| *** penick has quit IRC | 00:08 | |
| -openstackstatus- NOTICE: Restarted gerrit due to stuck stream-events connections. Events since 23:49 were missed and changes uploaded since then will need to be rechecked. | 00:43 | |
| *** cody-somerville has quit IRC | 01:00 | |
| *** weshay has quit IRC | 01:08 | |
| *** eghobo has quit IRC | 01:14 | |
| *** rhallisey has quit IRC | 01:31 | |
| *** Marga_ has quit IRC | 01:36 | |
| *** spzala has quit IRC | 01:41 | |
| *** derekh_ has joined #tripleo | 01:46 | |
| *** penick has joined #tripleo | 01:48 | |
| *** derekh has quit IRC | 01:50 | |
| *** penick_ has joined #tripleo | 01:51 | |
| *** penick has quit IRC | 01:52 | |
| *** penick_ is now known as penick | 01:52 | |
| *** yamahata has quit IRC | 02:22 | |
| *** mestery has joined #tripleo | 02:23 | |
| *** penick has quit IRC | 02:41 | |
| openstackgerrit | Merged openstack/diskimage-builder: Remove unused map_nbd function https://review.openstack.org/174084 | 02:50 |
|---|---|---|
| *** untriaged-bot has joined #tripleo | 03:00 | |
| untriaged-bot | Untriaged bugs so far: | 03:00 |
| untriaged-bot | https://bugs.launchpad.net/diskimage-builder/+bug/1451634 | 03:00 |
| openstack | Launchpad bug 1451634 in diskimage-builder "ISO image of deploy ramdisk doesn't have provision to log things to serial console" [Undecided,In progress] - Assigned to Ramakrishnan G (rameshg87) | 03:00 |
| untriaged-bot | https://bugs.launchpad.net/diskimage-builder/+bug/1449852 | 03:00 |
| openstack | Launchpad bug 1449852 in diskimage-builder "Buidling ramdisk with ironic-agent behind proxy fails" [Undecided,In progress] - Assigned to Ramakrishnan G (rameshg87) | 03:00 |
| untriaged-bot | https://bugs.launchpad.net/diskimage-builder/+bug/1449882 | 03:00 |
| untriaged-bot | https://bugs.launchpad.net/diskimage-builder/+bug/1449854 | 03:00 |
| openstack | Launchpad bug 1449882 in diskimage-builder "Deploying with ironic-agent ramdisk built from fedora 21 cloud image fails" [Undecided,In progress] - Assigned to Ramakrishnan G (rameshg87) | 03:00 |
| openstack | Launchpad bug 1449854 in diskimage-builder "Ironic agent ramdisk built using disk-image-create fails with iscsi_ilo driver" [Undecided,In progress] - Assigned to Ramakrishnan G (rameshg87) | 03:00 |
| *** untriaged-bot has quit IRC | 03:00 | |
| *** alop has quit IRC | 03:08 | |
| *** yamahata has joined #tripleo | 03:11 | |
| *** tzumainn_ has quit IRC | 03:25 | |
| *** lazy_prince has joined #tripleo | 03:35 | |
| *** lazy_prince has quit IRC | 03:38 | |
| *** lazy_prince has joined #tripleo | 03:39 | |
| *** links has joined #tripleo | 03:41 | |
| *** Marga_ has joined #tripleo | 03:49 | |
| *** Marga_ has quit IRC | 03:50 | |
| *** Marga_ has joined #tripleo | 03:51 | |
| *** sdake has joined #tripleo | 03:51 | |
| *** sdake_ has joined #tripleo | 03:57 | |
| *** sdake has quit IRC | 04:00 | |
| *** sdake has joined #tripleo | 04:01 | |
| *** eghobo has joined #tripleo | 04:02 | |
| *** sdake__ has joined #tripleo | 04:03 | |
| *** sdake_ has quit IRC | 04:04 | |
| *** sdake has quit IRC | 04:07 | |
| *** sdake__ is now known as sdake | 04:13 | |
| *** eghobo has quit IRC | 04:14 | |
| *** cwolferh has quit IRC | 04:40 | |
| *** regebro has quit IRC | 04:55 | |
| *** Marga_ has quit IRC | 05:01 | |
| *** yog_ has joined #tripleo | 05:05 | |
| *** sdake_ has joined #tripleo | 05:23 | |
| *** sdake has quit IRC | 05:27 | |
| *** subscope_ has joined #tripleo | 05:36 | |
| *** yog_ has quit IRC | 05:40 | |
| *** masco has joined #tripleo | 05:49 | |
| *** dobson has quit IRC | 05:52 | |
| *** rlandy has joined #tripleo | 05:57 | |
| *** dasm|afk is now known as dasm | 06:01 | |
| *** subscope_ has quit IRC | 06:02 | |
| *** jcoufal has joined #tripleo | 06:06 | |
| *** dobson has joined #tripleo | 06:07 | |
| *** ukalifon1 has joined #tripleo | 06:14 | |
| *** ishant has joined #tripleo | 06:23 | |
| *** aufi has joined #tripleo | 06:37 | |
| *** jprovazn has joined #tripleo | 06:39 | |
| *** tteggel has quit IRC | 06:41 | |
| *** bkero has quit IRC | 06:41 | |
| *** yamahata has quit IRC | 06:41 | |
| *** tteggel has joined #tripleo | 06:41 | |
| *** bkero has joined #tripleo | 06:43 | |
| *** athomas has joined #tripleo | 06:45 | |
| *** jayg|g0n3 has quit IRC | 06:46 | |
| *** jayg|g0n3 has joined #tripleo | 06:46 | |
| *** yamahata has joined #tripleo | 06:48 | |
| *** jcoufal_ has joined #tripleo | 06:49 | |
| *** jcoufal has quit IRC | 06:52 | |
| *** shardy_z is now known as shardy | 06:54 | |
| *** jtomasek has joined #tripleo | 06:59 | |
| *** mmagr has joined #tripleo | 07:00 | |
| *** yog_ has joined #tripleo | 07:07 | |
| *** eghobo has joined #tripleo | 07:36 | |
| *** panda|Zz is now known as panda|bfast | 07:38 | |
| *** ifarkas has joined #tripleo | 07:38 | |
| *** panda|bfast is now known as panda | 07:47 | |
| *** panda is now known as Guest56018 | 07:47 | |
| *** Guest56018 is now known as panda | 07:48 | |
| *** regebro has joined #tripleo | 07:57 | |
| *** gfidente has joined #tripleo | 07:57 | |
| *** jistr has joined #tripleo | 07:59 | |
| *** hasharMeeting is now known as hashar | 08:01 | |
| *** Goneri has joined #tripleo | 08:05 | |
| *** lucasagomes has joined #tripleo | 08:08 | |
| hewbrocca | jistr: Did you have a chance to check with shardy re the fence device config? | 08:10 |
| *** lsmola has joined #tripleo | 08:11 | |
| jistr | hewbrocca: not yet, but we might open the topic right now | 08:13 |
| * jistr digs up some info | 08:14 | |
| gfidente | jistr, hewbrocca that is on how to pass the fencing data via heat? | 08:16 |
| jistr | so, i think we support having different IPMI credentials for different nodes in overcloud, so i think we'll need to pass some structured data to heat re fencing params | 08:16 |
| hewbrocca | gfidente: yes -- this seems to be the long pole for fencing, right jistr ? | 08:17 |
| jistr | (unlike Astapor, which assumed the IPMI username and passwords are the same IIRC) | 08:17 |
| jistr | yeah | 08:18 |
| jistr | so i suggested making it a JSON param, similar to extra config https://github.com/openstack/tripleo-heat-templates/blob/48c3627baae7a061b22f273c3d55151577085895/overcloud-without-mergepy.yaml#L283-L319 | 08:18 |
| jistr | and we would pass in a JSON array of dictionaries, each dictionary describing one controller node's fencing information | 08:20 |
| hewbrocca | Seems like a reasonable idea | 08:21 |
| jistr | assuming that part ^^ sounds right, i'm wondering about two things | 08:21 |
| jistr | 1. how to integrate such solution with the lower levels, i.e. when user actually wants to deploy a real cloud, how do we feed that JSON parameter | 08:22 |
| *** chlong has quit IRC | 08:22 | |
| jistr | one possible solution for that in the short term is that we simply let the user write that thing manually | 08:23 |
| gfidente | jistr, so I remember we discussed some of this already the other day | 08:24 |
| jistr | in the long term and in an ideal world, this might be somehow fetched from ironic, because the user has already entered that info there | 08:24 |
| gfidente | I think, given we already have the data in the json file | 08:24 |
| gfidente | we might pass it along using the get_file | 08:25 |
| gfidente | so user doesn't have to enter it twice | 08:25 |
| jistr | (that's out of t-h-t/puppet scope, probably to be solved on tuskar or instack-undercloud level, but needs to be solved anyway) | 08:25 |
| gfidente | and we don't have to provide credentials for ironic in the overcloud | 08:25 |
| gfidente | I think I can work on this specific piece as I am more comfortable with it | 08:26 |
| gfidente | my problem was more on how to parse the json from the manifest | 08:26 |
| gfidente | it looks like we can match the mac address | 08:27 |
| *** bvandenh has quit IRC | 08:27 | |
| gfidente | because that seems to be the only 'unique' piece of information we have in the json to distinguish the nodes | 08:27 |
| *** Slower has quit IRC | 08:28 | |
| gfidente | but overall I am not sure how to work with the data | 08:28 |
| jistr | gfidente: yeah that was actually a question no. 2 i see as open -- how to match the fencing data with a particular node, mac addr seems like the best option | 08:28 |
