| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic: Adding a reference driver for the agent https://review.openstack.org/84795 | 00:01 |
|---|---|---|
| *** newell_ has quit IRC | 00:20 | |
| *** matsuhashi has joined #openstack-ironic | 00:27 | |
| *** matsuhashi has quit IRC | 00:40 | |
| *** matsuhashi has joined #openstack-ironic | 00:41 | |
| *** matsuhashi has quit IRC | 00:45 | |
| openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Implement API to get driver properties https://review.openstack.org/73005 | 00:47 |
| *** nosnos has joined #openstack-ironic | 00:48 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic-python-agent: Updated from global requirements https://review.openstack.org/88722 | 01:01 |
| *** rloo has quit IRC | 01:02 | |
| *** matsuhashi has joined #openstack-ironic | 01:36 | |
| *** rameshg87 has joined #openstack-ironic | 02:18 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic: Updated from global requirements https://review.openstack.org/92953 | 02:23 |
| *** eguz has quit IRC | 02:36 | |
| *** harlowja is now known as harlowja_away | 02:47 | |
| *** matsuhashi has quit IRC | 03:03 | |
| *** matsuhashi has joined #openstack-ironic | 03:11 | |
| *** nosnos has quit IRC | 03:15 | |
| *** coolsvap|afk is now known as coolsvap | 03:40 | |
| *** matsuhashi has quit IRC | 03:42 | |
| *** matsuhashi has joined #openstack-ironic | 03:44 | |
| *** nosnos has joined #openstack-ironic | 04:03 | |
| *** rameshg87 has quit IRC | 04:06 | |
| *** rameshg87 has joined #openstack-ironic | 04:22 | |
| *** eghobo has joined #openstack-ironic | 04:24 | |
| *** eghobo has quit IRC | 04:41 | |
| *** eghobo has joined #openstack-ironic | 04:43 | |
| *** eghobo has quit IRC | 04:48 | |
| *** rameshg87 has quit IRC | 04:55 | |
| *** rameshg87 has joined #openstack-ironic | 04:58 | |
| *** rameshg871 has joined #openstack-ironic | 05:17 | |
| *** rameshg87 has quit IRC | 05:20 | |
| *** Nisha has joined #openstack-ironic | 05:41 | |
| *** eghobo has joined #openstack-ironic | 05:52 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic: Imported Translations from Transifex https://review.openstack.org/88508 | 06:08 |
| *** rameshg871 has quit IRC | 06:10 | |
| *** nosnos has quit IRC | 06:20 | |
| *** nosnos has joined #openstack-ironic | 06:20 | |
| *** Mikhail_D_ltp has joined #openstack-ironic | 06:23 | |
| *** hemna_ has quit IRC | 06:29 | |
| *** lsmola has joined #openstack-ironic | 06:41 | |
| *** uberj has quit IRC | 06:54 | |
| *** uberj_ has joined #openstack-ironic | 06:55 | |
| *** eghobo has quit IRC | 07:06 | |
| *** mrda is now known as mrda-travel | 07:16 | |
| *** rameshg87 has joined #openstack-ironic | 07:18 | |
| *** matsuhashi has quit IRC | 07:24 | |
| *** matsuhashi has joined #openstack-ironic | 07:24 | |
| *** matsuhas_ has joined #openstack-ironic | 07:41 | |
| *** matsuhashi has quit IRC | 07:41 | |
| *** nosnos_ has joined #openstack-ironic | 07:42 | |
| *** nosnos has quit IRC | 07:42 | |
| *** jistr has joined #openstack-ironic | 07:48 | |
| *** derekh_ has joined #openstack-ironic | 08:06 | |
| *** romcheg1 has joined #openstack-ironic | 08:09 | |
| *** rameshg87 has quit IRC | 08:10 | |
| *** rameshg87 has joined #openstack-ironic | 08:15 | |
| *** foexle has joined #openstack-ironic | 08:17 | |
| *** foexle has quit IRC | 08:20 | |
| *** martyntaylor has joined #openstack-ironic | 08:27 | |
| *** lucasagomes has joined #openstack-ironic | 08:29 | |
| *** eghobo has joined #openstack-ironic | 08:39 | |
| *** Mikhail_D_ltp has quit IRC | 08:41 | |
| openstackgerrit | ChangBo Guo(gcb) proposed a change to openstack/ironic: Clean up openstack-common.conf https://review.openstack.org/91992 | 08:47 |
| *** geertj has joined #openstack-ironic | 08:48 | |
| *** ndipanov has joined #openstack-ironic | 08:56 | |
| *** athomas has joined #openstack-ironic | 09:15 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Remove non-existent 'pxe_default_format' parameter from patcher https://review.openstack.org/93012 | 09:33 |
| *** stephenpearson has joined #openstack-ironic | 09:33 | |
| *** nosnos_ has quit IRC | 09:38 | |
| *** matsuhas_ has quit IRC | 09:41 | |
| *** matsuhashi has joined #openstack-ironic | 09:48 | |
| *** nosnos has joined #openstack-ironic | 09:49 | |
| *** eghobo has quit IRC | 09:50 | |
| *** stephenpearson has quit IRC | 09:50 | |
| *** nosnos has quit IRC | 09:50 | |
| *** spearson has joined #openstack-ironic | 09:50 | |
| *** nosnos has joined #openstack-ironic | 09:51 | |
| openstackgerrit | A change was merged to openstack/ironic: Pin RPC client version min == max https://review.openstack.org/92723 | 09:54 |
| *** nosnos has quit IRC | 09:55 | |
| *** athomas has quit IRC | 09:56 | |
| *** athomas has joined #openstack-ironic | 10:01 | |
| *** matsuhashi has quit IRC | 10:12 | |
| *** rameshg87 has left #openstack-ironic | 10:18 | |
| *** geertj has quit IRC | 10:40 | |
| *** rameshg87 has joined #openstack-ironic | 10:41 | |
| *** coolsvap is now known as coolsvap|afk | 10:56 | |
| dtantsur | Nisha, hi, sorry for not replying - yesterday was holiday in Czech republic. Are you here? | 11:07 |
| dtantsur | and g'afternoon, Ironic :) | 11:07 |
| dtantsur | Nisha, when you want to check that some mocked function was called with appropriate arguments, you can use request_ilo_moc.assert_called_once_with(arg1, arg2, ...) | 11:08 |
| rameshg87 | hello dtantsur: | 11:09 |
| rameshg87 | nisha and i work in the same team. | 11:09 |
| rameshg87 | dtantsur: we were checking regarding your comment on using mock with assert_called_with() | 11:10 |
| rameshg87 | dtantsur: the problem seems to be that _request_ilo takes a xml.etree.ElementTree element as its argument | 11:13 |
| rameshg87 | dtantsur: https://review.openstack.org/#/c/89500/8/ironic/drivers/modules/ilo/ribcl.py | 11:14 |
| rameshg87 | dtantsur: the xml.etree.ElementTree object doesn't support comparison. http://stackoverflow.com/questions/7905380/testing-equivalence-of-xml-etree-elementtree | 11:14 |
| rameshg87 | dtantsur: so using assert_called_with() is difficult with this. | 11:15 |
| Shrews | lucasagomes: would be nice to have a better way to make sure that fields added to a patch are properly removed :/ | 11:15 |
| dtantsur | rameshg87, well... that becomes troublesome. Let me think. | 11:15 |
| rameshg87 | dtantsur: we were thinking if just checking whether request_ilo is called or not would suffice | 11:15 |
| lucasagomes | Shrews, indeed :/ | 11:15 |
| lucasagomes | Shrews, it will once we get the driver properties patch in | 11:15 |
| Shrews | cool | 11:16 |
| lucasagomes | Shrews, so we can ask the api what are the mandatory/optional options | 11:16 |
| lucasagomes | but right now we have to do it manually still | 11:16 |
| dtantsur | rameshg87, I see 2 possibilities: 1. check that request_ilo was called; should be mostly ok; 2. replace _request_ilo with fake function that checks XML tree. I'm ok with doing 1. | 11:17 |
