| *** cwolferh has quit IRC | 00:03 | |
| *** penick has quit IRC | 00:25 | |
| *** penick has joined #tripleo | 00:29 | |
| *** yuanying has joined #tripleo | 00:30 | |
| *** penick has quit IRC | 00:34 | |
| *** alop has quit IRC | 00:39 | |
| *** jcoufal has quit IRC | 00:46 | |
| *** Marga_ has quit IRC | 00:48 | |
| *** sdake has joined #tripleo | 00:53 | |
| openstackgerrit | Monty Taylor proposed openstack/diskimage-builder: Add ability to process Rackspace network info https://review.openstack.org/167003 | 00:55 |
|---|---|---|
| *** jerryz has joined #tripleo | 01:04 | |
| *** Marga_ has joined #tripleo | 01:16 | |
| *** Marga_ has quit IRC | 01:19 | |
| *** Marga_ has joined #tripleo | 01:25 | |
| *** eghobo has quit IRC | 01:31 | |
| *** rhallisey has quit IRC | 01:42 | |
| *** Marga_ has quit IRC | 02:13 | |
| *** dpaterson has quit IRC | 02:16 | |
| *** noslzzp has joined #tripleo | 02:18 | |
| *** sdake has quit IRC | 02:25 | |
| *** sdake has joined #tripleo | 02:32 | |
| *** kan__ has joined #tripleo | 02:32 | |
| *** penick has joined #tripleo | 02:34 | |
| *** MasterPiece has joined #tripleo | 02:48 | |
| *** Marga_ has joined #tripleo | 02:56 | |
| *** Marga_ has quit IRC | 02:59 | |
| *** Marga_ has joined #tripleo | 02:59 | |
| *** ChuckC has quit IRC | 03:00 | |
| *** spzala has quit IRC | 03:03 | |
| *** MasterPiece has quit IRC | 03:09 | |
| *** daneyon has joined #tripleo | 03:30 | |
| *** daneyon_ has quit IRC | 03:30 | |
| *** sdake has quit IRC | 03:34 | |
| *** daneyon_ has joined #tripleo | 03:34 | |
| *** bhunter_ has joined #tripleo | 03:36 | |
| *** bhunter has quit IRC | 03:37 | |
| *** bhunter_ is now known as bhunter | 03:37 | |
| *** daneyon has quit IRC | 03:37 | |
| *** eghobo has joined #tripleo | 03:53 | |
| openstackgerrit | Monty Taylor proposed openstack/diskimage-builder: Add ability to process Rackspace network info https://review.openstack.org/167003 | 03:57 |
| *** eghobo_ has joined #tripleo | 04:00 | |
| *** untriaged-bot has joined #tripleo | 04:00 | |
| untriaged-bot | Untriaged bugs so far: | 04:00 |
| untriaged-bot | https://bugs.launchpad.net/tripleo/+bug/1430445 | 04:00 |
| openstack | Launchpad bug 1430445 in tripleo "Horizon translations aren't compiled" [Undecided,In progress] - Assigned to Steve McLellan (sjmc7) | 04:00 |
| untriaged-bot | https://bugs.launchpad.net/diskimage-builder/+bug/1430599 | 04:00 |
| openstack | Launchpad bug 1430599 in diskimage-builder "ramdisk-image-create with ironic-discoverd-ramdisk element fails" [Undecided,Fix committed] - Assigned to Yuiko Takada (takada-yuiko) | 04:00 |
| untriaged-bot | https://bugs.launchpad.net/tripleo/+bug/1429902 | 04:00 |
| openstack | Launchpad bug 1429902 in tripleo "os-refresh-config fails during post-configure phase" [Undecided,New] | 04:00 |
| untriaged-bot | https://bugs.launchpad.net/tripleo/+bug/1434431 | 04:00 |
| openstack | Launchpad bug 1434431 in tripleo "Abnormal exit in building rhel images by diskimage-builder for system is not registered." [Undecided,New] | 04:00 |
| *** untriaged-bot has quit IRC | 04:00 | |
| *** eghobo has quit IRC | 04:04 | |
| *** eghobo has joined #tripleo | 04:04 | |
| *** noslzzp has quit IRC | 04:05 | |
| *** eghobo_ has quit IRC | 04:07 | |
| mordred | greghaynes: ok - latest patch works | 04:07 |
| openstackgerrit | Haomeng,Wang proposed openstack/diskimage-builder: Try 5 times for rmdir command call after umount https://review.openstack.org/137737 | 04:35 |
| *** eghobo has quit IRC | 04:39 | |
| *** pradeep has joined #tripleo | 04:44 | |
| *** Marga_ has quit IRC | 04:49 | |
| *** yamahata has quit IRC | 05:28 | |
| *** masco has joined #tripleo | 05:31 | |
| *** penick has quit IRC | 05:47 | |
| *** Marga_ has joined #tripleo | 05:52 | |
| *** ishant has joined #tripleo | 05:52 | |
| *** killer_prince has quit IRC | 06:14 | |
| *** pradeep has quit IRC | 06:16 | |
| *** killer_prince has joined #tripleo | 06:17 | |
| *** killer_prince is now known as lazy_prince | 06:17 | |
| *** Marga_ has quit IRC | 06:20 | |
| *** rlandy has joined #tripleo | 06:24 | |
| *** TheJulia has quit IRC | 06:29 | |
| *** dasm|afk is now known as dasm | 06:36 | |
| *** ukalifon has joined #tripleo | 06:42 | |
| *** pradeep has joined #tripleo | 06:44 | |
| *** bvandenh has joined #tripleo | 07:11 | |
| *** pradeep has quit IRC | 07:19 | |
| *** pradeep has joined #tripleo | 07:19 | |
| *** maufart has joined #tripleo | 07:29 | |
| *** mrunge has joined #tripleo | 07:32 | |
| *** jprovazn has joined #tripleo | 07:35 | |
| openstackgerrit | Merged openstack/tripleo-incubator: Set NeutronPublicInterface defaults to nic1 https://review.openstack.org/164350 | 07:37 |
| *** dsneddon has quit IRC | 07:42 | |
| *** rdopiera has joined #tripleo | 07:50 | |
| *** chlong has quit IRC | 07:52 | |
| *** dsneddon has joined #tripleo | 07:55 | |
| dasm | marios: thanks for explanation about nic1 vs eth0. i'll look at this. | 08:00 |
| *** pradeep1 has joined #tripleo | 08:01 | |
| *** pradeep has quit IRC | 08:04 | |
| *** yuanying has quit IRC | 08:05 | |
| *** dsneddon has quit IRC | 08:20 | |
| openstackgerrit | Yanis Guenane proposed openstack/tripleo-puppet-elements: overcloud-controller: Install Redis server https://review.openstack.org/167523 | 08:22 |
| openstackgerrit | Yanis Guenane proposed openstack/tripleo-puppet-elements: Enforce naming convention https://review.openstack.org/167524 | 08:26 |
| *** jistr has joined #tripleo | 08:38 | |
| *** lazy_prince has quit IRC | 08:46 | |
| *** dtantsur|afk is now known as dtantsur | 08:48 | |
| marios | dasm: np man | 08:51 |
| *** gfidente has joined #tripleo | 09:08 | |
| *** pradeep1 has quit IRC | 09:09 | |
| *** pradeep1 has joined #tripleo | 09:11 | |
| *** lucasagomes has joined #tripleo | 09:18 | |
| *** regebro has joined #tripleo | 09:33 | |
| *** kan__ has quit IRC | 09:44 | |
| *** mcornea has joined #tripleo | 09:47 | |
| openstackgerrit | Yanis Guenane proposed openstack/tripleo-heat-templates: Add support for Redis configuration https://review.openstack.org/167538 | 09:56 |
| *** untriaged-bot has joined #tripleo | 10:00 | |
| untriaged-bot | Untriaged bugs so far: | 10:00 |
| untriaged-bot | https://bugs.launchpad.net/tripleo/+bug/1430445 | 10:00 |
| openstack | Launchpad bug 1430445 in tripleo "Horizon translations aren't compiled" [Undecided,In progress] - Assigned to Steve McLellan (sjmc7) | 10:00 |
| untriaged-bot | https://bugs.launchpad.net/diskimage-builder/+bug/1430599 | 10:00 |
| openstack | Launchpad bug 1430599 in diskimage-builder "ramdisk-image-create with ironic-discoverd-ramdisk element fails" [Undecided,Fix committed] - Assigned to Yuiko Takada (takada-yuiko) | 10:00 |
| untriaged-bot | https://bugs.launchpad.net/tripleo/+bug/1429902 | 10:00 |
| openstack | Launchpad bug 1429902 in tripleo "os-refresh-config fails during post-configure phase" [Undecided,New] | 10:00 |
| untriaged-bot | https://bugs.launchpad.net/tripleo/+bug/1434431 | 10:00 |
| openstack | Launchpad bug 1434431 in tripleo "Abnormal exit in building rhel images by diskimage-builder for system is not registered." [Undecided,New] | 10:00 |
| *** untriaged-bot has quit IRC | 10:00 | |
| openstackgerrit | Yanis Guenane proposed openstack/tripleo-puppet-elements: overcloud-controller: Install Redis server https://review.openstack.org/167523 | 10:04 |
| *** pelix has joined #tripleo | 10:04 | |
| *** yog__ has joined #tripleo | 10:11 | |
| *** eghobo has joined #tripleo | 10:16 | |
| *** eghobo_ has joined #tripleo | 10:18 | |
| *** eghobo has quit IRC | 10:21 | |
| *** eghobo has joined #tripleo | 10:28 | |
| *** yog__ has quit IRC | 10:28 | |
| *** eghobo_ has quit IRC | 10:31 | |
| *** derekh has joined #tripleo | 10:33 | |
| *** yog__ has joined #tripleo | 10:33 | |
| *** pradeep1 has quit IRC | 10:38 | |
| gfidente | spredzy, the redis addition | 10:47 |
| *** dtantsur is now known as dtantsur|brb | 10:47 | |
| gfidente | do you plan to use that for something you're working on? | 10:47 |
| spredzy | gfidente, hey. Yes, I do. Ceilometer central agent in a HA setup | 10:48 |
| gfidente | it needs redis? | 10:48 |
| spredzy | So basically will use coordination ad using redis as a beckend for tooz | 10:48 |
| spredzy | redis will be HA using sentinels | 10:48 |