| gfidente | to get the fencing agents configured in pcmk | 08:28 |
| gfidente | jistr, yeah because ironic doesn't have any idea about the hostname | 08:29 |
| jistr | yeah re not sure how to work with the data -- this gets us back to question 1 :) | 08:29 |
| jistr | gfidente: yeah because hostname is only assigned on stack creation | 08:29 |
| *** eghobo has quit IRC | 08:29 | |
| gfidente | exactly so | 08:29 |
| gfidente | stdlib has something to parse json though, is this correct? | 08:30 |
| jistr | gfidente: so the problem i see with feeding instackenv.json or testenv.json to the stack is | 08:30 |
| lifeless | the kson module | 08:30 |
| lifeless | bah | 08:30 |
| lifeless | json | 08:30 |
| gfidente | hi lifeless | 08:30 |
| jistr | that the data in the json is not exactly what we want to configure fencing | 08:31 |
| gfidente | ack, I haven't investigated this | 08:31 |
| gfidente | if we miss something in that json | 08:32 |
| jistr | i don't think we should accept a JSON which has a different purpose into the template, and then do processing on top of that json inside the template or inside pupet | 08:32 |
| jistr | *puppet | 08:32 |
| jistr | e.g. translating ironic driver names into fence agent names etc. | 08:32 |
| jistr | i think heat should receive as pure parameters as possible and not do any computations on them | 08:32 |
| gfidente | well get_file won't do any computation | 08:33 |
| gfidente | so we'd be forced to do that in the manifet | 08:33 |
| jistr | yeah exactly :( | 08:33 |
| gfidente | not nice, I see | 08:33 |
| gfidente | but alternatively user should really paste twice 'same' data no? | 08:33 |
| gfidente | I mean, username, ip and all the other stuff is very much the same | 08:34 |
| jistr | yeah that's right | 08:35 |
| jistr | another issue is | 08:35 |
| jistr | i'm not really sure we'll have the instackenv file forever | 08:36 |
| hewbrocca | Something else to think about -- there are other kinds of fence agents/devices that are not IPMI | 08:36 |
| hewbrocca | we're not going to support them right now, but | 08:36 |
| hewbrocca | I'm not persuaded we'll be able to get all fence agent data from Ironic in all cases *anyway* | 08:37 |
| *** Marga_ has joined #tripleo | 08:37 | |
| gfidente | uhm.. if Ironic doesn't have those we are out of luck with tripleo entirely | 08:37 |
| hewbrocca | in which case, seems like having the fencing creds be in their own json data isn'ŧ necessarily redundant | 08:38 |
| jistr | yeah that sounds right to me. Which i think relates to what i wrote above and why i don't want to simply feed instackenv.json/testenv.json into the template -- they are not the same as the fencing config | 08:39 |
| hewbrocca | power supply fence devices? | 08:39 |
| hewbrocca | Yeah, that's kinda what I was thinking | 08:39 |
| hewbrocca | Anyway, I guess I wouldn't get too hung up on making the operator do the config twice, for the short term at least | 08:40 |
| jistr | we might be able to compute the fencing config from the info contained in those files for some cases, but even if we choose to do so, such computation should happen before heat/puppet gets into play. Heat should already receive final fencing params imho. | 08:41 |
| gfidente | so I think having to pass fence params will work | 08:42 |
| *** mcornea has joined #tripleo | 08:43 | |
| gfidente | I was trying to avoid it because, to me, there is no difference in the data Ironic needs and the data Pacemaker needs | 08:43 |
| *** mcornea has quit IRC | 08:43 | |
| gfidente | they are both trying to achieve same thing | 08:43 |
| gfidente | but maybe we should not make assumption at the stage of overcloud deployment on the undercloud config | 08:44 |
| gfidente | so as jistr said not rely on existence of the json file forever | 08:44 |
| *** mcornea has joined #tripleo | 08:44 | |
| hewbrocca | Seems sensible | 08:45 |
| gfidente | having data passed in its own param then is a way to decouple from the json, even though we might do the translation in the script | 08:45 |
| gfidente | jistr, before pasting it into heat, is this what you had in mind? | 08:45 |
| hewbrocca | Next step, put a proposal on the mailing list? | 08:45 |
| jistr | gfidente: +1 yes | 08:45 |
| *** bvandenh has joined #tripleo | 08:47 | |
| gfidente | ack, thanks for talking a little about :) | 08:47 |
| gfidente | jistr, things only cleared up to me here: computation should happen before heat/puppet gets into play | 08:51 |
| gfidente | I was not understanding that part initially :) | 08:51 |
| *** pdb_ has quit IRC | 08:56 | |
| *** pdb_ has joined #tripleo | 08:57 | |
| *** lsmola has quit IRC | 08:58 | |
| *** jtomasek has quit IRC | 08:58 | |
| *** morazi has quit IRC | 08:58 | |
| jistr | gfidente: the other bit which i still see as open, and this will probably require also some more investigation of code, is what is all the needed data for various fence agents, and how to map it to particular nodes. I think the mapping will have to happen via mac address as you suggested. | 08:58 |
| jistr | i see that astapor takes hostname->IP mappings as a parameter for ipmilan https://github.com/redhat-openstack/astapor/blob/master/puppet/modules/quickstack/manifests/pacemaker/common.pp#L38 | 08:59 |
| *** shardy_ has joined #tripleo | 08:59 | |
| jistr | this is something which we cannot take into Heat as a parameter, because as we said above, hostnames get only assigned when stack-create is already in progress | 09:00 |
| *** shardy has quit IRC | 09:00 | |
| gfidente | yeah and heat doesn't know about the mac address of a server probably | 09:00 |
| jistr | but we should be able to compute that part in Heat/Puppet | 09:00 |
| gfidente | maybe shardy_ can tell for sure? | 09:00 |
| *** untriaged-bot has joined #tripleo | 09:00 | |
| untriaged-bot | Untriaged bugs so far: | 09:00 |
| untriaged-bot | https://bugs.launchpad.net/diskimage-builder/+bug/1451634 | 09:00 |
| openstack | Launchpad bug 1451634 in diskimage-builder "ISO image of deploy ramdisk doesn't have provision to log things to serial console" [Undecided,In progress] - Assigned to Ramakrishnan G (rameshg87) | 09:00 |
| untriaged-bot | https://bugs.launchpad.net/diskimage-builder/+bug/1449852 | 09:00 |
| openstack | Launchpad bug 1449852 in diskimage-builder "Buidling ramdisk with ironic-agent behind proxy fails" [Undecided,In progress] - Assigned to Ramakrishnan G (rameshg87) | 09:00 |
| untriaged-bot | https://bugs.launchpad.net/diskimage-builder/+bug/1449882 | 09:00 |
| openstack | Launchpad bug 1449882 in diskimage-builder "Deploying with ironic-agent ramdisk built from fedora 21 cloud image fails" [Undecided,In progress] - Assigned to Ramakrishnan G (rameshg87) | 09:00 |
| untriaged-bot | https://bugs.launchpad.net/diskimage-builder/+bug/1449854 | 09:00 |
| openstack | Launchpad bug 1449854 in diskimage-builder "Ironic agent ramdisk built using disk-image-create fails with iscsi_ilo driver" [Undecided,In progress] - Assigned to Ramakrishnan G (rameshg87) | 09:00 |
| *** lsmola has joined #tripleo | 09:00 | |
| *** jtomasek has joined #tripleo | 09:00 | |
| *** morazi has joined #tripleo | 09:00 | |
| *** untriaged-bot has quit IRC | 09:00 | |
| gfidente | jistr, maybe with some introspection of the server resource we do get to its mac, not sure | 09:00 |
| jistr | gfidente: yeah we can get mac from puppet | 09:01 |
| gfidente | nah I meant in heat | 09:01 |
| gfidente | from the nova server resources | 09:01 |
| jistr | gfidente: maybe, and that would be nice. If not, i think we should still have a fallback solution at hand -- distribute the full fencing config JSON to all controllers, and let puppet pick the right bit from it based on local mac address. | 09:02 |
| gfidente | yeah | 09:02 |
| gfidente | heat probably won't be able to do any mapping anyway | 09:03 |
| *** shardy_ has quit IRC | 09:04 | |
| *** shardy has joined #tripleo | 09:05 | |
| openstackgerrit | Ramakrishnan G proposed openstack/diskimage-builder: Add grub2 element https://review.openstack.org/167929 | 09:08 |
| jistr | gfidente: yeah i'm not sure if heat can look inside a JSON parameter and e.g. extract only some parts from it (array or dictionary items), can't find any example of it right now | 09:08 |
