| devananda | jroll: so, just to throw this out there ... | 00:13 |
|---|---|---|
| * jroll listens | 00:13 | |
| devananda | jroll: I actually want the hash ring to be cached in the API service.. It just needs to be invalidate smartly. | 00:13 |
| devananda | jroll: this fix is fine for Juno as it keeps the behavior that alraedy existed in Juno | 00:14 |
| devananda | jroll: but what if it just used a time-based invalidation? | 00:14 |
| jroll | devananda: hmm, I'd rather just invalidate when a conductor joins the cluster, to be honest | 00:15 |
| devananda | call reset only if it wasn't called in the last N seconds | 00:15 |
| jroll | (or leaves) | 00:15 |
| devananda | well, sure | 00:15 |
| devananda | but how does an API know that? | 00:15 |
| jroll | right | 00:15 |
| jroll | naively, could add an API endpoint | 00:15 |
| jroll | or cache the driver list from the db and update when that changes | 00:15 |
| devananda | I tossed a WIP up a while back taht used an RPC broadcast for that | 00:15 |
| jroll | or that | 00:15 |
| devananda | but I dropped the ball on adding an RPC receiver to the API side | 00:15 |
| devananda | and it turned stale | 00:15 |
| devananda | anything based on a cache of the db will require polling the db on every API request, which is teh same barrel we're i nnow | 00:16 |
| jroll | yeah | 00:16 |
| jroll | true | 00:16 |
| devananda | either we do time-based refresh, or we use a notification AND time based refresh | 00:16 |
| devananda | because notifications are lossy | 00:16 |
| devananda | (well. they shouldn't be, but for the sake of argument ..) | 00:17 |
| jroll | right | 00:17 |
| jroll | they will be at some point at some scale | 00:17 |
| devananda | right | 00:17 |
| *** zz_jgrimm is now known as jgrimm | 00:17 | |
| *** rloo has quit IRC | 00:17 | |
| devananda | and having abberant routing on one aPI out of the pool just because of a network blip is hell to troubleshoot | 00:17 |
| devananda | i for one don't want to support that | 00:17 |
| jroll | oh yes | 00:18 |
| devananda | whereas knowing that, when a conductor gets added, things might route odly for N seconds -- this is trivial to handle in ops | 00:18 |
| *** Haomeng has quit IRC | 00:18 | |
| jroll | right, that's completely fine | 00:19 |
| *** Haomeng has joined #openstack-ironic | 00:20 | |
| jroll | devananda: we should do something better, I don't have the brainpower to come up with that tonight, so :) | 00:22 |
| *** bradjones has quit IRC | 00:22 | |
| *** bradjones has joined #openstack-ironic | 00:24 | |
| devananda | jroll: great :) neither do i | 00:24 |
| devananda | jroll: also, cheers for the RC fix | 00:24 |
| *** todd_dsm has joined #openstack-ironic | 00:25 | |
| jroll | devananda: np, would love a review tonight so I feel comfortable shipping it :P | 00:25 |
| jroll | (if you have the bandwidth) | 00:25 |
| devananda | updating bug and reviewing now | 00:28 |
| jroll | nice, thanks | 00:28 |
| JayF | jroll: I know it sucks you had to fix this, but thinking about how much this would be awful from an operation standpoint, I'm REALLY glad it's going to be fixed for our release :) | 00:33 |
| jroll | JayF: meh, it doesn't suck that I had to fix this, tis fine | 00:34 |
| *** lazy_prince has quit IRC | 00:36 | |
| *** killer_prince has joined #openstack-ironic | 00:37 | |
| *** killer_prince is now known as lazy_prince | 00:37 | |
| devananda | I'd love to see better functional tests in our own tree to possibly spot things like this | 00:37 |
| *** harlowja_ has quit IRC | 00:40 | |
| *** harlowja has joined #openstack-ironic | 00:41 | |
| *** krtaylor has quit IRC | 00:42 | |
| *** achanda has quit IRC | 00:45 | |
| *** todd_dsm has quit IRC | 01:00 | |
| * devananda finds food, returns tomorrow | 01:01 | |
| *** todd_dsm has joined #openstack-ironic | 01:01 | |
| *** dlaube has quit IRC | 01:04 | |
| *** krtaylor has joined #openstack-ironic | 01:05 | |
| *** ChuckC has quit IRC | 01:12 | |
| *** lazy_prince is now known as killer_prince | 01:25 | |
| *** marcoemorais has quit IRC | 01:25 | |
| *** corvus has quit IRC | 01:36 | |
| *** corvus has joined #openstack-ironic | 01:37 | |
| *** nosnos has joined #openstack-ironic | 01:51 | |
| *** ChuckC has joined #openstack-ironic | 02:00 | |
| *** todd_dsm has quit IRC | 02:01 | |
| *** todd_dsm has joined #openstack-ironic | 02:03 | |
| openstackgerrit | A change was merged to openstack/ironic: Clear hash ring cache in get_topic_for* https://review.openstack.org/126751 | 02:09 |
| *** todd_dsm has quit IRC | 02:15 | |
| *** jgrimm is now known as zz_jgrimm | 02:41 | |
| *** ramineni has joined #openstack-ironic | 02:43 | |
| *** harlowja is now known as harlowja_away | 02:44 | |
| *** achanda has joined #openstack-ironic | 03:02 | |
| *** pcrews has quit IRC | 03:04 | |
| *** ryanpetrello has quit IRC | 03:14 | |
| *** ryanpetrello has joined #openstack-ironic | 03:15 | |
| *** nosnos has quit IRC | 03:30 | |
| *** todd_dsm has joined #openstack-ironic | 03:33 | |
| *** rushiagr_away is now known as rushiagr | 03:37 | |
| *** rushiagr is now known as rushiagr_away | 03:51 | |
| *** Hefeweizen has quit IRC | 03:53 | |
| *** achanda_ has joined #openstack-ironic | 03:58 | |
| *** achanda has quit IRC | 04:01 | |
| *** vinbs has joined #openstack-ironic | 04:02 | |
| *** Poornima has joined #openstack-ironic | 04:03 | |
| *** lazy_prince has joined #openstack-ironic | 04:05 | |
| *** marcoemorais has joined #openstack-ironic | 04:14 | |
| *** marcoemorais1 has joined #openstack-ironic | 04:16 | |
| *** lazy_prince has quit IRC | 04:16 | |
| *** killer_prince is now known as lazy_prince | 04:18 | |
| *** nosnos has joined #openstack-ironic | 04:19 | |
| *** marcoemorais has quit IRC | 04:19 | |
| *** rushiagr_away is now known as rushiagr | 04:36 | |
| *** marcoemorais1 has quit IRC | 04:38 | |
| *** marcoemorais has joined #openstack-ironic | 04:46 | |
| *** todd_dsm has quit IRC | 04:48 | |
| *** mikal has quit IRC | 04:49 | |
| *** pensu has joined #openstack-ironic | 04:50 | |
| *** todd_dsm has joined #openstack-ironic | 05:11 | |
| *** rakesh_hs has joined #openstack-ironic | 05:11 | |
| *** todd_dsm has quit IRC | 05:13 | |
| *** Kui has quit IRC | 05:13 | |
| *** Kui has joined #openstack-ironic | 05:16 | |
| *** pradipta_away is now known as pradipta | 05:20 | |
| *** lazy_prince has quit IRC | 05:28 | |
| *** HenryG_ has joined #openstack-ironic | 05:29 | |
| *** HenryG has quit IRC | 05:30 | |
| *** achanda_ has quit IRC | 05:33 | |
| *** rakesh_hs has quit IRC | 05:35 | |
| *** rakesh_hs has joined #openstack-ironic | 05:36 | |
| *** pcrews has joined #openstack-ironic | 05:44 | |
| *** andreykurilin_ has joined #openstack-ironic | 05:44 | |
| *** loki184 has joined #openstack-ironic | 05:45 | |
| *** killer_prince has joined #openstack-ironic | 05:48 | |
| *** killer_prince is now known as lazy_prince | 05:49 | |
| *** pcrews has quit IRC | 05:54 | |
| *** Haomeng has quit IRC | 06:05 | |
| *** lazy_prince2 has joined #openstack-ironic | 06:05 | |
| *** pensu has quit IRC | 06:09 | |
| *** coolsvap|afk has quit IRC | 06:10 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic: Imported Translations from Transifex https://review.openstack.org/126811 | 06:10 |
| *** coolsvap has joined #openstack-ironic | 06:18 | |
| *** andreykurilin_ has quit IRC | 06:19 | |
| *** andreykurilin_ has joined #openstack-ironic | 06:20 | |
| *** lazy_prince2 has quit IRC | 06:21 | |
| *** lazy_prince has quit IRC | 06:23 | |
| *** lazy_prince has joined #openstack-ironic | 06:23 | |
| *** pensu has joined #openstack-ironic | 06:27 | |
| *** pensu1 has joined #openstack-ironic | 06:32 | |
| *** pensu has quit IRC | 06:33 | |
| *** shakamunyi has joined #openstack-ironic | 06:35 | |
| *** JayF has quit IRC | 06:47 | |
| *** andreykurilin_ has quit IRC | 06:52 | |
| *** ndipanov has joined #openstack-ironic | 07:00 | |
| *** ifarkas has joined #openstack-ironic | 07:01 | |