| openstackgerrit | Yanis Guenane proposed openstack/tripleo-heat-templates: Have a list of host IPs and hostnames per profile https://review.openstack.org/167555 | 10:51 |
| gfidente | I have no idea how that has to be setup :) | 10:54 |
| spredzy | Basically, one needs to setup the Redis Cluster, and then configure ceilometer with the proper coordination url, the rest will be taken care of | 10:55 |
| *** sdake has joined #tripleo | 10:55 | |
| spredzy | gfidente, Also, even if it is not for now, if we want to use sensu as a monitoring system, sensu requires Redi, so we will be able to use the same cluster | 10:56 |
| *** eghobo has quit IRC | 10:57 | |
| *** sdake__ has joined #tripleo | 10:59 | |
| *** sdake has quit IRC | 11:02 | |
| *** eghobo has joined #tripleo | 11:05 | |
| *** yog__ has quit IRC | 11:06 | |
| spredzy | jistr, can I get a review from you on this one https://review.openstack.org/#/c/167555 | 11:07 |
| spredzy | this is what I was asking yesterday (having a list of IPs), is it implemented the proper way ? | 11:08 |
| spredzy | gfidente, if you have time to take a look also I'll appreciate :) | 11:09 |
| *** spredzy is now known as spredzy|afk | 11:10 | |
| *** hewbrocca has joined #tripleo | 11:13 | |
| *** pelix has quit IRC | 11:13 | |
| gfidente | spredzy|afk, that change seems fine to me | 11:15 |
| gfidente | the thing is I am not sure i we should take the route of providing the lists of hosts per-service or per-role | 11:15 |
| gfidente | I think per-service is better because it allows for more granular config | 11:16 |
| gfidente | even though, at present time, we tie services to node types, we are trying to get away from that with little steps | 11:16 |
| gfidente | so the computes yaml already attempts that with RabbitHost, GlanceHost and stuff like that | 11:16 |
| gfidente | and I think that is where we should aim | 11:17 |
| *** pelix has joined #tripleo | 11:20 | |
| *** hewbrocca has quit IRC | 11:20 | |
| *** rhallisey has joined #tripleo | 11:28 | |
| *** yog__ has joined #tripleo | 11:33 | |
| *** athomas has joined #tripleo | 11:33 | |
| *** masco has quit IRC | 11:40 | |
| *** eghobo has quit IRC | 11:42 | |
| *** chlong has joined #tripleo | 11:50 | |
| jistr | spredzy|afk, gfidente: i did a review there. Per-service would be nice in giving more options but i sorta feel that making the per-role approach complete and working well should be a priority... | 11:55 |
| * jistr 's 2 cents :) | 11:55 | |
| jistr | so i +2'd | 11:56 |
| gfidente | jistr, per-service is what we tried to start with for rabbit | 11:58 |
| gfidente | and mysql in https://review.openstack.org/#/c/164983/ | 11:58 |
| gfidente | also, we expect a per-service Host parameter in the compute yaml | 11:58 |
| gfidente | why do you feel per-role is urgent? | 11:59 |
| gfidente | or a priority? | 11:59 |
| *** sdake__ has quit IRC | 11:59 | |
| *** dprince has joined #tripleo | 11:59 | |
| *** trown|outttypeww is now known as trown | 11:59 | |
| *** sdake has joined #tripleo | 12:00 | |
| gfidente | jistr, at present time, we're just adding the per-service params and computing those from the list of controllers | 12:01 |
| gfidente | the advantage I see with that is that the depending yamls don't make assumptions about where the services are | 12:01 |
| gfidente | sure we'll need to repeat calculation of the per-service hosts from the same list of controllers until we have more granular roles, but we'll be in better position when not all services are on same controllers | 12:02 |
| jistr | gfidente: i think per-role (making all controllers homogenous in what we run on them, all computes homogenous etc.) is easier to begin with, and will allow us to reach a useful working overcloud with puppet sooner than if we try to approach config per-service... So i'm more thinking in terms of what's less code, easier to comprehend, easier way forward. Even if we make it per-service on puppet level, the rest of the toolchain (instack, tuskar) | 12:04 |
| jistr | works only with roles AFAIK, so they wouldn't be able to use that power, no? | 12:04 |
| *** athomas has quit IRC | 12:04 | |
| *** openstackgerrit has quit IRC | 12:06 | |
| *** openstackgerrit has joined #tripleo | 12:06 | |
| *** pelix has quit IRC | 12:07 | |
| jistr | so i'm not against per-service, i'd rather say that i see it as more work and at this point i don't see the practical benefits very much... | 12:07 |
| gfidente | jistr, hehe okay but instead my point is that .pp should have no knowledge of the topology | 12:08 |
| gfidente | if we give it a list of hosts and expect it to know that is the list of rabbit nodes AND glance nodes | 12:08 |
| gfidente | we're doing it wrong | 12:08 |
| gfidente | so instead, I prefer to give it two lists | 12:08 |
| gfidente | and calculate them both, in the template, from the same list of controllers (which for now is the only thing we support, indeed) | 12:09 |
| gfidente | it really isn't much more work, we don't even need the list of hosts at present time, indeed I don't think we need to rely on that change for the moment | 12:09 |
| gfidente | and we're already taking steps into the per-service thing with the list of rabbit_nodes | 12:09 |
| gfidente | another scenario would be the one where we want to use external databases | 12:10 |
| gfidente | how do we configure the stuff for an external db if we can't even discern it from the list of controllers in the .pp? | 12:11 |
| *** pelix has joined #tripleo | 12:14 | |
| *** spredzy|afk is now known as spredzy | 12:14 | |
| gfidente | spredzy, ^^ look what a mess you caused! :) | 12:14 |
| spredzy | gfidente, ahah I was reading at the backlog :p | 12:15 |
| jistr | ah ok now i get what you meant, sorry :) so yeah not knowing the topology on the pp level rather than heat level makes more sense to make things future proof. hmm but still we have the .pp manifests hard-wired per role, so there's that... | 12:15 |
| gfidente | yeah that isn't optimal either and I can't think of how to cope with it if not with some enable/disable boolean per-service which we should use in the .pp | 12:17 |
| spredzy | jistr, gfidente thanks for the feedbacks. I do like gfidente idea of per-service, so for mongo create a mongo_nodes, etc... . But this will prevent the list of nodes to be precalculated. (ie. all controllers for example). For simpliciy's sake, I'd like to offer both, so if a service like mongo has a mongo_nodes then rely on it, if the customer assume it is installed on all nodes default to controller_nodes | 12:17 |
| spredzy | s/on all nodes/on all controller nodes/ | 12:18 |
| spredzy | I mean the two ideas are not exclusive but complementary | 12:18 |
| spredzy | So +1 for the per-service, but also +1 for the per-role (as a default/failover) | 12:18 |
| spredzy | does that make sense ? | 12:19 |
| * jistr reads more and thinks more | 12:20 | |
| jistr | :) | 12:20 |
| *** lucasagomes is now known as lucas-hungry | 12:21 | |
| *** jcoufal has joined #tripleo | 12:22 | |
| *** athomas has joined #tripleo | 12:22 | |
| gfidente | spredzy, not sure why it can't be precalculated? | 12:24 |
| spredzy | gfidente, if you don't specify mongo_nodes, how can you precalculate it ? You'll have to failover controller_nodes. Else there is not way to know on which hosts mongodb is expcted to be installed | 12:25 |
| jistr | spredzy: so to get back to why we're implementing this, you need a list of controller IP addresses for mongodb, right? | 12:25 |
| *** thrash|g0ne is now known as thrash | 12:26 | |
| spredzy | jistr, well mongodb needs it, ceph mon needs it (it does it differently atm but could use this patch) | 12:26 |
| gfidente | spredzy, today I would calculate mongo_nodes from the same list of controllers we use when calculating rabbit_nodes | 12:26 |
| spredzy | gfidente, which is the list of every controller nodes | 12:26 |
| gfidente | spredzy, tomorrow, if we get to isolate the mongo nodes, heat will know and we calculate mongo_nodes from that list instead | 12:26 |
| gfidente | spredzy, so today is the list of controllers, tomorrow it might not be and if it isn't, .pp won't need restructuring | 12:27 |
| gfidente | only templates will, which is fine to me because that is where orchestration should happen | 12:27 |