| *** yog_ has quit IRC | 09:09 | |
| jistr | shardy: good morning :) we've discussed how to do fencing on Heat+Puppet side for a while here with gfidente and hewbrocca. When you have some bandwidth available, could you please read through that and alert us if you think we're off somewhere? | 09:11 |
| jistr | shardy: i saw your connection dropped at one point, here's a full log http://eavesdrop.openstack.org/irclogs/%23tripleo/%23tripleo.2015-05-06.log | 09:11 |
| *** mmagr is now known as mmagr|afk | 09:12 | |
| openstackgerrit | Ramakrishnan G proposed openstack/diskimage-builder: Address comments on virtual media device label commit https://review.openstack.org/178672 | 09:13 |
| shardy | jistr: morning! | 09:13 |
| jistr | gfidente: thanks for going through that :) i feel like we might be a bit closer to fencing success at this point :D | 09:14 |
| shardy | Firstly, yes, heat can extract parts from a json parameter by path, via get_param | 09:14 |
| jistr | neat | 09:14 |
| shardy | jistr: the syntax works the same as for path-based get_attr | 09:14 |
| shardy | and IIRC it supports both key (map) and index (list) lookup | 09:15 |
| shardy | get_param: json_param, akey, 0, foo | 09:15 |
| shardy | etc | 09:15 |
| shardy | get_param: [json_param, akey, 0, foo] | 09:15 |
| shardy | I mean.. | 09:15 |
| shardy | jistr: and sure, I'll read through the fencing discussion | 09:16 |
| *** mmagr|afk is now known as mmagr | 09:16 | |
| jistr | shardy: thanks! | 09:19 |
| jistr | to be able to use such lookup on heat level we'd also probably need what gfidente wrote above -- be able to get a mac address via looking at OS::Nova::Server attributes somehow | 09:21 |
| gfidente | jistr, shardy I am also unsure how to match a list item from an attribute of the hash in the item | 09:22 |
| shardy | jistr: you should chat to dprince about that, AFAIK he's been looking into things related to nova node->MAC mappings | 09:23 |
| shardy | the problem is, you don't know that data ahead of time, and the discovery data is only accessible via Ironic, not Nova | 09:23 |
| jistr | gfidente: yeah i thought that might be too much as well :) maybe we could get around it by making it a hash where the keys would be MACs, instead of making it an array of hashes | 09:23 |
| gfidente | jistr, ack, doing the 'transformation' before pasting it into heat template | 09:24 |
| jistr | yup | 09:24 |
| *** pblaho has joined #tripleo | 09:25 | |
| shardy | jistr: So, I think we'll have to pass in a full mapping for all nodes as a parameter, then pass that into a SoftwareDeployment as an input | 09:31 |
| shardy | then the SoftwareConfig applied by the deployment can do the lookup and extract the info for that node, e.g by mac or whatever | 09:31 |
| shardy | atm I can't see a way to do that lookup ahead of time, because we don't know what node Nova will pick when scheduling the deployment | 09:32 |
| jistr | shardy: ack. So if i read that right, we'd extract the right key from the full mapping at Puppet level. | 09:33 |
| shardy | there probably is a way to clean that up in due course, but I wouldn't block on it, if passing the big list in will get the job done for now | 09:33 |
| shardy | jistr: yeah, puppet, or jq in a script, or whatever | 09:33 |
| jistr | shardy: right, thanks :) | 09:34 |
| shardy | jistr: we might be able to reduce the map a bit in the template, e.g by having the map categorized by role | 09:35 |
| shardy | so you pass in get_param: [JsonParamFoo, 'compute'] or something | 09:36 |
| jistr | shardy: i think at this point we'll only need controller nodes anyway, because they'd be the only ones in the pacemaker cluster | 09:36 |
| shardy | jistr: Ok, cool, makes sense, thanks! :) | 09:36 |
| *** athomas has quit IRC | 09:48 | |
| *** Marga_ has quit IRC | 09:50 | |
| *** athomas has joined #tripleo | 09:53 | |
| *** bvandenh has quit IRC | 10:04 | |
| *** akrivoka has joined #tripleo | 10:14 | |
| openstackgerrit | Giulio Fidente proposed stackforge/puppet-tripleo: Add support for service_ensure https://review.openstack.org/180484 | 11:01 |
| openstackgerrit | Giulio Fidente proposed stackforge/puppet-tripleo: Add support for service_ensure https://review.openstack.org/180484 | 11:03 |
| openstackgerrit | Giulio Fidente proposed stackforge/puppet-tripleo: Add support for ensure_enabled https://review.openstack.org/180484 | 11:05 |
| openstackgerrit | Giulio Fidente proposed openstack/tripleo-heat-templates: Ensure services are not enabled by puppet-tripleo when EnablePacemaker https://review.openstack.org/180490 | 11:07 |
| openstackgerrit | Marios Andreou proposed openstack/tripleo-heat-templates: Add metering agent as neutron-server plugin https://review.openstack.org/179370 | 11:13 |
| *** rhallisey has joined #tripleo | 11:13 | |
| *** morazi has quit IRC | 11:13 | |
| *** pdb_ has quit IRC | 11:28 | |
| *** yamahata has quit IRC | 11:30 | |
| *** hashar has quit IRC | 11:40 | |
| *** hashar has joined #tripleo | 11:40 | |
| *** hashar has quit IRC | 11:44 | |
| *** hashar has joined #tripleo | 11:44 | |
| *** shardy_ has joined #tripleo | 11:45 | |
| *** shardy has quit IRC | 11:46 | |
| gfidente | so I wasn't expecting github to go down so much | 11:47 |
| *** hashar has quit IRC | 11:49 | |
| *** weshay has joined #tripleo | 11:50 | |
| *** shardy_ has quit IRC | 11:50 | |
| *** shardy has joined #tripleo | 11:51 | |
| *** dprince has joined #tripleo | 11:51 | |
| openstackgerrit | Marios Andreou proposed openstack/tripleo-heat-templates: Add metering agent as neutron-server plugin https://review.openstack.org/179370 | 11:54 |
| openstackgerrit | Marios Andreou proposed openstack/tripleo-heat-templates: Adds neutron-server and agents as pacemaker resources https://review.openstack.org/178743 | 11:54 |
| marios | gfidente: thanks for pinging crag on that review - i updated ^^^ I think they should be good to go now | 11:54 |
| gfidente | they will fail on github though :P | 11:54 |
| marios | gfidente: jistr said it best 'single point of failure for it industry' | 12:00 |
| openstackgerrit | James Slagle proposed openstack/tripleo-puppet-elements: Rename $name variable to $module_name https://review.openstack.org/180499 | 12:01 |
| *** morazi has joined #tripleo | 12:06 | |
| *** sdake has joined #tripleo | 12:07 | |
| openstackgerrit | Giulio Fidente proposed openstack/tripleo-heat-templates: Add Glance as a Pacemaker resource when EnablePacemaker https://review.openstack.org/178152 | 12:08 |
| gfidente | jistr, ^^ that should do it | 12:09 |
| gfidente | with the logic in the depending change for puppet-glance | 12:09 |
| gfidente | I am going to attempt same change for puppet-keystone as well | 12:09 |
| gfidente | jdob, marios ^^ I think we might need changes in the keystone/neutron changes as well ;( | 12:10 |
| *** lucasagomes is now known as lucas-hungry | 12:11 | |
| *** sdake_ has quit IRC | 12:12 | |
| marios | gfidente: is this wrt db-sync and enabled_service or something ( i missed the details. db-sync doesn't run if enabled?) | 12:13 |
| gfidente | marios, yeah | 12:14 |
| *** lblanchard has joined #tripleo | 12:14 | |
| marios | gfidente: so would be needed in upstream puppet-neutron for example? or does this setting to 'false' fix for now? | 12:14 |
| gfidente | we'll have to check it | 12:14 |
| gfidente | depends on how the module uses the enabled and manage params | 12:14 |
| gfidente | give me a minute | 12:15 |
| jistr | so i wonder if $manage_service, $enabled, and $sync_db should be totally independent? or is one of them a prerequisite for some other? | 12:15 |
| jistr | $manage_service -- start/stop | 12:15 |
| jistr | $enabled -- start/stop during boot | 12:16 |
| jistr | $sync_db -- sync db :) | 12:16 |
| gfidente | jistr, I like the way rabbitmq copes with it | 12:16 |
| gfidente | it just doesn't define service if mange_service is false | 12:16 |
| gfidente | but I can live with the 'enabled' boolean the openstack modules have | 12:16 |
| gfidente | so long as it is consumed correctly, that is it won't prevent execution of db_sync | 12:17 |
| jdob | gfidente: whats the verdict on the keystone changes, do I need to do something or were you checking on it? | 12:17 |