| *** andreykurilin_ has joined #openstack-ironic | 07:02 | |
| *** k4n0 has joined #openstack-ironic | 07:08 | |
| *** ifarkas has quit IRC | 07:13 | |
| *** JayF has joined #openstack-ironic | 07:13 | |
| *** andreykurilin_ has quit IRC | 07:13 | |
| *** lazy_prince has quit IRC | 07:17 | |
| *** shakamunyi has quit IRC | 07:18 | |
| *** andreykurilin_ has joined #openstack-ironic | 07:18 | |
| *** ifarkas has joined #openstack-ironic | 07:18 | |
| *** shakamunyi has joined #openstack-ironic | 07:19 | |
| *** romcheg has joined #openstack-ironic | 07:21 | |
| *** romcheg has quit IRC | 07:22 | |
| *** athomas has joined #openstack-ironic | 07:25 | |
| *** Haomeng has joined #openstack-ironic | 07:34 | |
| *** romcheg has joined #openstack-ironic | 07:36 | |
| *** Krast has joined #openstack-ironic | 07:38 | |
| *** dtantsur|afk is now known as dtantsur | 07:42 | |
| dtantsur | Morning Ironic | 07:42 |
| Krast | afternoon now , I am in ShangHai | 07:43 |
| Krast | :) | 07:43 |
| yuriyz | morning Ironic dtantsur Krast | 07:49 |
| *** ndipanov has quit IRC | 07:49 | |
| *** andreykurilin_ has quit IRC | 07:50 | |
| GheRivero | morning @ironic | 07:55 |
| romcheg | Morning GheRivero! | 07:55 |
| romcheg | and yuriyz, dtantsur, Krast and everyone else too :) | 07:55 |
| Krast | where are you？ | 07:56 |
| *** viktors|afk is now known as viktors | 07:56 | |
| romcheg | me? | 07:56 |
| romcheg | I'm in Poland | 07:57 |
| viktors | morning Ironic | 07:57 |
| dtantsur | morning, GheRivero, romcheg, yuriyz , Krast, viktors :) | 07:57 |
| romcheg | Oh, I've seen that you're in ShangHai, good afternoon then :) | 07:57 |
| dtantsur | so many of us this morning :) | 07:57 |
| Krast | morning <romcheg> :) | 07:58 |
| *** ifarkas has quit IRC | 07:58 | |
| romcheg | Another problem the Diskworld would easily solve! | 07:59 |
| takadayuiko | Hi, Krast, yuriyz, GheRivero, romcheg, dtantsur, | 07:59 |
| dtantsur | takadayuiko, hi! | 08:00 |
| *** jcoufal has joined #openstack-ironic | 08:03 | |
| *** lucasagomes has joined #openstack-ironic | 08:03 | |
| *** ifarkas has joined #openstack-ironic | 08:03 | |
| *** foexle has joined #openstack-ironic | 08:06 | |
| *** derekh has joined #openstack-ironic | 08:07 | |
| *** jistr has joined #openstack-ironic | 08:10 | |
| *** jistr has quit IRC | 08:12 | |
| *** jistr has joined #openstack-ironic | 08:20 | |
| *** bradjones has quit IRC | 08:22 | |
| *** praneshp has quit IRC | 08:22 | |
| *** bradjones has joined #openstack-ironic | 08:23 | |
| *** shakamunyi has quit IRC | 08:24 | |
| *** kevinbenton has quit IRC | 08:28 | |
| *** lazy_prince has joined #openstack-ironic | 08:31 | |
| *** jcoufal has quit IRC | 08:35 | |
| *** kevinbenton has joined #openstack-ironic | 08:36 | |
| *** jcoufal has joined #openstack-ironic | 08:37 | |
| *** marcoemorais has quit IRC | 08:44 | |
| *** igordcard has joined #openstack-ironic | 08:46 | |
| *** shakamunyi has joined #openstack-ironic | 08:50 | |
| *** Krast has quit IRC | 08:52 | |
| *** shakamunyi has quit IRC | 08:55 | |
| *** tatyana has joined #openstack-ironic | 08:58 | |
| *** Krast has joined #openstack-ironic | 09:01 | |
| *** Krast has quit IRC | 09:04 | |
| *** Krast has joined #openstack-ironic | 09:05 | |
| *** cinerama` has quit IRC | 09:09 | |
| *** rushiagr is now known as rushiagr_away | 09:12 | |
| openstackgerrit | Yuriy Zveryanskyy proposed a change to openstack/ironic: Remove unused change_node_maintenance_mode from rpcapi https://review.openstack.org/125942 | 09:17 |
| *** cinerama has joined #openstack-ironic | 09:18 | |
| *** rushiagr_away is now known as rushiagr | 09:19 | |
| openstackgerrit | A change was merged to openstack/ironic: KeyError from AgentVendorInterface._heartbeat() https://review.openstack.org/126636 | 09:20 |
| *** romcheg has quit IRC | 09:39 | |
| *** tatyana has quit IRC | 09:41 | |
| *** tatyana has joined #openstack-ironic | 09:42 | |
| *** romcheg has joined #openstack-ironic | 09:47 | |
| *** nosnos has quit IRC | 10:09 | |
| *** killer_prince has joined #openstack-ironic | 10:15 | |
| *** pelix has joined #openstack-ironic | 10:15 | |
| *** jistr has quit IRC | 10:17 | |
| *** athomas has quit IRC | 10:21 | |
| *** Krast has quit IRC | 10:22 | |
| *** jistr has joined #openstack-ironic | 10:22 | |
| *** jmank has quit IRC | 10:23 | |
| *** jmank has joined #openstack-ironic | 10:24 | |
| *** killer_prince has quit IRC | 10:24 | |
| *** yuanying has quit IRC | 10:25 | |
| *** yuanying has joined #openstack-ironic | 10:26 | |
| *** athomas has joined #openstack-ironic | 10:27 | |
| *** yuanying has quit IRC | 10:30 | |
| *** lazy_prince has quit IRC | 10:30 | |
| *** killer_prince has joined #openstack-ironic | 10:32 | |
| *** killer_prince is now known as lazy_prince | 10:32 | |
| *** nosnos has joined #openstack-ironic | 10:38 | |
| *** yuanying has joined #openstack-ironic | 10:55 | |
| *** ramineni has quit IRC | 11:02 | |
| *** yuanying has quit IRC | 11:04 | |
| *** loki184 has quit IRC | 11:07 | |
| *** openstack has joined #openstack-ironic | 14:13 | |
| *** openstackstatus has joined #openstack-ironic | 14:14 | |
| *** ChanServ sets mode: +v openstackstatus | 14:14 | |
| rloo | morning Ironickers! | 14:19 |
| NobodyCam | morning rloo | 14:19 |
| dtantsur | rloo, morning | 14:19 |
| jroll | heya rloo | 14:19 |
| rloo | Hey, wrt 126500 (deprecating Exception names), does anyone know if we need to eg add a DocImpact or ?? How do we inform folks about deprecated stuff? | 14:20 |
| rloo | hi NobodyCam, dtantsur, jroll | 14:20 |
| *** rushiagr_away is now known as rushiagr | 14:21 | |
| *** rakesh_hs2 has quit IRC | 14:22 | |
| NobodyCam | rloo I'm not sure about execption names requiring a doc impact tag | 14:22 |
| NobodyCam | I tent to think not but very well could be wroung | 14:23 |
| *** rakesh_hs has joined #openstack-ironic | 14:23 | |
| NobodyCam | s/tent/tend/ | 14:23 |
| rloo | I think we should add a docimpact to at least track it. at the very least, "we're just not sure": https://wiki.openstack.org/wiki/Documentation/DocImpact | 14:23 |
| rloo | NobodyCam: so if you have code (like the out-of-tree driver that uses CatalogUnauthorized) and we remove that, your code will break. But how do you know that we're deprecating/removing it? | 14:24 |
| NobodyCam | ahh yes | 14:25 |
| * NobodyCam needs more coffee | 14:25 | |
| *** pcrews has joined #openstack-ironic | 14:25 | |
| lucasagomes | morning rloo | 14:31 |
| NobodyCam | morning lucasagomes | 14:31 |
| lucasagomes | NobodyCam, morning there! | 14:32 |
| dtantsur | brb | 14:32 |
| NobodyCam | :) | 14:32 |
| yuriyz | morning rloo | 14:32 |
| NobodyCam | morning yuriyz :) | 14:33 |
| *** zz_jgrimm is now known as jgrimm | 14:34 | |
| yuriyz | +1 for DocImpact | 14:34 |
| rloo | hi lucasagomes, yuriyz. | 14:34 |
| rloo | thx yuriy. was just adding another comment. | 14:34 |
| yuriyz | morning NobodyCam | 14:40 |
| NobodyCam | yuriyz: just checking to see if you'll have time for 103105? | 14:42 |
| *** rakesh_hs has quit IRC | 14:43 | |
| yuriyz | NobodyCam, I will work on it tomorrow | 14:43 |
| NobodyCam | :) awesome Thank you :) | 14:44 |
| *** shakamunyi has joined #openstack-ironic | 14:44 | |
| openstackgerrit | Amaury Medeiros proposed a change to openstack/ironic: Fix attributes ordering at common/imageutils.py https://review.openstack.org/126927 | 14:45 |
| *** shakamunyi has quit IRC | 14:50 | |
| *** amaurymedeiros has joined #openstack-ironic | 14:52 | |
| *** todd_dsm has quit IRC | 15:02 | |
| openstackgerrit | Yuriy Zveryanskyy proposed a change to openstack/ironic: Fix exceptions names and messages for Keystone errors https://review.openstack.org/126500 | 15:05 |
| NobodyCam | brb | 15:07 |
| *** pensu has quit IRC | 15:11 | |
| *** todd_dsm has joined #openstack-ironic | 15:15 | |
| *** shakamunyi has joined #openstack-ironic | 15:15 | |
| devananda | morning, all | 15:16 |