| spredzy | gfidente, I agree with that, hence why I think we should propose a mongo_nodes parameters but *default* to the the list of controller nodes ip | 12:29 |
| jistr | spredzy, gfidente: i'm not yet convinced we'll be able to make use of that flexibility, but thinking it through and looking through the code, it actually might not mean more work to go the way that gfidente suggested. | 12:29 |
| gfidente | spredzy, *default* in the .pp or in the .yaml ? | 12:29 |
| gfidente | jistr, yes I agree I am not sure it will ever be feasible to bring it into usable state, that is my biggest concern indeed :P | 12:30 |
| spredzy | gfidente, to me it will be in the yaml | 12:31 |
| spredzy | so in the .pp we always use hiera('mongo_nodes') | 12:31 |
| gfidente | spredzy, default in yaml is fine for me indee, it is the .pp I didn't want to make assumptions about it | 12:31 |
| jistr | gfidente: so if i understand correctly, you suggest making a new section like this https://github.com/openstack/tripleo-heat-templates/blob/master/puppet/all-nodes-config.yaml#L49-L54 | 12:31 |
| jistr | but make it for mongodb and feed it with IPs instead of names, right? | 12:31 |
| gfidente | jistr, exactly | 12:31 |
| spredzy | but based on the setup it will use the MongoDBNodes parameter or the list of controller ip | 12:31 |
| spredzy | gfidente, this is why the infra file is at the bottom of the hiera list | 12:31 |
| gfidente | spredzy, yeah ideally the resource-level yamls shouldn't need restructuring either | 12:32 |
| *** masco has joined #tripleo | 12:32 | |
| gfidente | spredzy, "MongoDBNodes parameter or the list of controller ip"? I need I pointer to understand that | 12:33 |
| jistr | spredzy: do you see some disadvantage in doing this? https://github.com/openstack/tripleo-heat-templates/blob/master/puppet/all-nodes-config.yaml#L49-L54 | 12:34 |
| spredzy | jistr, which param will you get ? get_param: mongo_nodes ? If the MongoDBNodes is not specified it will end up empty ? | 12:35 |
| *** tzumainn has quit IRC | 12:36 | |
| jistr | spredzy: we wouldn't have MongoDBNodes parameter right now, we would use mongo_nodes and it would be always computed from controller_ips for now | 12:36 |
| spredzy | I am probably missing something here ... :/ | 12:36 |
| spredzy | jistr, ok. When we will have MongDBNodes, how would that work ? | 12:37 |
| gfidente | spredzy, mongo_nodes is MongDBNodes as I read it | 12:37 |
| gfidente | isn't that the case? | 12:37 |
| *** jdob has joined #tripleo | 12:38 | |
| spredzy | gfidente, yes it is, hence to me here, if we provides MongoDBNodes, it should get_param: mongo_nodes (the heat parameter) | 12:40 |
| gfidente | spredzy, reading your comment on the review now | 12:41 |
| spredzy | gfidente, jistr I have to hope off for an hour | 12:41 |
| spredzy | will be back | 12:41 |
| *** spredzy is now known as spredzy|afk | 12:41 | |
| gfidente | so you meant to orchestrate that by overriding the values in the hiera hierarchy? | 12:41 |
| *** dprince has quit IRC | 12:42 | |
| *** morazi has joined #tripleo | 12:42 | |
| gfidente | jistr, ^^ I *think* that is why spredzy|afk referred to the case where MongoDBNodes is not provided by the user as input param | 12:42 |
| jistr | spredzy, gfidente: i think the value of what gfidente suggests would be if someone uses their custom role in the future (own heat template, own .pp file) to deploy their own mongo cluster *outside* of the regular controller nodes | 12:43 |
| gfidente | jistr, that or if we split the heat params per-service | 12:43 |
| gfidente | so that .pp won't make assumptions about the role of the system but optionally include things based on which service is | 12:44 |
| gfidente | "seen" enabled on that host | 12:44 |
| *** TheJulia has joined #tripleo | 12:44 | |
| gfidente | but yes this would require some concerted effott | 12:44 |
| jistr | spredzy|afk, gfidente: anyway we're probably talking a good deal into the future, but if we can do a bit of it right now at no additional cost (the patch might be even smaller than it is right now, originally i thought it would be otherwise...) then it might make sense to go for it | 12:46 |
| *** tzumainn has joined #tripleo | 12:48 | |
| jistr | spredzy|afk, gfidente: one thing i'm wondering though, why have separate hiera files for rabbit, mongo etc.? can we have a all_nodes_config.yaml which would contain the keys rabbit_nodes, mongo_nodes etc.? We wouldn't have to amend the hierarchy every time we add list of nodes for some service... | 12:49 |
| *** akrivoka has joined #tripleo | 12:49 | |
| gfidente | jistr, yep that I think would help! | 12:50 |
| gfidente | and be better then how it is now | 12:50 |
| *** akrivoka has quit IRC | 12:52 | |
| *** akrivoka has joined #tripleo | 12:54 | |
| *** rwsu has quit IRC | 12:56 | |
| *** dpaterson has joined #tripleo | 12:57 | |
| *** noslzzp has joined #tripleo | 12:59 | |
| *** akrivoka has quit IRC | 13:01 | |
| *** dprince has joined #tripleo | 13:01 | |
| *** noslzzp has quit IRC | 13:01 | |
| *** julim has joined #tripleo | 13:02 | |
| *** pblaho has joined #tripleo | 13:05 | |
| *** radez_g0n3 is now known as radez | 13:09 | |
| *** radez is now known as radez_g0n3 | 13:12 | |
| *** yamahata has joined #tripleo | 13:12 | |
| *** eghobo has joined #tripleo | 13:13 | |
| *** ishant has quit IRC | 13:18 | |
| *** spzala has joined #tripleo | 13:22 | |
| *** lucas-hungry is now known as lucasagomes | 13:24 | |
| *** weshay has joined #tripleo | 13:27 | |
| *** Marga_ has joined #tripleo | 13:28 | |
| *** Marga_ has quit IRC | 13:30 | |
| *** Marga_ has joined #tripleo | 13:31 | |
| *** radez_g0n3 is now known as radez | 13:34 | |
| *** openstackgerrit has quit IRC | 13:36 | |
| *** openstackgerrit has joined #tripleo | 13:36 | |
| jistr | spredzy|afk: i did the refactor i talked about above: https://review.openstack.org/#/c/167589 | 13:38 |
| *** lblanchard has joined #tripleo | 13:40 | |
| jistr | gfidente, spredzy|afk: also i think it would be nicer if we named variables named "rabbit_node_names" and "mongo_node_ips", because if we have "rabbit_nodes" and "mongo_nodes" it's not clear if it's names or IPs when you want to use the variable somewhere, and for some usecases it matters | 13:42 |
| jistr | s/named variables named/had variables named/ | 13:43 |
| jistr | spredzy|afk: and here's how the mongodb use case would look implemented on top of that refactoring: http://fpaste.org/202672/29141814/ does it seem ok? | 13:51 |
| jistr | cc gfidente ^^ | 13:52 |
| *** eghobo has quit IRC | 14:05 | |
| *** ChuckC has joined #tripleo | 14:09 | |
| *** rbrady is now known as rbrady-appt | 14:11 | |
| *** prad has joined #tripleo | 14:12 | |
| sdake | morning | 14:14 |
| *** dtantsur|brb is now known as dtantsur | 14:17 | |
| *** spredzy|afk is now known as spredzy | 14:18 | |
| gfidente | jistr, I'd +1 that | 14:25 |
| spredzy | gfidente, jistr I can go with that too | 14:25 |
| spredzy | thanks for the refactor | 14:26 |
| jistr | spredzy: ok so should i submit the code i have in fpaste ^? | 14:26 |
| spredzy | jistr, lgtm | 14:26 |
| * jistr wonders if separate patch or a new patchset in https://review.openstack.org/#/c/167555 -- spredzy any preference? | 14:27 | |
| spredzy | well don't know what you guys think, but on the review you mentioned, should we keep N_ips and N_names ? for future use or should we drop them now and implement them later ? | 14:28 |
| spredzy | gfidente, jistr ? | 14:28 |
| gfidente | the fpaste doesn't need to depend on 167555 | 14:29 |
| jistr | i'm inclinded for N_ips and N_names to keep things clear | 14:29 |
| jistr | gfidente, spredzy: yeah i meant the fpaste would replace 167555 | 14:29 |
| jistr | spredzy: oh now i see what you meant by asking "N_ips" and "N_names"... | 14:31 |
| jistr | spredzy, gfidente: so to answer *that*, i'd only implement things which we really need for something, so i'd reduce the scope to what's in the fpaste and implement the other things as we need them (it's possible we won't need some of the others at all) | 14:32 |
| gfidente | jistr, yeah that is what I thought as well | 14:33 |
| spredzy | jistr, gfidente well how I see it. A new patchset is needed for 16755, cleaning the infra section about the hiera file, letting - or not- the N_ips and N_names, a new patchset for https://review.openstack.org/#/c/165474/ to use what will be done in 167555 | 14:33 |