| gfidente | jdob, minute and will tell | 12:17 |
| jdob | kk | 12:17 |
| openstackgerrit | Giulio Fidente proposed openstack/tripleo-heat-templates: Use pacemaker::resource::service instead of ::systemd https://review.openstack.org/180506 | 12:18 |
| jdob | conceptually I actually mostly follow this \o/ | 12:18 |
| gfidente | jdob, so check https://review.openstack.org/#/c/180229/ | 12:18 |
| gfidente | I need to make the tests pass, but that is what we might need to do in puppet-keystone as well | 12:19 |
| jdob | ah, so it'll just be adding in some more flags | 12:19 |
| jdob | no rush, lemme know what you find | 12:19 |
| gfidente | not really adding a flag, but figuring how to consume the existing to achieve what we want | 12:19 |
| jdob | i just got in anyway | 12:19 |
| jistr | gfidente: hmm, that might be a way... i'm a bit concerned though that if we don't define the service resource at all and something wants to depend on it wrt ordering (via the arrows -> or ~>), then that breaks the puppet run and the deployment | 12:20 |
| gfidente | jistr, well that is how rabbitmq and mysql do it | 12:20 |
| gfidente | jistr, for the openstack modules I am fine with using $enabled instead of $manage_service | 12:21 |
| gfidente | so we do define the service class, but enabled goes to false | 12:21 |
| gfidente | we just need to make sure it won't prevent db_sync | 12:21 |
| gfidente | makes sense to you? | 12:21 |
| jistr | gfidente: ok enabled goes to false, but if you remove manage_service, then that defaults to true, no? | 12:22 |
| gfidente | remove manage_service? | 12:22 |
| openstackgerrit | Giulio Fidente proposed openstack/tripleo-heat-templates: Use pacemaker::resource::service instead of ::systemd https://review.openstack.org/180506 | 12:23 |
| jistr | like you did here gfidente: so the service wouldn't be enabled, but it would be started? | 12:23 |
| jistr | https://review.openstack.org/#/c/178152/7/puppet/manifests/overcloud_controller.pp | 12:23 |
| gfidente | jistr, ah no it won't go to running because of https://github.com/stackforge/puppet-glance/blob/master/manifests/registry.pp#L360-L378 | 12:25 |
| gfidente | am I reading it right? | 12:25 |
| jistr | ah ok, but it will go to stopped, right? (when enable = false and manage_service = true) | 12:26 |
| jistr | as in keystone's case https://github.com/stackforge/puppet-keystone/blob/master/manifests/init.pp#L775-L781 | 12:26 |
| jistr | which will break us too, unfortunately | 12:26 |
| *** jayg|g0n3 is now known as jayg | 12:26 | |
| gfidente | it goes to stopped yes, why it breaks? | 12:27 |
| jistr | because every puppet run then stops the service instead of leaving it as it was... | 12:29 |
| gfidente | jistr, that is on update | 12:29 |
| gfidente | but still, it's the module which should make enabled match the service->enable thing but not the service->ensure thing, right? | 12:30 |
| jistr | not only on update, also during the steps. E.g. in step 2 you do the db sync, service is ensure => stopped, which is fine. Step 3 you make pacemaker start the service, but that collides with the service ensure => stopped (could be mitigated if you do ordering of first stop the service, then create the pacemaker resource, but it's already a bad design to have contradictory settings in one puppet run) and in step 4 the service gets stopped | 12:31 |
| jistr | again and pacemaker will have to notice it's down and start it | 12:31 |
| gfidente | yeah so we should decouple db_sync AND enabled | 12:33 |
| jistr | yes | 12:33 |
| gfidente | and yet what would be the purpose of manage_service? | 12:33 |
| hewbrocca | I hate puppet starting and stopping services | 12:33 |
| hewbrocca | but, what do I know | 12:34 |
| gfidente | jistr, so in current status, if I pass manage_service and enabled to false we would get there | 12:34 |
| gfidente | that is for glance at least | 12:34 |
| gfidente | and we'd still need the change which decouples db_sync | 12:34 |
| jistr | yeah if i recall what jayg said, we shouldn't be starting the service manually, nor enabling it to start on boot, if we want to have the pacemaker setup go well | 12:36 |
| gfidente | jistr, yeah that is my intent | 12:36 |
| gfidente | problem was how to get there given logic in the mdule | 12:36 |
| jistr | gfidente: which means we'll probably want to have $enable => false and $manage_service => false and $sync_db => true, but for many modules that might not be a valid combination at the moment | 12:37 |
| jistr | yeah | 12:37 |
| jistr | exactly | 12:37 |
| gfidente | yeah so the change for decoupling db_sync is valid (and I need to make tests pass still( | 12:38 |
| gfidente | the change for glance needs manage_service -> false | 12:38 |
| jistr | yeah i think so | 12:38 |
| gfidente | and then we'll have to re-review keystone and neutron :( | 12:38 |
| *** dasm is now known as dasm|afk | 12:38 | |
| jistr | jayg: you mentioned that in a first step in Astapor you actually enable and/or start the services (i don't recall if either or both) to be able to perform db syncs. Did that need arise from the way the Puppet modules are done (it won't let you db_sync unless you enable the service)? Or do you recall some practical reason that some services really need to be running during the db sync? (that would be surprising but i thought i'd better ask) | 12:41 |
| jistr | jayg: and good morning to you btw :) | 12:41 |
| jistr | gfidente: ^^ might be interesting to know | 12:42 |
| gfidente | sure it is | 12:42 |
| gfidente | :) | 12:42 |
| jayg | jistr: good morning :) I don't recall precisely, I think that it may have been just due to puppet modules, but I cam poke back through quickstack and see if it jogs my memory | 12:46 |
| jistr | jayg: thanks | 12:47 |
| jayg | I also seem to recall that some setup of services required keystone running on initial config, though I may be remembering wrong | 12:47 |
| *** lblanchard has quit IRC | 12:47 | |
| jayg | same with rabbit and galera, which you probably realize | 12:48 |
| jayg | so it may be more that some things needed to be running, so it was just more consistent to start them all off running and then hand over control to pacemaker of service management | 12:48 |
| *** pdb_ has joined #tripleo | 12:49 | |
| *** jprovazn has quit IRC | 12:49 | |
| jistr | jayg: i'm a bit confused now... So it would be ok to start the services and then create the pacemaker resources while the servies are running? Or do you stop the services again before creating the pacemaker resources? | 12:52 |
| *** openstackgerrit has quit IRC | 12:53 | |
| jayg | no, we create the resources while the services are running | 12:53 |
| *** openstackgerrit has joined #tripleo | 12:53 | |
| jayg | the second run just gets puppet out of the way so they don't collide | 12:53 |
| *** jprovazn has joined #tripleo | 12:53 | |
| jayg | there are one or two exceptions (rabbit comes to mind), but that is the general pattern | 12:53 |
| jayg | really, any service using and RA actually | 12:54 |
| jayg | s/and/an | 12:54 |
| *** links has quit IRC | 12:54 | |
| jistr | yeah because the RA needs to take care of the start up (master election etc.) | 12:54 |
| jistr | ok then | 12:54 |
| jistr | so it's ok to create the resources while openstack services are running, but we shouldn't start/stop the services while they're already under pacemaker's control, and we shouldn't enable the services to start on boot | 12:56 |
| jistr | if i understood correctly | 12:56 |
| jistr | jayg: thanks :) | 12:57 |
| jayg | correct, pacemaker and constraints will take care of startup on boot, and if you were to stop the service without pacemaker, pacemaker would restart it almost immediately, causingcollisions | 12:58 |
| jayg | sure thing | 12:58 |
| *** spzala has joined #tripleo | 13:01 | |
| d0ugal | Any TripleO cores want to do some reviews? :) https://review.openstack.org/#/q/status:open+project:openstack/python-tuskarclient,n,z | 13:02 |
| d0ugal | We have lots of patches with one +2 | 13:03 |
| *** thrash|g0ne is now known as thrash | 13:03 | |
| *** trown|outttypeww is now known as trown | 13:05 | |
| gfidente | jistr, jayg in regard to start/stop I had this as well https://review.openstack.org/#/c/180490/ | 13:05 |