| NobodyCam | good morning devananda :) | 15:16 |
| lucasagomes | devananda, morning | 15:25 |
| arif-ali | chaps, found rc1 RPMs for nova which are still in rawhide, that have the relevant stuff for ironic; after updating to these i.e. version 2014.2-0.5rc1, I was able to start the nova services with ironic driver loaded | 15:25 |
| yuriyz | morning devananda | 15:25 |
| NobodyCam | arif-ali: sweet :) | 15:26 |
| lucasagomes | arif-ali, ! nice one | 15:29 |
| * NobodyCam needs to make a quick run to get a pack of smokes... bbiafm | 15:32 | |
| devananda | rloo: on 126500, I also don't see any reason for a DocImpact flag on that | 15:35 |
| devananda | rloo: it would be great if we did start using DocImpact more often, though it would also be great if we had an operations manual that was impacted :) | 15:35 |
| rloo | devananda: if we don't document somehow, that the exception is going to be deprecated, then we might as well get rid of it now. how is the user going to know unless they read the code? | 15:36 |
| devananda | rloo: who is affected by deprecating that exception? users? or developers of third-party drivers? | 15:37 |
| rloo | devananda: i'm assuming that one day there will be some operations-like manual for ironic. | 15:37 |
| rloo | devananda: isn't a developer of a third-party driver, a user? | 15:37 |
| rloo | devananda: esp if you're taking about have a base driver class that is packaged. | 15:38 |
| rloo | devananda: anyway, the docimpact twiki said 'if you aren't sure', and I wasn't sure what people thought at the time. it doesn't hurt to put docimpact. but if you're sure we don't need it, i won't ask for it in the future. | 15:39 |
| yuriyz | devananda, agree with rloo | 15:39 |
| devananda | rloo: you're right - it doesn't hurt to add it :) | 15:39 |
| devananda | rloo: thanks for explaining your reasoning there. I wasn't thinking about a doc for driver developers -- I'm assuming they would jsut read the code | 15:40 |
| rloo | ok, so let's see how useful it is, come end of Kilo ;) | 15:40 |
| *** jcoufal has quit IRC | 15:41 | |
| devananda | rloo: but yes, if we were to release the driver API as a separate lib, docs for that would be important | 15:41 |
| yuriyz | docs are important always :) | 15:42 |
| rloo | speaking of which, I suppose I should try to get our docn updated. I think I kind of volunteered at some point :-( | 15:42 |
| devananda | there's been a patch from "Vinay B S" up with some really good material in it | 15:43 |
| devananda | I see +2 / -1 -1 | 15:43 |
| rloo | devananda: yeah, vinay said he'd put up another revision soon. | 15:44 |
| rloo | devananda: you had a bunch of concerns about the user-guide (http://docs.openstack.org/developer/ironic/deploy/user-guide.html) | 15:44 |
| rloo | devananda: I need to find the bug etc... I'll be asking you questions soon I think... | 15:45 |
| devananda | cool. I hope so. would love to see Vinay's docs up | 15:45 |
| lucasagomes | devananda, my -1 on that patch is about setting the kernel/ramdisk id to the flavor | 15:45 |
| lucasagomes | like old way in a new doc | 15:45 |
| lucasagomes | afair | 15:45 |
| devananda | rloo: ack. I should be around. | 15:45 |
| *** spandhe has joined #openstack-ironic | 15:45 | |
| devananda | rloo: hm, on the DocImpact thing, that opens a bug, but it looks like Ironic isn't properly configured -- see https://bugs.launchpad.net/openstack-manuals/+bug/1376995 | 15:46 |
| * lucasagomes had too many internal stuff today and didn't progress on the boot interface >.< | 15:46 | |
| NobodyCam | and back | 15:47 |
| rloo | devananda: good catch. I didn't realize any configuration was needed. | 15:47 |
| devananda | rloo: since our docs aren't in openstack-manuals yet, I've pinged annegentle to ask her advice on what docimpact-group to use. | 15:47 |
| rloo | devananda: ok, then I guess we somehow try to find all the docimpact bugs and re-something them. | 15:49 |
| *** pcrews has quit IRC | 15:49 | |
| devananda | git log 2014.1..HEAD --oneline --no-merges | 15:50 |
| devananda | that gives a concise summary of all patches merged during Juno | 15:51 |
| *** pcrews has joined #openstack-ironic | 15:51 | |
| devananda | we only used DocImpact 5 times :( | 15:51 |
| rloo | devananda: that could be good too, we didn't do much that impacted our users :-) [Ok, I suspect we did more than 5 things.] | 15:52 |
| *** romcheg has quit IRC | 15:52 | |
| devananda | ahh, this is what we want: git log 2014.1..2014.2.rc1 --no-merges --grep DocImpact | 15:53 |
| devananda | yea, I'm fairly sure we did much more than those 5 patches that impacted our users :) | 15:53 |
| *** Poornima has joined #openstack-ironic | 15:54 | |
| NobodyCam | was there a sepc to go along with this BP: https://blueprints.launchpad.net/ironic/+spec/cinder-integration ??? | 15:54 |
| *** ifarkas has quit IRC | 15:54 | |
| devananda | NobodyCam: I dont see one | 15:54 |
| NobodyCam | ya | 15:55 |
| NobodyCam | :-p | 15:55 |
| openstackgerrit | A change was merged to openstack/ironic: Imported Translations from Transifex https://review.openstack.org/126811 | 15:56 |
| jroll | devananda: fwiw, that hash ring patch appears to work in our staging environment | 15:57 |
| jroll | hooray | 15:57 |
| devananda | \o/ | 15:57 |
| *** sambetts has quit IRC | 15:57 | |
| rloo | NobodyCam: speaking of cinder/ironic: http://docs.openstack.org/developer/ironic/_images/conceptual_architecture.png | 15:58 |
| NobodyCam | oh very nice rloo :) | 15:59 |
| rloo | NobodyCam: that's in our user guide today. I didn't do that. | 15:59 |
| NobodyCam | :-p | 15:59 |
| rloo | NobodyCam: it shows Cinder providing volumes to BM. | 15:59 |
| NobodyCam | yes it does | 16:00 |
| *** k4n0 has quit IRC | 16:00 | |
| Shrews | jroll: for IPA, i guess you guys aren't using translation operators yet, correct? _(), _LE(), etc | 16:00 |
| rloo | Anyone have any changed they'd like to have made in this diagram? | 16:00 |
| rloo | eg, should we add swift? | 16:00 |
| jroll | Shrews: we decided with deva that IPA strings aren't user facing, do not translate | 16:00 |
| rloo | Do we want this diagram? (deleting is easier than updating, thought I'd ask) | 16:00 |
| Shrews | jroll: ok. good point | 16:00 |
| *** rushiagr is now known as rushiagr_away | 16:01 | |
| devananda | rloo: yes, we need to add swift, which provides object storage for the iLO and agent drivers | 16:02 |
| devananda | rloo: cinder /can/ provide iSCSI volumes for nodes, but it's not documented anywhere afaik | 16:02 |
| Shrews | jroll: line 155 of https://review.openstack.org/#/c/113343/5/ironic_python_agent/agent.py... this Event object is added but unused. is that intentional for a future use? | 16:03 |
| devananda | rloo: IMHO, if the diagram is not in a master format (one we can all easily edit) it shouldn't be in our docs | 16:03 |
| rloo | devananda: most of the diagrams are in .png. One is in docx. | 16:04 |
| rloo | devananda: I can't easily edit any diagram, so it doesn't matter to me ;) | 16:05 |
| devananda | rloo: ah :) | 16:05 |
| JayF | There are tools (open source iirc) that have a translatable text format that renders into a graphic | 16:05 |
| * JayF trying to remember the name of the one he was thinking of | 16:06 | |
| * devananda steps away and goes in search of coffee | 16:06 | |
| JayF | graphviz | 16:06 |
| devananda | JayF: ++ | 16:06 |
| rloo | so -- start from scratch then? What do people suggest wrt diagrams. (I was so hoping that doc folks would do this.) | 16:06 |
| JayF | http://en.wikipedia.org/wiki/Graphviz since graphviz.org appears to be down | 16:06 |
| *** romcheg has joined #openstack-ironic | 16:09 | |
| rloo | inkscape SVG? http://lists.openstack.org/pipermail/openstack-docs/2014-July/004822.html | 16:10 |
| JayF | I mean, I'm personally not /opposed/ to SVG, but I find graphs that are done using "drawing tools" are harder to maintain than those that are generated using a markup language | 16:12 |
