne_wiebalck is disappointed TheJulia does not know the URL by heart given how long the patch is up | 13:28 | |
| arne_wiebalck | https://review.opendev.org/#/c/696156/ | 13:28 |
| patchbot | patch 696156 - ironic-python-agent - Software RAID: Add UEFI support - 11 patch sets | 13:28 |
| iurygregory | =OOOO | 13:28 |
| * TheJulia feels sad but URL memory was under allocated becuase my brain was from before the time of the intertubes as we know them today | 13:29 | |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic-python-agent master: WIP: add a deploy steps for writing an image https://review.opendev.org/716216 | 13:29 |
| arne_wiebalck | TheJulia: +1 ... anything missing, or just not sure about the patch? | 13:30 |
| TheJulia | arne_wiebalck: looking | 13:30 |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic master: WIP: Decompose core deploy step https://review.opendev.org/698309 | 13:31 |
| TheJulia | lgtm | 13:31 |
| TheJulia | At least I got sleep last night | 13:31 |
| * TheJulia starts the 3d printers back up before going and getting more coffee and then actually starting her workday | 13:32 | |
| *** ameya49 has quit IRC | 13:34 | |
| *** mkrai has joined #openstack-ironic | 13:38 | |
| *** jdandrea has joined #openstack-ironic | 13:38 | |
| *** jdandrea has quit IRC | 13:41 | |
| *** jdandrea has joined #openstack-ironic | 13:42 | |
| *** ociuhandu has quit IRC | 13:42 | |
| iurygregory | all the nodes are stuck after deploy_ramdisk =( | 13:50 |
| iurygregory | https://d6c4793bdf35783e53ec-8e6063eece8c96bdec38e25d6079d8b4.ssl.cf2.rackcdn.com/715411/5/check/networking-baremetal-multitenant-vlans/bf2fca4/controller/logs/ironic-bm-logs/node-3_no_ansi_2020-04-03-13%3A34%3A06_log.txt | 13:50 |
| *** tzumainn has joined #openstack-ironic | 13:50 | |
| arne_wiebalck | TheJulia: was there a doodle for the next white paper session? | 13:51 |
| dtantsur | if there was, I missed it as well | 13:54 |
| TheJulia | I didn't end up sending it out | 13:56 |
| TheJulia | Honestly, People's mental health seemed a little worrisome on Monday I didn't want to try and stress people further | 13:57 |
| openstackgerrit | Merged openstack/sushy master: Switch to the new canonical constraints URL on master https://review.opendev.org/717010 | 13:57 |
| hjensas | iurygregory: got the netmiko session log - https://storage.gra.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_9e4/717194/2/check/ironic-tempest-ipa-wholedisk-direct-tinyipa-multinode/9e4e005/controller/logs/netmiko_session_log.txt | 13:58 |
| *** Goneri has joined #openstack-ironic | 13:58 | |
| iurygregory | hjensas, awesome! | 13:58 |
| hjensas | iurygregory: only thing I notice is that it's alternatinv betweeb login from ::1 and from 198.72.124.21 | 14:01 |
| iurygregory | I need more coffee brb | 14:02 |
| *** kashifkhan has joined #openstack-ironic | 14:02 | |
| kashifkhan | Hello, I have a question, if node where IPA is running changes IP, and ironic triggers provisioning, will it fail? | 14:03 |
| arne_wiebalck | TheJulia: Let's discuss next Monday then how to proceed. There was quite some activity since our last meeting: contributions and reviews. | 14:03 |
| dtantsur | kashifkhan: that's quite likely | 14:11 |
| rpioso | Good morning, ironic | 14:13 |
| *** bnemec has joined #openstack-ironic | 14:13 | |
| rachit7 | Good morning rpioso | 14:13 |
| arne_wiebalck | dtantsur: can you elaborate? | 14:13 |
| iurygregory | morning rpioso | 14:14 |
| * arne_wiebalck would have answered "no" :) | 14:14 | |
| *** bnemec is now known as beekneemech | 14:14 | |
| arne_wiebalck | gm rpioso and rachit7 o/ | 14:14 |
| rpioso | arne_wiebalck, iurygregory, rachit7: Happy Friday! | 14:14 |
| rachit7 | arne_wiebalck rpioso 8) | 14:15 |
| kashifkhan | dtantsur: thanks for the answer. I am thinking about whether we need to move the dhcp leases from one cluster to the other during the move/pivot process. | 14:16 |
| *** Chaserjim has quit IRC | 14:16 | |
| kashifkhan | It would be really helpful for me if I get a bit more elaborated answer. Thanks in advance | 14:21 |
| *** mkrai has quit IRC | 14:24 | |
| *** cdearborn has joined #openstack-ironic | 14:25 | |
| *** ociuhandu has joined #openstack-ironic | 14:25 | |
| dtantsur | arne_wiebalck, kashifkhan, I mean, if the IP changes during provisioning, ironic probably won't be able to reach IPA. | 14:26 |
| arne_wiebalck | dtantsur: ah, I understood: the node is in available ... the IP changes in DHCP ... then the node is deployed | 14:27 |