| rameshg87 | dtantsur: i understand your point here. without checking anything, we cannot find out whether it triggered the usecase or not | 11:18 |
| rameshg87 | dtantsur: but for this, if we atleast make sure that request_ilo was called, then the mock would have been returned and then the parsing would have been done. | 11:19 |
| rameshg87 | dtantsur: so, does 1) sound fine to you ? | 11:19 |
| dtantsur | rameshg87, _ideally_ you should mock _request_ilo with function that checks at least basic properties of generated XML tree. But! this can be very complicated, I am not sure, so 1. sounds fine, yes. | 11:20 |
| rameshg87 | dtantsur: yes, doing that would complicate. we would need to get the xml passed to request_ilo, and then make sure if the child element exists properly. | 11:21 |
| rameshg87 | thanks dtantsur | 11:22 |
| dtantsur | ur welcome :) | 11:23 |
| dtantsur | Core folks, lucasagomes and others: I have any easy thing to +A: https://review.openstack.org/#/c/92367/ :) Already has 2 +2 and 1 +1, could you have a look :) | 11:24 |
| lucasagomes | dtantsur, heh it has my +2 already... as another core weight in I will just approve then | 11:25 |
| dtantsur | lucasagomes, you mean, the 3rd +2? | 11:25 |
| lucasagomes | dtantsur, no, I mean as we already have 2 +2 I will just go and approve it | 11:26 |
| lucasagomes | done | 11:26 |
| dtantsur | thanks ) | 11:26 |
| *** rameshg87 has left #openstack-ironic | 11:29 | |
| *** lucasagomes is now known as lucas-hungry | 11:29 | |
| openstackgerrit | ChangBo Guo(gcb) proposed a change to openstack/ironic: Clean up openstack-common.conf https://review.openstack.org/91992 | 11:30 |
| * devananda hates slideshare right now | 11:49 | |
| romcheg1 | Morning folks! | 11:49 |
| *** jbjohnso has joined #openstack-ironic | 11:51 | |
| NobodyCam | good morning Ironc, says the man make'n coffee | 11:51 |
| dtantsur | morning romcheg1, NobodyCam! | 11:52 |
| NobodyCam | morning devananda romcheg1 dtantsur and lucas-hungry | 11:52 |
| NobodyCam | :) | 11:52 |
| dtantsur | devananda, morning! Slides've never been easy :( | 11:52 |
| devananda | the slides be done. but slideshare doesn't seem to like google pres | 11:52 |
| NobodyCam | :( | 11:53 |
| NobodyCam | so devananda I found whats up with rebuild | 11:53 |
| NobodyCam | https://bugs.launchpad.net/ironic/+bug/1317647 | 11:53 |
| devananda | aaand i haven't figured out how to remoev an upload (or test their conversion ahead of time) | 11:53 |
| Shrews | NobodyCam: hot enough in Atl for ya? :) | 11:54 |
| * Shrews welcomes NobodyCam to the SE spring heat | 11:55 | |
| NobodyCam | I dont know.. have both a/c on so rv is nice and cool | 11:55 |
| Shrews | that's cheating | 11:55 |
| NobodyCam | cheating is useing all three a/c's :0p | 11:55 |
| * Shrews heads out for a morning run before the heat makes him melt. biab | 11:56 | |
| NobodyCam | but would have you know start the rv to use the one for the engine | 11:56 |
| devananda | NobodyCam: humm... that seems like a bug in parted. has lucas-hungry seen this yet? | 11:56 |
| NobodyCam | enjot Shrews :) | 11:56 |
| NobodyCam | no that I know of ... will getting wth him when he gets back | 11:57 |
| * NobodyCam need the coffee to become ready | 11:57 | |
| dtantsur | devananda, NobodyCam and other cores: mind having a look at BP https://blueprints.launchpad.net/ironic/+spec/nodes-tagging ? This is by request of #tuskar. (I now it should pass through new spec approval process, just want quick opinion) | 11:59 |
| NobodyCam | dtantsur: with out coffee... Is that not out extra field? | 12:00 |
| NobodyCam | s/out/our/ | 12:01 |
| dtantsur | NobodyCam, we can easily have an extra field, but we need also 1. filtering in API 2. filtering in scheduler (and as I wrote here, I'm not sure where it belongs: in Ironic or in Nova) | 12:02 |
| dtantsur | did I get you right (didn't have coffee today as well :( ) | 12:02 |
| NobodyCam | hehehe :) | 12:03 |
| romcheg1 | Might be interesting http://f.cl.ly/items/0V0V163F341x0x0e1B0b/output.html | 12:04 |
| NobodyCam | romcheg1: neet stuff | 12:05 |
| romcheg1 | Punishing IPMI server now for more real stuff | 12:05 |
| devananda | dtantsur: fwiw, this is on my agenda to talk about at the summit already, but there is a parallel discusion within nova about group vs tag | 12:07 |
| devananda | dtantsur: did you see the ML thread jaypipes started? | 12:07 |
| dtantsur | devananda, no, probably missed. Do you by chance have a link? | 12:07 |
| dtantsur | So I probably should wait for outcome of the Summit, right? | 12:08 |
| NobodyCam | :) | 12:08 |
| devananda | dtantsur: right. also, it would be good to have a description of what problem this feature is trying to solve | 12:10 |
| devananda | proposing a new feature without any (documented) reason does not allow others to think about whether this is an appropriate/optimal solution | 12:11 |
| dtantsur | lsmola, ping, can you think about a good "Why" section ^^^ ? My English lets me down today :( | 12:11 |
| devananda | dtantsur: here's the ML topic i mentioned: http://lists.openstack.org/pipermail/openstack-dev/2014-April/033746.html | 12:13 |
| lsmola | dtantsur: if I remember user stories from jcoufal, it should allow us to control what will be deployed where manually by tagging the nodes | 12:13 |
| lsmola | so e.g. tag=rack1234, then we add to flavor that it can deploy on rack1234 | 12:14 |
| devananda | lsmola: that is going to lead to an explosion of flavors | 12:14 |
| lsmola | the difference with groups was that node can be only in one group, but it can have several tags | 12:14 |
| devananda | and significantly increase complexity in your heat templates | 12:14 |
| lsmola | devananda: you should able to define multiple tags to flavor | 12:15 |
| devananda | yes, but | 12:15 |
| NobodyCam | ya I'm not I like it on the flavor | 12:15 |
| lsmola | devananda: you will have e.g. one compute flavor and list of tags it can deploy to, as well as other conditions if needed | 12:15 |
| devananda | lsmola: the "list of tags" doesn' tmake sense to me | 12:16 |
| NobodyCam | gah .. I really need to drink the coffee | 12:16 |
| lsmola | coffee is a good idea :-) | 12:16 |
| NobodyCam | I'm not sure I like taging on the flavor | 12:16 |
| devananda | lsmola: I'd like to see a write-up of how multiple tags on both flavor and node will be utilized, considering alternative usages for this capability | 12:17 |
| devananda | it seems like arbitrary N::M matching would be very fragile | 12:17 |
| lsmola | not sure if it is the best, but is is easily doable, it's basically matching of some extra specs of flavor with Ironic node | 12:18 |
| NobodyCam | maybe as a nova boot flag? but deva brings up a very good point | 12:18 |
| devananda | lsmola: would I need a different flavor for each rack? | 12:18 |
| devananda | - if yes, it's going to break heat | 12:19 |
| devananda | - if no, then why have a tag for rack at all? | 12:19 |
| lsmola | devananda: well I would say no, otherwise it would be crazy, but it would require OR condition in extra specs matching | 12:19 |