| rloo | up until now, I was happy if people submitted diagrams regardless of how they did it. I'm still happy, as long as I don't have to modify them myself ;) | 16:13 |
| JayF | I was looking at some diagrams I made that are onmetal specific to see if they would be useful as a starting point | 16:13 |
| JayF | and the answer is pretty much very no | 16:13 |
| rloo | but assuming that we might want to modify diagrams, I suppose we should agree on some thing(s). | 16:13 |
| rloo | i just asked on openstack-doc, but no reply (yet). | 16:14 |
| jroll | Shrews: doesn't look like it, not sure where that came from :/ feel free to -1 | 16:14 |
| *** amitpp has joined #openstack-ironic | 16:17 | |
| *** jistr has quit IRC | 16:22 | |
| *** coolsvap is now known as coolsvap|afk | 16:24 | |
| *** tatyana has quit IRC | 16:25 | |
| *** marcoemorais has joined #openstack-ironic | 16:28 | |
| rloo | JayF: so a doc person (from rackspace ;)), said they like omnigraffle. they tried inkscape and aren't too happy with that. | 16:29 |
| rloo | JayF: SVG output is required, regardless of which s/w is used. | 16:29 |
| JayF | omnigraffle is a commercial program | 16:29 |
| jroll | rloo: who said that? anne? | 16:29 |
| JayF | and omnigraffle is what I used to make our downstream diagrams | 16:29 |
| jroll | guess it doesn't matter anyhow | 16:29 |
| * jroll afk for a bit | 16:29 | |
| rloo | jroll: Sam-I-Am, some Dr. Seuss fan I guess ;) | 16:30 |
| JayF | I guess open format makes it palateable, but it seems like we should use open tools to generate our docs, even if they compile into open formats | 16:30 |
| jroll | huh | 16:30 |
| rloo | well, are you OK with SVG output? That is a requirement. | 16:30 |
| JayF | SVG output is awesome :) | 16:30 |
| rloo | Sam-I-Am said they're responsible for coming up with 'diagram standards' but it hasn't happened yet. So I guess you might be able to influence... | 16:31 |
| rloo | JayF: so if SVG output is awesome, does it matter what s/w you use? You can use whatever s/w you like as long as you save it in SVG output. And if the s/w works properly :-) | 16:32 |
| JayF | rloo: honestly, I'd probably interact with upstream docs teams via upstream, rackers or not :) | 16:32 |
| JayF | rloo: I guess so, I just know if you're able to put in a source file, it's *much* easier to make minor edits in the future | 16:32 |
| JayF | but I need to have less of an opinion before I accidentally volunteer for things | 16:32 |
| *** amitpp has quit IRC | 16:32 | |
| *** amitpp has joined #openstack-ironic | 16:33 | |
| rloo | JayF: I believe jroll offered to help with doc a while ago ;) | 16:33 |
| *** andreykurilin_ has joined #openstack-ironic | 16:35 | |
| *** amitpp has quit IRC | 16:38 | |
| *** shakamunyi has quit IRC | 16:46 | |
| *** todd_dsm has quit IRC | 16:49 | |
| NobodyCam | our next meeting appers to be a holiday in the US? are folks planning on working Columbus Day? | 16:51 |
| JayF | I know it's not a company holiday at Rackspace, although I personally have taken it off | 16:51 |
| NobodyCam | it is not a HP holiday either | 16:53 |
| NobodyCam | lucasagomes: still want Thoughts and comments on https://review.openstack.org/#/c/99426 (lucasagomes) on the agenda? | 16:55 |
| *** penick has joined #openstack-ironic | 16:58 | |
| *** derekh has quit IRC | 16:59 | |
| lucasagomes | NobodyCam, 1 sec, in a call | 17:00 |
| *** lucasagomes is now known as lucas-afk | 17:00 | |
| *** Haomeng has joined #openstack-ironic | 17:01 | |
| *** Haomeng|2 has quit IRC | 17:02 | |
| rloo | NobodyCam: holiday in Canada (Thanksgiving). I'm not working that day. | 17:02 |
| *** andreykurilin__ has joined #openstack-ironic | 17:03 | |
| *** andreykurilin_ has quit IRC | 17:03 | |
| NobodyCam | rloo: :) | 17:05 |
| NobodyCam | nice | 17:05 |
| *** Poornima has quit IRC | 17:06 | |
| lucas-afk | NobodyCam, I think we can remove it :) | 17:12 |
| lucas-afk | we can continue that on the patch itself and on IRC here | 17:12 |
| lucas-afk | I mean here in this channel | 17:12 |
| *** lucas-afk is now known as lucasagomes | 17:12 | |
| NobodyCam | :) | 17:13 |
| *** HenryG is now known as HenryG_afk | 17:20 | |
| *** todd_dsm has joined #openstack-ironic | 17:21 | |
| *** igordcard has quit IRC | 17:23 | |
| *** dtantsur is now known as dtantsur|afk | 17:28 | |
| *** praneshp has joined #openstack-ironic | 17:28 | |
| lucasagomes | aight folks I will call it a day | 17:34 |
| lucasagomes | have a great night | 17:34 |
| *** lucasagomes is now known as lucas-dinner | 17:34 | |
| NobodyCam | have a good night lucas-dinner | 17:40 |
| *** HenryG_afk is now known as HenryG | 17:44 | |
| *** Hefeweizen has joined #openstack-ironic | 17:49 | |
| *** krtaylor has quit IRC | 18:01 | |
| jroll | a PATCH on a locked node should 409, yeah? | 18:03 |
| NobodyCam | locked up how | 18:04 |
| NobodyCam | maintance = true prov_state = deploying? | 18:04 |
| JayF | reservation=conductor.hostname | 18:05 |
| JayF | NobodyCam: ^ | 18:05 |
| JayF | in that way :) | 18:05 |
| *** athomas has quit IRC | 18:06 | |
| NobodyCam | I think yes: | 18:09 |
| NobodyCam | https://github.com/openstack/ironic/blob/master/ironic/api/controllers/v1/node.py#L298 | 18:10 |
| rloo | jroll: yes, based on what I see in the code. | 18:10 |
| JayF | That's what our belief is too :) | 18:10 |
| jroll | huh. | 18:10 |
| jroll | ok | 18:11 |
| jroll | NobodyCam: wrong API endpoint, but yeah | 18:11 |
| jroll | I'm looking at https://github.com/openstack/ironic/blob/master/ironic/api/controllers/v1/node.py#L763 | 18:11 |
| rloo | jroll: although I looked at the exception file for NodeLocked ;) | 18:11 |
| jroll | cool | 18:12 |
| rloo | jroll: yeah, so the conductor/manager raises NodeLocked via rpc blah blah, and NodeLocked is a Conflict exception with has 409 | 18:12 |
| jroll | right | 18:12 |
| jroll | ty rloo | 18:12 |
| rloo | are you getting 409 or something else? | 18:12 |
| rloo | I suppose it might be useful to add docstrings to that v1/node.py file. | 18:13 |
| *** krtaylor has joined #openstack-ironic | 18:13 | |
| JayF | We're digging right now, give us a sec | 18:13 |
| rloo | and are we logging anything? | 18:14 |
| jroll | logging everything :) | 18:14 |
| NobodyCam | ya for detailed logs | 18:14 |
| jroll | I have a feeling this might not be an issue, hard to tell | 18:14 |
| *** harlowja_away is now known as harlowja | 18:14 | |
| *** todd_dsm has quit IRC | 18:19 | |
| *** athomas has joined #openstack-ironic | 18:23 | |
| *** praneshp has quit IRC | 18:29 | |
| *** praneshp has joined #openstack-ironic | 18:33 | |
| *** todd_dsm has joined #openstack-ironic | 18:34 | |
| *** pelix has quit IRC | 18:34 | |
| *** pensu has joined #openstack-ironic | 18:38 | |
| *** jistr has joined #openstack-ironic | 18:59 | |
| *** penick has quit IRC | 19:01 | |
| *** penick has joined #openstack-ironic | 19:13 | |
| *** romcheg1 has joined #openstack-ironic | 19:17 | |
| *** romcheg has quit IRC | 19:19 | |
| * NobodyCam steps away to look for some food stuffs before the NEXT all hands call :-p | 19:22 | |
| *** praneshp has quit IRC | 19:29 | |
| *** praneshp has joined #openstack-ironic | 19:39 | |
| *** krtaylor has quit IRC | 19:44 | |
| *** igordcard has joined #openstack-ironic | 19:48 | |
| *** tatyana has joined #openstack-ironic | 19:50 | |
| *** tatyana has quit IRC | 19:55 | |
| *** krtaylor has joined #openstack-ironic | 19:56 | |
| devananda | jroll: fix backported - https://review.openstack.org/#/c/126984/ | 19:57 |
| devananda | if anyone spots any other release critical bugs, pls let me know // target them to rc2 | 19:57 |
| devananda | https://bugs.launchpad.net/ironic/+milestone/juno-rc2 | 19:58 |
| devananda | we'll cut it tomorrow around 9AM pacific if that page doesn't have any more bugs on it | 19:58 |
| *** igordcard has quit IRC | 19:58 | |
| *** pensu has quit IRC | 20:02 | |
| NobodyCam | :) | 20:03 |
