Perl 6 - the future is here, just unevenly distributed

IRC log for #fuel-dev, 2016-03-23

| Channels | #fuel-dev index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:03 dims_ joined #fuel-dev
02:03 xarses joined #fuel-dev
02:09 lezbar joined #fuel-dev
02:48 ilbot3 joined #fuel-dev
02:48 Topic for #fuel-dev is now Fuel Development http://wiki.openstack.org/wiki/Fuel | Paste here http://paste.openstack.org/ | IRC logs http://irclog.perlgeek.de/fuel-dev/ | gerrit traffic @ #fuel-tracker
02:52 askb joined #fuel-dev
04:40 askb Hi I am trying setup of fuel on devstack, looks like the installation is stuck with a message for quite some time: http://pastebin.com/Wg6X72qS
04:40 askb Is there something, I could be missing here ?
04:43 askb I am using the deploy scripts from "fuel-virtualbox"
04:57 askb ok - looks like this not an issue, rather a message which is suggesting to the user if you have other nodes which have been booted and they could possibly be timed out while waiting for response, then reboot the nodes.
05:28 zerda joined #fuel-dev
05:40 fuel-ci joined #fuel-dev
05:44 dshulyak joined #fuel-dev
06:29 nadya joined #fuel-dev
06:58 e0ne joined #fuel-dev
07:02 gkibardin joined #fuel-dev
07:07 kozhukalov askb: yes, this is ok, once we start master node first time we build so called bootstrap images which is ramdisk that is used to boot slave nodes via PXE and to discover them
07:08 askb kozhukalov, hi
07:09 askb Is there a way to speed up process with building bootstrap image, like using the local image ... looks like this take about 1+ hours for me.
07:11 kozhukalov ok, it takes an hour or so, that probably means you have slow internet connection. if so, you could create a local Ubuntu mirror
07:11 kozhukalov but creating full mirror is going to take a lot more (60G)
07:12 kozhukalov but it is also possible to create partial mirror (that includes only required packages)
07:13 kozhukalov that could be done using fuel-mirror
07:15 askb oh ok - yes that makes sense
07:15 askb It would be really cool, if there was a way to cache the previous built bootstrap image (as an optional flag), just to speed up the process.
07:16 kozhukalov as far as I remember there was a way to skip building bootstrap image
07:17 kozhukalov then on the web interface there will be a notification that bootstrap image is not available
07:17 kozhukalov and slave nodes will not boot
07:18 kozhukalov then you could put an image manually and reboot slave nodes
07:19 kozhukalov btw, when you will try to install openstack cluster fuel will first try to build target OS image
07:20 kozhukalov it is also possible to put old image manually, but that is not regular user experience
07:21 askb aah, it sounds interesting option to try out, but has its set of cons :)
07:23 kozhukalov you mean skipping building bootstrap image? i don't think it has any cons, you just need to build it later or put your own image manually
07:23 kozhukalov for most people it is unnecessary because internet connections nowadays are fast
07:24 kozhukalov askb: yesterday I had a short conversation with RDO guys about deployment their packages using fuel
07:24 kozhukalov askb: currently they use tripleo mostly
07:25 kozhukalov askb: they are interested in making this happened, but they are a bit under resources
07:26 kozhukalov askb: i said them that the only major thing that needs to be implemented is the support of rpm images in bareon/fuel-agent
07:26 kozhukalov askb: and probably some minor things
07:27 kozhukalov askb: but there are no people who have  time to do this
07:27 kozhukalov askb: both from rdo side and from fuel side
07:28 askb ah true, we are short of good hands everywhere :)
07:29 askb kozhukalov, what does fuel use for provisioning the nodes (puppet) once the boot strap image is built ?
07:29 kozhukalov askb: as far as i know you are interested in contributing in fuel and fuel-agent and i wonder if it could be interesting for you to implement this feature (rdo deployment) with support from both sides fuel and rdo?
07:30 kozhukalov askb: bootstrap image has two important components installed into it. those are nailgun-agent (discovery agent) and fuel-agent (provisioning agent)
07:31 askb kozhukalov, sure, I could take a look, do you have bugID for that feature already opened ?
07:31 kozhukalov askb: once nodes are booted into bootstrap and you assigned roles to them and then click 'deploy' button,
07:32 salmon_ joined #fuel-dev
07:33 kozhukalov fuel builds OS target image on the master node and then tells to slaves nodes to download this OS image and copy it on local disks on slave nodes (that is how we install OS)
07:33 askb yup, i cleaned up the nodes and restarted the process before I could deploy them, as the controller node has only about 1.5Gib memory ... now have increased it to 4Gib and restarted the process.