| devananda | - and how is a "compute" flavor on rack 1[,2,3,...] different from a compute flavor on rack 4[,5,6,...] | 12:20 |
| lsmola | devananda: well it was maybe not a good idea, but the user story was that we shold be able to manually determine where we want to deploy stuff | 12:20 |
| devananda | AFAIUI, having >1 "compute" flavor is going to break heat templates in tripleo | 12:20 |
| devananda | lsmola: define "manually" please | 12:21 |
| lsmola | devananda: either by groups or tagging | 12:21 |
| devananda | tags on flavors and ironic nodes would be static -- so, not a manual user process | 12:21 |
| devananda | a user with admin privileges can already specify the node on which to boot an instance, using the "--availability-zone az:host:node" extension | 12:22 |
| lsmola | devananda: hm, didn't investigate this, if that is true, it should be enough | 12:24 |
| devananda | :) | 12:25 |
| NobodyCam | :) | 12:25 |
| dtantsur | ok, so we don't have an easy consensus on this :) For what I see in ML (thanks for link, devananda), interacting with Nova may be troublesome as well... | 12:26 |
| lsmola | hopefully jcoufal will join here and give us some more user stories for tags | 12:26 |
| devananda | dtantsur: see ^ ... if the user story is "manually pick the node to deploy to" -- there is another, existing solution. and we really shouldn't use tags for that | 12:26 |
| *** jcoufal has joined #openstack-ironic | 12:26 | |
| devananda | lsmola: that'd be great | 12:27 |
| lsmola | devananda: i think it was more like manually identify a group of nodes and deploy to them | 12:27 |
| lsmola | devananda: so host groups should be usable for that too | 12:27 |
| devananda | lsmola: well, nova host groups demarcate nova-compute hosts, not ironic nodes | 12:28 |
| devananda | there are several talks about the confusion between a compute host and a hypervisor_hostname | 12:29 |
| lsmola | devananda: hm I wonder if for TripleO those should be the same | 12:29 |
| devananda | and how this is a problem for all federated hypervisors | 12:29 |
| devananda | lsmola: sorry, which should be the same? | 12:29 |
| jcoufal | lsmola: how can I help? | 12:29 |
| lsmola | jcoufal: guys wanted to hear some user stories of Ironic node tags | 12:30 |
| lsmola | jcoufal: plus how we want to use it for scheduler filter | 12:30 |
| devananda | jcoufal: hi! can you point me at some descriptions / user stories for ironic and targeting nodes | 12:30 |
| devananda | jcoufal: I would rather start from the "why" instead of the "how - use tags" | 12:31 |
| jcoufal | lsmola: there is couple of advantages for them | 12:31 |
| jcoufal | first of all it is arbitrary grouping of nodes | 12:31 |
| jcoufal | Ironic can't know all the situations and options how the user might want to group the nodes, so this will help them | 12:32 |
| jcoufal | they can filter their views based on assigned tags | 12:32 |
| lsmola | so used for table views and chart views, right? | 12:33 |
| jcoufal | and for deployment reasons - similar situation - if user wants bigger restriction than just HW profile, he can restrict the group of nodes by the tag | 12:33 |
| spearson | Might tags be useful for recording failed status, or other lifecycle states (decommissioning etc)? | 12:33 |
| jcoufal | lsmola: yes, there it would be very useful | 12:33 |
| devananda | we already have a "maintenance" flag on nodes which does ^ | 12:34 |
| jcoufal | spearson: I think Ironic has special field for status | 12:34 |
| spearson | ok | 12:34 |
| devananda | a tag will not prevent Ironic's automated operations (health checks, etc) on nodes | 12:34 |
| devananda | even if a tag would give a scheduler hint to, say, avoid that node -- it wouldn't prevent it from direct use | 12:34 |
| jcoufal | devananda: by "direct use" you mean user getting to the node detail and provisioning it manually? | 12:35 |
| devananda | jcoufal: ironic exposes an API to manipulate the node -- while Nova is the most likely way that someone will interact with Ironic, well, there's an API, so any user (with admin credentials) can use it, too | 12:36 |
| devananda | jcoufal: so maintenance status needed to be tracked separately, so that ironic actually enforces a hands-off status on that node | 12:36 |
| lsmola | jcoufal: about the deployment, the story was deploy on nodes with all of these tags, with at least one of this tag, or a complex condition? | 12:36 |
| *** tatyana has joined #openstack-ironic | 12:37 | |
| devananda | jcoufal: as far as HW profile, do you envision the tags set on a flavor? or passed at "nova boot" time? | 12:37 |
| lsmola | devananda: i would keep it in TripleO scope rather, we can't probably cover all use cases | 12:37 |
| *** tatyana has quit IRC | 12:38 | |
| jcoufal | devananda: right, I agree on that | 12:38 |
| devananda | jcoufal: which :) | 12:38 |
| jcoufal | devananda: I envision it to be part of the flavor | 12:38 |
| devananda | ok | 12:38 |
| lsmola | devananda: we can use either flavors or scheduler hints, cause we use heat | 12:38 |
| devananda | so, I am cnocerned that tags statically assigned to flavor and node will be inflexible and likely to break any orchestration | 12:38 |
| jcoufal | the agreement was on the first statement :) | 12:38 |
| lsmola | devananda: btw. it should be dynamicaly assigned, tags would be somewhere in flavor metadata | 12:39 |
| lsmola | devananda: so you would just change it and run heat stack-update | 12:39 |
| openstackgerrit | A change was merged to openstack/ironic: Simplify calls to ImageCache in PXE module https://review.openstack.org/92367 | 12:39 |
| jcoufal | lsmola: regarding deployment - it depends on how complex condition we will implement. We should start with basics - deploy on node which is having this tag. | 12:39 |
| devananda | you seem to be assuming the user has admin privileges and can edit the nova flavor definition | 12:40 |
| devananda | yes? | 12:40 |
| lsmola | devananda: for tripleo yes | 12:40 |
| devananda | k | 12:40 |
| lsmola | devananda: we already use flavor extra specs for some other stuff | 12:40 |
| jcoufal | devananda: what exactly do you see as inflexibility and how it would break orchestration? | 12:41 |
| lsmola | devananda: like architecture, and link to deploy images | 12:41 |
| devananda | if I need to have multiple flavors that all represent a "compute" host, but each have some unique flags to indicate eg. which rack they're in | 12:41 |
| devananda | then my heat template is not portable between those racks | 12:42 |
| lsmola | devananda: yeah using one template for more deployments is not possible now | 12:43 |
| devananda | :( | 12:43 |
| *** lucas-hungry is now known as lucasagomes | 12:43 | |
| lucasagomes | NobodyCam, devananda romcheg morning | 12:43 |
| lsmola | devananda: though we might need to address it, it is one of the things we should talk about on Tuskar design session | 12:43 |
| lucasagomes | devananda, will take a look | 12:43 |
| NobodyCam | good mornig lucasagomes :) | 12:43 |