| *** marcoemorais has quit IRC | 20:03 | |
| *** marcoemorais has joined #openstack-ironic | 20:03 | |
| *** andreykurilin_ has joined #openstack-ironic | 20:04 | |
| *** marcoemorais has quit IRC | 20:04 | |
| *** andreykurilin__ has quit IRC | 20:05 | |
| *** marcoemorais has joined #openstack-ironic | 20:05 | |
| *** jistr has quit IRC | 20:08 | |
| jroll | devananda: nice, thanks | 20:14 |
| *** dprince has quit IRC | 20:14 | |
| *** morgabra has quit IRC | 20:15 | |
| *** pensu has joined #openstack-ironic | 20:18 | |
| *** romcheg1 has quit IRC | 20:25 | |
| *** morgabra has joined #openstack-ironic | 20:25 | |
| *** lucas-dinner has quit IRC | 20:26 | |
| *** todd_dsm has quit IRC | 20:27 | |
| *** todd_dsm has joined #openstack-ironic | 20:28 | |
| *** todd_dsm has quit IRC | 20:33 | |
| *** athomas has quit IRC | 20:50 | |
| *** pensu has quit IRC | 20:51 | |
| *** romcheg has joined #openstack-ironic | 20:52 | |
| JayF | WDYT about adding an API endpoint to break a lock? | 20:54 |
| JayF | It happens in edge cases around deploying, and when it does, having to hit the database is ... ugh | 20:55 |
| *** mrda has joined #openstack-ironic | 20:55 | |
| mrda | Good morning Ironic :) | 20:55 |
| devananda | morning, mrda! welcome back | 20:56 |
| mrda | hi devananda - glad to be back! | 20:57 |
| JayF | Did you put some of your vacation in a jar to share with the rest of us? :P | 20:57 |
| mrda | I haz photos, if that'll do :) | 20:57 |
| devananda | JayF: as an opts tool, that makes some sense. but from a user POV, itdoesn't. | 20:57 |
| JayF | devananda: our API is more ops-facing than user-facing | 20:58 |
| devananda | JayF: think what it says to users of our API: here is this end point where you can break things | 20:58 |
| JayF | devananda: Philosophically, I think if operators have to touch the DB we've deeply failed | 20:58 |
| JayF | and right now, that's absolutely something I'm doing | 20:58 |
| devananda | JayF: I agree with that. but I am not convinced an API for it is correct | 20:58 |
| devananda | ugh | 20:58 |
| JayF | Conductor gets restarted for a deploy. It had a lock on a node. How do you avoid this situation? | 20:58 |
| devananda | quiesce | 20:59 |
| JayF | devananda: we've talked some on the OnMetal team about wanting a way to "turn off the safety" | 20:59 |
| devananda | heh | 20:59 |
| JayF | like we added something to decom to forbid power actions during decom | 20:59 |
| JayF | but yet we find nodes sometimes get in a state in decom where automation has failed somehow and we have to unravel it | 20:59 |
| JayF | and right now that involves touching the BMC or DB in some cases | 20:59 |
| *** penick has quit IRC | 21:00 | |
| JayF | IDK the "right way" to fix stuff like that, but I do know the full set of tooling needed to operate ironic isn't available using existing abstractions | 21:01 |
| JayF | and while the preferred answer is generally "automation should be good enough to prevent things from getting stuck in bad states", I don't think that works well in the real world | 21:01 |
| JayF | (mainly because we aren't perfect) | 21:03 |
| devananda | hardware also isn't perfect | 21:03 |
| devananda | i agree loc breaking is needed | 21:03 |
| NobodyCam | ++ yes | 21:04 |
| devananda | the challenge is to prevent things which could physcaly damage hardware | 21:04 |
| rloo | no hammers allowed | 21:04 |
| devananda | so if decom is flashing firmware, and you think it takes too long and power cycle the box | 21:04 |
| devananda | now you can't get it to boot again ... | 21:04 |
| JayF | yeah we block that explicitly via the API right now in our downstream patch | 21:05 |
| devananda | in principle, that's why we have the task lock | 21:05 |
| JayF | but we do find that things happen even in decom state that require fixing | 21:05 |
| JayF | mostly around dhcp downtime and failing to boot an agent for various reasons | 21:05 |
| devananda | sure | 21:05 |
| *** penick has joined #openstack-ironic | 21:05 | |
| devananda | so we need a way to signal, within ironic (not ipa) whether a state is "unsafe to break" | 21:05 |
| JayF | I'm saying that even in those cases | 21:05 |
| JayF | the operator should hold the biggest hammer | 21:05 |
| * JayF references, yet again, the --i-am-a-dummy option to mysqld | 21:06 | |
| devananda | hah | 21:06 |
| JayF | it exists because some people need it, even though in 99.99999999% of cases it'll chew up your data and spit it out | 21:06 |
| *** marcoemorais has quit IRC | 21:09 | |
| *** marcoemorais has joined #openstack-ironic | 21:10 | |
| devananda | JayF: hwo about a CLI tool? | 21:14 |
| JayF | devananda: If it wants to connect directly to my DB, very no | 21:14 |
| JayF | devananda: I want to be able to secure ironic so that only conductors can talk to dbs | 21:14 |
| JayF | devananda: and only conductors can talk to bmcs | 21:14 |
| JayF | devananda: then be able to operate Ironic without having to have any code other than the conductor running on the conductor | 21:15 |
| devananda | so if a conductor is wedged or died | 21:15 |
| *** Kui has joined #openstack-ironic | 21:15 | |
| devananda | you want another conductor to take over its work | 21:15 |
| devananda | even if there was a lock left behind? | 21:16 |
| *** todd_dsm has joined #openstack-ironic | 21:16 | |
| JayF | So to put this into perspective, this is how defensively-coded our API is right now: | 21:16 |
| devananda | i need to think moer about this when i'm not on a conference | 21:16 |
| JayF | I cannot put a node into maintenance that has power_state: on target_power_state: off | 21:16 |
| JayF | even if the BMC is dead and that will never succeed | 21:16 |
| JayF | so I can't do a damn thing on that node until the power status loop fails the number of times and maints it automatically | 21:17 |
| JayF | which, as we discovered before, having a failed node in the power status loop hurts performance considerable | 21:17 |
| JayF | *considerably | 21:17 |
| JayF | I absolutely want to be able to say: This node is invalidly locked, unlock it | 21:17 |
| JayF | I don't ever think of it as conductor takeover, even though I should | 21:17 |
| JayF | since with None as the DHCP provider + IPA driver there's no local state on teh conductor | 21:17 |
| devananda | indeed | 21:18 |
| JayF | Also, fwiw, a lot of this *specific* situation is solved by pluggable locking + ephemeral locks | 21:19 |
| JayF | like in ZK where if the connection dies, all the locks made there die too | 21:19 |
| devananda | that ^ is terrible, fwiw, for other environments | 21:19 |
| devananda | if a conductor dies while a node it was working on is applying a firmware update using iLO, for example | 21:20 |
| *** Jmank has joined #openstack-ironic | 21:20 | |
| devananda | you really dont want another conductor taking over and immediately trying to do a deploy | 21:20 |
| devananda | because there's no lock | 21:20 |
| *** jjohnson2 has quit IRC | 21:21 | |
| JayF | It almost seems like | 21:21 |
| JayF | maintenance and reservation | 21:21 |
| JayF | are heavily overloaded and should have more metadata | 21:21 |
| devananda | ++ | 21:21 |
| devananda | totally | 21:21 |
| JayF | if that was fixed, then we can do smarter things, and enable the operator to make better decisions | 21:21 |
| JayF | but I'd still think we'd want to give the operator a ?i-am-a-dummy=true option to override the failsafe | 21:22 |
| JayF | and maybe even block that by default in shipped policy.json | 21:22 |
| devananda | + maintenance & reservation with more metadata (time started, type of thing in progress) | 21:23 |
| devananda | + configurable policy around different URI endpoints | 21:23 |
| jroll | there's also the option of only unbreaking a lock if it's been held for x minutes, where x > 60 or something ridiculous | 21:23 |
| devananda | + separate policy for node.driver_info | 21:23 |