| arne_wiebalck | dtantsur: for your scenario, I agree :) | 14:27 |
| dtantsur | if a node is available, it's usually powered off, the DHCP changes do not affect ironic anyhow | 14:28 |
| *** mkrai has joined #openstack-ironic | 14:28 | |
| arne_wiebalck | that's my point | 14:28 |
| kashifkhan | ok, I meant when IPA is running on a node, there is a change in IP and then Ironic triggers provisioning | 14:29 |
| dtantsur | kashifkhan: just to be clear: why IPA is running on a node before provisioning? are you using the fast-track option? | 14:30 |
| kashifkhan | what about the heartbit, would it somehow tell ironic to communicate to new IP? sorry my questions might seem foolish, I am a newbie in this field | 14:30 |
| etingof | TheJulia, yep, I think I resolved some already | 14:30 |
| dtantsur | kashifkhan: we have explicitly disallowed updating the IPA endpoint as a security improvement against potential hijacking | 14:31 |
| dtantsur | cc TheJulia | 14:31 |
| kashifkhan | dtantsur: yes | 14:31 |
| openstackgerrit | Harald Jensås proposed openstack/networking-baremetal master: Explicitly set ramdisk type https://review.opendev.org/714421 | 14:31 |
| dtantsur | TheJulia: do you remember if we really do anything about IP changes in fast-track? | 14:32 |
| dtantsur | like, if heartbeat insists that the node has a new IP, mark it as not fast-track-able? | 14:32 |
| dtantsur | and reboot the next provisioning? | 14:32 |
| * iurygregory fingers crosses for https://review.opendev.org/714421 | 14:32 | |
| patchbot | patch 714421 - networking-baremetal - Explicitly set ramdisk type - 4 patch sets | 14:32 |
| iurygregory | ironic-cores: https://review.opendev.org/#/c/717020/ would allow us to have ironic-tempest-ipa-wholedisk-direct-tinyipa-multinode voting again in our CI, just wondering if this is the approach we want | 14:34 |
| openstackgerrit | Tzu-Mainn Chen proposed openstack/ironic master: Use auth values from neutron conf when managing Neutron ports https://review.opendev.org/708678 | 14:34 |
| patchbot | patch 717020 - networking-generic-switch - Testing CI - 3 patch sets | 14:34 |
| TheJulia | dtantsur: ip can't really chagne with fasttrack | 14:36 |
| * arne_wiebalck hopes iurygregory has talked to Zuul re 714421 (crossing fingers is just superstition) | 14:36 | |
| TheJulia | dtantsur: because we don't allow heartbeats to change the address. We could add a knob to permit updates if the agent_token is avlid | 14:36 |
| iurygregory | arne_wiebalck, I've being talking a lot with zuul the problem is only the vms trying to boot =) | 14:36 |
| * TheJulia goes back to meeting | 14:36 | |
| kashifkhan | dtantsur: can you point some design doc or link where I can get a bit more info on this | 14:36 |
| iurygregory | so it's not his fault =) | 14:36 |
| dtantsur | kashifkhan: not from the top of my head, let's wait for TheJulia to return from meetings | 14:39 |
| kashifkhan | dtantsur: thanks | 14:40 |
| TheJulia | sorry, brain still in meeting | 14:42 |
| openstackgerrit | Harald Jensås proposed openstack/networking-generic-switch master: Add netmiko_session_logging https://review.opendev.org/717060 | 14:42 |
| * TheJulia looses her mind | 14:43 | |
| dtantsur | my examples shows that having a mind is overrated | 14:43 |
| dtantsur | you can go pretty far with decent feathers and an ability to swallow mice whole | 14:44 |
| iurygregory | a cat for TheJulia https://media.giphy.com/media/v6aOjy0Qo1fIA/giphy.gif | 14:44 |
| TheJulia | And you can turn your head like 210 degrees | 14:44 |
| TheJulia | single pivot attachment point | 14:44 |
| TheJulia | as opposed to humans with two | 14:44 |
| TheJulia | (which is why our range of movement is so restricted) | 14:44 |
| *** mkrai has quit IRC | 14:45 | |
| * TheJulia did some falcon rescue/rehab volunteer work when she was younger | 14:47 | |
| openstackgerrit | Ilya Etingof proposed openstack/ironic master: Support burning configdrive into boot ISO https://review.opendev.org/687250 | 14:49 |
| openstackgerrit | Ilya Etingof proposed openstack/ironic master: Add `network_data` field to ironic node object https://review.opendev.org/687910 | 14:52 |
| TheJulia | gah, meeting continues | 14:53 |
| openstackgerrit | Tzu-Mainn Chen proposed openstack/ironic master: Add node lessee field https://review.opendev.org/706864 | 14:54 |
| accakks | Final error I'm getting after package errors are done with while installing. This was the same one I was getting earlier too, on running further commands in process. | 14:56 |