| *** lucas-hungry is now known as lucasagomes | 13:06 | |
| * jayg looks | 13:07 | |
| openstackgerrit | Marios Andreou proposed openstack/tripleo-puppet-elements: Adds neutron-lbaas-package to overcloud-controller element packages https://review.openstack.org/180530 | 13:07 |
| gfidente | jayg, and the dep | 13:08 |
| openstackgerrit | Merged openstack/python-tuskarclient: Updated from global requirements https://review.openstack.org/179874 | 13:09 |
| jistr | d0ugal: wanna get some more +1s from CLI-involved folks on the first one in the chain? https://review.openstack.org/#/c/179213/ | 13:09 |
| d0ugal | jistr: Sure, I'll ask | 13:10 |
| jayg | gfidente: right, the only question is if there is any issue because of service interdependencies or puppet modules configuration, but so long as the services are configure din the correct order, it may be just fine | 13:10 |
| jistr | d0ugal: i can review from the code point of view but not as much from "does this feature correspond with how we want the CLI to look like", as i'm not as much involved with CLI things | 13:10 |
| *** mcornea has quit IRC | 13:11 | |
| d0ugal | jistr: Gotcha, that makes sense. It is tricky as there are no CLI folk with tripleo core. | 13:11 |
| *** radez_g0n3 is now known as radez | 13:12 | |
| *** mcornea has joined #tripleo | 13:12 | |
| *** mcornea has quit IRC | 13:14 | |
| *** mcornea has joined #tripleo | 13:14 | |
| jistr | d0ugal: damn, i sent the requirements bot change in first cause i thought "sure why not", but didn't look at the rest first so i didn't realize they're making requirements changes, and now there's merge conflicts all over. Sorry about that :( I suspect that it will be just the first patch that needs amending, but we might need to re-collect the votes on the rest too :/ Not sure now if gerrit will preserve the votes on the dependent patches if | 13:15 |
| jistr | the first one changes. | 13:15 |
| *** jprovazn_ has joined #tripleo | 13:16 | |
| d0ugal | jistr: np, rebasing :) | 13:16 |
| openstackgerrit | Dougal Matthews proposed openstack/python-tuskarclient: Implement list plans for the OpenStack client https://review.openstack.org/179216 | 13:16 |
| openstackgerrit | Dougal Matthews proposed openstack/python-tuskarclient: Implement create Plan for the OpenStack client https://review.openstack.org/179217 | 13:16 |
| openstackgerrit | Dougal Matthews proposed openstack/python-tuskarclient: Implement delete Plan for the OpenStack client https://review.openstack.org/179218 | 13:16 |
| openstackgerrit | Dougal Matthews proposed openstack/python-tuskarclient: Implement Update Plan for the OpenStack client https://review.openstack.org/179219 | 13:16 |
| openstackgerrit | Dougal Matthews proposed openstack/python-tuskarclient: Implement show Plan for the OpenStack client https://review.openstack.org/179220 | 13:16 |
| openstackgerrit | Dougal Matthews proposed openstack/python-tuskarclient: Implement Plan add Role for the openstack client https://review.openstack.org/179221 | 13:16 |
| openstackgerrit | Dougal Matthews proposed openstack/python-tuskarclient: Implement Plan remove Role for the openstack client https://review.openstack.org/179222 | 13:16 |
| openstackgerrit | Dougal Matthews proposed openstack/python-tuskarclient: Implement download Plan for the openstack client https://review.openstack.org/179223 | 13:16 |
| openstackgerrit | Dougal Matthews proposed openstack/python-tuskarclient: Add initial python-openstackclient plugin https://review.openstack.org/179213 | 13:16 |
| openstackgerrit | Dougal Matthews proposed openstack/python-tuskarclient: Implement list Roles for the OpenStack client https://review.openstack.org/179215 | 13:16 |
| openstackgerrit | Dougal Matthews proposed openstack/python-tuskarclient: Add stub commands and tests for all OpenStack client commands https://review.openstack.org/179214 | 13:16 |
| d0ugal | jistr: I only lost one review from pblaho in the process, not bad :) | 13:17 |
| openstackgerrit | Marios Andreou proposed openstack/tripleo-heat-templates: Add neutron-lbaas as neutron-server plugin & neutron-lbaas-agent https://review.openstack.org/180535 | 13:17 |
| jistr | d0ugal: cool, gerrit preserved the votes on the others, at least that :) | 13:17 |
| pblaho | d0ugal: I will review the rest... | 13:18 |
| *** jprovazn has quit IRC | 13:19 | |
| *** tzumainn_ has joined #tripleo | 13:19 | |
| jistr | pblaho: just the first one in the chain is needed to re-review | 13:19 |
| pblaho | jistr: thanks | 13:19 |
| d0ugal | pblaho: thanks :) | 13:19 |
| marios | d0ugal: i can promise to make these my review run for tomorrow morning | 13:20 |
| d0ugal | marios: Great, thanks! | 13:20 |
| *** masco has quit IRC | 13:20 | |
| gfidente | dprince, works for me in its current form passing both enabled and manage_service to false | 13:23 |
| gfidente | dprince, I will soon update the glance change to use it so we get the CI results | 13:24 |
| openstackgerrit | Giulio Fidente proposed openstack/tripleo-heat-templates: Add Glance as a Pacemaker resource when EnablePacemaker https://review.openstack.org/178152 | 13:25 |
| gfidente | jistr, ^^ works for me | 13:25 |
| gfidente | let's see CI results | 13:25 |
| pblaho | d0ugal: +2 | 13:26 |
| pblaho | pblaho: at first I was afraid of non-alphabetical order of requirements but that check was dropped, right? | 13:27 |
| d0ugal | pblaho: asking yourself? :) I'm not sure. | 13:28 |
| d0ugal | pblaho: in requirements.txt? I didn't know that was ever checked. | 13:28 |
| pblaho | d0ugal: I run all tests included pep8 and it passed... | 13:28 |
| pblaho | d0ugal: I remember it was there year ago :-) | 13:29 |
| pblaho | d0ugal: I was bitten by fedora locale/sorting bug once | 13:29 |
| *** BadCub_Hiding is now known as BadCub | 13:37 | |
| *** openstackgerrit has quit IRC | 13:38 | |
| *** openstackgerrit has joined #tripleo | 13:38 | |
| *** jprovazn_ has quit IRC | 13:42 | |
| *** lifeless has quit IRC | 13:42 | |
| *** sdake_ has joined #tripleo | 13:42 | |
| *** sdake__ has joined #tripleo | 13:43 | |
| jistr | gfidente: yeah i think all the openstack puppet modules should work like what you did in puppet-glance https://review.openstack.org/#/c/180229/3 | 13:44 |
| *** sdake has quit IRC | 13:44 | |
| gfidente | jistr so EmilienM suggested to use a bug to track this | 13:45 |
| gfidente | where we add all modules which need attention | 13:45 |
| gfidente | I've added -keystone for now of which I am sure | 13:46 |
| jistr | ack | 13:47 |
| *** sdake_ has quit IRC | 13:48 | |
| openstackgerrit | Pino Toscano proposed openstack/diskimage-builder: epel: support centos element https://review.openstack.org/180547 | 13:48 |
| openstackgerrit | Pino Toscano proposed openstack/diskimage-builder: centos/centos7: switch to epel element https://review.openstack.org/180548 | 13:48 |
| openstackgerrit | Pino Toscano proposed openstack/diskimage-builder: epel: support centos element https://review.openstack.org/180547 | 13:50 |
| openstackgerrit | Pino Toscano proposed openstack/diskimage-builder: centos/centos7: switch to epel element https://review.openstack.org/180548 | 13:50 |
| *** athomas_ has joined #tripleo | 13:51 | |
| *** athomas has quit IRC | 13:52 | |
| *** athomas_ has quit IRC | 13:53 | |
| *** athomas has joined #tripleo | 13:54 | |
| *** jehb has quit IRC | 13:54 | |
| openstackgerrit | Giulio Fidente proposed stackforge/puppet-tripleo: Add support for service_manage https://review.openstack.org/180484 | 13:57 |
| *** mcornea has quit IRC | 13:57 | |
| *** lblanchard has joined #tripleo | 14:03 | |
| *** mcornea has joined #tripleo | 14:09 | |
| *** shakamunyi has joined #tripleo | 14:09 | |
| mmagr | dprince, I found the cause why Keystone domain for Heat patch does not work | 14:10 |
| *** lifeless has joined #tripleo | 14:10 | |
| openstackgerrit | Merged openstack/tripleo-image-elements: Fix ubuntu grub-pc install failure https://review.openstack.org/168208 | 14:11 |
| mmagr | dprince, python-keystoneclient has a bug ... domains are not filtered correctly via 'id' hence keystone-init then tries sometimes (depends on how domains are sorted) to search for user with name admin and domain heat_stack | 14:12 |