| devananda | some of this is captured in my old async-api spec, i think | 21:23 |
| devananda | ....the metadata, time started, stuff | 21:24 |
| jroll | I would love for maintenance to be a text field, node is in maintenance if not locked | 21:24 |
| jroll | but that's right on the cmdb line | 21:24 |
| devananda | JayF: the difference between an API for controlling a node and an API for controlling a conductor's state is significant, IMO | 21:24 |
| *** marcoemorais has quit IRC | 21:25 | |
| JayF | I agree, but I'm saying these are clearly operability problems, and we have to have a better answer than people going to the DB | 21:25 |
| devananda | totally agree with that | 21:25 |
| JayF | I am not claiming my solution is the best solution | 21:26 |
| devananda | :) | 21:26 |
| JayF | just generally as an operator, I *hate* when something refuses to do something | 21:26 |
| JayF | if I say to do X, yes I know the node is locked, it's been locked since August, f$%#$ing do it | 21:26 |
| JayF | hence why I want a bigger hammer ;) | 21:27 |
| devananda | that the lock feels arbitrary is the problem | 21:27 |
| devananda | if the lock actually protected you from something harmful MOST of the time | 21:27 |
| JayF | I think for me, that may be partially driven by locks being taken on heartbeat | 21:27 |
| JayF | for agent right now | 21:27 |
| JayF | so we probably cycle locks a *lot* more than folks using other drivers | 21:27 |
| JayF | that's my hunch, at least | 21:27 |
| *** marcoemorais has joined #openstack-ironic | 21:27 | |
| devananda | sure. but that will hit other drivers too | 21:27 |
| devananda | so the problem is | 21:27 |
| devananda | locks DONT actually indicate "dangerous thing is in progress -- keep your hands off" | 21:28 |
| NobodyCam | just have the cli PRINT out a mysql command and tell operators to cut and paste this command if you really want to do this | 21:28 |
| NobodyCam | lol ... j/k | 21:28 |
| jroll | NobodyCam: don't make me borrow monty's cats | 21:28 |
| devananda | the lock indicates "system may or may not be busy, but you acn't really tell" | 21:28 |
| NobodyCam | hehehe | 21:28 |
| devananda | which is silly and frustrating | 21:28 |
| JayF | NobodyCam: you jest, but I'm really likely to be writing a script that'll find nodes with long-stuck-locks and print out sql commands to run to fix them | 21:28 |
| JayF | yeah, I want to know *why* the node is locked | 21:29 |
| NobodyCam | :/ | 21:29 |
| JayF | similar idea with maintenance | 21:29 |
| devananda | yup | 21:29 |
| JayF | last_error is supposed to be that | 21:29 |
| *** Ugallu has quit IRC | 21:29 | |
| Shrews | NobodyCam: let's get really silly and have each conductor serve a web page displaying internal state and clickable action buttons :) | 21:29 |
| JayF | but surprise: if the power status loop fails, you just lost your error | 21:29 |
| JayF | and it is clobbered by ironic with another error | 21:29 |
| JayF | I kinda want a node_errors table like Nova has an instance_errors table | 21:29 |
| * JayF going to ponder that further | 21:29 | |
| devananda | with history | 21:29 |
| devananda | yes | 21:30 |
| devananda | last_error should be the latest entry in that table | 21:30 |
| NobodyCam | Shrews: thats kinda GUI like for operators... but if it was a app for the phone then i can see it | 21:30 |
| jroll | notifications would also help with this :) | 21:30 |
| devananda | notifications is an overloaded word | 21:30 |
| JayF | I also want maintenance_reason | 21:30 |
| JayF | but I'll take my 'out of scope' medicine on that one if needed | 21:30 |
| jroll | devananda: notifications in the nova sense of the word | 21:31 |
| JayF | because I know my motivation is to use that field for cmdb-ish things (I'd rather Ironic be canonical for why a node is in maintenance than a CMDB) | 21:31 |
| jroll | there is an error, put it on the notifications bus | 21:31 |
| *** todd_dsm has quit IRC | 21:32 | |
| *** ChuckC has quit IRC | 21:33 | |
| *** ChuckC has joined #openstack-ironic | 21:34 | |
| *** romcheg has quit IRC | 21:37 | |
| *** romcheg1 has joined #openstack-ironic | 21:37 | |
| *** foexle has quit IRC | 21:38 | |
| NobodyCam | brb | 21:39 |
| *** romcheg has joined #openstack-ironic | 21:41 | |
| *** romcheg1 has quit IRC | 21:41 | |
| *** romcheg1 has joined #openstack-ironic | 21:44 | |
| *** todd_dsm has joined #openstack-ironic | 21:44 | |
| *** igordcard has joined #openstack-ironic | 21:44 | |
| *** romcheg has quit IRC | 21:45 | |
| openstackgerrit | Vladyslav Drok proposed a change to openstack/ironic: Fix markup-related issues in documentation https://review.openstack.org/127034 | 21:51 |
| *** pcrews has quit IRC | 21:53 | |
| *** romcheg1 has quit IRC | 21:56 | |
| *** romcheg has joined #openstack-ironic | 21:57 | |
| *** dlaube has joined #openstack-ironic | 22:00 | |
| *** penick has quit IRC | 22:00 | |
| NobodyCam | oh did we get a patch up for the spec templates | 22:02 |
| *** praneshp has quit IRC | 22:02 | |
| *** marcoemorais1 has joined #openstack-ironic | 22:03 | |
| *** marcoemorais has quit IRC | 22:03 | |
| rloo | NobodyCam: I looked into it. The only way that I could figure out, is to put a dummy file (I think it has to be .rst but I didn't try) in specs/kilo/. Is it worth doing? | 22:03 |
| *** marcoemorais1 has quit IRC | 22:04 | |
| dlaube | Hey guys, is there some trick to using disk-image-create to make images for baremetal? | 22:04 |
| *** rwsu has quit IRC | 22:05 | |
| *** marcoemorais has joined #openstack-ironic | 22:05 | |
| dlaube | I ran "disk-image-create -a amd64 -o ubuntu-amd64 ubuntu" but I'm seeing some references to qemu-img and other paravirt like stuff | 22:05 |
| *** igordcard has quit IRC | 22:06 | |
| NobodyCam | can we land with jenkins -1'ing? | 22:09 |
| NobodyCam | not that i am trying to land anything/ just checking that we can | 22:09 |
| *** andreykurilin_ has quit IRC | 22:16 | |
| *** rwsu has joined #openstack-ironic | 22:17 | |
| *** praneshp has joined #openstack-ironic | 22:18 | |
| openstackgerrit | Ruby Loo proposed a change to openstack/ironic-specs: Set up for kilo specs https://review.openstack.org/127047 | 22:18 |
| rloo | NobodyCam: take a look at that. I don't like it and maybe someone else that knows more about sphinx etc can figure something out. | 22:19 |
| rloo | NobodyCam: otherwise, to get jenkins to pass, all the spec patches will need to include the changes in that patch (except the dummy .rst I put in). | 22:19 |
| NobodyCam | rloo: aawesome :) we'll get some eyes on it | 22:20 |
| *** penick has joined #openstack-ironic | 22:20 | |
| NobodyCam | but ya. I don't think spec writers should have to do that | 22:20 |
| rloo | NobodyCam: yeah, I don't think we should add more hurdles for the spec writers. | 22:21 |
| rloo | NobodyCam: I gotta run. later maybe ;) | 22:21 |
| JayF | rloo: does the dir need a .rst file in it specifically? | 22:21 |
| *** rloo is now known as rloo_afk | 22:21 | |
| NobodyCam | have a great night rloo_afk | 22:21 |
| rloo_afk | JayF: unfortunately yes. And not only that, cuz of the unit tests, it has to have all the sections in the template.rst!!! | 22:21 |
| *** shakamunyi has joined #openstack-ironic | 22:21 | |
| JayF | hah | 22:22 |
| JayF | okay, ty | 22:22 |
| NobodyCam | :) | 22:22 |
| arif-ali | is there an ironic driver for cobbler by any chance? | 22:26 |
| *** romcheg has quit IRC | 22:28 | |
| devananda | arif-ali: what would that look like? | 22:29 |
| devananda | arif-ali: ironic talks directly to the hardware via IPMI, iLO, DRAC, etc... | 22:29 |
| arif-ali | afaik, cobbler is a bare-metal provisioning tool | 22:30 |
| arif-ali | just looking up some stuff on it at the moment | 22:30 |
| devananda | arif-ali: ironic is also a bare metal provisioning tool | 22:31 |
| arif-ali | hmm, that means I need to read up on ironic more then | 22:32 |
| *** pcrews has joined #openstack-ironic | 22:32 | |
