| JayF | jroll: found either a bug in your docs or in the devstack stuff | 00:00 |
|---|---|---|
| adam_g | okay, well all of this will enable similar kinds of sideways migration testing later on, so we should be able to test that easily in grenade later | 00:00 |
| JayF | jroll: nevermind that for now, I'm pretty sure I just fubar'd and didn't download your correct branch | 00:00 |
| JayF | whoops | 00:01 |
| devananda | wow, pennicilin sucks | 00:06 |
| devananda | i mean, it also saved countless lives | 00:06 |
| devananda | possibly mine | 00:06 |
| devananda | but it still sucks | 00:06 |
| * bigjools concurs with devananda ;) | 00:08 | |
| JayF | side effects or something? | 00:09 |
| devananda | ohai! ltns bigjools :) | 00:09 |
| devananda | JayF: instant nausea | 00:09 |
| * JayF has been told he was allergic to pennicillin since he was too young to remember | 00:09 | |
| bigjools | devananda: indeed! | 00:09 |
| devananda | sorta like instant cuponoodles. but less salty. | 00:09 |
| * bigjools currently taking 4 different antibiotics... fml | 00:09 | |
| devananda | :-/ | 00:10 |
| *** hemna has quit IRC | 00:10 | |
| devananda | JayF: I'm going to abandon it shortly, as I never got back around to working on it after atlanta, but | 00:12 |
| NobodyCam | devananda: :( | 00:12 |
| devananda | JayF: you might want to cannibalize the problem description here https://review.openstack.org/#/c/94925/4/specs/juno/better-bootloader-options.rst | 00:12 |
| JayF | Heh. Sure :) Not sure I want to write the spec but I'd rather do that than continue to crush dreams of firmware settings apis in every third spec :P | 00:13 |
| devananda | heh | 00:14 |
| devananda | yea. i think something like this will cover a fair part of the needs | 00:14 |
| devananda | and fit into the mgmt iface that we have now | 00:14 |
| JayF | I actually think it's a decent litmus | 00:14 |
| devananda | which we didn't have when i wrote it | 00:14 |
| JayF | for if something belongs in ironic or not | 00:14 |
| JayF | if you can't abstract it down to a statement about the capabilities of the machine you want | 00:14 |
| JayF | it's probably snowflake-configuration rather than cloud-configuration | 00:14 |
| devananda | yep | 00:14 |
| devananda | especially clear when the justification is "a power user might want to tweak X" | 00:15 |
| JayF | there are no power users of ironic | 00:15 |
| JayF | there are admins of ironic | 00:15 |
| JayF | and users of nova | 00:15 |
| *** adam_g has left #openstack-ironic | 00:15 | |
| devananda | ++ | 00:15 |
| *** adam_g has joined #openstack-ironic | 00:15 | |
| JayF | (arguably, at least, although I know you have ambitions of salt/ansible/ etc) | 00:15 |
| devananda | aye | 00:15 |
| JayF | although I've found 'would this make sense for a vm in nova' to be a decent standard | 00:16 |
| JayF | and I don't see nova exposing exact hypervisor configuration options down to the user in any form except flavors | 00:16 |
| JayF | and capabilities | 00:16 |
| devananda | or put another way - how would nova expose this? | 00:16 |
| devananda | i think everything we do today (and in the forseeable future) breaks down into two categories | 00:17 |
| devananda | - things ironic does, unique to hardware, and never exposed to users. (may be exposed to admins, sometimes) | 00:18 |
| devananda | - things nova exposes to users, which it consumes from ironic in some way | 00:18 |
| * devananda wanders afk for the evening | 00:26 | |
| * NobodyCam thinks a the spec review team should be renamed "SnowBlowers" | 00:26 | |
| NobodyCam | have good night devananda, and try and get *some* rest | 00:27 |
| JayF | -9001 | 00:28 |
| JayF | oh, I get it | 00:28 |
| JayF | still don't love the name :P | 00:28 |
| NobodyCam | lol I just said I thought it, never said it was good | 00:29 |
| NobodyCam | :-p | 00:29 |
| *** NobodyCam has quit IRC | 00:37 | |
| *** NobodyCam has joined #openstack-ironic | 00:37 | |
| NobodyCam | lol there is difference between /quit and /part in irssi | 00:38 |
| NobodyCam | doh | 00:38 |
| *** igordcard has quit IRC | 00:59 | |
| *** lazy_prince has quit IRC | 01:21 | |
| *** killer_prince has joined #openstack-ironic | 01:24 | |
| *** killer_prince is now known as lazy_prince | 01:24 | |
| *** cinerama has joined #openstack-ironic | 01:38 | |
| *** nosnos has joined #openstack-ironic | 01:45 | |
| *** jgrimm has joined #openstack-ironic | 01:45 | |
| *** eghobo has quit IRC | 01:53 | |
| openstackgerrit | Stephanie Miller proposed a change to openstack/ironic: Add posix_ipc to requirements https://review.openstack.org/112746 | 01:59 |
| *** vinbs has joined #openstack-ironic | 02:08 | |
| *** penick has joined #openstack-ironic | 02:12 | |
| *** penick has quit IRC | 02:15 | |
| *** vinbs has quit IRC | 02:27 | |
| *** mitz has joined #openstack-ironic | 02:30 | |
| *** nosnos has quit IRC | 03:04 | |
| *** vinbs has joined #openstack-ironic | 03:06 | |
| *** eghobo has joined #openstack-ironic | 03:36 | |
| *** ramineni has joined #openstack-ironic | 03:48 | |
| *** vinbs has quit IRC | 04:00 | |
| *** romcheg1 has joined #openstack-ironic | 04:05 | |
| *** nosnos has joined #openstack-ironic | 04:08 | |
| *** vinbs has joined #openstack-ironic | 04:14 | |
| *** MattMan has quit IRC | 04:42 | |
| *** MattMan has joined #openstack-ironic | 04:43 | |
| *** romcheg1 has quit IRC | 04:49 | |
| *** rakesh_hs has joined #openstack-ironic | 04:57 | |
| *** Poornima has joined #openstack-ironic | 04:59 | |
| *** SpamapS has quit IRC | 05:26 | |
| *** bmahalakshmi has joined #openstack-ironic | 05:31 | |
| *** SpamapS has joined #openstack-ironic | 05:38 | |
| *** pcrews has quit IRC | 05:43 | |
| *** bvivek has joined #openstack-ironic | 05:46 | |
| *** Mikhail_D_ltp has joined #openstack-ironic | 05:54 | |
| *** dtantsur|afk is now known as dtantsur | 05:55 | |
| dtantsur | Morning Ironic, TGIF! | 05:55 |
| lifeless | o/ | 05:58 |
| *** mrda-away is now known as mrda | 06:00 | |
| *** k4n0 has joined #openstack-ironic | 06:02 | |
| *** Mikhail_D_ltp has quit IRC | 06:06 | |
| *** Mikhail_D_ltp has joined #openstack-ironic | 06:09 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic: Imported Translations from Transifex https://review.openstack.org/112210 | 06:11 |
| *** rakesh_hs has quit IRC | 06:11 | |
| *** rakesh_hs has joined #openstack-ironic | 06:12 | |
| *** pradipta_away is now known as pradipta | 06:35 | |
| dtantsur | Folks, could you remind me how we work with requirements? I'm asking because of https://review.openstack.org/#/c/112746 | 06:40 |
| dtantsur | brb | 06:44 |
| openstackgerrit | A change was merged to openstack/ironic: backport reviewer comments on nova.virt.ironic.patcher https://review.openstack.org/112691 | 06:53 |
| openstackgerrit | Ramakrishnan G proposed a change to openstack/ironic-specs: iLO Virtual Media IPA Deploy Driver https://review.openstack.org/108445 | 06:54 |
| *** athomas has joined #openstack-ironic | 07:23 | |
| *** bvivek has quit IRC | 07:43 | |
| *** rakesh_hs2 has joined #openstack-ironic | 07:47 | |