| * accakks sent a long message: < https://matrix.org/_matrix/media/r0/download/matrix.org/hSoAwiKQMxyjSQhPIRDZpIXZ > | 14:56 | |
| accakks | * | 14:57 |
| accakks | Failed to discover available identity versions when contacting http://192.168.1.108/identity. Attempting to parse version from URL. | 14:57 |
| accakks | Could not find versioned identity endpoints when attempting to authenticate. Please check that your auth_url is correct. Internal Server Error (HTTP 500) | 14:57 |
| rpittau | accakks: I suggest you use an external link to point to logs, like http://paste.openstack.org/ | 14:58 |
| accakks | arne_wiebalck: | 14:58 |
| openstackgerrit | Bob Fournier proposed openstack/sushy stable/train: Ignore failing message registry download https://review.opendev.org/717318 | 14:58 |
| openstackgerrit | Tzu-Mainn Chen proposed openstack/python-ironicclient master: Add node lessee https://review.opendev.org/716011 | 14:58 |
| accakks | <accakks "Final error I'm getting after pa"> http://paste.openstack.org/show/791593/ | 14:59 |
| accakks | > <@accakks:matrix.org> Final error I'm getting after package errors are done with while installing. This was the same one I was getting earlier too, on running further commands in process. | 14:59 |
| accakks | * Here: http://paste.openstack.org/show/791593/ | 14:59 |
| TheJulia | ugh, I guess there is a matrix to irc gateway that doesn't fully grok how irc works? | 14:59 |
| TheJulia | and out of meeting \o/ | 15:00 |
| TheJulia | kashifkhan: Hi! | 15:00 |
| TheJulia | kashifkhan: so fast track was intended to operate on static and edge networks, not networks where machines are switched around physical network to physical network, in part because of security and additionally because of the variation of race condition issues when we try and test such actions in CI. The bulk of our CI failures are actually that same sort of race condition | 15:05 |
| TheJulia | well, aside form patches that break things accidently | 15:05 |
| dtantsur | TheJulia: we could probably detect this situation (attempt to change IP address) and make the node not-fast-track-able | 15:07 |
| TheJulia | i thought we had a conditional wrapped up in if we could based upon networking | 15:08 |
| TheJulia | or maybe we did early on | 15:09 |
| TheJulia | yeah, I think the initial iteration had it but it came out | 15:10 |
| TheJulia | I guess we need to delineate, what is changing the network | 15:11 |
| TheJulia | because if ironic is doing it, we should know and be able to clear any flags fairly easily. If agent_url attempts to be updated with a valid token, we may want to accept it? or enable accepting it? | 15:11 |
| TheJulia | I guess there are many possible paths too | 15:11 |
| TheJulia | I'm looking for the rfe now | 15:12 |
| dtantsur | given that we have agent tokens and hashing for rescue password, we may want to relax this requirement again | 15:13 |
| TheJulia | yeah | 15:13 |
| *** yoctozepto has quit IRC | 15:17 | |
| TheJulia | GAH, too many terminal windows consumed by the 3d slicing tools | 15:18 |
| *** Mahendra22 has joined #openstack-ironic | 15:18 | |
| TheJulia | kashifkhan: fyi https://storyboard.openstack.org/#!/story/2004965 | 15:18 |
| *** yoctozepto8 has joined #openstack-ironic | 15:18 | |
| openstackgerrit | Bob Fournier proposed openstack/sushy stable/train: Make MessageRegistryFile.Registry attribute non-required https://review.opendev.org/717322 | 15:18 |
| *** Mahendra22 has quit IRC | 15:18 | |
| TheJulia | bfournie: you mentioned a manual backport the other day, do you remember what it was because I totally got distracted and forgot :( | 15:18 |
| bfournie | TheJulia: the IPMO debug log change perhaps? | 15:19 |
| bfournie | TheJulia: IPMI | 15:19 |
| TheJulia | yeah! | 15:19 |
| dtantsur | I also mentioned a manual backport of https://review.opendev.org/#/c/714511/ | 15:20 |
| patchbot | patch 714511 - ironic - Make deploy step failure logging indicate the error (MERGED) - 2 patch sets | 15:20 |
| dtantsur | I think the IPMI one has landed, no? | 15:20 |
| bfournie | TheJulia, dtantsur: yes and we decided the back to keep debug false by default so no changes in backport | 15:21 |
| *** hjensas is now known as hjensas|afk | 15:21 | |
| *** gyee has joined #openstack-ironic | 15:22 | |
| TheJulia | yeah, I'm running local tests on it now and will upload it shortly | 15:24 |
| TheJulia | sorry! :( | 15:24 |
| TheJulia | accakks: your error, if your not running in stand-alone mode, has me wondering if you have a proxy in-front of keystone and that keystone is just not responding | 15:25 |