| spredzy | ok for N_ips and N_names, majority wins ! | 14:34 |
| gfidente | I am not touching anything of this :P | 14:34 |
| gfidente | spredzy, we had controller_names already so I think it is good to have controller_ips | 14:34 |
| gfidente | and maybe we won't need to add the others anytime soon | 14:34 |
| *** ChuckC has quit IRC | 14:35 | |
| spredzy | gfidente, jistr ack | 14:37 |
| spredzy | will update the review | 14:37 |
| jistr | spredzy: yeah that sounds good re new patchsets needed. So i'll submit the fpaste as a new patchset for 16755 (i already have the code changes made on top of https://review.openstack.org/#/c/167589 which is i think what we want), and then if we want to amend that patchset further in case https://review.openstack.org/#/c/165474/ would require something more, we can | 14:37 |
| spredzy | ack jistr let me know when its done | 14:37 |
| jistr | spredzy: ack | 14:37 |
| *** jprovazn has quit IRC | 14:39 | |
| openstackgerrit | Jiri Stransky proposed openstack/tripleo-heat-templates: Implement mongo_node_ips hiera key https://review.openstack.org/167555 | 14:41 |
| jistr | spredzy: done ^^ :) | 14:42 |
| openstackgerrit | Monty Taylor proposed openstack/diskimage-builder: Add a yum-minimal element that just uses yum https://review.openstack.org/166640 | 14:42 |
| openstackgerrit | Monty Taylor proposed openstack/diskimage-builder: Port centos-minimal to yum-minimal https://review.openstack.org/166641 | 14:42 |
| openstackgerrit | Monty Taylor proposed openstack/diskimage-builder: Split the debootstrap functions into an element https://review.openstack.org/166650 | 14:42 |
| openstackgerrit | Monty Taylor proposed openstack/diskimage-builder: Add ability to process Rackspace network info https://review.openstack.org/167003 | 14:42 |
| spredzy | jistr, +1 thanks ! | 14:44 |
| spredzy | gfidente, jistr just so we're clear, so we do have a per-service hiera variable, but no per-service heat variable for now | 14:45 |
| spredzy | ? | 14:45 |
| gfidente | spredzy, ahahah you're right to hate me but yes | 14:46 |
| gfidente | I think that is a step into the direction | 14:47 |
| *** Marga_ has quit IRC | 14:47 | |
| spredzy | gfidente, no hate here ;) just clarifying the current state so if someone ask we're clear on the line, will add it to the review | 14:48 |
| jistr | spredzy: ack :) | 14:49 |
| *** subscope has joined #tripleo | 14:50 | |
| gfidente | but in my secret thoughts, I think jistr is right we might still have a hard time getting there | 14:51 |
| gfidente | I mean getting to a point where we can turn this into an actual 'feature' of some kind | 14:53 |
| spredzy | Well we will have to get there sooner or later. Imagine a a MongoDB or RabbitMQ cluster is not on the controller but on its own nodes. We would need to be able to point the appropriate configuration there | 14:55 |
| *** ChuckC has joined #tripleo | 14:55 | |
| * gfidente dreams of resources per-service in the template | 14:55 | |
| spredzy | so it is a *needed* feature IMO | 14:55 |
| spredzy | today we put everything on the controller node, but it might not and should nt be 'mandatory' | 14:55 |
| spredzy | jistr, do you agree with the last comment ? | 14:56 |
| gfidente | also the .pp in general should not make assumptions about the list of controllers matching the list of ntp servers, or mongo servers | 14:57 |
| *** dsneddon has joined #tripleo | 14:57 | |
| jistr | spredzy: yeah i do. I think Tuskar should be helping on the front of customization in the future (letting people have their custom roles and actually generating the Heat template instead of having our current hardcoded overcloud-without-mergepy). And when that happens i think we'll make use of the per-service variables we're defining today... | 14:58 |
| jistr | so yeah it's a good bit into the future but if it comes at basically no extra cost, then why not have it that way now | 14:58 |
| gfidente | I still dream of per-service configs in the template | 15:00 |
| gfidente | bounded to nodes to build custom roles | 15:00 |
| gfidente | spredzy, a bit like we did for ceph, but for every service | 15:00 |
| spredzy | gfidente, so you mean building a MongoDB role, a MySQL role, etc... the same way we did CephStorage ? | 15:01 |
| spredzy | ie. not having a monolithic role | 15:02 |
| gfidente | spredzy, yes | 15:02 |
| gfidente | a single resource config with its own params | 15:02 |
| gfidente | which is attached to a server | 15:02 |
| gfidente | so that which of those are attached to which node, defines a role | 15:03 |
| gfidente | (a single resource per-service) | 15:03 |
| *** masco has quit IRC | 15:04 | |
| spredzy | I def. +1 the idea | 15:04 |
| jistr | yeah if we can deploy a golden image, something like that should be doable (but still it would be nice if we could make things work in the monolithic approach first :D) | 15:04 |
| spredzy | I think it we want to be flexible on the infra we deploy we will need that any wayt | 15:04 |
| spredzy | jistr, well just to defend this idea, even if I agree we should have something work first, having service role will make development/test easier | 15:05 |
| spredzy | isolating at its core a feature | 15:06 |
| *** eghobo has joined #tripleo | 15:07 | |
| *** openstackgerrit has quit IRC | 15:08 | |
| *** openstackgerrit has joined #tripleo | 15:09 | |
| *** jprovazn has joined #tripleo | 15:22 | |
| *** eghobo_ has joined #tripleo | 15:23 | |
| *** eghobo has quit IRC | 15:26 | |
| *** noslzzp has joined #tripleo | 15:32 | |
| *** eghobo_ has quit IRC | 15:34 | |
| *** akrivoka has joined #tripleo | 15:35 | |
| *** noslzzp has quit IRC | 15:35 | |
| *** yamahata has quit IRC | 15:37 | |
| *** noslzzp has joined #tripleo | 15:40 | |
| *** yog__ has quit IRC | 15:44 | |
| *** stendulker has joined #tripleo | 15:44 | |
| *** trown is now known as trown|lunch | 15:47 | |
| *** alop has joined #tripleo | 15:49 | |
| *** rwsu has joined #tripleo | 15:53 | |
| *** spzala has quit IRC | 15:56 | |
| *** yamahata has joined #tripleo | 15:58 | |
| *** jang1 has quit IRC | 15:59 | |
| *** jang has quit IRC | 15:59 | |
| *** untriaged-bot has joined #tripleo | 16:00 | |
| untriaged-bot | Untriaged bugs so far: | 16:00 |
| untriaged-bot | https://bugs.launchpad.net/tripleo/+bug/1430445 | 16:00 |
| openstack | Launchpad bug 1430445 in tripleo "Horizon translations aren't compiled" [Undecided,In progress] - Assigned to Steve McLellan (sjmc7) | 16:00 |
| untriaged-bot | https://bugs.launchpad.net/diskimage-builder/+bug/1430599 | 16:00 |
| openstack | Launchpad bug 1430599 in diskimage-builder "ramdisk-image-create with ironic-discoverd-ramdisk element fails" [Undecided,Fix committed] - Assigned to Yuiko Takada (takada-yuiko) | 16:00 |
| untriaged-bot | https://bugs.launchpad.net/tripleo/+bug/1429902 | 16:00 |
| openstack | Launchpad bug 1429902 in tripleo "os-refresh-config fails during post-configure phase" [Undecided,New] | 16:00 |
| untriaged-bot | https://bugs.launchpad.net/tripleo/+bug/1434431 | 16:00 |
| openstack | Launchpad bug 1434431 in tripleo "Abnormal exit in building rhel images by diskimage-builder for system is not registered." [Undecided,New] | 16:00 |
| *** untriaged-bot has quit IRC | 16:00 | |
| *** jang has joined #tripleo | 16:06 | |
| *** jang1 has joined #tripleo | 16:06 | |
| *** rbrady-appt is now known as rbrady | 16:09 | |
| *** jang has quit IRC | 16:11 | |
| *** jang1 has quit IRC | 16:11 | |
| *** rhallisey has quit IRC | 16:12 | |
| *** rlandy has quit IRC | 16:12 | |
| *** Marga_ has joined #tripleo | 16:12 | |
| *** rhallisey has joined #tripleo | 16:14 | |
| *** daneyon_ has quit IRC | 16:20 | |
| *** daneyon has joined #tripleo | 16:20 | |
| *** cwolferh has joined #tripleo | 16:21 | |
| *** chlong has quit IRC | 16:22 | |
| *** maufart has quit IRC | 16:25 | |
| *** ukalifon has quit IRC | 16:26 | |
| *** jprovazn has quit IRC | 16:29 | |
| *** chlong has joined #tripleo | 16:33 | |
| openstackgerrit | Merged stackforge/kolla: Refactor nova-network https://review.openstack.org/166420 | 16:37 |
| *** chlong has quit IRC | 16:39 | |
| ccrouch | rhallisey: jpeeler: quick kolla question | 16:40 |
| sdake | ccrouch join #kolla | 16:42 |