| *** jistr has joined #openstack-ironic | 07:47 | |
| *** rakesh_hs has quit IRC | 07:47 | |
| *** nosnos has quit IRC | 07:52 | |
| *** nosnos has joined #openstack-ironic | 07:54 | |
| *** ndipanov has joined #openstack-ironic | 08:02 | |
| *** bvivek has joined #openstack-ironic | 08:05 | |
| *** dtantsur is now known as dtantsur|lunch | 08:15 | |
| *** lucasagomes has joined #openstack-ironic | 08:37 | |
| *** rameshg87 has joined #openstack-ironic | 08:52 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Add iPXE support for Ironic https://review.openstack.org/99318 | 09:07 |
| *** rameshg87 has quit IRC | 09:09 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Add create() and destroy() to Port object https://review.openstack.org/112323 | 09:12 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Remove direct calls to dbapi's get_node_by_instance https://review.openstack.org/112595 | 09:12 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Add create() and destroy() to Chassis object https://review.openstack.org/112290 | 09:12 |
| *** bvivek has quit IRC | 09:12 | |
| *** vdrok has joined #openstack-ironic | 09:12 | |
| soren | Uh... I'm wanting to look up a node by its mac address. Is there really no way to do that? | 09:18 |
| soren | I can't even look up a port by MAC. I have to get the list and iterate over it on the client side? | 09:18 |
| soren | ...and then when I get the port, how can I get the corresponding node? | 09:18 |
| soren | Oh. I have to .get() the port again. | 09:22 |
| lucasagomes | soren, GET v1/ports?address= | 09:22 |
| lucasagomes | the node of the correspond port is in the node_uuid field | 09:22 |
| soren | lucasagomes: That doesn't seem to be exposed in the python api. | 09:31 |
| soren | lucasagomes: The searching bit, I mean. | 09:32 |
| lucasagomes | soren, oh, in the client? | 09:32 |
| lucasagomes | hmmm yeah it's possible it's not there :( | 09:32 |
| lucasagomes | soren, mind submitting a patch for it? should be straight forward | 09:32 |
| *** MattMan has quit IRC | 09:36 | |
| *** pelix has joined #openstack-ironic | 09:41 | |
| lucasagomes | jroll, I hit the same problem you did in a fresh devstack install | 09:49 |
| lucasagomes | jroll, something may change on devstack that is making it call cleanup before configure | 09:49 |
| lucasagomes | -.- | 09:49 |
| *** MattMan has joined #openstack-ironic | 09:50 | |
| *** aswadr has joined #openstack-ironic | 09:56 | |
| *** bvivek has joined #openstack-ironic | 10:08 | |
| *** MattMan has quit IRC | 10:09 | |
| *** nosnos has quit IRC | 10:16 | |
| *** nosnos has joined #openstack-ironic | 10:17 | |
| *** killer_prince has joined #openstack-ironic | 10:22 | |
| *** rwsu has quit IRC | 10:23 | |
| *** davidlenwell has quit IRC | 10:23 | |
| *** lazy_prince has quit IRC | 10:23 | |
| *** Shrews has quit IRC | 10:23 | |
| *** killer_prince is now known as lazy_prince | 10:23 | |
| *** Shrews has joined #openstack-ironic | 10:23 | |
| *** davidlenwell has joined #openstack-ironic | 10:25 | |
| *** nosnos has quit IRC | 10:26 | |
| *** rwsu has joined #openstack-ironic | 10:27 | |
| *** MattMan has joined #openstack-ironic | 10:47 | |
| *** foexle has joined #openstack-ironic | 10:59 | |
| *** ramineni has quit IRC | 11:04 | |
| *** vdrok has quit IRC | 11:04 | |
| lucasagomes | dtantsur|lunch, jroll https://review.openstack.org/#/c/99318/ was rebased, conflicted at deploy_utils.py because of the oslo.config patch | 11:05 |
| lucasagomes | can i get a re-approval on that? | 11:05 |
| *** lucasagomes is now known as lucas-hungry | 11:07 | |
| *** bvivek has quit IRC | 11:17 | |
| *** dtantsur|lunch is now known as dtantsur | 11:21 | |
| dtantsur | lucas-hungry, sure, lemme see | 11:21 |
| *** ndipanov has quit IRC | 11:21 | |
| *** ndipanov has joined #openstack-ironic | 11:23 | |
| *** rakesh_hs2 has quit IRC | 11:41 | |
| *** Poornima has quit IRC | 12:03 | |
| *** pradipta is now known as pradipta_away | 12:03 | |
| openstackgerrit | A change was merged to openstack/ironic: Add create() and destroy() to Chassis object https://review.openstack.org/112290 | 12:11 |
| *** ndipanov has quit IRC | 12:13 | |
| *** ndipanov has joined #openstack-ironic | 12:27 | |
| *** Mikhail_D_ltp has quit IRC | 12:28 | |
| *** Mikhail_D_ltp has joined #openstack-ironic | 12:28 | |
| *** lucas-hungry is now known as lucasagomes | 12:31 | |
| *** bmahalakshmi has quit IRC | 12:31 | |
| *** vinbs has quit IRC | 12:32 | |
| *** andreykurilin has quit IRC | 12:33 | |
| *** andreykurilin has joined #openstack-ironic | 12:34 | |
| *** romcheg has quit IRC | 12:34 | |
| *** romcheg has joined #openstack-ironic | 12:34 | |
| *** dhellmann_ is now known as dhellmann | 12:47 | |
| openstackgerrit | A change was merged to stackforge/pyghmi: Force non-numeric for compact sensor records https://review.openstack.org/112301 | 12:52 |
| *** ramineni has joined #openstack-ironic | 12:56 | |
| *** matty_dubs|gone is now known as matty_dubs | 13:01 | |
| *** agordeev has quit IRC | 13:09 | |
| openstackgerrit | Andrey Kurilin proposed a change to openstack/ironic: Use timeutils from one place https://review.openstack.org/112303 | 13:16 |
| lucasagomes | dtantsur, that's pretty weird but jenkins seems to not have pick the approved ipxe patch | 13:21 |
| lucasagomes | it's not in zuul or anywhere | 13:21 |
| dtantsur | lucasagomes, lol wow. let me redo it | 13:21 |
| lucasagomes | dtantsur, yeah I don't know if I haven't seem that before | 13:21 |
| dtantsur | lucasagomes, ok, this attempt seems to succeed | 13:22 |
| lucasagomes | dtantsur, hah thanks | 13:22 |
| *** agordeev has joined #openstack-ironic | 13:24 | |
| *** ramineni1 has joined #openstack-ironic | 13:29 | |
| *** ramineni has quit IRC | 13:31 | |
| *** pcrews has joined #openstack-ironic | 13:33 | |
| *** ramineni1 has quit IRC | 13:36 | |
| *** ramineni has joined #openstack-ironic | 13:37 | |
| jroll | morning y'all | 13:38 |
| *** eghobo has quit IRC | 13:39 | |
| *** ramineni has quit IRC | 13:42 | |
| dtantsur | jroll, morning! | 13:46 |
| jroll | hi dtantsur :) | 13:46 |
| *** ndipanov is now known as ndipanoff | 13:50 | |
| *** russellb is now known as rustlebee | 13:50 | |
| *** jasondotstar has joined #openstack-ironic | 13:52 | |
| *** jasondotstar has quit IRC | 13:54 | |
| *** jasondotstar has joined #openstack-ironic | 13:54 | |
| openstackgerrit | A change was merged to openstack/ironic: Add iPXE support for Ironic https://review.openstack.org/99318 | 13:57 |
| openstackgerrit | A change was merged to openstack/ironic: Add posix_ipc to requirements https://review.openstack.org/112746 | 13:57 |
| dtantsur | lucasagomes, ^^^! | 14:01 |
| lucasagomes | yayayay | 14:02 |
| jroll | \o/ | 14:03 |
| *** chenglch has joined #openstack-ironic | 14:18 | |
| devananda | morning, all | 14:34 |
| dtantsur | devananda, morning | 14:35 |
| NobodyCam | morning | 14:35 |
| openstackgerrit | Ilya Pekelny proposed a change to openstack/ironic: Test migrations with Alembic, using Oslo.db https://review.openstack.org/111984 | 14:36 |