| rachit7 | Hi TheJulia this patch needs last review : https://review.opendev.org/#/c/710852/ . please add this patch in your review list :) | 15:26 |
| patchbot | patch 710852 - ironic - DRAC: Fix a failure to create virtual disk bug - 5 patch sets | 15:26 |
| *** k_mouza has joined #openstack-ironic | 15:27 | |
| openstackgerrit | Julia Kreger proposed openstack/ironic stable/stein: Move ipmi logging to a separate option https://review.opendev.org/717323 | 15:29 |
| *** accakks_ has joined #openstack-ironic | 15:29 | |
| *** k_mouza has quit IRC | 15:31 | |
| *** igordc has joined #openstack-ironic | 15:32 | |
| *** k_mouza has joined #openstack-ironic | 15:34 | |
| * dtantsur watches the in-band deploy steps patch going green in zuul with an expression of disbelief | 15:36 | |
| TheJulia | is it too early to ship it? | 15:39 |
| TheJulia | Maybe we can begin warming up a steam ship | 15:39 |
| TheJulia | I have a friend that is actually the chief engineer of a steam ship | 15:42 |
| rpittau | bye! have a nice weekend! o/ | 15:43 |
| *** rpittau is now known as rpittau|afk | 15:43 | |
| TheJulia | bye rpittau|afk, stay safe! | 15:44 |
| openstackgerrit | Ilya Etingof proposed openstack/ironic master: Add `network_data` field to ironic node object https://review.opendev.org/687910 | 15:44 |
| TheJulia | time to close windows | 15:44 |
| openstackgerrit | Andreas Jaeger proposed openstack/ironic master: Cleanup Python 2.7 support https://review.opendev.org/715454 | 15:49 |
| *** JayF is now known as JasonF | 15:55 | |
| *** JasonF is now known as JayF | 15:55 | |
| *** lucasagomes has quit IRC | 16:00 | |
| dtantsur | TheJulia: steam ship, wow | 16:03 |
| TheJulia | https://www.virginiav.org/ | 16:07 |
| *** ociuhandu has quit IRC | 16:08 | |
| *** Lucas_Gray has joined #openstack-ironic | 16:11 | |
| openstackgerrit | Julia Kreger proposed openstack/ironic stable/train: Make deploy step failure logging indicate the error https://review.opendev.org/717329 | 16:11 |
| dtantsur | oh, cool | 16:12 |
| *** dking_desktop has joined #openstack-ironic | 16:12 | |
| *** Lucas_Gray has quit IRC | 16:20 | |
| *** rachit7 has quit IRC | 16:21 | |
| iurygregory | networking-baremetal is also fixed \o/ | 16:42 |
| iurygregory | https://review.opendev.org/#/c/714421/ https://review.opendev.org/717020 | 16:42 |
| patchbot | patch 714421 - networking-baremetal - Explicitly set ramdisk type - 4 patch sets | 16:42 |
| patchbot | patch 717020 - networking-generic-switch - Testing CI - 3 patch sets | 16:42 |
| TheJulia | woot | 16:43 |
| iurygregory | I forgot to test using tinyipa... | 16:43 |
| *** yoctozepto8 is now known as yoctozepto | 16:43 | |
| iurygregory | I will update the ngs with a release note so we can go ahead | 16:44 |
| iurygregory | and than I will run to the supermarket hehehe | 16:44 |
| dking_desktop | Is the noop network_interface the best option if you want to provision servers which will afterwards live on their own network, and not the provisioning or tenant networks? | 16:45 |
| TheJulia | dking_desktop: noop if the networking is entirely already staticly configured and there is an external dhcp server. You should be able to change the network interface after provisioning if you need to | 16:50 |
| TheJulia | Just... naturally, you may break the ability to clean/redeploy if needed | 16:50 |
| openstackgerrit | Verification of a change to openstack/ironic failed: Do not autoescape all Jinja2 templates https://review.opendev.org/716845 | 16:53 |
| dking_desktop | So, here's what I'm trying to do with my new project: I would like some way to be able to provision servers and manage servers. I want to be able to start those servers off with some basic network configuration, including bonds and VLANs. While those servers are running, I would like them to not have access to any shared provisioning network, but I would like to monitor them to ensure that they are up, and I | 16:56 |
| dking_desktop | ity to reboot, clean, and redeploy them. I don't mind handling the networking separately if necessary, but of course I would appreciate any tool which can help with that. How could Ironic fit into this particular use case? What networking interface would be best for that? | 16:56 |
| openstackgerrit | Iury Gregory Melo Ferreira proposed openstack/networking-generic-switch master: Fix ngs https://review.opendev.org/717020 | 17:03 |
| *** Lucas_Gray has joined #openstack-ironic | 17:06 | |