| ccrouch | if i cloned kolla now, do we have enough services working in fig to stand up a trusty based instance in nova say? | 16:42 |
| ccrouch | ah ha! | 16:42 |
| sdake | hi :) | 16:42 |
| jpeeler | what he said | 16:42 |
| sdake | answer is yes, bottom to horizon | 16:42 |
| *** eghobo has joined #tripleo | 16:43 | |
| *** cwolferh has quit IRC | 16:47 | |
| openstackgerrit | Giulio Fidente proposed openstack/tripleo-incubator: Add support for building Ceph nodes and configure Cinder to use it https://review.openstack.org/137031 | 16:57 |
| gfidente | guys ^^ this is yet a needed rebase | 16:57 |
| *** bvandenh has quit IRC | 16:57 | |
| *** jerryz has quit IRC | 16:58 | |
| *** jistr has quit IRC | 16:58 | |
| *** eghobo_ has joined #tripleo | 17:04 | |
| *** alop has quit IRC | 17:04 | |
| *** eghobo has quit IRC | 17:07 | |
| *** alop has joined #tripleo | 17:12 | |
| *** dpaterson has quit IRC | 17:13 | |
| *** pdb has quit IRC | 17:14 | |
| *** eghobo_ has quit IRC | 17:14 | |
| *** pdb has joined #tripleo | 17:14 | |
| bnemec | dprince: Know who I could poke to get https://github.com/puppetlabs/puppetlabs-rabbitmq/pull/334 merged ASAP? | 17:17 |
| greghaynes | mordred: ah, you found the typo I was banging my head against for way too long | 17:17 |
| greghaynes | mordred: for some reason I had to prove to myself that environment.d worked durring root.d phase before noticing that DIB_RELEASE was mistyped | 17:18 |
| crinkle | bnemec: merged | 17:18 |
| dprince | bnemec: I see you found the right person, thanks crinkle! | 17:19 |
| bnemec | crinkle: Awesome, thanks! | 17:19 |
| greghaynes | mordred: also, you have a patch loop in gerrit | 17:19 |
| *** eghobo has joined #tripleo | 17:19 | |
| mordred | greghaynes: AWESOME | 17:20 |
| greghaynes | SpamapS: https://review.openstack.org/#/c/166995/ - what specifically would you liked changed to make you :) | 17:21 |
| greghaynes | mordred: ^ im guessing that is the intended patch stack bottom? | 17:21 |
| mordred | greghaynes: yes. that shoudl be the bottom | 17:22 |
| mordred | greghaynes: btw - rax centos did not work | 17:22 |
| mordred | greghaynes: I'm retesting hp to see if I screwed something up in the most recent rebase/cleanup | 17:22 |
| SpamapS | greghaynes: nothing, I wanted us to think about it is all. | 17:23 |
| greghaynes | ah, ok | 17:23 |
| SpamapS | greghaynes: mkdir -p and not having cloud-init in base might be a more valid approach really. | 17:23 |
| mordred | SpamapS: I could go either way - I do not feel strongly that one is better than the other | 17:24 |
| greghaynes | Yea, that makes sense - once we add the dep its harder to remove it but no one can be depending on the side effect of cloud-init being pulled in by that element | 17:24 |
| greghaynes | until we add the dep | 17:24 |
| mordred | ++ | 17:24 |
| mordred | do you want me to update that and repush the stack? | 17:25 |
| greghaynes | yesplz | 17:25 |
| mordred | greghaynes, SpamapS: mkdir -p /etc/cloud/cloud.cfg.d ok? | 17:26 |
| SpamapS | mordred: or if [ -d /etc/cloud/cloud.cfg.d ] ; then .... | 17:27 |
| SpamapS | but that might have ordering issues | 17:27 |
| SpamapS | in case cloud-init gets pulled in at install.d/11* | 17:27 |
| mordred | nod | 17:27 |
| SpamapS | mordred: so yeah, mkdir -p seems the simplest approach. | 17:27 |
| greghaynes | WFM | 17:28 |
| SpamapS | because that is basically "hey cloud-init, if you're here, feel free to manage localhost in /etc/hosts" | 17:28 |
| openstackgerrit | Monty Taylor proposed openstack/diskimage-builder: Add a yum-minimal element that just uses yum https://review.openstack.org/166640 | 17:28 |
| openstackgerrit | Monty Taylor proposed openstack/diskimage-builder: Port centos-minimal to yum-minimal https://review.openstack.org/166641 | 17:28 |
| openstackgerrit | Monty Taylor proposed openstack/diskimage-builder: Handle non-cloud-init installs https://review.openstack.org/166995 | 17:28 |
| openstackgerrit | Monty Taylor proposed openstack/diskimage-builder: Split the debootstrap functions into an element https://review.openstack.org/166650 | 17:28 |
| openstackgerrit | Monty Taylor proposed openstack/diskimage-builder: Split dib-init-system into its own element https://review.openstack.org/167005 | 17:28 |
| openstackgerrit | Monty Taylor proposed openstack/diskimage-builder: Add ability to process Rackspace network info https://review.openstack.org/167003 | 17:28 |
| *** alop has quit IRC | 17:29 | |
| openstackgerrit | Daneyon Hansen proposed stackforge/kolla: Extends Nova Metadata Service Parameterization and Composition https://review.openstack.org/167362 | 17:29 |
| SpamapS | oh that also fixed the outdated dep, good | 17:30 |
| mordred | yay! | 17:30 |
| * mordred is helpful | 17:30 | |
| SpamapS | and plucky | 17:31 |
| *** trown|lunch is now known as trown | 17:31 | |
| *** Marga_ has quit IRC | 17:32 | |
| mordred | ok | 17:35 |
| mordred | so - it still works on HP | 17:35 |
| mordred | which means that the centos mechanism for running the script is valid | 17:35 |
| mordred | but something on rackspace has made it unhappy | 17:36 |
| greghaynes | mordred: did you look at https://github.com/dprince/os-net-config when you were making your process-config-drive deal? | 17:39 |
| mordred | nope | 17:39 |
| mordred | but I couldn't possible useit | 17:39 |
| mordred | https://github.com/dprince/os-net-config/blob/master/requirements.txt | 17:39 |
| greghaynes | :( | 17:39 |
| * mordred goes to look now | 17:39 | |
| mordred | part of the issue is that I need to boot test nodes for openstack | 17:39 |
| mordred | which means that programs that install things that openstack also wants to install | 17:40 |
| mordred | just to get an ip address | 17:40 |
| mordred | are overkill | 17:40 |
| mordred | and destructive | 17:40 |
| SpamapS | Why oh why does it use eventlet?! | 17:40 |
| greghaynes | ah, ok, so you dont just not want the deps you have to avoid them | 17:40 |
| mordred | righ | 17:40 |
| mordred | well, I mean, I really want to avoid it | 17:40 |
| mordred | I could do something tricky | 17:40 |
| *** alop has joined #tripleo | 17:41 | |
| mordred | but really - I don't need _any_ deps with process-config-drive | 17:41 |
| greghaynes | yea, I remember that deal now - you want valid tests | 17:41 |
| mordred | the code in os-net-config does look nice though | 17:41 |
| SpamapS | dprince: hey, why does os-net-config have eventlet as a dep? | 17:42 |
| SpamapS | +from os_net_config.openstack.common import processutils | 17:42 |
| dsneddon | SpamapS, I can answer that. os-net-config only requres eventlet because processutils from openstack common requires it. | 17:43 |
| SpamapS | yeah I'm digging in | 17:43 |
| *** mcornea has quit IRC | 17:43 | |
| *** spzala has joined #tripleo | 17:44 | |
| SpamapS | ah so now it's oslo.concurrency | 17:44 |
| SpamapS | for... processutils | 17:44 |
| SpamapS | I don't think that's a thing we need in os-net-config | 17:45 |
| SpamapS | I mean, processutils.execute() is a monster of a function, when all that is really needed is 'subprocess.check_call()' | 17:46 |
| greghaynes | mordred: on 167003 do we really need to merge dhcp-all-interfaces into the new element? Seems like we could leave it and make the new element depend on it | 17:48 |
| greghaynes | or, curious what the motivation for that wat | 17:48 |
| greghaynes | s/wat/was | 17:49 |
| greghaynes | It has the downside of now people who try to use dhcp-all-interfaces have to depend on the new simple-config element | 17:49 |
| *** eghobo has quit IRC | 17:51 | |
| SpamapS | <klaxon>interface change alert!</klaxon> | 17:52 |
| *** derekh has quit IRC | 17:56 | |
| *** Marga_ has joined #tripleo | 17:59 | |
| *** eghobo has joined #tripleo | 17:59 | |
| *** cwolferh has joined #tripleo | 18:00 | |
| mordred | greghaynes: well, I made a new dhcp-all-interfaces that depends on the simple-config | 18:02 |
| mordred | SpamapS, greghaynes: however, I dont' feel strogly - I mainly wanted workig-now | 18:02 |
| mordred | I think if we wanted to not combine them, we could make dhcp-all-interfaces pluggable | 18:02 |
| mordred | although I kinda think that I'd like to refactor the network part of process-config-drive to also operate on being able to pass in an interface name like dhcp-all-interfaces does | 18:03 |