| openstackgerrit | Ilya Pekelny proposed a change to openstack/ironic: Correct `op.drop_constraint` parameters https://review.openstack.org/112896 | 14:36 |
| dtantsur | and you, NobodyCam! where is your TGIF? :) | 14:36 |
| lucasagomes | NobodyCam, devananda morning | 14:36 |
| NobodyCam | morning lucasagomes dtantsur devananda jroll | 14:37 |
| NobodyCam | and TGIF | 14:37 |
| jroll | ^^ morning | 14:37 |
| NobodyCam | :) | 14:37 |
| NobodyCam | so going for me this morning as I for to turn on hte heater lastnight so it darn cold in here this morning | 14:38 |
| NobodyCam | s/so/slow/ | 14:38 |
| *** k4n0 has quit IRC | 14:39 | |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Correct `op.drop_constraint` parameters https://review.openstack.org/112896 | 14:39 |
| dtantsur | devananda, maybe rebase it ^^^ on top of master? it's pretty annoying... | 14:41 |
| devananda | dtantsur: ?? | 14:41 |
| devananda | I just fixed a typo in the commit message | 14:41 |
| dtantsur | devananda, ah sorry, didn't notice. I encountered the bug yesterday and was too glad to see a fix :) | 14:41 |
| *** mdorman has joined #openstack-ironic | 14:42 | |
| *** annegentle is now known as gentlesaurus | 14:43 | |
| dtantsur | ah, and I confused "depends on" and "needed by" >_< | 14:43 |
| dtantsur | seems like it's Friday evening for me :) | 14:44 |
| devananda | hmm | 14:44 |
| devananda | why didn't we hit https://bugs.launchpad.net/ironic/+bug/1354463 before? | 14:44 |
| devananda | are we not running migration tests in the gate again? | 14:44 |
| dtantsur | seems like that... | 14:46 |
| devananda | :-( | 14:47 |
| * devananda rebuilds venv | 14:47 | |
| dtantsur | the problematic change is not very recent btw | 14:47 |
| lucasagomes | yeah we really should start running migration in gate >.< | 14:47 |
| devananda | I thought we were ... | 14:48 |
| devananda | after we switched to oslo.db | 14:48 |
| dtantsur | also, is it desired that ironic-dbsync downgrade raises a NotImplementedError on the last step? | 14:48 |
| devananda | probably not? | 14:48 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Add create() and destroy() to Port object https://review.openstack.org/112323 | 14:49 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Remove direct calls to dbapi's get_node_by_instance https://review.openstack.org/112595 | 14:49 |
| dtantsur | devananda, ironic/db/sqlalchemy/alembic/versions/2581ebaf0cb2_initial_migration.py line #105 | 14:50 |
| lucasagomes | devananda, I remember we blocked oslo.db at the beggining because it had the same problem of running migration tests in parallel | 14:50 |
| dtantsur | "Downgrade from initial migration is not supported" | 14:50 |
| *** mdorman has quit IRC | 14:51 | |
| devananda | lucasagomes: right. I tracked that bug with the oslo.db team. afaict, it was fixed | 14:51 |
| *** mdorman has joined #openstack-ironic | 14:51 | |
| devananda | dtantsur: ahh. right. that seems reasonable actually | 14:51 |
| devananda | dtantsur: but I don't know what other projects do | 14:51 |
| lucasagomes | devananda, hmm maybe do we need to enable something somewhere to get the migration tests running on gate | 14:51 |
| dtantsur | well yes, to me as well, but looks scary :) | 14:51 |
| * lucasagomes not familiar with that part of the code | 14:52 | |
| lucasagomes | lemme dig a bit into it | 14:52 |
| dtantsur | also, if downgrade does not succeed, can we run tests against it? only up to first revision? | 14:52 |
| devananda | dtantsur: so Nova does exactly the same thing | 14:53 |
| dtantsur | I see | 14:53 |
| devananda | nova/db/sqlalchemy/migrate_repo/versions/216_havana.py | 14:53 |
| devananda | 1555 def downgrade(migrate_engine): | 14:53 |
| devananda | 1556 raise NotImplementedError('Downgrade from Havana is unsupported.') | 14:53 |
| *** Srinivas has joined #openstack-ironic | 14:54 | |
| dtantsur | devananda, oh, do they squash all migration from the same release into one big? | 14:55 |
| devananda | yes | 14:55 |
| *** Amarnath has joined #openstack-ironic | 14:56 | |
| devananda | and trim old releases' migrations after n-2 | 14:56 |
| devananda | *after 2 releases | 14:56 |
| dtantsur | interesting practice. are we going to do it? | 14:57 |
| Amarnath | hi, i have few queries about ironic-python-agent | 14:58 |
| devananda | dtantsur: if we have that many migrations at some point, sure | 14:59 |
| jroll | Amarnath: hi! ask away | 15:01 |
| Amarnath | thx Jim, request to have a look at our BP https://blueprints.launchpad.net/ironic/+spec/hba-discovery-driver | 15:01 |
| devananda | dtantsur: how are you hitting that migration bug? | 15:01 |
| devananda | dtantsur: when I run migration tests locally against mysql, it works | 15:02 |
| dtantsur | devananda, try ironic-dbsync --config-file blah-blah downgrade | 15:02 |
| Amarnath | we want to write hba-discovery function in hardware.py | 15:02 |
| Amarnath | is that fine ? | 15:02 |
| jroll | Amarnath: do you know about the new spec process? | 15:02 |
| Amarnath | not sure, pls share some documents | 15:03 |
| jroll | yes, looking | 15:03 |
| jroll | Amarnath: https://wiki.openstack.org/wiki/Ironic/Specs_Process | 15:03 |
| jroll | Amarnath: we've already frozen juno for new specs, it will have to wait for kilo or you will need a freeze exception | 15:04 |
| jroll | Amarnath: this HBA stuff may or may not be in scope of ironic, as well, but it's hard to tell from that blueprint | 15:04 |
| jroll | devananda: ^^ feel free to comment here :) | 15:04 |
| devananda | dtantsur: ooh, i see. so why is that not being tested? :( | 15:04 |
| dtantsur | I don't understand :( our database is still somewhat cryptic to me, at least what concerns migrations | 15:05 |
| jroll | Amarnath: we've changed one thing on that spec process recently as well: http://www.mail-archive.com/openstack-dev@lists.openstack.org/msg31127.html | 15:05 |
| devananda | GheRivero: hi! around? have a moment for oslo db question? | 15:06 |
| devananda | jroll: Amarnath: I question this assertion in the BP: "This information is required to utilize the required HBA for FC SAN boot operations." | 15:07 |
| devananda | Amarnath: certainly you need WWN details. But why should Ironic store model, vendor name, firmware version, etc? | 15:07 |
| devananda | some of this info should be exposed to the scheduler | 15:08 |
| Amarnath | For Ex: if a user want to boot from 8 GB or 16 GB card, controller should be able to know card specs | 15:08 |
| devananda | "controller" ? | 15:08 |
| Amarnath | I meant Ironic | 15:09 |
| devananda | So, no. These should be exposed as a capability to the nova scheduler. | 15:09 |
| devananda | then when a user requests to boot an instance with flavor "has HBA with 8 GB cache" (or what ever), Nova will select the right node. | 15:10 |
| devananda | Ironic doesn't need to use that information. It just needs to expose it to Nova. | 15:10 |
| Amarnath | so that is later, may be another BP | 15:11 |
| devananda | why? | 15:11 |
| Amarnath | but now, we just try to discover HBA info and store in ironic DB in this BP | 15:11 |