| openstackgerrit | Iury Gregory Melo Ferreira proposed openstack/ironic master: Change multinode job to voting https://review.opendev.org/717348 | 17:06 |
| iurygregory | time to go o/ bye everyone | 17:08 |
| *** jlvillal is now known as jlvillal_laptop | 17:10 | |
| *** jlvillal_laptop is now known as jlvill-travel | 17:10 | |
| *** jlvill-travel is now known as jlvacation | 17:11 | |
| *** jlvacation is now known as jlv-sick | 17:11 | |
| *** jlv-sick is now known as jlviva-viva | 17:11 | |
| *** jlviva-viva is now known as jlvillal | 17:11 | |
| *** ociuhandu has joined #openstack-ironic | 17:12 | |
| TheJulia | dking_desktop: if your using neutron, that is exactly what the neutron network interface is for, however if your not using neutron, you'll need to create a similar network interface to integrate with your network ?automation? | 17:14 |
| arne_wiebalck | bye everyone o/ | 17:15 |
| openstackgerrit | Merged openstack/sushy stable/train: Ignore failing message registry download https://review.opendev.org/717318 | 17:15 |
| dking_desktop | So, from what I was reading, it seemed that with the neutron interface, the created servers would be using a tenant network that would need the neutron server up and running. Is that the case? I was hoping that once the servers were running, that they wouldn't necessarily be dependent upon any outside servers. Also, I was hoping that those servers could use VXLANs, or any other arbitrary networking we decide | 17:18 |
| TheJulia | dking_desktop: so, tenant network if the tenant has access for something to be bound to that network, by design neutron will want to perform IPAM and DHCP management, but it doesn't have to do routing, and I think even dhcp might be able to be turned off if you have static DHCP assignments/management | 17:21 |
| TheJulia | ports can also not have IP addresses assigned at all, so it becomes a port attachment to a network attachment concept | 17:22 |
| dking_desktop | I'm trying to get my head around that. What I'm going for here is a bit like TripleO but not. So, with Ironic, I would use the neutron network interface, which would manage the switch, and then use that to deploy, redeploy, and clean servers, but while a server is running, it could set the networking so that it doesn't communicate with openstack at all, except for the IPMI reaching out to it over the separate | 17:30 |
| dking_desktop | the baremetal servers can have their networking setup in anyway they want, including the ability to use VXLANs and trunk ports, etc.? | 17:30 |
| JayF | dking_desktop: you basically described how neutron works :) -- it's a control plane tool for configuring networks, then usually gets the hell out of the way for data plane stuff | 17:34 |
| JayF | that's especially true for the Ironic use case | 17:34 |
| JayF | Let me lay it out this directly: there should be no situation in which your OpenStack control plane going down should impact your data plane *unless* you're using network booting on your deployed instances (e.g. you always pxe from conductor instead of local booting) | 17:35 |
| *** k_mouza has quit IRC | 17:40 | |
| dking_desktop | JayF: Okay. I really like that. Thanks for clarifying. I'm apologize for asking silly questions. I think I had some wrong ideas about how things would work. So, another silly question: When a baremetal node is provisioned, by default, is the OS actually installed on the server, rather than requiring a PXE boot? | 17:43 |
| dking_desktop | I know that I'm using a glance image currently when I do "openstack server create ...", but that doesn't seem like it would be something that would necessarily be installed onto the server. | 17:44 |
| *** dking_desktop has left #openstack-ironic | 17:45 | |
| *** dking_desktop has joined #openstack-ironic | 17:45 | |
| JayF | So the answer to what the default is, I'm not a great person to ask | 17:45 |
| JayF | but generally speaking there are two modes: netboot and local boot. With local boot, you either have a bootloader baked into a whole disk image, or ironic puts down a bootloader with your partition image | 17:46 |
| JayF | with netboot, you continue to boot into your *local installed OS* via pxe and the configured pxe kernel/initrd | 17:46 |
| *** bdodd has quit IRC | 17:48 | |
| dking_desktop | So, even if you are using a Glance image, it's possible to do local boot, where Ironic installs a bootloader which then uses that image? | 17:48 |
| *** derekh has quit IRC | 17:52 | |