| mordred | and really, there's no real good reason that the two need to be separate - given the scope of what they need to do | 18:04 |
| greghaynes | mordred: yea, I worry about someone who currently depends on dhcp-all-interfaces and does their own sort of initial networking config magic | 18:04 |
| greghaynes | hrm | 18:05 |
| greghaynes | i should probably read more of the code then | 18:05 |
| greghaynes | gerrit is less than useful for that when its a bunch of file renames | 18:05 |
| SpamapS | Long term, I'm pretty sure os-net-config has gotten out of hand at this point. We should reign it in, drop the unnecessarily complicated deps, and consider adding process-config-drive's magic to it. | 18:05 |
| SpamapS | Short term, I'm with mordred. Make it go. | 18:06 |
| mordred | yes short term | 18:06 |
| *** Marga_ has quit IRC | 18:06 | |
| SpamapS | Do we track bugs for dib-utils in diskimage-builder ? | 18:06 |
| mordred | also - I kinda like that process-config-drive is a python script that can just be copied and not installed, fwiw | 18:06 |
| SpamapS | the header requirements in non-dib-run-parts scripts needs to go away. | 18:06 |
| mordred | so that you don't need to install setuptools | 18:07 |
| mordred | because otherwise you need to install setuptools | 18:07 |
| greghaynes | SpamapS: I think so | 18:07 |
| SpamapS | mordred: agree on the self-contained part | 18:07 |
| mordred | and, again, it turns out that devstack runs an "install setuptools" script - so I'd love to make sure that works and sstuff | 18:07 |
| *** Marga_ has joined #tripleo | 18:07 | |
| mordred | we may be SEVERAL steps away from that though | 18:07 |
| *** ChuckC has quit IRC | 18:08 | |
| *** ChuckC has joined #tripleo | 18:09 | |
| SpamapS | so far I haven't found any deps of os-net-config that are really worth it | 18:09 |
| *** lucasagomes is now known as lucas-dinner | 18:10 | |
| SpamapS | Except PyYAML | 18:10 |
| SpamapS | and pbr is hard to make optional. ;) | 18:10 |
| *** pblaho has quit IRC | 18:10 | |
| SpamapS | mordred: I think if we made up our minds that os-net-config was going to work w/o setuptools, we could do it with minimal surgery | 18:11 |
| SpamapS | just make yaml optional and allow json | 18:11 |
| SpamapS | the bit that netaddr is doing is a tiny convenience. | 18:11 |
| SpamapS | oslo.utils.strutils is just doing 'bool_from_string' | 18:12 |
| SpamapS | but yeah, long term | 18:12 |
| * SpamapS unclenches and goes back to reviewing | 18:12 | |
| *** Marga_ has quit IRC | 18:13 | |
| *** Marga_ has joined #tripleo | 18:13 | |
| *** stendulker has quit IRC | 18:14 | |
| greghaynes | ah, ISWYD for dhcp-all-interfaces | 18:15 |
| greghaynes | I think im fine with that | 18:15 |
| greghaynes | that script (dhcp-all-interfces) is now named pretty poorly | 18:16 |
| mordred | yes | 18:17 |
| mordred | I could not come up with a better name | 18:18 |
| mordred | I was originally not going to rename it and just leave it as dhcp-all-interfaces and let it be named poorly | 18:18 |
| greghaynes | network-hammer | 18:19 |
| greghaynes | :p | 18:19 |
| greghaynes | ok, I might try and write up a better readme for what that element actually does | 18:20 |
| greghaynes | to make sure I have a good handle on it | 18:20 |
| openstackgerrit | Merged stackforge/kolla: Extends Nova Metadata Service Parameterization and Composition https://review.openstack.org/167362 | 18:21 |
| *** pelix has quit IRC | 18:27 | |
| greghaynes | Yea, so the interface change is that if someone is currently using dhcp-all-interfaces on a vm with a config drive without cloud-init config drive data source we now assume they want the config-drive networking config rather than to just dhcp everywhere reguardless | 18:29 |
| greghaynes | its a pretty small niche, so we might not care | 18:29 |
| *** dtantsur is now known as dtantsur|afk | 18:29 | |
| greghaynes | a simple fix might be so make a guard around the process-config-drive bit in dhcp-all-interfaces.sh that gets set by the dhcp-all-interfaces element, then someone depending from that element still gets the old behavior | 18:32 |
| SpamapS | greghaynes: I think it's a non-existant niche | 18:33 |
| SpamapS | I base my data on the fact that all these things you said add up to a rather insane situation to be silently consuming diskimage-builder from | 18:34 |
| SpamapS | s/data/conclusion/ | 18:34 |
| greghaynes | heh | 18:34 |
| *** spzala has quit IRC | 18:35 | |
| greghaynes | another thing - how does the process-config-drive.py interact if cloud-init is enabled with config-drive datasource? | 18:36 |
| greghaynes | because it is possible for us to disable that explicitly in the cloud-instance-simple-config element | 18:37 |
| *** Marga_ has quit IRC | 18:38 | |
| *** akrivoka has quit IRC | 18:39 | |
| *** Marga_ has joined #tripleo | 18:39 | |
| SpamapS | greghaynes: it should be messy. | 18:42 |
| mordred | they should really conflict | 18:42 |
| SpamapS | greghaynes: my gut says cloud-init will run before these bits and configure everything. | 18:42 |
| mordred | is there a way that we can mark elements as conflicting? | 18:43 |
| SpamapS | assuming the version of cloud-init knows how to read this particular config drive | 18:43 |
| SpamapS | mordred: yeah they can both provide the same thing IIRC | 18:43 |
| mordred | we might want to try that | 18:43 |
| mordred | beacuse honestly, combining these two things is a recipe for crazy | 18:43 |
| greghaynes | alternatively - why not just disable the config-drive data source, if cloud-init is not installed itll be a noop and if it is then they wont conflict | 18:43 |
| SpamapS | Building elements: base ubuntu fedora | 18:43 |
| SpamapS | Expanded element dependencies to: cloud-init-datasources rpm-distro fedora redhat-common dib-run-parts manifests cache-url pkg-map base ubuntu yum source-repositories package-installs dpkg dkms | 18:43 |
| SpamapS | Building in /tmp/image.sgC8o7hX | 18:44 |
| SpamapS | maybe not | 18:44 |
| SpamapS | There is a duplicated hook in your elements: ubuntu/bin/map-services | 18:44 |
| mordred | I think greghaynes may have it | 18:44 |
| SpamapS | if they both have the same file looksl ike | 18:44 |
| SpamapS | greghaynes: yeah I think thats the way to go. | 18:44 |
| SpamapS | though I do think it would be a reasonable interface change to make it impossible to include two elements that provide the same thing. | 18:45 |
| mordred | SpamapS: re: your comment about emitting a sstatic-network-up event | 18:45 |
| SpamapS | because, you know, diskimage-builder needs to maintain feature parity with dpkg. ;) | 18:45 |
| greghaynes | SpamapS: ++ | 18:45 |
| mordred | haha | 18:45 |
| mordred | SpamapS: if we split it up so that process-config-drive _also_ takes an interface as an argument | 18:46 |
| *** Marga_ has quit IRC | 18:46 | |
| mordred | SpamapS: and then have the event emitting look at the full list of static interfaces from config-drive and emit only when it can test that they're all up | 18:47 |
| mordred | the last instance of process-config-drive to run should be able to see that all of the interfaces have been ifupd, no? | 18:47 |
| *** Marga_ has joined #tripleo | 18:48 | |
| SpamapS | yeah thats basically how the ifupdown script works now | 18:49 |
| SpamapS | but it assumes you configured everything before boot | 18:50 |
| SpamapS | so what dhcp-all-interfaces is supposed to do, is configure them all, not just one | 18:50 |
| SpamapS | we got hung up on race conditions and settled on one at a time IIRC | 18:50 |
| SpamapS | Which meant the first one emits static-network-up | 18:50 |
| SpamapS | but still, it at least emits it _after_ the ifup | 18:50 |
| mordred | and that's not what the script dose now? | 18:51 |
| SpamapS | all that to say, just make sure you configure everything before ifup is run, and it works itself out. | 18:51 |
| SpamapS | or if you want to run ifup, then you have to emit static-network-up | 18:51 |
| mordred | it's possible I do not follow how it all hangs together | 18:51 |
| mordred | I up in the emit so that the "waiting for static-networking" would go away | 18:51 |
| mordred | put | 18:51 |