| romcheg1 | Morning lucasagomes! | 12:43 |
| NobodyCam | hey hey romcheg1 when does your flight leave? | 12:44 |
| romcheg1 | The day after tomorrow in the morning | 12:44 |
| NobodyCam | :) | 12:44 |
| romcheg1 | But since I follow the sun I will arrive the same day in the US :) | 12:45 |
| NobodyCam | and good morning (if I missed you earlier) | 12:45 |
| lucasagomes | NobodyCam, wow that's a nice error haha | 12:45 |
| lucasagomes | NobodyCam, I will give it some investigation | 12:45 |
| NobodyCam | oh ya | 12:45 |
| NobodyCam | wanta chat with you about it | 12:45 |
| NobodyCam | when you have a free miinutes | 12:45 |
| lucasagomes | sure | 12:45 |
| jcoufal | devananda: actually that's a purpose. It depends how you will label the nodes, but if I label them rack1 and I include this tag into my flavor definition, then I specifically want the node to be in that rack. | 12:46 |
| jcoufal | devananda: it doesn't solve a situation when you say "I want one compute node per rack" - that's a different story | 12:46 |
| lsmola | lucasagomes: morning | 12:47 |
| NobodyCam | lucasagomes: I was thinking we could set the uuid to a quique value before we ran parted if the nodes disk volumes already have a label, Or just remove the labels? or... | 12:47 |
| lucasagomes | lsmola, morning | 12:47 |
| devananda | jcoufal: right -- so the community has a need for [anti]affinity scheduling in general | 12:48 |
| NobodyCam | s/uuid/label/ | 12:48 |
| lucasagomes | NobodyCam, yeah, I'm trying to reproduce it here, see if I can tell parted to ignore it | 12:48 |
| NobodyCam | :) | 12:48 |
| devananda | jcoufal: which this sounds like it doesn't really solve | 12:48 |
| lucasagomes | or in the last case we should workaround relabeling it yea | 12:48 |
| devananda | jcoufal: let me ask, from another angle, how useful will these manual tags be at very large scale? | 12:48 |
| *** radsy has joined #openstack-ironic | 12:49 | |
| *** radsy has joined #openstack-ironic | 12:49 | |
| lsmola | devananda: depends on customers :-) | 12:49 |
| jcoufal | devananda: yeah, I don't think that tags would solve [anti]affinity issues | 12:49 |
| NobodyCam | lucasagomes: what I got from the parted guys was that parted checks for: if /dev/sdbX shows up in /proc/mounts or /etc/mtab | 12:49 |
| jcoufal | devananda: as for large scales - it depends how the tagging will be used - if I tag larger groups or smaller ones | 12:50 |
| NobodyCam | but it also check for /dev/disk/by-label/<label> which is what was causing the error | 12:50 |
| lucasagomes | NobodyCam, :( it should make sure that the device is different | 12:51 |
| NobodyCam | my proc/mount have /dev/disk/by-label/cloudimg-rootfs | 12:51 |
| lucasagomes | sounds like a bug of their part | 12:51 |
| jcoufal | devananda: it doesn't have to necessarily be useful for compute nodes which are in larger scales, but for example for specific controller nodes which are really somehow dedicated | 12:51 |
| lsmola | devananda: what i like is matching of extra specs of flavors to ironic nodes, which can be used for tags if people wants | 12:51 |
| NobodyCam | is it really thou. we are the ones assigin the same label to both disks | 12:51 |
| lsmola | devananda: but allows people to mark their hardware with anything and then use it to matching | 12:52 |
| jcoufal | devananda: scheduling itself is (I would say) secondary usage which gives user more control over deployment | 12:52 |
| jcoufal | devananda: primary usage for tags is grouping and filtering | 12:52 |
| devananda | jcoufal: "grouping and filtering" in the UI -- this makes sense to me | 12:52 |
| lucasagomes | NobodyCam, hmm yeah, is it wrong to have the same label on diff devices? I think I gotta get my head about this before | 12:53 |
| devananda | jcoufal: but scheduling is what you have been talking about so far, and I think there are some larger issues there -- what if I want to say | 12:53 |
| devananda | *what if I want to schedule some instances with flavor tags == node tags | 12:53 |
| devananda | and another instance with node tags CONTAINS (some tag) | 12:54 |
| devananda | and ... so on. this requires restarting the scheduler, no? | 12:54 |
| devananda | basically, changing the logical matching of nova flavor tag <=> ironic node tag, is not (afaik) programatic at deploy time | 12:55 |
| lsmola | devananda: i would assume you would have to use another filter maybe and restart it yes | 12:55 |
| devananda | right | 12:55 |
| lsmola | devananda: I would say we should be fine with exact match for now | 12:56 |
| lsmola | devananda: because for anything else we would need more dynamic scheduler filters | 12:56 |
| jcoufal | I am sorry, I have to leave for now (meeting), bbl | 12:57 |
| lsmola | jcoufal: thanks, see ya | 12:57 |
| NobodyCam | thank you for the info jcoufal | 12:57 |
| NobodyCam | brb... | 12:58 |
| jcoufal | np, I am sorry for leaving you guys here, but we can catch up on this topic at summit the latest | 12:58 |
| *** geertj has joined #openstack-ironic | 12:58 | |
| devananda | these powercables aren't working and I need to save battery ... | 12:59 |
| devananda | lsmola, jcoufal - thanks for the discussion. let's definitely continue @summit | 12:59 |
| * devananda drops offline again | 13:00 | |
| lsmola | devananda: ok, see you there | 13:00 |
| dtantsur | ok, folks, so I leave BP as it is, waiting for your feedback from summit :) thank you | 13:05 |
| spearson | The "maintenance" flag on an ironic node is of type bool. That doesn't give you a lot of flexibility to record lifecycle state. Suppose we had a node in production that we wanted to indicate was to be scrapped at some point, but not immediately. I can't see an obvious way to do that currently. | 13:05 |
| *** matty_dubs|gone is now known as matty_dubs | 13:06 | |
| *** linggao has joined #openstack-ironic | 13:07 | |
| NobodyCam | enjoy devananda we'll see you at the summit in a few days :) | 13:08 |
| NobodyCam | morning matty_dubs :) | 13:18 |
| matty_dubs | Morning NobodyCam | 13:20 |
| NobodyCam | :) | 13:21 |
| lucasagomes | NobodyCam, on ur /proc/mount u see the label or the device path? | 13:22 |
| NobodyCam | lucasagomes: yep: /dev/disk/by-label/cloudimg-rootfs / ext4 rw,relatime,data=ordered 0 0 | 13:22 |
| lucasagomes | hmm | 13:22 |
| NobodyCam | lucasagomes: I know it comeing from the fact that all my images are duild with DIB | 13:24 |
| NobodyCam | s/duild/built | 13:25 |
| lucasagomes | NobodyCam, yeah, trying to reproduce it here | 13:25 |
| lucasagomes | NobodyCam, we can workaround it easily at the partition time | 13:25 |
| lucasagomes | but it's an interisting error, idk how parted should handle it | 13:25 |
| *** geertj has quit IRC | 13:25 | |
| NobodyCam | yea its a good one | 13:26 |
| NobodyCam | :-p | 13:26 |
| matty_dubs | https://review.openstack.org/#/projects/openstack/ironic,dashboards/important-changes:review-inbox-dashboard -- slick | 13:29 |
| NobodyCam | neat-o | 13:30 |