| mmagr | dprince, which obviously does not exist | 14:12 |
| mmagr | >>> client.domains.list(id='default') | 14:13 |
| mmagr | [<Domain description=Contains users and projects created by heat, enabled=True, id=29f4f3f567f943eb9769329352753b89, links={u'self': u'http://192.0.2.5:35357/v3/domains/29f4f3f567f943eb9769329352753b89'}, name=heat_stack>, <Domain description=Owns users and tenants (i.e. projects) available on Identity API v2., enabled=True, id=default, links={u'self': u'http://192.0.2.5:35357/v3/domains/default'}, name=Default>] | 14:13 |
| mmagr | >>> | 14:13 |
| mmagr | dprince, ^ | 14:13 |
| *** shakamunyi has quit IRC | 14:13 | |
| mmagr | dprince, we might change os_cloud_config/keystone.py to filter by name instead | 14:14 |
| openstackgerrit | Merged openstack/tripleo-incubator: Add logic to select L3 agents configuration https://review.openstack.org/174308 | 14:14 |
| dprince | mmagr: sure, a change to os-cloud-config would be reasonable I think | 14:16 |
| dprince | mmagr: might be worth filing a keystone bug on this too though | 14:16 |
| *** ishant has quit IRC | 14:16 | |
| *** lazy_prince has quit IRC | 14:16 | |
| openstackgerrit | Giulio Fidente proposed openstack/tripleo-heat-templates: Ensure services are not enabled by puppet-tripleo when EnablePacemaker https://review.openstack.org/180490 | 14:16 |
| mmagr | dprince, yes I definitely will | 14:17 |
| openstackgerrit | Pino Toscano proposed openstack/diskimage-builder: package-installs: fix error case for Python 2.6 https://review.openstack.org/180561 | 14:19 |
| *** jprovazn has joined #tripleo | 14:23 | |
| *** gfidente has quit IRC | 14:23 | |
| openstackgerrit | Merged openstack/tuskar-ui: Make use of parameter_type and constraints https://review.openstack.org/180211 | 14:23 |
| *** gfidente has joined #tripleo | 14:23 | |
| openstackgerrit | Merged openstack/os-collect-config: Use non-versioned auth_url for keystone https://review.openstack.org/138246 | 14:27 |
| openstackgerrit | Martin Mágr proposed openstack/os-cloud-config: Filter domain by name https://review.openstack.org/180563 | 14:30 |
| *** shakamunyi has joined #tripleo | 14:35 | |
| *** Marga_ has joined #tripleo | 14:36 | |
| *** shakamunyi has quit IRC | 14:36 | |
| *** shakamunyi has joined #tripleo | 14:36 | |
| openstackgerrit | Martin Mágr proposed openstack/tripleo-heat-templates: [resubmitted] Keystone domain for Heat https://review.openstack.org/180566 | 14:39 |
| *** lazy_prince has joined #tripleo | 14:41 | |
| openstackgerrit | Giulio Fidente proposed openstack/tripleo-heat-templates: Use $service_name from module when adding resource to pcmk https://review.openstack.org/180570 | 14:50 |
| *** untriaged-bot has joined #tripleo | 15:00 | |
| untriaged-bot | Untriaged bugs so far: | 15:00 |
| untriaged-bot | https://bugs.launchpad.net/diskimage-builder/+bug/1451634 | 15:00 |
| openstack | Launchpad bug 1451634 in diskimage-builder "ISO image of deploy ramdisk doesn't have provision to log things to serial console" [Undecided,In progress] - Assigned to Ramakrishnan G (rameshg87) (rameshg87) | 15:00 |
| untriaged-bot | https://bugs.launchpad.net/diskimage-builder/+bug/1449852 | 15:00 |
| openstack | Launchpad bug 1449852 in diskimage-builder "Buidling ramdisk with ironic-agent behind proxy fails" [Undecided,In progress] - Assigned to Ramakrishnan G (rameshg87) (rameshg87) | 15:00 |
| untriaged-bot | https://bugs.launchpad.net/diskimage-builder/+bug/1449882 | 15:00 |
| openstack | Launchpad bug 1449882 in diskimage-builder "Deploying with ironic-agent ramdisk built from fedora 21 cloud image fails" [Undecided,In progress] - Assigned to Ramakrishnan G (rameshg87) (rameshg87) | 15:00 |
| untriaged-bot | https://bugs.launchpad.net/diskimage-builder/+bug/1449854 | 15:00 |
| openstack | Launchpad bug 1449854 in diskimage-builder "Ironic agent ramdisk built using disk-image-create fails with iscsi_ilo driver" [Undecided,In progress] - Assigned to Ramakrishnan G (rameshg87) (rameshg87) | 15:00 |
| *** untriaged-bot has quit IRC | 15:00 | |
| jistr | gfidente: do you have a link for the bug re db syncs? | 15:09 |
| gfidente | https://bugs.launchpad.net/puppet-glance/+bug/1452278 | 15:10 |
| openstack | Launchpad bug 1452278 in puppet-glance "db_sync should be executed based on sync_db but not enabled nor manage_service" [Medium,In progress] - Assigned to Giulio Fidente (gfidente) | 15:10 |
| gfidente | I checked neutron and that seems to be doing fine | 15:10 |
| gfidente | keystone wasn't | 15:11 |
| *** jeh has joined #tripleo | 15:12 | |
| gfidente | cinder isn't either | 15:12 |
| gfidente | will add to bug | 15:12 |
| gfidente | nova instead is doing fine | 15:12 |
| jistr | gfidente: i wonder if we can get the fixes merged in the upstream puppet modules in a reasonable time frame, or if we should do the same hack as Astapor | 15:13 |
| jistr | gfidente: enable service on one step, disable it in the next | 15:13 |
| gfidente | well we have neutron and nova doing well already | 15:13 |
| gfidente | we can push a little for the others to fix their behaviour and make them consistent | 15:14 |
| gfidente | at least, that I was trying to do | 15:14 |
| jistr | gfidente: ack, i'm tempted to join the effort here, sounds ok? | 15:14 |
| *** sdake has joined #tripleo | 15:14 | |
| gfidente | jistr, oh sure | 15:14 |
| gfidente | I am sure about cinder needing a change | 15:14 |
| jistr | ack i'll take that one | 15:14 |
| gfidente | ack | 15:15 |
| openstackgerrit | Martin Mágr proposed openstack/os-cloud-config: Filter domain by name https://review.openstack.org/180563 | 15:15 |
| *** sdake__ has quit IRC | 15:17 | |
| *** rlandy has quit IRC | 15:19 | |
| *** MasterPiece has joined #tripleo | 15:24 | |
| *** lazy_prince has quit IRC | 15:26 | |
| *** lazy_prince has joined #tripleo | 15:27 | |
| *** aufi is now known as aufi_afk | 15:32 | |
| *** mmagr has quit IRC | 15:34 | |
| *** rhallisey has quit IRC | 15:36 | |
| jayg | gfidente jistr: this is the puppet-pacemaker PR, if you have not seen it - https://github.com/redhat-openstack/puppet-pacemaker/pull/45 | 15:36 |
| *** openstackgerrit has quit IRC | 15:37 | |
| *** openstackgerrit has joined #tripleo | 15:37 | |
| *** cody-somerville has joined #tripleo | 15:39 | |
| *** ukalifon1 has quit IRC | 15:40 | |
| *** rhallisey has joined #tripleo | 15:40 | |
| *** lazy_prince has quit IRC | 15:44 | |
| *** yamahata has joined #tripleo | 15:45 | |
| jistr | jayg, gfidente: i'm kinda on the fence re making a breaking change right now. We don't have that many pacemaker resources in the controller manifest as of now https://github.com/openstack/tripleo-heat-templates/blob/master/puppet/manifests/overcloud_controller.pp | 15:50 |
| jistr | but there are pending patches https://review.openstack.org/#/q/status:open+project:openstack/tripleo-heat-templates+message:pacemaker,n,z | 15:50 |
| jayg | the thing is, I believe puppet-pacemaker as is is insufficient, which is why we had a bunch of wrappers in quickstack | 15:52 |
| *** alop has joined #tripleo | 15:57 | |
| *** cwolferh has joined #tripleo | 15:58 | |
| *** shakamunyi has quit IRC | 15:58 | |
| *** ifarkas has quit IRC | 16:05 | |
| *** eghobo has joined #tripleo | 16:05 | |
| jistr | gfidente: looking at why $enabled is needed for db_sync to happen, i think it's just historical reasons. Initially db_sync happened always, then it was moved into the $enabled block to have a way of disabling it, but it was coupled with disabling the service. The $sync_db param was introduced fairly recently, but the change didn't decouple it from $enabled. | 16:06 |
| jistr | at least that's the case for puppet-cinder ^ | 16:06 |
| gfidente | given sync_db is bottom list of params for keystone, it could be about the same for keystone | 16:08 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/os-cloud-config: Updated from global requirements https://review.openstack.org/178418 | 16:08 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/os-collect-config: Updated from global requirements https://review.openstack.org/178419 | 16:08 |
| openstackgerrit | Merged openstack/tripleo-incubator: devtest_seed.sh add sbin in $PATH https://review.openstack.org/177192 | 16:16 |