| *** vinbs has joined #openstack-ironic | 15:11 | |
| devananda | Amarnath: why? if you're not scheduling based on that, what's the point in storing it? | 15:12 |
| jroll | hrm, but does that actually accomplish anything? | 15:12 |
| devananda | Ironic is not a CMDB ... | 15:12 |
| Amarnath | ironic should be aware of compute node configuration | 15:12 |
| dtantsur | devananda, we need to put "Ironic is not a CMDB" to the main wiki page and to the topic of this channel :) | 15:12 |
| devananda | dtantsur: srsly :) | 15:12 |
| devananda | Amarnath: for what purpose? | 15:13 |
| dtantsur | I have one more Smart Plan(tm): create an OpenStack CMDB Project, make a couple of commits there, send people there and quit :D | 15:14 |
| jroll | lol | 15:14 |
| jroll | dtantsur: I want to call it 'alanis', and tightly integrate with ironic, because it's a list of all things ironic | 15:15 |
| devananda | hah! | 15:15 |
| jroll | :) | 15:15 |
| NobodyCam | :) | 15:15 |
| Amarnath | is wwn information currently available ? | 15:17 |
| Amarnath | if a user want to boot a baremetal compute node using boot from san, wwn is required | 15:17 |
| Amarnath | and this BP is to register this information may be during enrollment time or when there is an API call to the ironic-python-agent | 15:18 |
| dtantsur | Calling it a day. Have a nice weekend! | 15:19 |
| *** dtantsur is now known as dtantsur|afk | 15:19 | |
| devananda | dtantsur|afk: ciao! | 15:19 |
| NobodyCam | have a good weekend dtantsur|afk | 15:20 |
| jroll | Amarnath: sure, but: 1) why is model number etc required, and 2) why shouldn't the operator enter WWN info at enrollment time? | 15:20 |
| jroll | see ya dtantsur|afk :) | 15:20 |
| *** Mikhail_D_ltp has quit IRC | 15:21 | |
| *** chenglch has quit IRC | 15:21 | |
| Amarnath | if there are 100's of baremetal nodes in cloud, difficult for user to maintain list of WWNs and so dynamic discovery helps | 15:21 |
| Amarnath | vendor name, model, driver info etc are required for R&D teams where the use case required specific model of card | 15:22 |
| jroll | CMDBs are for maintaining lists of hardware info, ironic is not a cmdb | 15:23 |
| pquerna | people who have history, already have one, people doing cleanroom new things, might need one. even the people who have one, they tend to be old and horirble. | 15:24 |
| pquerna | they also tended to be... integrated with workflows for getting new servers | 15:24 |
| pquerna | so, mentally, there is the overlap. | 15:24 |
| pquerna | but even then, ironic still has some kind of integration with them, eg, put things in maintance mode, $magic happens. | 15:25 |
| jroll | yes, it's a fine line | 15:26 |
| *** eglynn has joined #openstack-ironic | 15:26 | |
| Amarnath | i agree that ironic is not CMDB for now, but having all compute node specs is essential for cloud providers | 15:27 |
| jroll | agree, but ironic is not the place to store those :| | 15:27 |
| *** vinbs has quit IRC | 15:29 | |
| * jroll bbiab | 15:29 | |
| *** romcheg1 has joined #openstack-ironic | 15:32 | |
| devananda | lucasagomes: around? | 15:33 |
| lucasagomes | devananda, yup | 15:33 |
| *** eghobo has joined #openstack-ironic | 15:34 | |
| *** romcheg1 has quit IRC | 15:37 | |
| *** romcheg1 has joined #openstack-ironic | 15:38 | |
| Amarnath | so, should we store the baremetal compute node specs in nova db | 15:39 |
| NobodyCam | devananda: (and all spec cores) should we have a spec review jam today so we can we are ready for mondays meeting? and final J cut off? | 15:40 |
| devananda | probably | 15:41 |
| devananda | though I think we've only got one or two left that we will get in | 15:41 |
| NobodyCam | ack !!! | 15:42 |
| *** romcheg1 has quit IRC | 15:42 | |
| *** comstud is now known as bearhands | 15:45 | |
| devananda | hm, anyone seen rloo lately? | 15:53 |
| devananda | she needs to lift some -2's for code that we've approved the spec for | 15:54 |
| Shrews | devananda: vacation, iirc | 15:54 |
| openstackgerrit | A change was merged to openstack/ironic: Use timeutils from one place https://review.openstack.org/112303 | 15:54 |
| devananda | Shrews: i see. know when she gets back? | 15:54 |
| Shrews | nope | 15:54 |
| *** romcheg1 has joined #openstack-ironic | 15:57 | |
| devananda | :-/ | 16:00 |
| NobodyCam | devananda: can you override as ptl? | 16:00 |
| devananda | not afaik | 16:00 |
| NobodyCam | want me pop into infra and see they can? | 16:00 |
| devananda | nah | 16:01 |
| devananda | infra-core can | 16:01 |
| devananda | but it's not that urgent | 16:01 |
| NobodyCam | ack | 16:01 |
| devananda | if she's not back in a week or two, then yes | 16:01 |
| *** Mikhail_D_wk has joined #openstack-ironic | 16:06 | |
| *** Amarnath has quit IRC | 16:09 | |
| *** Srinivas has quit IRC | 16:10 | |
| Shrews | adam_g: shouldn't this also set suspend=False ? https://review.openstack.org/109300 | 16:14 |
| *** matty_dubs is now known as matty_dubs|lunch | 16:14 | |
| NobodyCam | brb | 16:22 |
| lucasagomes | yeah haven't see rloo lately :/ | 16:22 |
| * devananda updates meeting agenda for monday | 16:24 | |
| NobodyCam | :) | 16:29 |
| *** Hefeweiz1n has quit IRC | 16:31 | |
| *** aswadr has quit IRC | 16:32 | |
| Shrews | grrr. our ironic-specs unit test doesn't easily allow for template changes | 16:34 |
| Shrews | if you make a change to the template, it will retroactively check already merged specs... which will fail | 16:35 |
| *** MattMan has left #openstack-ironic | 16:36 | |
| NobodyCam | oh thats a new one: "grep: memory exhausted" | 16:36 |
| lucasagomes | NobodyCam, lol, ouch haven't seem that | 16:38 |
| NobodyCam | :-p | 16:38 |
| *** Hefeweizen has joined #openstack-ironic | 16:40 | |
| *** ellenh has joined #openstack-ironic | 16:46 | |
| *** pelix has quit IRC | 16:48 | |
| devananda | yes | 16:56 |
| devananda | Shrews: yes | 16:56 |
| devananda | Shrews: we should not be changing the template except at cycle boundaries | 16:56 |
| devananda | eg, when we open ironic-specs for Kilo, any changes to the template we want to make should be the first thing that lands | 16:57 |
| *** ndipanoff has quit IRC | 16:58 | |
| *** penick has joined #openstack-ironic | 16:58 | |
| lucasagomes | aight I will call it a day | 17:02 |
| lucasagomes | have a good night everybody, enjoy the weekend | 17:02 |
| NobodyCam | have a good weekend lucasagomes | 17:03 |
| *** Chandra has joined #openstack-ironic | 17:03 | |
| *** Mikhail_D_wk has quit IRC | 17:03 | |
| *** athomas has quit IRC | 17:04 | |
| NobodyCam | cd .. | 17:07 |
| NobodyCam | doh | 17:07 |
| *** dvorak has quit IRC | 17:09 | |
| *** jistr has quit IRC | 17:09 | |
| JayF | If no objections in the next 15 minutes or so I'm going to land https://review.openstack.org/#/c/108445 | 17:11 |
| adam_g | Shrews, hmm. yea, i think | 17:12 |
| *** dvorak has joined #openstack-ironic | 17:13 | |
| NobodyCam | JayF: giving it a final read over | 17:14 |
| JayF | NobodyCam: cool | 17:15 |
| *** romcheg1 has quit IRC | 17:19 | |
| *** matty_dubs|lunch is now known as matty_dubs | 17:21 | |