| TheJulia | dking_desktop: yes, we have support for full disk images and partition images where a bootloader is installed | 17:54 |
| TheJulia | Whole disk naturally is good if your using things that are not linux | 17:54 |
| dking_desktop | Looking at the documentation, I only see PXE boot when I'm looking at "Boot interfaces". It says that it is the default for IPMI. | 17:54 |
| TheJulia | that should really be titled "boot interface for deployment" | 17:54 |
| dking_desktop | Okay. I'm using CentOS, but like was said before, I want these servers, my data plane, to not be dependent at all upon my control plane while they are running. Preferably, I would like it to be possible for one of these servers to be able to reboot even if my openstack instance were to be down for some reason. | 17:55 |
| dking_desktop | Yeah. I don't even know where to look for boot options for after deployment. | 17:56 |
| TheJulia | That is totally doable with the local boot option setting | 17:58 |
| TheJulia | which I think is _now_ the default instead of netboot | 17:58 |
| dking_desktop | That's great to hear! What is "now"? Train? | 17:58 |
| TheJulia | train just change default_boot_option in ironic.conf to "local" | 17:59 |
| TheJulia | gah, still hasn't merged | 17:59 |
| dking_desktop | Great! We're pretty much all on Train now, so that would be perfect. | 17:59 |
| TheJulia | dtantsur: https://review.opendev.org/#/c/492114/ | 17:59 |
| patchbot | patch 492114 - ironic - Change [deploy]/default_boot_option to local - 32 patch sets | 17:59 |
| dking_desktop | Oh. Well, now that I know the option, I can set it manually. | 18:00 |
| TheJulia | yup | 18:00 |
| dking_desktop | So, for the practicals, then. For this use case, setup Ironic baremetal nodes using the neutron network_interface, and also set default_boot_option to local in ironic.conf. Then, I can perhaps setup the actual network configuration, including bonds and VXLANs, VLANs, etc., either in the glance image? When does Neutron "get out of the way" for that? After the deploy is done would it just set the physical switc | 18:04 |
| dking_desktop | ort and remove any VLANs it setup for the deploy? | 18:04 |
| TheJulia | dtantsur: looks like the change for inspector to default to netboot timed out right at the end of the job | 18:08 |
| dtantsur | :( | 18:09 |
| TheJulia | dking_desktop: yeah, I don't see why not boot wise. You will want to use configuration drive to embed the network configuration data | 18:09 |
| dking_desktop | The configuration drive? | 18:09 |
| TheJulia | neutron would get out of the way after deployment has completed | 18:09 |
| TheJulia | would you be using nova? | 18:09 |
| dking_desktop | Excellent, and then I assume that neutron would re-configure and get back involved if it ever needed to clean or redeploy with Ironic. So, that's what I'm wanting. | 18:10 |
| TheJulia | yes | 18:11 |
| JayF | dking_desktop: to give you extra confidence: we used ironic for rackspace.com/onmetal (a previous job), and had MULTIPLE incidents where we lost control plane access but never had openstack-related issues take our data plane down | 18:11 |
| dking_desktop | Yes, I believe that I would be using Nova. | 18:11 |
| dking_desktop | JayF: That's great to hear! | 18:11 |
| TheJulia | configuration drive is a format for passing instance data/configuration to the machine. If you want to do super special things, you could likely use user-data which nova would embed in the configuration drive which it would send to ironic for the node. | 18:12 |
| TheJulia | dtantsur: I rechecked the job | 18:12 |
| dtantsur | ++ | 18:13 |
| *** ociuhandu has quit IRC | 18:17 | |
| *** ociuhandu has joined #openstack-ironic | 18:17 | |
| TheJulia | its end of cycle, I suspect we'll need weekend rechecks because we're starting to see the weirdest timout failures | 18:18 |
| dking_desktop | Okay, so I'm looking at https://docs.openstack.org/ironic/latest/install/configdrive.html. I wasn't aware of this before, so that's handy. That way, the image could just have some startup or configuration scripts built in which could then pull instance specific information from that mount to help configure the networks. | 18:21 |
| TheJulia | Yup, super useful :) | 18:22 |
| *** ociuhandu has quit IRC | 18:23 | |
| JayF | dking_desktop: `glean` is a software product that already exists that does that, if you don't want to use the more traditional `cloud-init` to read that info and do configs | 18:23 |