07:33 kozhukalov that is just a base OS (there are no openstack components installed)
07:34 kozhukalov then nodes are rebooted into target OS and we launch deployment task (run puppet manifests on slave nodes) that install all necessary openstack packages and configures them
07:34 askb so the openstack components only get installed on the node when you deploy
07:35 kozhukalov yes
07:35 kozhukalov they are not installed into OS images
07:36 kozhukalov but that is, in fact, configurable
07:39 kozhukalov you can chose a list of packages that are to be installed into image
07:40 kozhukalov but we can not run puppet or any other configuration tool while building images
07:41 kozhukalov askb: there is no BP yet for rdo stuff. if you could take care of this, it would be great
08:04 pbelamge joined #fuel-dev
08:14 igajsin joined #fuel-dev
08:15 igajsin left #fuel-dev
08:17 askb kozhukalov, this is something I could look at once I am comfy with fuel env
08:23 kozhukalov askb: please don't hesitate to ask any questions, and btw, if you have questions about fuel-agent/bareon you can ask in #openstack-bareon
08:23 askb kozhukalov, sure thing thanks :)
08:33 xek joined #fuel-dev
08:35 ekosareva joined #fuel-dev
08:37 DaveJ__ joined #fuel-dev
08:46 HeOS joined #fuel-dev
09:03 e0ne joined #fuel-dev
09:03 askb zigo, Hi, I am looking into the bug you recently filed, and was hoping to get some input from your side on reproducing the issue or actual error message when the issue occured
09:03 askb https://bugs.launchpad.net/fuel/+bug/1552346
09:33 zefciu joined #fuel-dev
09:42 ekosareva joined #fuel-dev
09:43 bogdando please review https://review.openstack.org/#/c/294979/ and confirm if swift-proxies shall be deployed after storage nodes
09:55 zigo askb: You wont be able to reproduce it unless you attempt to install the Debian port of the Fuel master node.
09:55 zigo askb: I can guide you through it though.
09:56 zigo askb: That's maybe something we should fix for Newton, as we're in feature freeze.
10:00 dims joined #fuel-dev
10:12 gkibardin joined #fuel-dev
10:19 rvyalov_ joined #fuel-dev
10:21 nurla joined #fuel-dev
10:24 bookwar joined #fuel-dev
10:29 nadya joined #fuel-dev
10:38 msemenov joined #fuel-dev
10:45 e0ne joined #fuel-dev
10:45 dshulyak joined #fuel-dev
10:55 agordeev zigo, could you attach any debian friendly grub.conf which was generated by fuel-agent to the bug? that will help a lot.
11:05 dims_ joined #fuel-dev
11:51 askb joined #fuel-dev
11:56 askb joined #fuel-dev
11:56 ekosareva joined #fuel-dev
11:57 evgenyl ikalnitsky: bookwar don't know if you've seen it, but jfyi http://lists.openstack.org/pipermail/openstack-dev/2016-March/090052.html
11:59 askb zigo, ping, thanks for clarifying, so in this case you installed the master node on debian port right ?
12:15 jaypipes joined #fuel-dev
12:26 bookwar evgenyl: yes, ack, thank you for pinging
12:41 ikalnitsky evgenyl: thanks for pinging. however, i think fuel-web won't be affected since we create database in conftest.py for each worker (we run tests using 4 workers). :)
13:01 evgenyl ikalnitsky: it's not only about db creation, the packages (postgresql and mysql) will not be installed by default, when the change to infra is merged.
13:07 ikalnitsky evgenyl: why do you think so? according to the ML, i got understanding that they won't create openstackci_test database by default and that's it.
13:09 evgenyl ikalnitsky: "Removing this by default, allows to only install those packages that are really needed by projects" which means no pre-installed os packages, so each project will have to define explicit dependencies, or use -db gates to get database installed and configured.
13:09 evgenyl ikalnitsky: but I may be wrong.
13:09 evgenyl ikalnitsky: that is what I got from today's discussion with infra folks.
13:09 ikalnitsky evgenyl: it starts with 'The OpenStack Infra team is removing soon access to the openstack_citest MySQL/PostgreSQL database for unittests by default.'
13:10 ikalnitsky so probably i got wrong understanding
13:10 ikalnitsky evgenyl: ok then
13:10 ikalnitsky bookwar: Do I need to prepare a patch to openstack infra for fuel-web? or fuel infra team will do it?