| *** jgrimm has joined #openstack-ironic | 13:38 | |
| NobodyCam | lucasagomes: you the failure in check-tempest-dsvm-virtual-ironic on https://review.openstack.org/#/c/91992/ ? actually may be a real error ... i'm rechecking | 13:48 |
| lucasagomes | NobodyCam, right... I think that test is broken :( it failed in pretty much all the new checks I saw today | 13:50 |
| Shrews | NobodyCam, lucasagomes: a fix for that is up for review: https://review.openstack.org/92883 | 13:54 |
| lucasagomes | Shrews, ah thanks | 13:55 |
| lucasagomes | merged failed :( | 13:55 |
| *** radsy has quit IRC | 13:55 | |
| Shrews | i'm sure adam_g will fix that soon | 13:56 |
| NobodyCam | :) | 14:01 |
| NobodyCam | lucasagomes: been able to reproduce the reuild error? | 14:02 |
| lucasagomes | NobodyCam, no :( I might try with devtest then | 14:04 |
| lucasagomes | NobodyCam, http://paste.openstack.org/show/79732/ | 14:04 |
| lucasagomes | but my /proc/mount is not showing the label, it shows the partition path | 14:04 |
| lucasagomes | I'm now looking at the parted code, and it seems to check on 3 places to see if the device is present | 14:04 |
| lucasagomes | return _mount_table_search( "/proc/mounts", dev) | 14:05 |
| lucasagomes | || _mount_table_search( "/proc/swaps", dev) | 14:05 |
| lucasagomes | || _mount_table_search( "/etc/mtab", dev); | 14:05 |
| NobodyCam | :) | 14:05 |
| lucasagomes | idk hw to get proc/mount to show the label | 14:05 |
| NobodyCam | use ubuntu | 14:06 |
| lucasagomes | maybe it's only because it was booted with ROOT=LABEL | 14:06 |
| lucasagomes | yeah or it's some ubuntuish thing | 14:06 |
| NobodyCam | the parted guys said it had to do with mtab and ubuntu... | 14:06 |
| lucasagomes | hmm | 14:07 |
| NobodyCam | I think it is the way ubuntu is mounting the root fs | 14:07 |
| lucasagomes | yeah | 14:07 |
| NobodyCam | on cloud images | 14:07 |
| lucasagomes | also, we wouldn't see the error if we wiped the data on the tenant disk before dpeloying it | 14:07 |
| lucasagomes | as suggested by https://bugs.launchpad.net/ironic/+bug/1174153 | 14:07 |
| * lucasagomes thinks about the best way to work it around for now | 14:08 | |
| NobodyCam | label do get reset on rebuild... execpt for EP if you are using --preserve-ephemeral | 14:09 |
| NobodyCam | we could just wire any label | 14:09 |
| NobodyCam | *wipe | 14:10 |
| lucasagomes | we have to maintain the ephemeral partition label | 14:10 |
| NobodyCam | yea | 14:11 |
| NobodyCam | in my testing with preserve-ephemeral it ofcourse did not get reset | 14:11 |
| *** geertj has joined #openstack-ironic | 14:12 | |
| lucasagomes | right yeah | 14:12 |
| lucasagomes | heh so tricky | 14:12 |
| lucasagomes | that was def a great catch hah idk I bet it took awhile to figure that out | 14:12 |
| NobodyCam | oh ya... was drving me up the wall | 14:13 |
| NobodyCam | :-p | 14:13 |
| NobodyCam | I ended up in the #parted channnel | 14:13 |
| * NobodyCam makes more coffee | 14:15 | |
| lucasagomes | hah awesome! good stuff | 14:16 |
| matty_dubs | lucasagomes: So I hit a weird issue last night. After maybe 2 minutes of running, Ironic would lose all of its drivers. ironic driver-list would return an empty set. | 14:17 |
| matty_dubs | Restarting conductor would fix it... for a couple minutes. | 14:18 |
| matty_dubs | Ever seen anything like that? Or have any idea how I'd even debug it? | 14:19 |
| NobodyCam | hart beat issue? | 14:19 |
| matty_dubs | If it makes a difference, I registered 80 VMs with Ironic | 14:20 |
| lucasagomes | yeah sounds like a hear beat issue | 14:20 |
| lucasagomes | there's a periodic tasks in the conductors | 14:20 |
| lucasagomes | to keep pining (pretty much touch a field in the db) to say it's alive | 14:20 |
| matty_dubs | Ooh! It takes forever to poll all 80. I wonder if that causes it to miss the heartbeat? | 14:20 |
| NobodyCam | matty_dubs: if so that a real bug | 14:21 |
| NobodyCam | thats* | 14:21 |
| lucasagomes | yeah | 14:22 |
| matty_dubs | How would I run this down? MariaDB [(none)]> select * from ironic.conductors; shows the conductor, with the right drivers | 14:22 |
| matty_dubs | Oh, but last updated 6 minutes ago | 14:22 |
| lucasagomes | matty_dubs, to pool u mean to list them? | 14:23 |
| NobodyCam | thats it | 14:23 |
| matty_dubs | lucasagomes: No, in the Conductor log, when it goes to refresh status | 14:23 |
| lucasagomes | oh i c | 14:23 |
| NobodyCam | sounds like heat beat should have its own thread | 14:24 |
| matty_dubs | Ha, so my scalability testing turned up something useful! | 14:24 |
| NobodyCam | so other tasks dont impact it | 14:24 |
| matty_dubs | Just, not at all what I was trying to test. ;) | 14:24 |
| lucasagomes | matty_dubs, https://github.com/openstack/ironic/blob/master/ironic/conductor/manager.py#L472-L474 | 14:24 |
| NobodyCam | but a great catch | 14:24 |
| lucasagomes | that's the periodic task | 14:24 |
| lucasagomes | +1 good catch | 14:25 |
| matty_dubs | The interesting thing here is that I didn't even _do_ anything. I just set up devstack with Ironic and had it set up 80 VMs. | 14:26 |
| matty_dubs | Makes me think there could be more problems to come | 14:26 |
| Shrews | nah, that's probably the only one | 14:26 |
| * Shrews goes to fill his half full coffee cup | 14:27 | |
| NobodyCam | oh Shrews cursed it | 14:27 |
| NobodyCam | lol there will be a ton more now | 14:27 |
| NobodyCam | lol | 14:27 |
| NobodyCam | hehehe | 14:27 |
| matty_dubs | haha | 14:27 |
| lucasagomes | so does periodic tasks run in parallel? | 14:28 |
| lucasagomes | if not we def should dedicate a greenthread to the heart bit | 14:29 |
| matty_dubs | Yeah, they ought to be parallel, but it seems like they might not be | 14:30 |
| NobodyCam | lucasagomes: with the critical nature of Hart beats I would think it SHOULD have it own green thread | 14:30 |
| matty_dubs | You can watch them just scroll by in the log | 14:30 |
| lucasagomes | :( | 14:31 |
| lucasagomes | def run_periodic_tasks(self, context, raise_on_error=False): | 14:32 |
| lucasagomes | """Tasks to be run at a periodic interval.""" | 14:32 |
| lucasagomes | idle_for = DEFAULT_INTERVAL | 14:32 |
| lucasagomes | for task_name, task in self._periodic_tasks: | 14:32 |
| lucasagomes | yeah... doesn't look to be in parallel | 14:32 |
| lucasagomes | >.< | 14:32 |
| NobodyCam | :( | 14:32 |
| lucasagomes | I assume it would, sounds logical to just a thread pool on the periodic tasks | 14:32 |
| matty_dubs | Should I file a bug? | 14:33 |
| lucasagomes | matty_dubs, yeah please | 14:33 |
| *** killer_prince has joined #openstack-ironic | 14:33 | |
| NobodyCam | matty_dubs: YES! | 14:33 |
| NobodyCam | :) | 14:33 |
| lucasagomes | brb will make some tea | 14:34 |
| jroll | matty_dubs: https://bugs.launchpad.net/ironic/+bug/1308680 | 14:34 |
| jroll | morning ironic | 14:34 |
| jroll | happy friday! | 14:34 |
| NobodyCam | good morning jroll TGIF!!!! | 14:34 |
| NobodyCam | brb | 14:35 |
| matty_dubs | jroll: Ah-ha, a step ahead of me, I see | 14:35 |
| jroll | :) | 14:35 |