| dking_desktop | I see that it mentions passing options to "nova boot", and also mentions a --user-data option. My typical process is to do "baremetal node create..." "baremetal port create ...", manage, inspect, provide, then "openstack server create ..." Where would I modify that to add in these options? Are they part of "openstack server create ..."? | 18:24 |
| TheJulia | dtantsur: I'm not sure I grok your comment on https://review.opendev.org/#/c/714212/8/ironic/common/pxe_utils.py | 18:27 |
| patchbot | patch 714212 - ironic - "dual stack" support for PXE/iPXE - 8 patch sets | 18:27 |
| dtantsur | TheJulia: imagine dhcp_options_for_instance(task, ipxe_enabled, dual_stack=True) | 18:28 |
| dtantsur | that returns you options for both | 18:28 |
| dtantsur | rather than doing two calls and merging on the caller side | 18:28 |
| TheJulia | dking_desktop: no modification beyond that | 18:28 |
| dtantsur | it's mostly a nit | 18:28 |
| TheJulia | dtantsur: so.. make it semi-recursive :) | 18:28 |
| TheJulia | ? | 18:28 |
| * TheJulia ducks | 18:28 | |
| dtantsur | lol | 18:28 |
| dtantsur | not necessary, changing a couple of 'if' may do :) | 18:29 |
| TheJulia | I'm kind of afraid to do that, just based on us breaking this stuff before with older neutron | 18:29 |
| TheJulia | but... | 18:29 |
| dking_desktop | JayF: That looks perfect. I'm gong to look into that. I suppose that's what I can use to build the image once I figure out how to get the config-drive data in there. | 18:29 |
| TheJulia | hjensas|afk tested neutron's interpretation of dual stack dhcp stuff so things seem a lot better now | 18:29 |
| dking_desktop | TheJulia: I meant, where do I put that image and make it, and how do I modify my current commands to know to use it. | 18:30 |
| TheJulia | which image, I'm not sure I understand | 18:30 |
| dking_desktop | I meant the config-drive image, or files. I see that it is mostly called an image in the standalone section. | 18:34 |
| TheJulia | so nova will make one on its own using all available data | 18:34 |
| dtantsur | have a wonderful weekend | 18:36 |
| *** dtantsur is now known as dtantsur|afk | 18:36 | |
| *** rmart04 has quit IRC | 18:38 | |
| TheJulia | o/ | 18:40 |
| dking_desktop | Nice. I'm not sure I understand where it gets that data. Also, when I do "server create ..." I currently add in a network. If I'm wanting all my networking to be handled by my glance image and the config-drive, would I just not use a network option there? | 18:41 |
| *** alexmcleod has quit IRC | 18:43 | |
| openstackgerrit | Merged openstack/networking-generic-switch master: Add netmiko_session_logging https://review.opendev.org/717060 | 18:44 |
| dking_desktop | Again, I apologize for the silly questions. I know that there's a pretty big gap in my understanding somewhere. I'm just trying to figure out where that's at. | 18:45 |
| JayF | OpenStack is a huge distributed system, it's hard to keep the details internalized even if you work on it every day. Nobody is going to judge you based on your questions. | 18:46 |
| openstackgerrit | Tzu-Mainn Chen proposed openstack/ironic master: Add node lessee field https://review.opendev.org/706864 | 18:46 |
| JayF | Now if you don't hang around and answer these same questions for the next person, that's when the judgement comes in :D | 18:47 |
| dking_desktop | lol. Yep. I don't just plan to answer the questions. If I'm having trouble, I expect others are, too, and if everything goes well, that means that there will be a lot of opportunities in writing some documentation. | 18:53 |
| JayF | There is /always/ opportunity for improving docs, thanks :D | 18:57 |
| JayF | dking_desktop: I'm curious if you wanna share the overall problem you're solving, it's always fun to hear about folks' use cases | 18:57 |
| dking_desktop | Sure. And I would love to learn more about your experience. Right now, I have two objectives. My primary objective at the moment is to use Ironic to deploy servers which will in turn be used to run OpenStack. | 19:01 |
| * JayF works on a private cloud deployment of OpenStack, including Ironic, at Verizon Media FKA Yahoo | 19:02 | |
| dking_desktop | We're wanting to create a system that can automatically deploy isolated OpenStack instances for clients. | 19:02 |
| JayF | ah, okay, there was a business that did that once for a while | 19:02 |
| JayF | bluebox maybe? I don't have a great memory | 19:02 |
| dking_desktop | I don't remember the name, but I heard there was another that attempted it for a while, but they eventually learned that the technology wasn't yet mature enough at the time, and had to give up. | 19:03 |