| NobodyCam | JayF: it now has a +2 from /me.. I did not +a :) | 17:22 |
| JayF | NobodyCam: coming in for a landing | 17:23 |
| openstackgerrit | A change was merged to openstack/ironic-specs: iLO Virtual Media IPA Deploy Driver https://review.openstack.org/108445 | 17:25 |
| NobodyCam | JayF: have you reviewed the drac power spec, I think that and the pdu-power are the only ones left on the Ironic Priorities gdoc | 17:25 |
| JayF | I'm not sure I've seen either of those | 17:26 |
| JayF | NobodyCam: I see two DRAC specs, one is h/w discovery, one is drac vendor passthru for raid | 17:26 |
| JayF | both of which are not getting into J because of obvious reasons | 17:27 |
| NobodyCam | ack | 17:27 |
| JayF | and I don't see any pdu power driver specs either | 17:27 |
| JayF | imbw but I think those two landed a while ago? | 17:27 |
| JayF | actually looking at the list of specs, I think we're pretty damn close to done with landing specs for j | 17:27 |
| NobodyCam | yep!!!! | 17:28 |
| jroll | \o/ | 17:29 |
| jroll | now we just need code >:) | 17:29 |
| *** eguz has joined #openstack-ironic | 17:29 | |
| *** lazy_prince has quit IRC | 17:30 | |
| devananda | and reviews! | 17:30 |
| jroll | heh | 17:32 |
| jroll | I meant code to be landed, not proposed :) | 17:32 |
| devananda | i also don't see any more specs up that I wanted us to land | 17:33 |
| devananda | sweet | 17:33 |
| devananda | so, process qusetion | 17:33 |
| devananda | i'll bring it up on monday too, but ya'll are here now, so ... | 17:33 |
| devananda | is there any benefit at this point to -2'ing all the specs? | 17:34 |
| *** eghobo has quit IRC | 17:34 | |
| devananda | or just send an email out early next week saying "specs are closed for J, will announce when they're open for K" | 17:34 |
| openstackgerrit | A change was merged to openstack/ironic: Correct `op.drop_constraint` parameters https://review.openstack.org/112896 | 17:35 |
| NobodyCam | can we bulk move the j specs to k folder? | 17:36 |
| JayF | -1 | 17:36 |
| JayF | I don't think we should do that | 17:36 |
| JayF | because at least a third, if not more, of the specs will need to be massively rewritten | 17:36 |
| JayF | because most of them involve discovery, node updates, auto enrollment, or firmware settings | 17:36 |
| JayF | all of which are probably going to be talked about /a lot/ at the design summit | 17:36 |
| NobodyCam | so -2 them and send email to have authors move/rewrite into k | 17:37 |
| JayF | or move them all into a specs graveyard or something | 17:37 |
| NobodyCam | JayF: I agree with your point | 17:37 |
| * JayF wants workflow -2 like the ML was talking about | 17:37 | |
| NobodyCam | oh thats a instresting idea | 17:37 |
| jroll | ellenh: reviewed 112351, and I think some of the notes might be related to your tempest failures :) | 17:38 |
| jroll | JayF: would workflow -2 be 'abandon'? | 17:38 |
| JayF | in the case of specs, force-abandoning them isn't a bad idea | 17:38 |
| jroll | right | 17:38 |
| jroll | I just missed that thread so I'm curious | 17:38 |
| JayF | but you might workflow -2 a code review, for instance, that doesn't have a spec merged | 17:39 |
| *** killer_prince has joined #openstack-ironic | 17:39 | |
| jroll | I read email through you, you know this | 17:39 |
| *** killer_prince is now known as lazy_prince | 17:39 | |
| jroll | what's wrong with CR -2 for that? | 17:39 |
| JayF | so you can separate the 'quality of code / fittingness of code' from the 'should this merge now' | 17:39 |
| jroll | mmmmmmmm | 17:39 |
| devananda | so we can alraedy force-abandon reviews | 17:40 |
| devananda | W-2 is a proposed change to differentiate process from technical blocking | 17:40 |
| devananda | NobodyCam: we definitely should not bulk move proposed specs to the K folder. that's up to the proposers | 17:41 |
| devananda | also, as JayF pointed out, I think a lot of the currently-proposed specs are out of scope | 17:41 |
| *** eglynn has quit IRC | 17:41 | |
| NobodyCam | ack | 17:41 |
| NobodyCam | yes | 17:41 |
| devananda | but we need to clearly communicate that scope better than we are now | 17:41 |
| devananda | also, specs for K will open /before/ the summit | 17:42 |
| devananda | as design sessions will be linkable to proposed specs | 17:42 |
| NobodyCam | to the new short intro/idea specs | 17:43 |
| NobodyCam | ? | 17:43 |
| NobodyCam | or full detail versions | 17:43 |
| *** penick has quit IRC | 17:44 | |
| *** bmahalakshmi has joined #openstack-ironic | 17:50 | |
| *** romcheg1 has joined #openstack-ironic | 17:51 | |
| jroll | devananda: I don't think -2 with a good comment would be a bad thing | 17:53 |
| JayF | NobodyCam: I'd suggest for some things (like a general 'capabilities' piece being the way to communicate what raid/firmware/etc configs you want via nova flavor) doing the partial spec first would be good, hash out the design at the design summit, then write the rest based on that | 17:54 |
| JayF | hell have whoever writes the notes for the session do it in spec format | 17:54 |
| NobodyCam | oh actually finish the spec at the summit | 17:55 |
| NobodyCam | with everyone in the same room | 17:55 |
| NobodyCam | would be awesome if we could get every to the mid cycle meetup too | 17:56 |
| NobodyCam | gah: s/every/everyone/ | 17:56 |
| jroll | indeed | 17:57 |
| jroll | maybe we can con hp into funding travel for everyone :P | 17:57 |
| * jroll ducks | 17:57 | |
| *** eghobo has joined #openstack-ironic | 17:58 | |
| NobodyCam | lol good luck... write to MEG@..... | 17:58 |
| NobodyCam | lol | 17:58 |
| *** eguz has joined #openstack-ironic | 17:58 | |
| jroll | ha | 18:00 |
| *** penick has joined #openstack-ironic | 18:01 | |
| *** eghobo has quit IRC | 18:02 | |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic: Add docs for agent driver with devstack https://review.openstack.org/112136 | 18:03 |
| *** penick has quit IRC | 18:04 | |
| jroll | JayF: pls to remove your -1 here? :) https://review.openstack.org/#/c/112143/ | 18:04 |
| *** scubacuda has joined #openstack-ironic | 18:05 | |
| devananda | JayF: another thing to consider in our discussions about capabilities -- what if I'mnot using Nova? | 18:07 |
| jroll | does it matter? | 18:08 |
| jroll | you can still query capabilities | 18:08 |
| devananda | actually, you can't query BY capability | 18:08 |
| devananda | you can get the capabilities for a given node | 18:08 |
| jroll | hrm, true | 18:09 |
| devananda | nova-scheduler works by caching a complete list of all available nodes and filtering in python | 18:09 |
| jroll | same with properties though | 18:09 |
| devananda | (yay performance!) | 18:09 |
| devananda | so i think we'll need to stash those in a new DB table and support some form of querying | 18:09 |
| jroll | regardless, if you want to do any sort of scheduling right now, with or without nova, you have to get all nodes | 18:09 |
| jroll | which yes is not ideal | 18:09 |
| devananda | which might create some sanity around the set of capabilities we support, in fact | 18:09 |