| *** gfidente has quit IRC | 16:19 | |
| jayg | jistr: btw, if you did not already, I would encourage you to voice any concerns on that PR so they are noted and we can work through them | 16:23 |
| jayg | (the puppet-pacemaker one) | 16:23 |
| jistr | jayg: i'm off for today, i'll try to review that in a bit more detail tomorrow, but tbh that level of detail is a bit beyond my current knowledge of pacemaker. | 16:26 |
| jistr | if that change is what it takes to get to a better HA architecture, then we'll need to go with it at some point | 16:27 |
| *** jcoufal_ has quit IRC | 16:27 | |
| jistr | btw i got a $sync_db fix for cinder https://review.openstack.org/#/c/180610/1 | 16:28 |
| * jistr off, ttyl | 16:29 | |
| *** jistr has quit IRC | 16:29 | |
| dprince | shardy: Just tried to do a 'type: {get_param: InternalAPIType}' and was bummed it won't work: Resource InternalApiPort type type must be string | 16:31 |
| dprince | shardy: I suppose I've gone too far in trying to make meta templates here | 16:32 |
| shardy | dprince: yeah, you can't dynamically assign types, only map via the resource_registry | 16:33 |
| dprince | shardy: Yeah. I vaguely remember trying this once before. Easy to forget... | 16:34 |
| *** Goneri has quit IRC | 16:34 | |
| shardy | dprince: Yeah, it's probably not impossible, but making that work with the current heat implementation would probably be pretty hard | 16:34 |
| shardy | we parse the stack, including instantiating all the object types, before evaluating parameter references atm | 16:35 |
| *** regebro has quit IRC | 16:41 | |
| *** shakamunyi has joined #tripleo | 16:48 | |
| *** lucasagomes is now known as lucas-dinner | 16:54 | |
| *** sdake_ has joined #tripleo | 16:56 | |
| *** sdake has quit IRC | 17:00 | |
| *** derekh_ has quit IRC | 17:01 | |
| dprince | shardy: So I'm torn. Should I create one nested stack like this per role: https://review.openstack.org/#/c/177845/6/network/ports/ports.yaml,cm | 17:02 |
| dprince | shardy: or.... inline those ports in all the role templates themselves | 17:02 |
| *** Lotus907efi has joined #tripleo | 17:03 | |
| shardy | dprince: I think if the nested stack is tied to the role implementation, do inline type: foo.yaml, otherwise keep it all in the resource_registry | 17:04 |
| dprince | shardy: either of the approaches would use the resource registry | 17:05 |
| dprince | shardy: because I need to enable/disable the ports that way | 17:05 |
| dprince | shardy: all I'm really asking is an organizational question | 17:05 |
| *** shakamunyi has quit IRC | 17:05 | |
| dprince | shardy: is there value in having all the port stuff containerized in a nested stack which we can use to provide clean outputs. Or just inline it in the role templates. | 17:06 |
| dprince | shardy: ports.yaml could become: compute_ports.yaml, controller_ports.yaml.... | 17:06 |
| shardy | dprince: ah, I see - I'd say it may be clearer if we just put it all in the role template | 17:06 |
| shardy | unless that becomes really unweildy | 17:06 |
| dprince | shardy: its just more patches, :) | 17:07 |
| shardy | I don't have a strong opinion, but we're already getting to a pretty massive tree of nested stacks.. | 17:07 |
| * dprince gets rebase taxed | 17:07 | |
| dprince | shardy: yeah, well. What can I say. We need them | 17:07 |
| dprince | shardy: in this case they are optional. So I'll go inline | 17:08 |
| dprince | shardy: thanks, I was on the fence on this one | 17:08 |
| shardy | dprince: Agreed, I'm just leaning towards the inline approach, but like I said not a strong opinion :) | 17:08 |
| shardy | anywhere we can significantly reduce duplication the nested stack approach makes sense | 17:08 |
| shardy | but if it's just boiler-plate resource definitions vs actual logic we have to maintain, I'm not sure it's all that worth it | 17:09 |
| *** sdake has joined #tripleo | 17:10 | |
| *** athomas has quit IRC | 17:10 | |
| *** Marga_ has quit IRC | 17:11 | |
| dprince | shardy: Exactly. My initial understanding of the problem was just that. I was like ports.yaml... a nice way to decrease duplication. But then dsneddon was like: "what if I don't want a tenant port on my storage node". So yeah, just a bit of refactoring here to support this more flexable network architecture. | 17:11 |
| *** yamahata has quit IRC | 17:12 | |
| *** sdake_ has quit IRC | 17:14 | |
| *** sdake__ has joined #tripleo | 17:14 | |
| *** sdake has quit IRC | 17:18 | |
| dprince | shardy: I guess the one downside is testability. I (previously) was testing all my networks and ports by using these nested stacks directly. | 17:19 |
| dprince | shardy: if I inline it... I'll just need to fire up overcloud-without-mergepy directly I guess | 17:19 |
| *** shakamunyi has joined #tripleo | 17:19 | |
| dprince | shardy: probably not a bit deal once it is all in place I guess | 17:19 |
| shardy | dprince: Yeah, I guess we'll need to figure out a way to easily test the whole thing anyway | 17:20 |
| dprince | shardy: Yeah. Well the devtest stuff for that is mostly done (posted) as well | 17:20 |
| dprince | shardy: I didn't have that initially though | 17:20 |
| *** Slower has joined #tripleo | 17:20 | |
| shardy | dprince: cool, I think I've looked at most of that, but I'll try to pull and test it tomorrow | 17:21 |
| *** shardy has quit IRC | 17:22 | |
| *** mcornea has quit IRC | 17:29 | |
| *** pelix has quit IRC | 17:38 | |
| *** akrivoka has quit IRC | 17:42 | |
| *** Marga_ has joined #tripleo | 17:46 | |
| openstackgerrit | Merged openstack/diskimage-builder: Turn docs warnings into errors and fix issues https://review.openstack.org/177801 | 17:48 |
| openstackgerrit | greghaynes proposed openstack/diskimage-builder: Support building ACIs https://review.openstack.org/155484 | 17:51 |
| *** akrivoka has joined #tripleo | 17:54 | |
| *** cody-somerville has quit IRC | 17:57 | |
| *** saurabhs has joined #tripleo | 18:03 | |
| *** Marga_ has quit IRC | 18:10 | |
| *** Marga_ has joined #tripleo | 18:11 | |
| *** sdake has joined #tripleo | 18:20 | |
| *** jeh has quit IRC | 18:22 | |
| openstackgerrit | greghaynes proposed openstack/diskimage-builder: Test elements have README.rst file https://review.openstack.org/177832 | 18:23 |
| *** sdake_ has joined #tripleo | 18:23 | |
| *** sdake__ has quit IRC | 18:24 | |
| alop | darn that devstack-trusty again... | 18:25 |
| greghaynes | alop: the infra change for that merged not long ago | 18:25 |
| alop | oh good | 18:25 |
| greghaynes | and it takes up to 45mins to get deployed | 18:25 |
| alop | I was just about to ask where I would start looking | 18:25 |
| greghaynes | so, new jobs should start testing with the "fix" | 18:25 |
| alop | what was the fix? 'apt-get update' ? | 18:26 |
| greghaynes | The big change though is we now control dep installation, so if theres more issues we can fix them quickly rather than try and chase after infra to merge a fix | 18:26 |
| greghaynes | yea | 18:26 |
| *** sdake has quit IRC | 18:26 | |
| alop | the infra/ci side is truly amazing | 18:26 |
| greghaynes | yea, I am going to add some tests for the other distros and im super excited that were super close to having tests for the whole combinatoric build all the distros on all the distros | 18:28 |
| *** yamahata has joined #tripleo | 18:33 | |
| *** lblanchard has quit IRC | 18:33 | |
| *** lblanchard has joined #tripleo | 18:34 | |
| *** Marga_ has quit IRC | 18:35 | |
| *** Marga_ has joined #tripleo | 18:36 | |
| *** cwolferh has quit IRC | 18:44 | |
| *** openstackgerrit has quit IRC | 18:47 | |
| *** tzumainn_ has quit IRC | 18:47 | |
| *** lsmola has quit IRC | 18:47 | |
| *** jtomasek has quit IRC | 18:47 | |
| *** yamahata has quit IRC | 18:48 | |
| *** zaneb has quit IRC | 18:48 | |
| *** morganfainberg has quit IRC | 18:48 | |
| *** rbrady has quit IRC | 18:48 | |
| *** vipul has quit IRC | 18:48 | |
| *** devananda has quit IRC | 18:48 | |
| *** dobson has quit IRC | 18:48 | |
| *** yamahata has joined #tripleo | 18:49 | |
| *** dobson has joined #tripleo | 18:49 | |
| *** zaneb has joined #tripleo | 18:49 | |
| *** devananda has joined #tripleo | 18:49 | |
| *** vipul has joined #tripleo | 18:49 | |
| *** rbrady has joined #tripleo | 18:49 | |