| *** accakks_ has quit IRC | 19:05 | |
| *** k_mouza has joined #openstack-ironic | 19:06 | |
| dking_desktop | From what I've heard, the challenge is that the standard OpenStack use case is to have one large system which is handling a large number of small tenants, where we want to create a system that deploys a large number of small open stacks with a small number of tenants. | 19:08 |
| *** stevebaker has quit IRC | 19:11 | |
| *** k_mouza has quit IRC | 19:21 | |
| TheJulia | dking_desktop: sorry, meeting and food() interupted me. I'm just not familiar with the network option in your case as you've stated that it seems like you want to do native vxlan things. I believe passing it will just create a standard port attachment and I guess in your config you'll likely want to manually create the ports so nova is not doing it for you... in that nova has certian use/interaction | 19:29 |
| TheJulia | expectations like network addresses for instances. | 19:29 |
| TheJulia | JayF: I don't remember what Bluebox did... for provisioning. | 19:30 |
| TheJulia | dking_desktop: and what JayF says! | 19:30 |
| TheJulia | Yes, OpenStack is designed such that you cloudify your physical resources, but there should be exactly nothing that prevents you firing up openstacks with-in those infrastuctures to handle specific things. The key all really comes down to how you interact and abstract along with your processes. TripleO or was it QuintupleO was the closest at these ideas of a deployment with-in a deployment. The downside of | 19:32 |
| TheJulia | tripleo is that it is lots of networking is configured/managed outside of the openstack processes and contexties | 19:32 |
| dking_desktop | That's fair. So, perhaps it would be best to ask the Nova folks on details about how I could pass in config-drive data so that when it's done, it doesn't rely on any actual neutron ports, etc.? | 19:33 |
| dking_desktop | And I did start with TripleO earlier, but it mostly seemed like overkill. | 19:34 |
| TheJulia | so nova largely generates the configuration drive data for you | 19:34 |
| TheJulia | and embeds what neutron has | 19:34 |
| TheJulia | Your case is almost more towards infrastucture management so maybe directly asking ironic to deploy physical nodes might be more ideal | 19:34 |
| TheJulia | because then you could contrain clusters with extra infrastructure knowledge | 19:34 |
| TheJulia | but nova largely allows user data to be passed in, and cloud-init can grok data from that and take additional actions | 19:35 |
| TheJulia | You can also embed files if memory serves | 19:35 |
| dking_desktop | I was thinking about just doing the deploy, but I do want it to be able to monitor the services long term, and I would like for it to more easily be able to redeploy if necessary. I'm just trying to decide on the best way to do that. | 19:40 |
| dking_desktop | And if nova allows user data to be passed in, then it sounds like it might be a great option. I like the idea of using glance images with local boot, so all of that sounds appealing. | 19:41 |
| *** Lucas_Gray has quit IRC | 19:52 | |
| *** Lucas_Gray has joined #openstack-ironic | 19:57 | |
| *** ociuhandu has joined #openstack-ironic | 19:58 | |
| *** Lucas_Gray has quit IRC | 20:13 | |
| *** Lucas_Gray has joined #openstack-ironic | 20:17 | |
| TheJulia | dking_desktop: awesome | 20:23 |
| *** Lucas_Gray has quit IRC | 20:24 | |
| *** ociuhandu has quit IRC | 20:46 | |
| *** ociuhandu has joined #openstack-ironic | 20:47 | |
| *** ociuhandu has quit IRC | 20:52 | |
| *** ociuhandu has joined #openstack-ironic | 21:00 | |
| *** ociuhandu has quit IRC | 21:04 | |
| tzumainn | TheJulia, hi! thanks for the earlier +2s on my patches; just a note that I had to rebase them all due to merge conflicts | 21:32 |
| *** k_mouza has joined #openstack-ironic | 21:43 | |
| *** k_mouza has quit IRC | 21:47 | |
| *** k_mouza has joined #openstack-ironic | 21:48 | |
| *** rloo has quit IRC | 22:04 | |
| openstackgerrit | Sean McGinnis proposed openstack/sushy-cli master: Use unittest.mock instead of third party mock https://review.opendev.org/717408 | 22:08 |
| *** igordc has quit IRC | 22:36 | |
| *** igordc has joined #openstack-ironic | 22:36 | |
| *** tzumainn has quit IRC | 22:37 | |
| *** rh-jelabarre has quit IRC | 22:38 | |
| *** ociuhandu has joined #openstack-ironic | 23:36 | |
| *** ociuhandu has quit IRC | 23:41 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!