| devananda | arif-ali: what did you think ironic was? | 22:33 |
| arif-ali | to be able to control bare-metal nodes | 22:34 |
| arif-ali | I presume then ironic dumps an image from glance to a node, and not from kickstart/autoyast? | 22:34 |
| *** mikal has joined #openstack-ironic | 22:35 | |
| devananda | correct | 22:35 |
| devananda | for lack of a better word, "cloudiness" | 22:35 |
| *** shakamunyi has quit IRC | 22:35 | |
| devananda | ironic deploys binary images to cattle. it doesn't install an OS on a pet. | 22:36 |
| arif-ali | ok, we need to be able to install bare-metal via templates such as kickstart/autoyast/preseed | 22:36 |
| devananda | then ironic isn't a good fit | 22:36 |
| dlaube | can someone confirm if I should be using disk-image-create or ramdisk-image-create to begin creating an image for ironic? | 22:36 |
| devananda | unless you first create images for each of your templates | 22:37 |
| devananda | dlaube: yes | 22:37 |
| dlaube | I used disk-image-create but it looks…. "cloudy" | 22:37 |
| dlaube | ;D | 22:37 |
| arif-ali | That's why I was looking at xCAT's implementation of ironic, as I have been using xCAT for ~8years | 22:37 |
| devananda | dlaube: ramdisk-i-c is for the deploy ramdisk. it's short-lived, but that's how we copy the disk image onto the node | 22:37 |
| NobodyCam | arif-ali: I know I have worked with you on and off this week is there anything missing from our doc's that would have helped you sooner? | 22:37 |
| dlaube | ahh ok | 22:37 |
| devananda | dlaube: disk-i-c can be used to customize your chosen vendor's cloud image | 22:37 |
| dlaube | thank you devananda | 22:37 |
| *** jgrimm is now known as zz_jgrimm | 22:37 | |
| dlaube | I'll keep playing with this | 22:38 |
| arif-ali | devananda, I have already created an image for nodes using xCAT's KVM implementation, and I was able to drop that as an image into glance and import into an instance very easily | 22:38 |
| arif-ali | NobodyCam, I had been using RDO, so the only thing for me was that I needed the new test RPMs from fedora's CI build | 22:39 |
| arif-ali | NobodyCam, once I did that, it was good to go, and I was able to run the ironic commands | 22:40 |
| NobodyCam | arif-ali: :) | 22:40 |
| arif-ali | NobodyCam: The only thing I thing that I would have liked was maybe systemd or init.d scripts samples, and also creating the ironic user | 22:41 |
| NobodyCam | the ironic user for keystone? | 22:42 |
| arif-ali | no, system user, for the service to run, as all other services run as their own users | 22:42 |
| NobodyCam | not sure what user | 22:42 |
| NobodyCam | oh | 22:42 |
| NobodyCam | got it | 22:43 |
| JayF | arif-ali: are you running ironic-api directly or behind apache|nginx | 22:43 |
| arif-ali | JayF, running directly | 22:44 |
| JayF | You probably want to put it behind apache | 22:44 |
| NobodyCam | for ANYHITNG production yes!!! | 22:45 |
| arif-ali | NobodyCam, the only other difference was the admin_tenant_name was services instead of service on my setup with packstack | 22:45 |
| NobodyCam | :) | 22:46 |
| arif-ali | JayF, thanks for that, I was wondering why there was a folder for apache, and the same port numbers being used. That's now clarified ;) | 22:46 |
| jroll | NobodyCam: lol | 22:47 |
| jroll | NobodyCam: we don't have it behind anything right now D: | 22:47 |
| NobodyCam | zomg... | 22:47 |
| jroll | it's been fine, fwiw | 22:47 |
| JayF | except logging | 22:47 |
| JayF | and that we've likely had to overscale it :) | 22:47 |
| jroll | yeah. | 22:47 |
| jroll | yeah. | 22:47 |
| arif-ali | NobodyCam, the doc assumes that the init.d/systemd scripts are already there, obviously I did it from source; and just to be clear, I used http://docs.openstack.org/developer/ironic/deploy/install-guide.html#setup-the-drivers-for-bare-metal-service | 22:48 |
| dlaube | when I run "./bin/ramdisk-image-create -a amd64 -o ubuntu-amd64 ubuntu deploy.ramdisk deploy-ironic" I'm seeing "ERROR: Element 'deploy.ramdisk' not found in '/root/image-work/tmpy/diskimage-builder/bin/../elements:/root/image-work/tmpy/diskimage-builder/bin/../elements'" | 22:48 |
| dlaube | does deploy.ramdisk live in tripleo or something? | 22:48 |
| JayF | it lives in tripleo-image-elements iirc | 22:48 |
| JayF | It will likely be moved into an ironic ramdisk builder repository. Figuring that out is on our list for things to figure out at the summit. | 22:49 |
| dlaube | cool | 22:49 |
| dlaube | I'll grab that | 22:49 |
| NobodyCam | dlaube: sounds like ELEMENTS_PATH or TRIPLEO_ROOT not set | 22:49 |
| JayF | because Ironic users shouldn't have to use tripleo pieces to use ironic :) | 22:49 |
| dlaube | thanks JayF | 22:49 |
| dlaube | NobodyCam: I'll set those env vars now that I have tripleo image elements | 22:49 |
| dlaube | ;) | 22:50 |
| arif-ali | any interesting talks/hands-on to do with ironic at the summit, that may be useful | 22:51 |
| JayF | Myself and JoshNang are talking about cleaning up after bare metal tenants | 22:51 |
| NobodyCam | arif-ali: ya tues and wed we have sessions | 22:51 |
| JayF | basically presenting on a patch we've had downstreamed for a couple of months, that some form of will be in Ironic in K (I hope :D) | 22:51 |
| JayF | jroll has a talk on scaling ironic | 22:51 |
| jroll | I've got a talk about running it at scale | 22:51 |
| *** todd_dsm has quit IRC | 22:52 | |
| NobodyCam | we also have a slot in the operators track | 22:52 |
| jroll | yessss | 22:53 |
| jroll | excited for that | 22:53 |
| arif-ali | cool, looking forward to it | 22:53 |
| jroll | and then we have like 3-4 hours on friday or something right? | 22:53 |
| jroll | just open time? | 22:53 |
| * NobodyCam wounders what side the J's will be on.. op or dev :-p | 22:53 | |
| arif-ali | ah, wont be there for Thu-Fri | 22:53 |
| jroll | ah :( | 22:54 |
| jroll | NobodyCam: I tend to think those are the same thing :) | 22:54 |
| NobodyCam | :) | 22:54 |
| jroll | or have a large overlap, at least | 22:54 |
| NobodyCam | ++ | 22:54 |
| JayF | NobodyCam: I mean, we all rotate on call? All of us develop things, all of us operate things | 22:54 |
| jroll | like... that backport last night was from me deploying and digging into our staging environment, not some ops guy calling me :P | 22:54 |
| jroll | s/backport/bugfix/ | 22:55 |
| NobodyCam | :) | 22:55 |
| * NobodyCam thinks thats just awesome! he really does | 22:55 | |
| jroll | :) | 22:56 |
| * NobodyCam :) | 22:57 | |
| *** penick has quit IRC | 22:58 | |
| devananda | JayF: is that (consolidate ironic image build bits) on the schedule doc? | 22:58 |
| JayF | devananda: yes? | 22:58 |
| devananda | JayF: I mean - is that something you added to | 22:58 |
| JayF | devananda: although now you make me doubt | 22:59 |
| * JayF checks | 22:59 | |
| jroll | I think I saw it | 22:59 |
| JayF | row 5 | 22:59 |
| devananda | yup found it | 22:59 |
| dlaube | NobodyCam: I've tried export ELEMENTS_PATH and TRIPLEO_ROOT but something still isn't right | 23:00 |
| dlaube | http://pastie.org/private/6eajucaazonejzeh0dr45a | 23:00 |
| *** praneshp has quit IRC | 23:01 | |
| *** penick has joined #openstack-ironic | 23:01 | |
| arif-ali | damn, I've registered for the summit incorectly | 23:02 |
| devananda | arif-ali: ? | 23:02 |
| NobodyCam | oh wait | 23:03 |
| arif-ali | I went for the keynote+expo, thinking it was also the general sessions as well | 23:03 |
| *** marcoemorais has quit IRC | 23:03 | |
| devananda | arif-ali: ah :( | 23:03 |
| *** marcoemorais has joined #openstack-ironic | 23:03 | |
| arif-ali | it's not really clear on the website | 23:04 |
| NobodyCam | arif-ali: what are you buildnig here. the deployment ramdisk or a image | 23:04 |
| *** marcoemorais has quit IRC | 23:04 | |
| *** marcoemorais has joined #openstack-ironic | 23:05 | |
| *** praneshp has joined #openstack-ironic | 23:05 | |