| SpamapS | Well right now net-device-added (udev) triggers upstart job network-interface starting, which triggers dhcp-all-interfaces ... | 18:52 |
| SpamapS | once dhcp-all-interfaces exits, network-interface starts, and runs 'ifup $INTERFACE' | 18:52 |
| mordred | oh | 18:52 |
| mordred | so you're saying we do not need the emit at all | 18:52 |
| mordred | if it's working properly | 18:52 |
| SpamapS | which triggers /etc/network/if-up.d/upstart which checks ifquery, and if we haven't yet emitted static-network-up, and all interfaces are up, emits it. | 18:53 |
| SpamapS | This is all super crazy amounts of hoops to jump through to try and make it so that arbitrary amounts of statically configured, dynamically present adapters can work. | 18:54 |
| SpamapS | For bridges/etc, after all the filesystems are mounted, we just run 'ifup -a' which will run /etc/network/if-up.d/upstart for every interface it ends up bringing up. | 18:55 |
| SpamapS | (all the non-network filesystems) | 18:55 |
| SpamapS | zomg | 18:55 |
| SpamapS | my head just popped | 18:55 |
| SpamapS | I'm going to go ponder on what you're attempting over some indian buffet | 18:55 |
| SpamapS | bbiab | 18:55 |
| *** dpaterson has joined #tripleo | 18:59 | |
| openstackgerrit | Monty Taylor proposed openstack/diskimage-builder: Add a yum-minimal element that just uses yum https://review.openstack.org/166640 | 19:01 |
| openstackgerrit | Monty Taylor proposed openstack/diskimage-builder: Port centos-minimal to yum-minimal https://review.openstack.org/166641 | 19:01 |
| openstackgerrit | Monty Taylor proposed openstack/diskimage-builder: Split the debootstrap functions into an element https://review.openstack.org/166650 | 19:01 |
| openstackgerrit | Monty Taylor proposed openstack/diskimage-builder: Add ability to process Rackspace network info https://review.openstack.org/167003 | 19:01 |
| *** cwolferh has quit IRC | 19:02 | |
| *** spzala has joined #tripleo | 19:03 | |
| *** alop has quit IRC | 19:07 | |
| *** alop has joined #tripleo | 19:12 | |
| *** Marga_ has quit IRC | 19:14 | |
| *** Marga_ has joined #tripleo | 19:15 | |
| *** eghobo has quit IRC | 19:15 | |
| *** alop_ has joined #tripleo | 19:16 | |
| *** alop has quit IRC | 19:17 | |
| *** alop_ is now known as alop | 19:17 | |
| *** radez is now known as radez_g0n3 | 19:19 | |
| *** radez_g0n3 is now known as radez | 19:22 | |
| openstackgerrit | Mathieu Gagné proposed openstack/diskimage-builder: Download of translations not properly disabled in APT https://review.openstack.org/167741 | 19:24 |
| mordred | mgagne: ^^ ooh. | 19:24 |
| mgagne | mordred: *evil eyes* | 19:25 |
| mordred | mgagne: thank you - I support getting that right | 19:25 |
| mordred | (also, sorry for getting it wrong) | 19:25 |
| *** Marga_ has quit IRC | 19:25 | |
| mgagne | mordred: dib already disables translation during build phase but "restore" the default behavior after. | 19:25 |
| mgagne | mordred: but it uses the right config name at that time =) | 19:25 |
| mgagne | https://github.com/openstack/diskimage-builder/blob/master/elements/dpkg/root.d/99-trim-dpkg | 19:27 |
| * mgagne is now off to figure out why apt-get update isn't idempotent | 19:28 | |
| *** sdake__ has joined #tripleo | 19:28 | |
| *** sdake has quit IRC | 19:32 | |
| *** rlandy has joined #tripleo | 19:37 | |
| *** eghobo has joined #tripleo | 19:38 | |
| *** mikeit has joined #tripleo | 19:42 | |
| *** lucas-dinner has quit IRC | 19:46 | |
| *** eghobo_ has joined #tripleo | 19:53 | |
| *** radez is now known as radez_g0n3 | 19:53 | |
| *** eghobo has quit IRC | 19:56 | |
| *** akrivoka has joined #tripleo | 19:56 | |
| *** rlandy has quit IRC | 19:57 | |
| *** alop has quit IRC | 20:00 | |
| *** daneyon_ has joined #tripleo | 20:00 | |
| *** pblaho has joined #tripleo | 20:01 | |
| *** daneyon has quit IRC | 20:03 | |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: Add doc for the minimal environment vars required per container https://review.openstack.org/167777 | 20:07 |
| openstackgerrit | Steven Dake proposed stackforge/kolla: Add some documentation around running the exec_replace driver https://review.openstack.org/167781 | 20:11 |
| *** akrivoka has quit IRC | 20:11 | |
| *** rlandy has joined #tripleo | 20:18 | |
| *** dsneddon has quit IRC | 20:19 | |
| SpamapS | any word on the TripleO CI coming back? | 20:20 |
| SpamapS | Would hate for diskimage-builder to break something while it is down. | 20:21 |
| *** dsneddon has joined #tripleo | 20:22 | |
| openstackgerrit | Steven Dake proposed stackforge/kolla: Add some documentation around running the exec_replace driver https://review.openstack.org/167781 | 20:23 |
| *** akrivoka has joined #tripleo | 20:25 | |
| *** Marga_ has joined #tripleo | 20:26 | |
| *** Marga_ has quit IRC | 20:31 | |
| *** lblanchard has quit IRC | 20:33 | |
| *** alop has joined #tripleo | 20:35 | |
| mordred | SpamapS, greghaynes: what do we need to do to get 156433 in? it would let me drop some things from teh ansible role... | 20:49 |
| greghaynes | well, right now, fix tripleo CI might be part of whats needed | 20:53 |
| greghaynes | mordred: other than that I think its good to go | 20:53 |
| *** Marga_ has joined #tripleo | 20:54 | |
| *** sdake has joined #tripleo | 20:55 | |
| *** sdake has joined #tripleo | 20:55 | |
| *** sdake__ has quit IRC | 20:58 | |
| *** trown is now known as trown|outttypeww | 21:00 | |
| mordred | oh. eek | 21:01 |
| greghaynes | yea :( | 21:02 |
| mordred | greghaynes, SpamapS: well, in that case, I'm going to update the process-config-drive stuff | 21:02 |
| mordred | to better integrate the two | 21:02 |
| mordred | basically, I've added an "interface" arg to process-config-drive so that we can call it in the loop with the rest of the things | 21:03 |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: Add doc for the minimal environment vars required per container https://review.openstack.org/167777 | 21:05 |
| *** jeblair has joined #tripleo | 21:07 | |
| *** eghobo_ has quit IRC | 21:10 | |
| SpamapS | I'm torn between waiting for TripleO CI to come back and just moving forward. | 21:10 |
| SpamapS | Perhaps if its not back by Monday? | 21:10 |
| *** dprince has quit IRC | 21:12 | |
| greghaynes | We should probably talk with the other cores about that, but id be fine with that for the patches we have up now | 21:12 |
| greghaynes | they arent doing to much change to existing behavior | 21:12 |
| greghaynes | s/to/too | 21:13 |
| SpamapS | greghaynes: the dhcp-all-interfaces stuff is _tricky_ | 21:13 |
| *** rlandy has quit IRC | 21:13 | |
| greghaynes | yes, but I dont think tripleo uses it | 21:13 |
| greghaynes | dont they os-net-config now? | 21:13 |
| *** julim has quit IRC | 21:14 | |
| greghaynes | er, at least not in ci | 21:14 |
| greghaynes | oh, no | 21:14 |
| greghaynes | im wrong | 21:14 |
| greghaynes | it does | 21:15 |
| greghaynes | well, thats unfortunate | 21:16 |
| greghaynes | so yea, at a minimum we probably want some of them to sign off on that change | 21:16 |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: Add doc for the minimal environment vars required per container https://review.openstack.org/167777 | 21:19 |
| *** Marga_ has quit IRC | 21:23 | |
| SpamapS | greghaynes: we could smoke test | 21:23 |
| SpamapS | greghaynes: just run the toci job once with the change applied. | 21:23 |
| *** daneyon_ has quit IRC | 21:26 | |
| *** Marga_ has joined #tripleo | 21:26 | |
| *** daneyon has joined #tripleo | 21:26 | |
| *** rlandy has joined #tripleo | 21:29 | |
| openstackgerrit | Merged stackforge/kolla: Add doc for the minimal environment vars required per container https://review.openstack.org/167777 | 21:30 |
| openstackgerrit | Monty Taylor proposed openstack/diskimage-builder: Add a yum-minimal element that just uses yum https://review.openstack.org/166640 | 21:31 |
| openstackgerrit | Monty Taylor proposed openstack/diskimage-builder: Port centos-minimal to yum-minimal https://review.openstack.org/166641 | 21:31 |
| openstackgerrit | Monty Taylor proposed openstack/diskimage-builder: Split the debootstrap functions into an element https://review.openstack.org/166650 | 21:31 |
| openstackgerrit | Monty Taylor proposed openstack/diskimage-builder: Add ability to process Rackspace network info https://review.openstack.org/167003 | 21:31 |