| devananda | yep | 18:10 |
| *** eglynn has joined #openstack-ironic | 18:11 | |
| NobodyCam | devananda: that will move ironic one small step closer to being a cmdb no. if we are actually doing the scheduling of a node? | 18:14 |
| *** jasondotstar has quit IRC | 18:15 | |
| NobodyCam | s/no./no?/ | 18:15 |
| devananda | NobodyCam: I don't follow either of those statements | 18:17 |
| NobodyCam | if we allow searching for capabilities from ironic then we are moving closer to actually scheduling which node to deploy to .. which I (kinda) feel is taking us closer to being a cmdb? | 18:19 |
| devananda | exposing an interface by which one can find nodes matching some criteria isn't scheduling | 18:19 |
| devananda | it's searching | 18:19 |
| *** Mikhail_D_wk has joined #openstack-ironic | 18:19 | |
| NobodyCam | ok | 18:20 |
| devananda | for example: GET /v1/nodes?num_cpus=8 | 18:21 |
| devananda | we already support some limited searching, eg. GET /v1/nodes?maintenance_mode=False&associated=True | 18:22 |
| NobodyCam | but what we search on now it not anything in .properties | 18:24 |
| devananda | indeed | 18:24 |
| NobodyCam | s/it/is/ | 18:24 |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic: Add docs for agent driver with devstack https://review.openstack.org/112136 | 18:27 |
| devananda | NobodyCam: my initial point was, to use ironic without nova, we either expose some better search mechanism OR expect the user to do what nova does (pull everything, filter locally) | 18:28 |
| jroll | devananda: then we can make the compute side actually scale maybe | 18:29 |
| jroll | :) | 18:29 |
| devananda | right :) | 18:29 |
| jroll | I like | 18:29 |
| jroll | execute | 18:29 |
| jroll | :) | 18:29 |
| devananda | and secondly, if we were to expose a better search mechanism, that would inform how we model capabilities | 18:29 |
| devananda | as they would need to be searchable | 18:30 |
| devananda | jroll: you'd have to completely rewrite the nova scheduler, i think | 18:30 |
| devananda | nothing about it today understands "take these parameters and QUERY that service over there to get some potential matches" | 18:30 |
| NobodyCam | I would have a easier time with "filter" as opposed to search, though in reallity that would do the same thing. | 18:34 |
| jroll | devananda: ok, sounds good to me | 18:34 |
| jroll | :P | 18:35 |
| devananda | :) | 18:35 |
| jroll | although I think (with my limited knowledge of the scheduler) you could hack it in | 18:35 |
| *** eguz has quit IRC | 18:37 | |
| NobodyCam | brb | 18:51 |
| *** ellenh has quit IRC | 18:54 | |
| *** bmahalakshmi has quit IRC | 19:05 | |
| *** ellenh has joined #openstack-ironic | 19:17 | |
| *** romcheg1 has quit IRC | 19:27 | |
| *** eglynn has quit IRC | 19:32 | |
| openstackgerrit | A change was merged to openstack/ironic: Imported Translations from Transifex https://review.openstack.org/112210 | 19:42 |
| *** eghobo has joined #openstack-ironic | 19:42 | |
| *** tatyana has joined #openstack-ironic | 19:44 | |
| *** tatyana has quit IRC | 19:50 | |
| iron1 | Jroll, I knew that IPA driver landed in Juno. What about IPA agent? Has it landed inJuno? | 19:55 |
| JayF | hey iron1 | 19:56 |
| jroll | iron1: IPA is a separate project that does not have releases like the others | 19:56 |
| iron1 | JayF, hi | 19:56 |
| JayF | IPA has been a part of the Ironic program since the Icehouse mid-cycle meetup | 19:56 |
| jroll | it is released similar to clients, as in whenever we decide | 19:56 |
| JayF | IPA is more-or-less continually released | 19:56 |
| jroll | that said, we have never tagged a release | 19:56 |
| JayF | since we don't have any intention of publishing a pip artifact | 19:56 |
| * jroll wonders if that's beneficial | 19:57 | |
| jroll | ah true | 19:57 |
| JayF | because it doesn't neccessarily make sense | 19:57 |
| JayF | we don't want someone to pip install a rest IPA for wiping their disks :P | 19:57 |
| jroll | lolllll | 19:57 |
| JayF | s/IPA/API/ | 19:57 |
| JayF | oh god | 19:57 |
| JayF | iron1: If you're curious about experimenting with IPA, I'd suggest looking at the reviews in progress for devstack support | 19:58 |
| JayF | summarized here: http://lists.openstack.org/pipermail/openstack-dev/2014-August/042357.html | 19:58 |
| iron1 | JayF: ty. | 19:58 |
| JayF | iron1: if you have any trouble getting the stuff running, let us know in here. Also if you have any general questions about Ironic+IPA once you get it checked out, please ask :) | 19:59 |
| *** penick has joined #openstack-ironic | 20:00 | |
| iron1 | JayF: sure. Thanks! | 20:07 |
| jroll | NobodyCam: did you want to abandon 107778? | 20:12 |
| *** coolsvap has quit IRC | 20:16 | |
| *** openstackgerrit has quit IRC | 20:20 | |
| NobodyCam | humm | 20:23 |
| *** coolsvap has joined #openstack-ironic | 20:23 | |
| NobodyCam | thought I had :-p | 20:24 |
| NobodyCam | have done so now | 20:24 |
| NobodyCam | TY jroll | 20:24 |
| *** matty_dubs is now known as matty_dubs|gone | 20:27 | |
| jroll | NobodyCam: ha no problem :) | 20:36 |
| jroll | devananda, NobodyCam, there's a lot of open questions here, but wdyt about this spec for juno? https://review.openstack.org/#/c/102405/ | 20:47 |
| jroll | JayF: ^^ biased opinions welcome :) | 20:47 |
| jroll | (joking) | 20:47 |
| jroll | also, we should push this through so we can get some nova improvements in while those patches are still in review: https://review.openstack.org/#/c/112610/ | 20:48 |
| jroll | cores ^^ | 20:48 |
| JayF | I mean, we have the code written | 20:48 |
| JayF | we've gotten general agreement on this being an OK thing to do | 20:48 |
| JayF | I wouldn't mind landing it for j-3, but obviously it's not an 'official' priority | 20:48 |
| jroll | JayF: meh, there's way more to consider there unless you punt everything to the operator | 20:49 |
| JayF | TBH we should probably get tempest checks running for IPA and Ironic :( | 20:49 |
| jroll | we have the basic "if node is on, leave it on" | 20:49 |
| jroll | E_WAITING_ON_REVIEWS | 20:49 |
| JayF | I'll review that when I get done with what I'm doing atm | 20:49 |
| *** eghobo has quit IRC | 20:51 | |
| devananda | jroll: https://review.openstack.org/#/c/112610/4/ironicclient/v1/node_shell.py | 20:51 |
| devananda | if args.detail: | 20:51 |
| devananda | what if the option was 'false' ? | 20:52 |
| jroll | oh shit, that's not converted to bool, is it | 20:52 |
| devananda | i dont think so, no | 20:52 |
| devananda | nodes = cc.node.list(**params) | 20:52 |
| jroll | sec, will fix | 20:52 |
| devananda | passes the values to ironic's API which convers them to bools | 20:53 |
| *** Chandra has quit IRC | 20:53 | |
| jroll | devananda: vvv | 20:54 |
| devananda | ? | 20:54 |
| jroll | oh cmon bot | 20:54 |
| jroll | it's dead | 20:54 |
| jroll | https://review.openstack.org/#/c/112610 | 20:54 |
| jroll | updated | 20:54 |
| jroll | but we have no bot :( | 20:54 |
| devananda | ugh | 20:54 |
| jroll | devananda: also good news, apparently all of devstack-core is on vacation \o/ | 20:56 |
| devananda | haha | 20:56 |