| *** tteggel has quit IRC | 14:39 | |
| *** tteggel has joined #openstack-ironic | 14:41 | |
| NobodyCam | thats the way jroll rolls :-p | 14:41 |
| matty_dubs | Punny! | 14:43 |
| NobodyCam | lucasagomes: we could use blkid to get the lable of the volumes and then just reset the ones we don't touch (ie. EP) for reuilds? | 14:44 |
| NobodyCam | matty_dubs: I was going to say "thats the way j rolls.." but I hit tab by default | 14:45 |
| NobodyCam | :-p | 14:45 |
| lucasagomes | NobodyCam, yeah, or simply reset the labels of the partitions we are going to overwirte | 14:45 |
| lucasagomes | heh j rolls | 14:45 |
| NobodyCam | what if the conductor had an EP | 14:45 |
| NobodyCam | like the under cloud should | 14:46 |
| NobodyCam | I did my testing with seed | 14:46 |
| lucasagomes | NobodyCam, what we could do in this case is, reset all the labels | 14:46 |
| lucasagomes | if preserve_ephemeral is set | 14:47 |
| lucasagomes | instead of doing a mkfs on the ephemeral partition | 14:47 |
| lucasagomes | we just rewrite the label there | 14:47 |
| NobodyCam | any volume we rebuild will get reset in the rebuild. so we only need to reset the ones we dont rebuild | 14:47 |
| NobodyCam | yea that would work too | 14:48 |
| jroll | lol NobodyCam | 14:48 |
| *** radsy has joined #openstack-ironic | 14:48 | |
| *** radsy has joined #openstack-ironic | 14:48 | |
| NobodyCam | jroll: lol :) | 14:48 |
| *** jcoufal_ has joined #openstack-ironic | 14:49 | |
| *** jcoufal has quit IRC | 14:50 | |
| NobodyCam | lucasagomes: ya I think your idea is the cleanest code path for this.. | 14:54 |
| lucasagomes | yeah :( but still an ugly workaround to have | 14:55 |
| NobodyCam | ya but I think we are going to need it. | 14:55 |
| NobodyCam | :( | 14:55 |
| NobodyCam | lucasagomes: https://review.openstack.org/#/c/91992/ failed even worse on recheck | 14:58 |
| *** uberj_ is now known as uberj | 14:59 | |
| lucasagomes | NobodyCam, :/ check-tempest-dsvm-virtual-ironic is broken | 14:59 |
| NobodyCam | :( | 14:59 |
| NobodyCam | :-p | 14:59 |
| lucasagomes | the other one failed @ the beginning of the tests :/ | 14:59 |
| lucasagomes | I don't think that the change is causing it tho | 15:00 |
| lucasagomes | jroll, so that's fixed just by tweaking the rpc_thread_pool_size? | 15:00 |
| NobodyCam | this was my concern: http://logs.openstack.org/92/91992/5/check/check-tempest-dsvm-virtual-ironic/fd8f1cd/console.html#_2014-05-09_14_11_12_775 | 15:01 |
| jroll | lucasagomes: it mitigates it, but I think it may slow things down a bit | 15:01 |
| lucasagomes | NobodyCam, https://review.openstack.org/#/c/92883/ | 15:02 |
| NobodyCam | :) ahh TY | 15:02 |
| lucasagomes | jroll, what you think about having a dedicated greenthread to run the keepalive? | 15:02 |
| matty_dubs | That's not unreasonable, but IMHO missing heartbeat is just a symptom of a larger problem, not _the_ problem | 15:03 |
| jroll | ^ | 15:04 |
| jroll | lucasagomes: I agree with matt but I think it might be a good fix in the meantime | 15:04 |
| lucasagomes | right, I'm worried that periodic tasks don't run in parallel | 15:04 |
| lucasagomes | jroll, right, mind if I give it a go/test? | 15:05 |
| jroll | lucasagomes: all yours :) | 15:05 |
| lucasagomes | cheers | 15:05 |
| * jroll bbiab | 15:05 | |
| NobodyCam | bbt...brb | 15:06 |
| Shrews | wow, lots of open reviews | 15:09 |
| * Shrews slows Earth's rotation to make the day 36 hours long | 15:10 | |
| NobodyCam | oh and on a friday ... wont that make the work day longer too :-p | 15:11 |
| Shrews | NobodyCam: only for you | 15:12 |
| NobodyCam | :-p | 15:12 |
| Shrews | hehe | 15:12 |
| * NobodyCam thinks he'll need more coffee | 15:12 | |
| *** rameshg87 has joined #openstack-ironic | 15:27 | |
| *** rameshg87 has left #openstack-ironic | 15:28 | |
| *** geertj has quit IRC | 15:28 | |
| NobodyCam | lucasagomes: are you poking at the label stuff? | 15:30 |
| lucasagomes | NobodyCam, not in the moment, I'm testing the greenthread stuff heh | 15:30 |
| NobodyCam | ack | 15:30 |
| *** killer_prince has quit IRC | 15:31 | |
| lucasagomes | NobodyCam, almost done with ^ | 15:31 |
| NobodyCam | :) | 15:33 |
| jbjohnso | now that's a fun mistake, used wrong list name in a place, produced effectively 'for a in mylist: mylist.append(a)' | 15:40 |
| NobodyCam | the never ending loop | 15:41 |
| NobodyCam | sounds like a movie | 15:41 |
| *** eghobo has joined #openstack-ironic | 15:41 | |
| *** comstud is now known as bearhands | 15:42 | |
| NobodyCam | brb | 15:48 |
| *** matty_dubs is now known as matty_dubs|lunch | 15:57 | |
| *** geertj has joined #openstack-ironic | 16:03 | |
| *** lsmola has quit IRC | 16:05 | |
| *** derekh_ has quit IRC | 16:07 | |
| *** radsy has quit IRC | 16:11 | |
| *** ndipanov has quit IRC | 16:21 | |
| *** martyntaylor has quit IRC | 16:25 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Run keepalive in a dedicated thread https://review.openstack.org/93083 | 16:25 |
| lucasagomes | matty_dubs|lunch, ^ when u get a time can you give it a go in ur env with 80 vms? | 16:25 |
| *** jistr is now known as jistr|gone | 16:27 | |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic: Adding a reference driver for the agent https://review.openstack.org/84795 | 16:29 |
| spearson | See you all next week. | 16:34 |
| NobodyCam | see ya spearson :) enjoy the travel | 16:34 |
| spearson | Flying Delta, so should be extra fun :) | 16:34 |
| NobodyCam | lol | 16:34 |
| *** spearson has quit IRC | 16:35 | |
| lucasagomes | I booked with airfrance but it's going to be Delta as well :( | 16:35 |
| NobodyCam | :( | 16:35 |
| lucasagomes | partnership | 16:35 |
| lucasagomes | at least I got a directly flight! yay | 16:36 |
| NobodyCam | yep. start aligence or some such | 16:36 |
| lucasagomes | NobodyCam, how is the weather over there? | 16:36 |
| lucasagomes | better now? | 16:36 |
| NobodyCam | warm and muggie | 16:36 |
| *** zdin0bot has joined #openstack-ironic | 16:36 | |
| *** dwalleck has joined #openstack-ironic | 16:37 | |
| NobodyCam | but ok.. no storms | 16:37 |
| lucasagomes | :) | 16:37 |
| lucasagomes | atlanta seems pretty nice from what I saw in the street view | 16:37 |
| NobodyCam | :) I've been watching thing bait car Atlanta, and Cops Atlanta...lol seen a different side | 16:38 |
| NobodyCam | :-p | 16:38 |
| NobodyCam | s/thing/things/ | 16:38 |
| NobodyCam | s/things/things like/ | 16:39 |
| * NobodyCam deff tell it is friday | 16:39 | |
| lucasagomes | heh | 16:40 |
| lucasagomes | NobodyCam, well you gotta watch the news in rio de janeiro then | 16:40 |
| NobodyCam | hehehehe | 16:40 |
| -openstackstatus- NOTICE: New contributors can't complete enrollment due to https://launchpad.net/bugs/1317957 (Gerrit is having trouble reaching the Foundation Member system) | 16:42 | |
| *** ChanServ changes topic to "New contributors can't complete enrollment due to https://launchpad.net/bugs/1317957 (Gerrit is having trouble reaching the Foundation Member system)" | 16:42 | |