13:10 mgariepy joined #fuel-dev
13:11 bookwar igorbelikov: will you take care about it? ^
13:12 igorbelikov bookwar: ok, I’ll prepare a patch to switch fuel-web to -db gates
13:23 askb actually one of the nodes, i was unable to unable to boot up ubuntu_bootstrap, however the default bootstrat worked which I selected the options manually, but this booted the node to centos7
13:24 askb kozhukalov, ^^
13:24 askb I am wondering if this ok to proceed to best to clean up and restart the process
13:27 kozhukalov sylwesterB: the thing is that the doc directory in fuel-web contains the documentation not only about nailgun, so we moved everything to fuel-docs with the intention to move pages related to nailgun back to fuel-web, pages related to, let say, fuel-menu to fuel-menu repo. and component teams are resposible for doing this.
13:27 nurla joined #fuel-dev
13:27 kozhukalov the whole directory was moved by infra team
13:28 kozhukalov bookwar: ^ could you please give more details if necessary on docs move to fuel-docs?
13:30 kozhukalov askb: yes, in 8.0 we have two bootstrap options 1) centos 7 which is built together with ISO and installed on the master node as rpm package 2) ubuntu 14.04 bootstrap image which is to be built in runtime on the master node immediately after fuel is deployed
13:30 sylwesterB kozhukalov: so there is a plan for 'split' and move Nailgun related docs back to fuel-web - that sounds good. When it will be done (iteration/release)? Is there a bp for this activity
13:31 kozhukalov askb: so if something went wrong with ubuntu bootstrap you can use centos7 based one, if it works for you
13:32 kozhukalov askb: the same set of discovery and provisioning components are installed inside centos7 based bootstrap, so the functionality is the same
13:32 askb sure, thatz cool, now that the nodes are up, let me try to deploy the node before wiping it clean again.
13:32 kozhukalov askb: however, we got rid of pre-built centos 7 bootstrap in 9.0
13:35 kozhukalov askb: if we have volunteer to implement rpm based images in fuel-agent/bareon in 10.0 then rpm based bootstrap will be an option again
13:35 lezbar joined #fuel-dev
13:36 kozhukalov sylwesterB: I am not sure about BP for this, but as I said component teams are to be responsible for actualizing their docs
13:37 kozhukalov sylwesterB: if you would like to help with this, you are welcome
13:37 kozhukalov sylwesterB: probably ikalnitsky could lead this
13:39 ikalnitsky kozhukalov: sylwesterB: yes, it's in my todo: to make docs actual and publish it to docs.openstack.org
13:44 nurla joined #fuel-dev
13:49 nurla joined #fuel-dev
13:50 sylwesterB ikalnitsky: so after the publishing, we will be able to work on actuall split and move back to nailgun and other components or the docs you're going to publish will be already splitted?
13:51 bookwar ikalnitsky: we actually already publish them
13:51 bookwar igorbelikov enabled it
13:51 bookwar but we need cleanup and a bit of layout reorganization
13:51 ikalnitsky bookwar: could you please show me the link? i didn't found one
13:52 bookwar ikalnitsky: we need these patches to align theme and restore diagrams https://review.openstack.org/#/q/project:openstack/fuel-docs+owner:%22Dzmitry+Moisa+%253Cdmoisa%2540mirantis.com%253E%22+status:open
13:53 ikalnitsky bookwar: it's about fuel-docs, not about fuel-web
13:53 bookwar current link is this one http://docs.openstack.org/developer/fuel-docs/
13:53 ikalnitsky we need to have developer docs
13:53 ikalnitsky in fuel-web
13:53 bookwar yes, but all docs were moved there
13:54 bookwar i think we need to update dev docs first before we start developing separate one
13:54 ikalnitsky bookwar: i don't think it's a good idea to have *nailgun-development-documentation* in fuel-docs repo. i'd prefer to see fuel-docs as User Guide
13:55 bookwar i'd rather get a reasonable common dev guide to start, before splitting resources on more detailed per-project topics
13:56 bookwar we'd get better overview of  docs structure
13:56 bookwar and then define what goes where
13:57 evgenyl The thing is we have upcoming doc whose audience will be Nailgun developers, don't see reasons to start writing it as a part of fuel-docs.
13:59 kozhukalov evgenyl: that was a plan to split the whole documentation bunch and adopt necessary pages into components
13:59 kozhukalov evgenyl: ui docs should be put into fuel-ui, nailgun docs into fuel-web, etc.
14:00 ikalnitsky kozhukalov: correct. but why do we need to write new nailgun docs in fuel-docs?
14:00 ikalnitsky i want it in fuel-web/docs
14:00 evgenyl kozhukalov: but we have a document to write (nailgun extensions) for developers, what we should do now? Should we put document into fuel-docs? Or should we write it as a part of fuel-web repo?