| jroll | except dtroyer | 20:56 |
| *** penick has quit IRC | 20:58 | |
| NobodyCam | nice | 21:02 |
| devananda | re: long running deploy spec, I'd prefer to punt it | 21:02 |
| devananda | not only a lot of unresolved qusetions about api and/or conf changes | 21:03 |
| jroll | boooooooo (totally understand) | 21:03 |
| devananda | but a breaking change to the driver API | 21:03 |
| devananda | that, at this point, probably wouldn't land until the last moments of Juno3 | 21:03 |
| jroll | yeah | 21:04 |
| *** penick has joined #openstack-ironic | 21:04 | |
| jroll | well, it could be backwards compat | 21:04 |
| devananda | i dont think so | 21:04 |
| jroll | make it an optional param | 21:05 |
| devananda | take a driver that works today. move it out of tree. make this change. call that driver.validate() with an extra param -- it'll barf | 21:05 |
| jroll | method=None | 21:05 |
| devananda | no, because the driver won't have been updated | 21:05 |
| jroll | if method in (None, 'deploy'): # do deploy validation | 21:05 |
| jroll | right | 21:05 |
| jroll | but like | 21:05 |
| jroll | if that driver just does validation and doesn't check the method | 21:05 |
| jroll | then it does the same thing as it does today | 21:06 |
| jroll | OH | 21:06 |
| jroll | never mind, I'm an idiot | 21:06 |
| devananda | i mean python with barf | 21:06 |
| jroll | ignore me | 21:06 |
| devananda | wrong number of params to function cal | 21:06 |
| jroll | yeah | 21:06 |
| jroll | I honestly don't think it's a huge deal to break that in juno 3 (e.g. before feature freeze) but that's just me | 21:07 |
| *** eghobo has joined #openstack-ironic | 21:08 | |
| NobodyCam | jroll 112610 lgtm waiting on mr J | 21:08 |
| jroll | woohoo | 21:09 |
| jroll | thanks | 21:09 |
| NobodyCam | mr j still running ... brb | 21:16 |
| jroll | yeah no rush, I'm going to start hacking up the actual patch | 21:16 |
| JayF | jroll: JoshNang: we should all probably start drafting our agent-feature specs for k :P | 21:17 |
| jroll | oh gawd | 21:17 |
| jroll | I think they mostly exist already :) | 21:17 |
| JayF | by like k2, we should be running ironic master in onmetal, amirite? | 21:17 |
| jroll | other than network isolation things | 21:17 |
| jroll | yes | 21:17 |
| jroll | pls | 21:17 |
| jroll | make everyone's job easier | 21:17 |
| JoshNang | JayF: jroll: image pre caching and support for torrents? | 21:18 |
| JoshNang | and +100 to master | 21:18 |
| JayF | JoshNang: I'm thinking more what we're currently running | 21:18 |
| JayF | JoshNang: i.e. long running agents, decom, the perf fixes we haven't fully upstreamed | 21:19 |
| jroll | JoshNang: those aren't ironic :) | 21:19 |
| jroll | JoshNang: those are nova and glance things, specifically | 21:19 |
| jroll | maybe s/glance/swift | 21:19 |
| * jroll keeps working on this perf thing | 21:19 | |
| JoshNang | i think pre caching is going to need some ironic involvement. but yeah mostly scheduler and a pre caching scheduler | 21:20 |
| JayF | jroll: long running agents == if you could setup the gate to run that way, I bet tempest woudl be faster | 21:20 |
| jroll | JoshNang: it will, yeah, you're right | 21:20 |
| jroll | JayF: notrly, because it provisions those vms on the fly afaik | 21:20 |
| jroll | although it might do that ahead of time | 21:20 |
| JayF | okie | 21:20 |
| jroll | unclear | 21:20 |
| NobodyCam | for a week in which I have rested so much ... I am wiped out this friday | 21:31 |
| jroll | yay for contagious diseases | 21:33 |
| ellenh | jroll: stylistic question - should i docstring abstract methods, or their implementations, or both? | 21:39 |
| *** marzif_ has joined #openstack-ironic | 21:43 | |
| jroll | ellenh: generally both | 21:44 |
| ellenh | cool, thanks | 21:47 |
| *** eghobo has quit IRC | 21:53 | |
| NobodyCam | jroll: +2'd 112610 | 22:01 |
| jroll | yay! | 22:01 |
| NobodyCam | see comment as to why no +a | 22:01 |
| NobodyCam | :-p | 22:01 |
| jroll | and I'm running tests, then submitting the patch it depends on :) | 22:02 |
| jroll | oh | 22:02 |
| jroll | wat | 22:02 |
| jroll | oh well | 22:04 |
| NobodyCam | :( | 22:04 |
| jroll | lucasagomes: when you're around tomorrow please check out https://review.openstack.org/#/c/112610/ :) | 22:04 |
| NobodyCam | it will land monday mornign | 22:04 |
| jroll | yeah | 22:04 |
| jroll | then we can cut a new client | 22:04 |
| jroll | then I have a couple ironic patches to land :) | 22:04 |
| NobodyCam | :) | 22:04 |
| jroll | too many patches in my queue, need to land them all | 22:05 |
| jroll | like this one https://review.openstack.org/#/c/108545/ | 22:06 |
| *** penick has quit IRC | 22:07 | |
| jroll | and this one https://review.openstack.org/#/c/113036/ | 22:09 |
| jroll | and this one https://review.openstack.org/#/c/105590/ | 22:09 |
| jroll | and when devstack people are back I have like 5 more | 22:10 |
| jroll | so many open patches :( | 22:10 |
| NobodyCam | 108545 & 113036 are to nova driver and would cause deva to stab us if we landed them | 22:13 |
| NobodyCam | 105590 looks like it may need a rev or two more | 22:14 |
| *** penick has joined #openstack-ironic | 22:14 | |
| NobodyCam | :( | 22:14 |
| NobodyCam | thou I like the cache idea | 22:15 |
| NobodyCam | would love a way to validate the cache with a single checksum check | 22:16 |
| NobodyCam | not sure how it would work just thought it would be neet | 22:16 |
| NobodyCam | what did we come up with reguard to soft powering off nodes on rebuild?? anything? | 22:28 |
| jroll | NobodyCam: deva *wants* them to land because it's a huge perf improvement | 22:28 |
| jroll | NobodyCam: as in, it was unusable in our production environment without those patches | 22:28 |
| jroll | NobodyCam: 105590 is good to go except that we need to fix the client bug, which landed recently, just needs a release | 22:29 |
| NobodyCam | jroll: which your cache patches or the soft power offf | 22:29 |
| * NobodyCam assumes cache | 22:29 | |
| jroll | nova patches | 22:29 |
| jroll | yeah | 22:29 |
| jroll | on ~1000 nodes everything grinds to a halt without that cache | 22:30 |
| jroll | and even with that cache it's still kinda slow :| | 22:30 |
| NobodyCam | :( | 22:31 |
| NobodyCam | line 448 of driver.py on 108545, why not add it to the cache too? | 22:33 |
| jroll | the cache gets cleared every run of update_available_resource | 22:34 |
| jroll | so honestly that case should almost never happen | 22:35 |
| jroll | maybe never | 22:35 |
| jroll | but anyhow it won't look it up until the next run, so | 22:35 |
| jroll | it wouldn't help | 22:35 |
| NobodyCam | ack | 22:35 |
| jroll | s/look it up/look it up again/ | 22:35 |
| NobodyCam | jroll: got a second to take a look at :https://review.openstack.org/#/c/107864 <- disreguard commit message and failed checks, please look more at the idea of what it is doing | 22:36 |
| jroll | NobodyCam: should I ignore "THIS CHANGE SHOULD PROBABLY NEVER LAND." :P | 22:37 |
| devananda | jroll: I what? | 22:37 |
| NobodyCam | lol yes please. | 22:37 |