| mordred | SpamapS, greghaynes: https://review.openstack.org/167003 <-- has the mildly reworked process-config-drive to integrate it into the flow a little more | 21:32 |
| mordred | also, I split the ssh-key handling out into its own script, because WOW that's not related AT ALL | 21:32 |
| *** rdopiera has quit IRC | 21:33 | |
| *** cwolferh has joined #tripleo | 21:34 | |
| openstackgerrit | greghaynes proposed openstack/diskimage-builder: Short circuit qemu-img convert for raw images https://review.openstack.org/167804 | 21:35 |
| *** penick has joined #tripleo | 21:36 | |
| mordred | greghaynes: is that goign to work with mv ? | 21:37 |
| mordred | greghaynes: if you have more than one type in the type list? | 21:37 |
| greghaynes | that is a good question | 21:37 |
| greghaynes | yea | 21:37 |
| greghaynes | boom | 21:37 |
| greghaynes | cp it is | 21:37 |
| mordred | well ... | 21:37 |
| mordred | maybe sort the list | 21:37 |
| greghaynes | ooo | 21:37 |
| mordred | so that you always do raw last | 21:37 |
| SpamapS | mordred: ok, CirrOS won't help btw.. it runs entirely from initrd's pid1 which is why it boots so damn fast. ;) | 21:37 |
| * mordred should have said that on the review so he gets dib review credit points | 21:37 | |
| mordred | SpamapS: nice | 21:38 |
| * greghaynes awards mordred 5 dib points | 21:38 | |
| * mordred puts them in a safe place | 21:38 | |
| mordred | I -1'd the patch just on principle | 21:39 |
| greghaynes | hah | 21:39 |
| openstackgerrit | Clint 'SpamapS' Byrum proposed openstack/diskimage-builder: Add a CirrOS element https://review.openstack.org/167807 | 21:41 |
| * SpamapS WIP's that good | 21:41 | |
| lifeless | WIP it hard? | 21:42 |
| SpamapS | WIP it | 21:42 |
| SpamapS | into shape | 21:42 |
| mordred | WIP it REAL good | 21:42 |
| SpamapS | When a patchset comes along | 21:43 |
| *** mikeit has quit IRC | 21:43 | |
| SpamapS | YOU must WIP it | 21:43 |
| mordred | you know - I love how I can delete images by name from hp which runs glance v1 | 21:44 |
| mordred | but on rax, where it's glance v2, I have to do them by id | 21:45 |
| mordred | it's like they are two cmpletely unrelated clouds | 21:45 |
| *** rhallisey has quit IRC | 21:45 | |
| greghaynes | it is amazing how different they are | 21:45 |
| mordred | SO DIFFERENT | 21:45 |
| greghaynes | the whole swift uploading process is amazing | 21:46 |
| greghaynes | in a not amazing kind of way | 21:46 |
| mordred | greghaynes: I LOVE my ansible playbook | 21:47 |
| mordred | btw | 21:47 |
| mordred | like, in a dirty kinda gross way | 21:47 |
| greghaynes | I saw that, its funy because I have a very similar ansible playbook | 21:47 |
| mordred | :) | 21:47 |
| greghaynes | but mine doesnt actually use a cloud, it just rsync's an image over and libvirt's | 21:47 |
| mordred | o_0 | 21:48 |
| greghaynes | yeaaaa | 21:48 |
| mordred | greghaynes: oh - also - https://github.com/emonty/puppet-diskimage-builder | 21:48 |
| greghaynes | hahaha | 21:48 |
| SpamapS | mordred: what happened there? Did the glance v1 team just stop reviewing, and the v2 team was like a Bolivian Tree Lizard, laying its v2 egg and eating v1? | 21:48 |
| mordred | to just make everything full circule | 21:48 |
| mordred | SpamapS: yes | 21:48 |
| mordred | SpamapS: semver happened | 21:48 |
| mordred | "it's ok that everything is different, we bumped the major number" | 21:49 |
| mordred | literally. this was said by the PTL | 21:49 |
| mordred | of course, he works for coreos now | 21:49 |
| SpamapS | This happens a lot.. the thing that is simplest and least problematic gets railroaded by WTF every time. | 21:49 |
| *** dpaterson has quit IRC | 21:50 | |
| mordred | so - to be fair - actually transiting the data payload through the glance api when there is a swift api already | 21:50 |
| mordred | is kinda daft | 21:50 |
| mordred | because it means you're uploading a large file through a rest api without the benefit of the swift scale out model | 21:50 |
| greghaynes | sure, the client should just be smart enough to do this though | 21:50 |
| mordred | YES | 21:51 |
| mordred | that is the big problem | 21:51 |
| SpamapS | I mean, 302's are a thing, right? | 21:51 |
| SpamapS | "your PUT goes over there" .. done. | 21:51 |
| greghaynes | hrm, thats an interesting idea | 21:52 |
| SpamapS | aaaand we've just invented v3 | 21:52 |
| greghaynes | hah | 21:52 |
| SpamapS | MogileFS worked like this | 21:52 |
| SpamapS | you'd say "Hey I have a file I want to name abc123" and it would say "GREAT! Upload it to http://server3/uuid-12356678234910 kthxbai" | 21:53 |
| greghaynes | that makes way too much sense | 21:53 |
| SpamapS | Not "Hey I have a URL i want you to hang on to" | 21:53 |
| *** penick has quit IRC | 21:59 | |
| *** untriaged-bot has joined #tripleo | 22:00 | |
| untriaged-bot | Untriaged bugs so far: | 22:00 |
| untriaged-bot | https://bugs.launchpad.net/tripleo/+bug/1430445 | 22:00 |
| openstack | Launchpad bug 1430445 in tripleo "Horizon translations aren't compiled" [Undecided,In progress] - Assigned to Steve McLellan (sjmc7) | 22:00 |
| untriaged-bot | https://bugs.launchpad.net/diskimage-builder/+bug/1430599 | 22:00 |
| openstack | Launchpad bug 1430599 in diskimage-builder "ramdisk-image-create with ironic-discoverd-ramdisk element fails" [Undecided,Fix committed] - Assigned to Yuiko Takada (takada-yuiko) | 22:00 |
| untriaged-bot | https://bugs.launchpad.net/diskimage-builder/+bug/1436523 | 22:00 |
| openstack | Launchpad bug 1436523 in diskimage-builder "Download of translations not properly disabled in APT" [Undecided,In progress] - Assigned to Mathieu Gagné (mgagne) | 22:00 |
| untriaged-bot | https://bugs.launchpad.net/tripleo/+bug/1429902 | 22:00 |
| openstack | Launchpad bug 1429902 in tripleo "os-refresh-config fails during post-configure phase" [Undecided,New] | 22:00 |
| untriaged-bot | https://bugs.launchpad.net/tripleo/+bug/1434431 | 22:00 |
| openstack | Launchpad bug 1434431 in tripleo "Abnormal exit in building rhel images by diskimage-builder for system is not registered." [Undecided,New] | 22:00 |
| *** untriaged-bot has quit IRC | 22:00 | |
| *** mrunge has quit IRC | 22:01 | |
| *** penick has joined #tripleo | 22:03 | |
| *** thrash is now known as thrash|g0ne | 22:03 | |
| *** rlandy has quit IRC | 22:09 | |
| *** akrivoka has quit IRC | 22:10 | |
| *** weshay has quit IRC | 22:10 | |
| *** sdake__ has joined #tripleo | 22:14 | |
| *** jcoufal has quit IRC | 22:16 | |
| *** noslzzp has quit IRC | 22:16 | |
| *** prad has quit IRC | 22:17 | |
| *** sdake has quit IRC | 22:17 | |
| *** sdake has joined #tripleo | 22:17 | |
| *** Marga_ has quit IRC | 22:19 | |
| *** Marga_ has joined #tripleo | 22:20 | |
| *** sdake__ has quit IRC | 22:21 | |
| *** tzumainn has quit IRC | 22:24 | |
| *** athomas has quit IRC | 22:27 | |
| *** eghobo has joined #tripleo | 22:31 | |
| *** jcoufal has joined #tripleo | 22:31 | |
| *** jdob has quit IRC | 22:33 | |
| *** morazi has quit IRC | 22:34 | |
| *** tzumainn has joined #tripleo | 22:37 | |
| *** jcoufal has quit IRC | 22:40 | |
| *** akrivoka has joined #tripleo | 22:42 | |
| *** akrivoka has quit IRC | 22:50 | |
| openstackgerrit | greghaynes proposed openstack/diskimage-builder: Short circuit qemu-img convert for raw images https://review.openstack.org/167804 | 22:50 |
| openstackgerrit | greghaynes proposed openstack/diskimage-builder: Short circuit qemu-img convert for raw images https://review.openstack.org/167804 | 22:50 |
| *** Marga_ has quit IRC | 22:51 | |
| *** Marga_ has joined #tripleo | 22:52 | |
| *** penick has quit IRC | 23:02 | |
| *** penick has joined #tripleo | 23:04 | |
| *** Marga_ has quit IRC | 23:06 | |
| *** Marga_ has joined #tripleo | 23:07 | |
| *** chlong has joined #tripleo | 23:10 | |
| openstackgerrit | greghaynes proposed openstack/diskimage-builder: Add generic devuser element https://review.openstack.org/153439 | 23:13 |
| *** alop has quit IRC | 23:13 | |
| *** ChuckC has quit IRC | 23:18 | |
| *** rhallisey has joined #tripleo | 23:21 | |
| *** ChuckC has joined #tripleo | 23:32 | |
| *** chlong has quit IRC | 23:37 | |
| *** chlong has joined #tripleo | 23:37 | |
| *** rhallisey has quit IRC | 23:54 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!