| NobodyCam | lucasagomes: you made comment that removing the laels would be easy. did you have a spot in mind for that | 16:42 |
| NobodyCam | oh thats a good thing the friday before the summit :-p | 16:43 |
| lucasagomes | NobodyCam, we could run e2label or tune2fs to clean up the labels | 16:43 |
| *** geertj has quit IRC | 16:44 | |
| *** newell_ has joined #openstack-ironic | 16:44 | |
| *** geertj has joined #openstack-ironic | 16:45 | |
| *** mkerrin has left #openstack-ironic | 16:45 | |
| NobodyCam | have you had luck removing labels with e2label I've always had to use tune2fs -L | 16:45 |
| NobodyCam | run e2label or tune2fs to clean up the labels | 16:46 |
| NobodyCam | 16:45 | NobodyCam > have you had luck remo | 16:46 |
| NobodyCam | huh | 16:46 |
| NobodyCam | http://paste.openstack.org/show/6cW5WVnfuTdJ8wxp1eYq/ | 16:46 |
| NobodyCam | sorry auto cut went wacko.. /me needs a new more battery | 16:47 |
| NobodyCam | s/more/mouse/ | 16:47 |
| *** harlowja_away is now known as harlowja | 16:50 | |
| *** harlowja is now known as harlowja_away | 16:51 | |
| lucasagomes | NobodyCam, hmm lemme try with e2label | 16:51 |
| lucasagomes | but if tune2fs does it nicely we can use that too | 16:52 |
| openstackgerrit | Jarrod Johnson proposed a change to stackforge/pyghmi: Avoid looping select() on sockets repeatedly https://review.openstack.org/93086 | 16:52 |
| NobodyCam | ya tune2fs has always worked for me | 16:53 |
| NobodyCam | tune2fs -L "" <device> | 16:54 |
| openstackgerrit | A change was merged to stackforge/pyghmi: Correct return value of set_bootdev https://review.openstack.org/92888 | 16:55 |
| lucasagomes | [stack@localhost devstack]$ sudo e2label /dev/vdc1 | 16:55 |
| lucasagomes | testlabel | 16:55 |
| lucasagomes | [stack@localhost devstack]$ sudo e2label /dev/vdc1 '' | 16:55 |
| lucasagomes | [stack@localhost devstack]$ sudo e2label /dev/vdc1 | 16:55 |
| lucasagomes | [stack@localhost devstack]$ | 16:55 |
| lucasagomes | NobodyCam, ^ works the same way | 16:55 |
| *** harlowja_away is now known as harlowja | 16:56 | |
| NobodyCam | lucasagomes: http://paste.openstack.org/show/ZXCteaSMJxfJzNMkZNRp/ | 16:56 |
| *** athomas has quit IRC | 16:57 | |
| NobodyCam | wait | 16:57 |
| openstackgerrit | A change was merged to stackforge/pyghmi: Avoid looping select() on sockets repeatedly https://review.openstack.org/93086 | 16:58 |
| NobodyCam | oh brb | 16:58 |
| lucasagomes | NobodyCam, http://paste.openstack.org/show/79750/ | 16:59 |
| lucasagomes | NobodyCam, if tune2fs works for both we can use it then | 16:59 |
| lucasagomes | but it's weird that e2label doesnt work in ubuntu | 16:59 |
| NobodyCam | yes | 16:59 |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic: Factoring out PXE and TFTP functions https://review.openstack.org/90233 | 17:01 |
| lucasagomes | NobodyCam, all right, we can talk more about this problem and fix it between the sessions in the summit? | 17:01 |
| lucasagomes | I think I should start printing my stuff and packing my bag | 17:01 |
| lucasagomes | make sure I won't forget anything here | 17:01 |
| lucasagomes | my flight is tomorrow morning, NobodyCam if ur around there already maybe we could catchup during the weekend for a beer | 17:02 |
| lucasagomes | there I mean atlanta | 17:02 |
| NobodyCam | cool deal lucasagomes ... I'll prob do up a patch today | 17:07 |
| NobodyCam | its a blocker for me atm | 17:07 |
| *** dwalleck has quit IRC | 17:08 | |
| lucasagomes | NobodyCam, oh alright :) | 17:09 |
| lucasagomes | good stuff :) | 17:09 |
| NobodyCam | so I'll add a removelabels to disk_partitioner.py and call it in make_partitions | 17:09 |
| lucasagomes | NobodyCam, so I see u soon | 17:09 |
| lucasagomes | NobodyCam, sounds good | 17:10 |
| NobodyCam | :) | 17:10 |
| lucasagomes | add a note about it as well, seems I couldn't reproduce in other OSs | 17:10 |
| lucasagomes | s/seems/since | 17:10 |
| jbjohnso | I will say I sorely miss decent profiling as I do more python development... | 17:11 |
| NobodyCam | ack | 17:11 |
| *** dwalleck has joined #openstack-ironic | 17:11 | |
| adam_g | lucasagomes, that test will be fixed once this lands in tempest https://review.openstack.org/#/c/92883/ | 17:11 |
| jbjohnso | heapy is great for memory, but cProfile and yappi just aren't no where near as nice as I'd like.. | 17:12 |
| lucasagomes | adam_g, :) good stuff | 17:12 |
| lucasagomes | yeah I saw that patch | 17:12 |
| lucasagomes | it was approved but the merge failed :( | 17:12 |
| *** lucasagomes is now known as lucas-packing | 17:12 | |
| adam_g | ya :| | 17:13 |
| *** matty_dubs|lunch is now known as matty_dubs | 17:18 | |
| *** eghobo has quit IRC | 17:24 | |
| *** eghobo has joined #openstack-ironic | 17:24 | |
| NobodyCam | ok need to move... bbaib... | 17:26 |
| *** dwalleck has quit IRC | 17:36 | |
| *** Mikhail_D_ltp has joined #openstack-ironic | 17:54 | |
| NobodyCam | move done | 18:06 |
| *** jistr|gone has quit IRC | 18:12 | |
| *** athomas has joined #openstack-ironic | 18:12 | |
| *** annegentle_ has joined #openstack-ironic | 18:12 | |
| openstackgerrit | Jarrod Johnson proposed a change to stackforge/pyghmi: Handle packet loss correctly in logout/logon https://review.openstack.org/93098 | 18:14 |
| *** annegentle has quit IRC | 18:17 | |
| Shrews | ooh, fun gerrit bug: http://lists.openstack.org/pipermail/openstack-dev/2014-May/034843.html | 18:21 |
| NobodyCam | ieek that is a good one | 18:25 |
| openstackgerrit | A change was merged to stackforge/pyghmi: Handle packet loss correctly in logout/logon https://review.openstack.org/93098 | 18:29 |
| *** dwalleck has joined #openstack-ironic | 18:32 | |
| *** dwalleck_ has joined #openstack-ironic | 18:53 | |
| *** dwalleck_ has quit IRC | 18:53 | |
| *** zdin0bot1 has joined #openstack-ironic | 18:56 | |
| adam_g | whats the quickest way to abort a provision thats in wait call-back? | 18:56 |
| *** zdin0bot has quit IRC | 18:56 | |
| *** dwalleck has quit IRC | 18:57 | |
| Shrews | power off? | 18:59 |
| Shrews | as in, your laptop :) | 18:59 |
| *** geertj has quit IRC | 18:59 | |
| Shrews | adam_g: think there's any chance of getting https://review.openstack.org/92883 pushed through today before everyone heads off to the summit? | 19:02 |
| *** romcheg1 has quit IRC | 19:02 | |
| adam_g | Shrews, pinged #openstack-qa. fingers crossed. :) | 19:02 |
| *** dwalleck has joined #openstack-ironic | 19:08 | |
| *** dwalleck has quit IRC | 19:09 | |
| *** dwalleck has joined #openstack-ironic | 19:10 | |
| adam_g | how about recovering from a provision error? | 19:10 |
| *** openstackgerrit has quit IRC | 19:19 | |
| *** openstackgerrit has joined #openstack-ironic | 19:20 | |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic: Consider free disk space before downloading images into cache https://review.openstack.org/92625 | 19:28 |
| *** dwalleck has quit IRC | 19:31 | |
| *** epim has joined #openstack-ironic | 19:31 | |
| *** dwalleck has joined #openstack-ironic | 19:36 | |