| *** morganfainberg has joined #tripleo | 18:49 | |
| *** openstackgerrit has joined #tripleo | 18:49 | |
| *** tzumainn_ has joined #tripleo | 18:49 | |
| *** lsmola has joined #tripleo | 18:49 | |
| *** jtomasek has joined #tripleo | 18:49 | |
| *** akrivoka has quit IRC | 18:50 | |
| openstackgerrit | Merged openstack/diskimage-builder: debootstrap: fix syntax issues https://review.openstack.org/180066 | 18:50 |
| *** cwolferh has joined #tripleo | 18:57 | |
| *** MasterPiece has quit IRC | 19:00 | |
| jdob | dprince: i think I found the first of the image patches you were talking about and kicked CI to rerun. the next one up has a merge conflict: https://review.openstack.org/#/c/173645/ | 19:06 |
| dprince | jdob: yep. got it. I can rebase and push a new series there too | 19:07 |
| dprince | jdob: just digging out of a hole today | 19:07 |
| jdob | and the one after it | 19:07 |
| jdob | no problem, just following up since I said I'd see about getting that series landed | 19:07 |
| dprince | jdob: yeah, build-images needs to land first | 19:08 |
| dprince | jdob: shouldn't be conflicts for that one as its new though | 19:08 |
| jdob | hopefully CI will pass and we can +A it | 19:08 |
| jdob | no conflicts, just a failing run | 19:08 |
| dprince | good ol' Fedora mirrors today it looks like | 19:08 |
| jdob | \o/ | 19:09 |
| *** yamahata has quit IRC | 19:18 | |
| *** zaneb has quit IRC | 19:18 | |
| *** morganfainberg has quit IRC | 19:18 | |
| *** rbrady has quit IRC | 19:18 | |
| *** vipul has quit IRC | 19:18 | |
| *** devananda has quit IRC | 19:18 | |
| *** yamahata has joined #tripleo | 19:19 | |
| *** zaneb has joined #tripleo | 19:19 | |
| *** devananda has joined #tripleo | 19:19 | |
| *** vipul has joined #tripleo | 19:19 | |
| *** rbrady has joined #tripleo | 19:19 | |
| *** morganfainberg has joined #tripleo | 19:19 | |
| *** zaneb has quit IRC | 19:19 | |
| *** zbitter has joined #tripleo | 19:19 | |
| *** morganfainberg has quit IRC | 19:20 | |
| *** morganfainberg has joined #tripleo | 19:21 | |
| *** zbitter is now known as zaneb | 19:21 | |
| openstackgerrit | Jay Dobies proposed openstack/tripleo-heat-templates: Add Keystone as a Pacemaker resource when EnablePacemaker https://review.openstack.org/178694 | 19:22 |
| *** Marga_ has quit IRC | 19:24 | |
| *** cody-somerville has joined #tripleo | 19:31 | |
| *** sdake has joined #tripleo | 19:36 | |
| *** sdake_ has quit IRC | 19:39 | |
| *** lucas-dinner has quit IRC | 19:42 | |
| *** sdake_ has joined #tripleo | 19:44 | |
| *** sdake has quit IRC | 19:47 | |
| openstackgerrit | Doug Hellmann proposed openstack/python-tuskarclient: Drop use of 'oslo' namespace package https://review.openstack.org/180710 | 19:48 |
| openstackgerrit | Doug Hellmann proposed openstack/tuskar: Drop use of 'oslo' namespace package https://review.openstack.org/180721 | 20:09 |
| *** jcoufal has joined #tripleo | 20:11 | |
| *** jcoufal has quit IRC | 20:11 | |
| *** jayg is now known as jayg|g0n3 | 20:14 | |
| *** yamahata has quit IRC | 20:14 | |
| *** aufi_afk has quit IRC | 20:23 | |
| *** Marga_ has joined #tripleo | 20:25 | |
| *** jprovazn has quit IRC | 20:26 | |
| *** aufi_afk has joined #tripleo | 20:30 | |
| *** Marga_ has quit IRC | 20:30 | |
| *** aufi_afk has quit IRC | 20:31 | |
| *** radez is now known as radez_g0n3 | 20:35 | |
| *** Marga_ has joined #tripleo | 20:46 | |
| *** Marga_ has quit IRC | 20:52 | |
| *** Marga_ has joined #tripleo | 20:57 | |
| *** MasterPiece has joined #tripleo | 20:58 | |
| *** dsneddon has quit IRC | 20:59 | |
| *** untriaged-bot has joined #tripleo | 21:00 | |
| untriaged-bot | Untriaged bugs so far: | 21:00 |
| untriaged-bot | https://bugs.launchpad.net/diskimage-builder/+bug/1452400 | 21:00 |
| untriaged-bot | https://bugs.launchpad.net/diskimage-builder/+bug/1449852 | 21:00 |
| untriaged-bot | https://bugs.launchpad.net/diskimage-builder/+bug/1449854 | 21:00 |
| openstack | Launchpad bug 1452400 in diskimage-builder "debian-minimal builds still busted, now on locale issues" [Undecided,New] | 21:00 |
| *** untriaged-bot has quit IRC | 21:00 | |
| openstack | Launchpad bug 1449852 in diskimage-builder "Buidling ramdisk with ironic-agent behind proxy fails" [Undecided,In progress] - Assigned to Ramakrishnan G (rameshg87) (rameshg87) | 21:00 |
| openstack | Launchpad bug 1449854 in diskimage-builder "Ironic agent ramdisk built using disk-image-create fails with iscsi_ilo driver" [Undecided,In progress] - Assigned to Ramakrishnan G (rameshg87) (rameshg87) | 21:00 |
| *** trown is now known as trown|outttypeww | 21:04 | |
| *** dsneddon has joined #tripleo | 21:04 | |
| *** MasterPiece has quit IRC | 21:06 | |
| greghaynes | alop: hrm, I think infra is having issues right now deploying jenkins changes so I think the fix still hasnt gone out :( | 21:10 |
| alop | sad trombone | 21:10 |
| *** cody-somerville has quit IRC | 21:12 | |
| *** rhallisey has quit IRC | 21:12 | |
| *** lblanchard has quit IRC | 21:14 | |
| *** sdake has joined #tripleo | 21:14 | |
| *** sdake_ has quit IRC | 21:18 | |
| *** david-lyle has quit IRC | 21:29 | |
| *** jtomasek has quit IRC | 21:49 | |
| openstackgerrit | Merged openstack/tripleo-image-elements: Make delorean.repo overrideable https://review.openstack.org/176824 | 21:52 |
| *** sdake has quit IRC | 21:53 | |
| *** sdake has joined #tripleo | 21:53 | |
| *** thrash is now known as thrash|g0ne | 21:56 | |
| *** eghobo_ has joined #tripleo | 21:59 | |
| *** eghobo has quit IRC | 22:00 | |
| *** sdake has quit IRC | 22:17 | |
| *** sdake has joined #tripleo | 22:17 | |
| *** sdake has quit IRC | 22:17 | |
| *** sdake has joined #tripleo | 22:17 | |
| *** dprince has quit IRC | 22:20 | |
| dsneddon | devtest doesn't seem to work on Fedora 20 (at least for me). What is the best Fedora-based distro/version to use? | 22:31 |
| *** morazi has quit IRC | 22:58 | |
| *** chlong has joined #tripleo | 23:16 | |
| *** panda has quit IRC | 23:17 | |
| *** panda has joined #tripleo | 23:18 | |
| Lotus907efi | is this the right chat room to ask if a limitation I discovered today with diskimage-builder is a bug or not? | 23:22 |
| *** weshay has quit IRC | 23:22 | |
| Lotus907efi | or is there an IRC channel for diskimage-builder? | 23:22 |
| *** alop has quit IRC | 23:39 | |
| *** Marga_ has quit IRC | 23:45 | |
| *** david-lyle has joined #tripleo | 23:47 | |
| greghaynes | This is the right channel | 23:50 |
| SpamapS | Lotus907efi: how can we help? | 23:53 |
| Lotus907efi | ok, so I discovered if I write a script to be executed from the post-install.d directory from an element I am creating and the script is called something like 43-rc.local then it will not be run at all | 23:53 |
| Lotus907efi | if I move the script to 43-rclocal it works fine | 23:53 |
| SpamapS | Lotus907efi: ding ding, sounds like a bug in dib-run-parts | 23:54 |
| Lotus907efi | hmmm | 23:54 |
| SpamapS | targets=$(find $target_dir -maxdepth 1 -xtype f -executable -printf '%f\n' | grep -E "$allowed_regex" | LANG=C sort -n || echo "") | 23:55 |
| SpamapS | Lotus907efi: allowed_regex=${RUN_PARTS_REGEX:-"^[0-9A-Za-z_-]+$"} | 23:55 |
| Lotus907efi | so something with a dot in the script name might not be allowed_regex ? | 23:55 |
| Lotus907efi | this held me up in doing something today for quite a while | 23:56 |
| Lotus907efi | trying to figure out why the script was not running | 23:56 |
| SpamapS | its the allowed_regex | 23:56 |
| SpamapS | . not allowed | 23:56 |
| Lotus907efi | yeah | 23:57 |
| Lotus907efi | I see that | 23:57 |
| SpamapS | Lotus907efi: we should fail loudly in this case | 23:57 |
| SpamapS | Or maybe not | 23:57 |
| SpamapS | I think it's used to store stuff in the .d dirs | 23:57 |
| Lotus907efi | the .d dirs??? | 23:57 |
| Lotus907efi | what .d dirs? | 23:57 |
| SpamapS | http://paste.ubuntu.com/11001428/ | 23:58 |
| Lotus907efi | oh | 23:59 |
| Lotus907efi | ok | 23:59 |
| SpamapS | Not sure it's "right" | 23:59 |
| SpamapS | but agree it is _confusing_ | 23:59 |
| Lotus907efi | so those scripts for dhcp-all-interfaces work ok, right? | 23:59 |
| SpamapS | have never liked that stuff gets left in the .d dirs | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!