| jroll | devananda: want these nova changes: https://review.openstack.org/#/c/108545/ and its sibling | 22:37 |
| NobodyCam | devananda: ok also can you take a look at https://review.openstack.org/#/c/107864 | 22:37 |
| jroll | the /nodes/detail thing and not doing a bunch of node.get | 22:37 |
| devananda | jroll: right. I want those AND i'm going to stab you | 22:38 |
| jroll | right :) | 22:38 |
| NobodyCam | lol | 22:38 |
| JayF | Are you sure it's not mrda whose going to be doing the stabbing? | 22:39 |
| jroll | so sorry I ran your code with a little scale :P | 22:39 |
| devananda | jroll: that was very rude of you ;) | 22:39 |
| *** openstackgerrit has joined #openstack-ironic | 22:39 | |
| jroll | haha | 22:39 |
| NobodyCam | devananda: http://www.amazon.com/s/?ie=UTF8&keywords=throwing+knive&tag=googhydr-20&index=aps&hvadid=32823487709&hvpos=1t1&hvexid=&hvnetw=g&hvrand=17509291247490150068&hvpone=&hvptwo=&hvqmt=b&hvdev=c&ref=pd_sl_86glpxos1t_b | 22:40 |
| NobodyCam | lol | 22:40 |
| jroll | NobodyCam: it seems I've looked at this soft thing before... seems mostly fine | 22:41 |
| NobodyCam | so your ok if I where to clean that patch up? maybe change the commit message | 22:42 |
| jroll | NobodyCam: it might be nice to have a way to hard reboot if the soft reboot fails for some reason | 22:42 |
| jroll | yeah, I'm fine with it | 22:42 |
| devananda | NobodyCam: http://www.coldsteel.com/Product/TANTO.20_SERIES/TANTO_SERIES.aspx | 22:42 |
| jroll | NobodyCam: I like it better than your original patch with the "try soft first" method | 22:42 |
| jroll | oh god this is gonna hurt | 22:43 |
| NobodyCam | oh that is much nicer...hehehehe | 22:43 |
| *** jdonalds has joined #openstack-ironic | 22:44 | |
| NobodyCam | devananda: is the idea of Ng's soft power off ok to move forward with? if so I'll clean up that patch | 22:45 |
| devananda | if i had a home town, cold steel is in it | 22:46 |
| devananda | NobodyCam: reviewing it. several -1's. probably a -2 coming up | 22:50 |
| *** jdonalds has quit IRC | 22:50 | |
| jroll | :| why -2? (not that I care about rebuild :P ) | 22:51 |
| NobodyCam | it needs alot of clean up | 22:51 |
| devananda | NobodyCam: comments posted | 23:00 |
| devananda | NobodyCam: most notably, see https://review.openstack.org/#/c/107864/2/ironic/drivers/modules/pxe.py | 23:00 |
| jroll | ouch | 23:02 |
| devananda | so the whole premise of "use SOFT OFF for rebuild()" is flawed, IMO | 23:03 |
| jroll | the whole premise of "rebuild" is flawed, IMO | 23:03 |
| jroll | :) | 23:03 |
| NobodyCam | i agree that I think any service who's data can not with stand a hardpower off should be shut down by the orchestration layer | 23:05 |
| NobodyCam | but I can also kinda understand the desire to attempt a soft power off | 23:05 |
| devananda | it's the wrong layer | 23:05 |
| NobodyCam | it's = we're (ironic_ | 23:06 |
| devananda | this is a problem tripleo's running into all over the place -- it's essentially a CMS | 23:06 |
| NobodyCam | ) | 23:06 |
| JayF | I honestly never want a soft power off via ironic api | 23:06 |
| devananda | JayF: thank you | 23:06 |
| JayF | anything that could prevent or delay a node from powering off when I tell it to turn off | 23:06 |
| *** penick has quit IRC | 23:07 | |
| devananda | a CMS that is restarting a host after, say, a kernel upgrade should do it via local commands | 23:07 |
| JayF | would make me unhappy :) | 23:07 |
| devananda | if that FAILS, it should invoke ironic | 23:07 |
| JayF | yeah, exactly | 23:07 |
| JayF | even in the agent decom stuff we're using, we've mocked out ways to have the agent soft-reboot itself in band in some cases | 23:07 |
| JayF | (we aren't using that yet, because it's also kinda gross, but that's how strongly I feel that soft power offs should only be done in-band) | 23:07 |
| devananda | this proposal is mixing those two layers -- a CMS wants Ironic to perform an upgrade of the node, but that CMS (tripleo) is refusing to reach into the instance, shut down services, and power it off | 23:08 |
| NobodyCam | JayF: (unless your using ilo-pxe driver) | 23:08 |
| *** f13o_ has joined #openstack-ironic | 23:08 | |
| devananda | tripleo alraedy has means to restart services on an instance (os-refresh-config) | 23:08 |
| JayF | NobodyCam: that's the thing I hated most about that spec, and why I want ilo-ipa to get more love than ilo-iscsi | 23:09 |
| NobodyCam | JayF: it has to :) | 23:09 |
| JayF | devananda: would you accept a spec for K to rename pxe driver to iscsi driver? Not sure I want to do all the things that'd have to be done, but I reallly hate that it's called pxe :( | 23:09 |
| JayF | especially now that you can ipxe with the pxe driver | 23:09 |
| NobodyCam | devananda: JayF: jroll: is that out stand : we're the wrong layer | 23:09 |
| JayF | plus when we were onboarding to Ironic as a project.... that's just a super confusing name | 23:09 |
| JayF | I agree Ironic should not be in the business of worrying about soft-powering off nodes | 23:10 |
| NobodyCam | s/out/our/ | 23:10 |
| JayF | I don't mind if we added an api endpoint to do it | 23:10 |
| devananda | JayF: i would like to see the current driver split into boot and deploy | 23:10 |
| JayF | but I do mind if we made normal power off/on calls use it by default | 23:10 |
| devananda | s/driver/interface/ | 23:10 |
| JayF | devananda: so we'd have like ipxe_agent_ipmitool | 23:10 |
| JayF | devananda: or pxe_iscsi_ipminative | 23:10 |
| JayF | for drivers? | 23:11 |
| devananda | JayF: we can also give the combinations more interesting names | 23:11 |
| devananda | like "unicorn" and "zebra" | 23:11 |
| devananda | :) | 23:11 |
| JayF | plus that'd be a good time to introduce experimental tags on driver | 23:11 |
| JayF | ugh oh please no | 23:11 |
| jroll | jroll_superdriver = ipxe + agent + ipmitool | 23:11 |
| jroll | jroll_superdriver = cloud_config + agent + ipmitool | 23:11 |
| JayF | I hate cutesy names. I actually like that we name things based on what they do | 23:11 |
| jroll | more like | 23:11 |
| jroll | so | 23:11 |
| jroll | something to keep in mind | 23:11 |
| NobodyCam | oh no... you just know we'll end up with the "pinkSpottedUnicorn" driver | 23:12 |
| jroll | is that today drivers have a limit of 15 chars in the name | 23:12 |
| devananda | right | 23:12 |
| devananda | though we can change that column def easily enough | 23:12 |
| jroll | noooooooo | 23:12 |
| devananda | ? | 23:12 |
| jroll | EHATE migrations | 23:12 |
| devananda | hah | 23:12 |
| jroll | brings back bad memories | 23:12 |
| jroll | in reality, should be fine | 23:12 |
| * jroll actually just hates migrations on mysql | 23:13 | |
| *** mdorman has quit IRC | 23:25 | |
| *** foexle has quit IRC | 23:26 | |
| openstackgerrit | Dan Prince proposed a change to openstack/ironic: Add method for deallocating networks on reschedule https://review.openstack.org/113047 | 23:39 |
| *** f13o_ has quit IRC | 23:44 | |
| jroll | huh | 23:48 |
| *** aswadr has joined #openstack-ironic | 23:55 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!