| *** epim has quit IRC | 19:37 | |
| openstackgerrit | Jarrod Johnson proposed a change to stackforge/pyghmi: Fix attribute error on particularly bad connections https://review.openstack.org/93117 | 19:54 |
| adam_g | Shrews, were you working on some partitioning issues earlier this week or was that someone else? | 19:55 |
| *** epim has joined #openstack-ironic | 19:55 | |
| Shrews | adam_g: that was NobodyCam and lucas-packing | 19:55 |
| adam_g | Shrews, do you remember the bug by chance? | 19:55 |
| NobodyCam | could be me | 19:56 |
| NobodyCam | :-p | 19:56 |
| Shrews | adam_g: https://bugs.launchpad.net/ironic/+bug/1317647 | 19:56 |
| adam_g | NobodyCam, http://paste.openstack.org/show/79772/ that look farmiliar? | 19:56 |
| NobodyCam | yep... testing a patch right now | 19:57 |
| NobodyCam | adam_g: its the nodes volume lables | 19:58 |
| NobodyCam | labels even | 19:58 |
| openstackgerrit | A change was merged to stackforge/pyghmi: Fix attribute error on particularly bad connections https://review.openstack.org/93117 | 19:58 |
| adam_g | NobodyCam, ah! | 19:58 |
| adam_g | NobodyCam, changing the label will likely break boot tho? /etc/fstab boots by label | 19:59 |
| NobodyCam | depends this is a rebuild? | 19:59 |
| openstackgerrit | Jarrod Johnson proposed a change to stackforge/pyghmi: Fix unregister_keepalive when no keepalives registered https://review.openstack.org/93120 | 19:59 |
| NobodyCam | the label will get reset | 19:59 |
| *** dwalleck has quit IRC | 19:59 | |
| adam_g | NobodyCam, im not doing rebuild yet. just trying to boot from the seed | 20:00 |
| *** geertj has joined #openstack-ironic | 20:00 | |
| *** jcoufal_ has quit IRC | 20:01 | |
| *** Mikhail_D_ltp has quit IRC | 20:01 | |
| NobodyCam | does the nodes disk have labels already.. | 20:01 |
| *** dwalleck has joined #openstack-ironic | 20:02 | |
| adam_g | NobodyCam, it might? im not sure. the seed node does. i get the same partitioning error during provisioning, after the ramdisk and conductor have initiated the iscsi connection | 20:02 |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic: Adding a reference driver for the agent https://review.openstack.org/84795 | 20:03 |
| NobodyCam | I should have a patch here in a few min | 20:04 |
| adam_g | NobodyCam, ready to test! | 20:04 |
| *** jcoufal has joined #openstack-ironic | 20:04 | |
| openstackgerrit | A change was merged to stackforge/pyghmi: Fix unregister_keepalive when no keepalives registered https://review.openstack.org/93120 | 20:08 |
| *** jcoufal has quit IRC | 20:19 | |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic: Factoring out PXE and TFTP functions https://review.openstack.org/90233 | 20:20 |
| dtantsur | Shrews, left a comment on https://review.openstack.org/#/c/92340/ . I do realize that recently-introduced ImageCache can lead to confusions :) | 20:26 |
| *** annegentle_ has quit IRC | 20:27 | |
| Shrews | dtantsur: thx. going through that code now | 20:27 |
| *** dwalleck has quit IRC | 20:28 | |
| NobodyCam | humm We don't know where your gerrit is. | 20:30 |
| *** jcoufal has joined #openstack-ironic | 20:30 | |
| Shrews | where was the last place you used it??? | 20:31 |
| Shrews | :) | 20:31 |
| dtantsur | lol | 20:31 |
| Shrews | dtantsur: changed my vote | 20:31 |
| Shrews | NobodyCam: fresh clone? | 20:32 |
| NobodyCam | yes... no its me.. I cant ssh to gerrit | 20:37 |
| Shrews | i had trouble accessing it earlier too. it cleared up quickly though | 20:37 |
| Shrews | NobodyCam: probably related to the gerrit bug in the topic | 20:39 |
| NobodyCam | yea I'm getting Permission denied (publickey). | 20:39 |
| NobodyCam | yea.. checking | 20:39 |
| Shrews | ok, calling it a week. see you all IRL soon | 20:41 |
| NobodyCam | have a good weekend Shrews see ya in atalanta | 20:54 |
| *** ChanServ changes topic to "OpenStack Bare Metal Provisioning | Docs: http://docs.openstack.org/developer/ironic/ | Bugs: https://bugs.launchpad.net/ironic | Status: https://etherpad.openstack.org/p/IronicWhiteBoard" | 20:55 | |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic: Factoring out PXE and TFTP functions https://review.openstack.org/90233 | 20:57 |
| *** linggao has quit IRC | 20:57 | |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic: Adding a reference driver for the agent https://review.openstack.org/84795 | 20:58 |
| *** epim has quit IRC | 20:59 | |
| *** eguz has joined #openstack-ironic | 21:03 | |
| *** annegentle has joined #openstack-ironic | 21:06 | |
| *** eghobo has quit IRC | 21:07 | |
| *** Nisha has quit IRC | 21:10 | |
| *** epim has joined #openstack-ironic | 21:15 | |
| *** zdin0bot1 has quit IRC | 21:16 | |
| *** matty_dubs is now known as matty_dubs|gone | 21:19 | |
| *** epim has quit IRC | 21:25 | |
| openstackgerrit | Chris Krelle proposed a change to openstack/ironic: Workaround for conflicting volume labels https://review.openstack.org/93133 | 21:27 |
| NobodyCam | adam_g: ^^^^ | 21:28 |
| *** harlowja has quit IRC | 21:29 | |
| adam_g | NobodyCam, nice! will test shortly | 21:30 |
| NobodyCam | adam_g: I dont expect it will land in current form.. but it did work in my testing | 21:31 |
| adam_g | NobodyCam, cool.w ill definitely give it a shot | 21:31 |
| *** epim has joined #openstack-ironic | 21:34 | |
| *** epim has quit IRC | 21:37 | |
| * NobodyCam wanders afk for a bit... | 21:38 | |
| NobodyCam | adam_g: please let me know your results...I'll be back to check | 21:39 |
| adam_g | NobodyCam, booting now | 21:40 |
| *** athomas has quit IRC | 21:42 | |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic: Adding a reference driver for the agent https://review.openstack.org/84795 | 21:43 |
| adam_g | NobodyCam, http://paste.openstack.org/show/79782/ | 21:49 |
| NobodyCam | adam_g: humm let me add a try around tune2fs call... | 21:54 |
| NobodyCam | is after I eat ok :-p | 21:54 |
| adam_g | NobodyCam, yeah, i was just adding a comment | 21:55 |
| *** openstackgerrit has quit IRC | 22:05 | |
| *** openstackgerrit has joined #openstack-ironic | 22:06 | |
| openstackgerrit | Chris Krelle proposed a change to openstack/ironic: Workaround for conflicting volume labels https://review.openstack.org/93133 | 22:09 |
| *** jcoufal has quit IRC | 22:24 | |
| NobodyCam | adam_g: ^^ think that may work for ya | 22:27 |
| *** dtantsur has quit IRC | 22:41 | |
| *** eghobo has joined #openstack-ironic | 22:44 | |
| *** eghobo has quit IRC | 22:45 | |
| *** eghobo has joined #openstack-ironic | 22:46 | |
| *** geertj has quit IRC | 22:47 | |
| *** harlowja has joined #openstack-ironic | 22:54 | |
| lucas-packing | ok it's pretty late here, have a good night everyone... see some of you next week :) | 23:03 |
| JayF | see you monday | 23:04 |
| *** lucas-packing has quit IRC | 23:06 | |
| *** killer_prince has joined #openstack-ironic | 23:11 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic: Updated from global requirements https://review.openstack.org/92953 | 23:49 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!