| arif-ali | NobodyCam, at the moment, I don't have anything built on my PoC environment; my plan first is to get a qcow2 image that I have generated, and then deploy that | 23:05 |
| dlaube | NobodyCam: I think you mean me… maybe | 23:05 |
| NobodyCam | oh yes | 23:06 |
| NobodyCam | arif-ali: sorry | 23:06 |
| arif-ali | nvm | 23:06 |
| NobodyCam | dlaube: yes you are correct | 23:06 |
| dlaube | I trying to build anything so that I can get ironic-node validate to succeed | 23:06 |
| dlaube | I'd like to provision a node and have it deploy anything | 23:06 |
| dlaube | heh | 23:06 |
| NobodyCam | dlaube: ok so you need a deployment kerneel and ramdisk | 23:07 |
| NobodyCam | and a image to deploy | 23:07 |
| devananda | dlaube: which driver_info you need to suppy the node with depends on which driver you're using | 23:09 |
| devananda | dlaube: node-validate should inform you of the required / missing parameters | 23:09 |
| NobodyCam | dlaube: to use DIB to create a deployment ramdisk try: ramdisk-image-create -a i386 -o deploy.ramdisk deploy-ironic ubuntu | 23:09 |
| devananda | I wonder what would happen if I started tweeting hints about all the docs we're missing | 23:10 |
| jroll | lol | 23:10 |
| dlaube | NobodyCam: ok, so I guess just need to figure out why 'deploy.ramdisk' not found per http://pastie.org/private/6eajucaazonejzeh0dr45a | 23:10 |
| devananda | like how to use driver-list and node-validate to discover what configuration you need to pass in | 23:10 |
| dlaube | devananda: yessir, I'm using the pxe driver and I've got a handle on what im missing | 23:10 |
| NobodyCam | dlaube: to create a image for deployment type: TRIPLEO_ROOT/diskimage-builder/bin/disk-image-create ubuntu stackuser -a i386 -o LocalTestImage | 23:11 |
| dlaube | Can not validate PXE bootloader. The following parameters were not passed to ironic: ['pxe_root_gb', 'pxe_image_source', 'pxe_deploy_kernel', 'pxe_deploy_ramdisk'] | | 23:11 |
| dlaube | but this has helped me I think -> https://ask.openstack.org/en/question/30707/failed-to-deploy-error-kernel_id/ | 23:11 |
| dlaube | now at least I know I need a deploy ramdisk and an image before I can get back to trying to validate/provision a node | 23:12 |
| NobodyCam | dlaube: on the paste you showed the DIB build command was incorrect it combined the two | 23:12 |
| devananda | wait a sec | 23:12 |
| devananda | why does anyone need tripleo-image-elements to build the deploy ramdisk? | 23:12 |
| dlaube | good question ;) | 23:12 |
| devananda | it shouldn't | 23:13 |
| dlaube | I'd love to help you guys sort out docs as I discover issues | 23:13 |
| devananda | it's not a service image | 23:13 |
| devananda | and looking in t-i-e righ tnow, I dont see it | 23:13 |
| devananda | pretty sure it's NOT there | 23:13 |
| adam_g | yea, all the elements we need are in diskimage-builder's elements | 23:13 |
| NobodyCam | I thought dlaube was using DIB | 23:13 |
| devananda | dlaube: where did you see the requirement for "deploy.ironic" ? | 23:13 |
| devananda | er | 23:13 |
| JayF | devananda: dlaube: Whoops, sorry for being wrong :) | 23:13 |
| devananda | i mean "deploy.ramdisk" | 23:13 |
| NobodyCam | dlaube: that is a mashup of ramdisk create and image create commands | 23:14 |
| devananda | JayF: np. I am. Quite often. | 23:14 |
| JayF | devananda: pretty sure I said it was in t-i-e in the googdoc as well | 23:14 |
| JayF | summit planning and such | 23:14 |
| dlaube | https://pypi.python.org/pypi/diskimage-builder/0.1.31 | 23:14 |
| dlaube | and here https://github.com/openstack/diskimage-builder | 23:14 |
| devananda | dlaube: youre missing the "-o" | 23:14 |
| devananda | which means out file name | 23:14 |
| dlaube | ahhhh | 23:15 |
| devananda | :) | 23:15 |
| dlaube | well, the example in both of those still throws an error for me | 23:16 |
| dlaube | root@lab5:~# ramdisk-image-create -o deploy.ramdisk deploy-ironic | 23:16 |
| dlaube | Building elements: ramdisk base deploy-ironic | 23:16 |
| dlaube | ERROR: Please include an operating system element. | 23:16 |
| dlaube | im going to try what NobodyCam mentioned above and see if that works | 23:16 |
| devananda | JayF: updated gdoc to remove t-i-e reference | 23:16 |
| JayF | ty | 23:16 |
| devananda | dlaube: it should automaticaly assume your current base os. but yes, specifying it is helpful | 23:17 |
| dlaube | ok, looks like this is building me an image "root@lab5:~/image-work/tmpy/diskimage-builder# ./bin/disk-image-create ubuntu stackuser -a i386 -o LocalTestImage" | 23:18 |
| dlaube | but I still need to build a ramdisk? | 23:19 |
| NobodyCam | yes that is the image for deployment | 23:19 |
| devananda | yep. that'll give you a basic 32-bit cloud image withthe stackuser account | 23:19 |
| devananda | you stil need the ramdisk | 23:19 |
| NobodyCam | yep | 23:19 |
| JayF | devananda: should we build pxe ramdisks and publish them like we do for IPA images? | 23:19 |
| NobodyCam | try: ramdisk-image-create -a i386 -o deploy.ramdisk deploy-ironic ubuntu for that | 23:20 |
| JayF | devananda: seems like this is an onerous process to require someone to do to use ironic :) | 23:20 |
| devananda | JayF: yes | 23:20 |
| devananda | JayF: frankly i dont know why dib isn't doing that | 23:20 |
| JayF | dlaube: fwiw; if you ever choose to use the agent deploy driver, we publish premade kernel/ramdisk on every commit here: http://tarballs.openstack.org/ironic-python-agent/coreos/files/ | 23:20 |
| dlaube | NobodyCam: sweet! looks like that's building | 23:20 |
| devananda | well, besides me not having done it ... heh | 23:20 |
| dlaube | oh nice JayF! | 23:21 |
| NobodyCam | sould we come up with a build images script that could download DIb and run the command to build images? | 23:21 |
| * jroll takes JayF's marketing hat away | 23:21 | |
| devananda | JayF: feel inclined to add that post-commit job to dib? | 23:21 |
| NobodyCam | we have a tools folder | 23:21 |
| JayF | devananda: I've never built anything using dib :x (except for trying to run some of the builders that were WIP for IPA) | 23:22 |
| *** penick has quit IRC | 23:22 | |
| devananda | ack | 23:22 |
| dlaube | now that I've got images and a ramdisk to play with, should I pick up with steps beginning with "glance image-create" from https://wiki.openstack.org/wiki/Baremetal ? | 23:23 |
| dlaube | or are there better docs somewhere that I should follow | 23:23 |
| NobodyCam | dlaube: yup... :) | 23:23 |
| dlaube | awesomesauce | 23:23 |
| JayF | dlaube: NobodyCam: That wiki is listed as deprecated? | 23:23 |
| NobodyCam | oh wait | 23:24 |
| JayF | dlaube: NobodyCam: like it's for nova-bm, not ironic? | 23:24 |
| NobodyCam | ya | 23:24 |
| dlaube | yeah, that's why I asked | 23:24 |
| JayF | I think https://wiki.openstack.org/wiki/Ironic is right for us | 23:24 |
| NobodyCam | just saw the url | 23:24 |
| NobodyCam | ++++ | 23:24 |
| devananda | oh, right | 23:24 |
| * devananda deletes it | 23:24 | |
| NobodyCam | :) | 23:24 |
| dlaube | but there are no glance image creation docs in the proper wiki… is there? | 23:25 |
| * devananda looks at the page before deleting it, then regrets having done that | 23:26 | |
| dlaube | I saved the old wiki to a PDF if you need to recover ;) | 23:27 |
| devananda | dlaube: i'm not removing history :) | 23:28 |
| dlaube | ahh right, you can pick the previous revision | 23:28 |
| *** penick has joined #openstack-ironic | 23:30 | |
| NobodyCam | devananda: what that was two or more years ago | 23:30 |
| *** takadayuiko has joined #openstack-ironic | 23:34 | |
| *** marcoemorais has quit IRC | 23:42 | |
| *** marcoemorais has joined #openstack-ironic | 23:49 | |
| *** praneshp has quit IRC | 23:55 | |
| *** marcoemorais has quit IRC | 23:58 | |
| *** marcoemorais1 has joined #openstack-ironic | 23:58 | |
| *** marcoemorais1 has quit IRC | 23:58 | |
| *** marcoemorais has joined #openstack-ironic | 23:59 | |
| *** penick has quit IRC | 23:59 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