14:01 ikalnitsky and there should be topics like, how to prepare development environment, what API do we have and how to implement new nailgun extensions, etc
14:01 bookwar ikalnitsky: can you start a thread on a mailing list where outline like "from current dev-docs we take smth and smth, and add smth else"
14:03 bookwar ikalnitsky: and we'd need to have working docs before summit, so i am thinking maybe should postpone major restructuring, because i am not sure how much time it would take
14:04 dims joined #fuel-dev
14:10 lezbar__ joined #fuel-dev
14:11 spyzalski_ joined #fuel-dev
14:16 ikalnitsky bookwar: I don't know whether do we want to put something back to fuel-web/docs :) I'd prefer to write them from scratch for fuel-web.
14:16 xarses joined #fuel-dev
14:17 spyzalski joined #fuel-dev
14:18 dims_ joined #fuel-dev
14:20 nurla joined #fuel-dev
14:20 gongysh joined #fuel-dev
14:20 salmon_1 joined #fuel-dev
14:21 omolchanov joined #fuel-dev
14:26 yottatsa could you please review one-line bugfix https://review.openstack.org/296480 Fix l2_bond prefetch
14:39 nurla joined #fuel-dev
14:50 kozhukalov bookwar: ikalnitsky since we build ISO from mitaka repositories, this https://github.com/openstack/fuel-web/blob/master/nailgun/nailgun/fixtures/openstack.yaml#L1949-L1950 seems to be a bug, right?
14:52 nurla joined #fuel-dev
15:10 openstackstatus joined #fuel-dev
15:23 askb joined #fuel-dev
15:27 nurla joined #fuel-dev
15:42 zigo askb: Yes, using only Debian packages and a tiny setup script.
15:42 zigo agordeev: I'm busy with finishing the Mitaka RC1 release, I can't work on Fuel again right now, but will do once RC1 passes my tempest tests.
15:43 zigo agordeev: A "debian friendly" grub.conf is one with a *single* root=, not *twice*. :)
15:45 agordeev zigo, nice. it sounds like some of grub scripts appended root= for the second time? as for the first time it was fuel-agent who did that.
15:45 zigo agordeev: I'm not sure, really, I'd have to investigate further.
15:46 zigo agordeev: But for me, removing this in manager.py solved my problem:
15:46 zigo grub.append_kernel_params('root=UUID=%s ' % mount2uuid['/'])
15:46 zigo (around line 680)
15:47 zigo FYI guys, I'm completely done with packaging RC1, the only things that are remaining are Horizon plugins, which still have unit test failures in Sid.
15:48 zigo All of these are uploaded: http://paste.openstack.org/show/491612/
15:49 agordeev zigo: well done!
16:00 askb zigo, thanks for updating on the issue
16:02 askb zigo, which version of debian were you using ? would you have any errors messages as seen while hitting the issue
16:07 jaypipes joined #fuel-dev
16:14 fuel-ci joined #fuel-dev
16:20 dshulyak joined #fuel-dev
16:50 yottatsa merge plz one-line bugfix https://review.openstack.org/296480 Fix l2_bond prefetch
16:52 zigo askb: I'm using Jessie.
16:52 zigo (aka Debian 8)
16:58 nurla joined #fuel-dev
17:06 dims joined #fuel-dev
17:29 ikalnitsky kozhukalov: sorry, i missed your message :( correct. but it can't be changes simply without change to fuel-library
17:34 tlbr joined #fuel-dev
17:39 krobzaur joined #fuel-dev
17:51 askb zigo, I am not very familiar with deb, but how is diff from ubuntu 14.04 grub.cfg, as ubuntu should have only a single line root=UUID as a part of its menuentry
17:54 angdraug joined #fuel-dev
18:38 jaypipes joined #fuel-dev
18:41 yottatsa please review nailgun-agent one-line fix https://review.openstack.org/296666 Fixed SR-IOV VF device ID parsing
18:49 angdraug joined #fuel-dev
18:54 izinovik- joined #fuel-dev
18:54 fBigBro_ joined #fuel-dev
18:55 dkaigarodsev joined #fuel-dev
18:55 MiroslavAnashkin joined #fuel-dev
18:56 akislitsky_ joined #fuel-dev
18:56 SergK joined #fuel-dev
18:57 elemoine_ joined #fuel-dev
19:00 mirrorbox joined #fuel-dev
19:02 dancn joined #fuel-dev
19:02 izinovik joined #fuel-dev
19:07 openstackstatus joined #fuel-dev
19:11 akislitsky__ joined #fuel-dev
19:12 ikalnitsky_ joined #fuel-dev
19:12 tkhno_ joined #fuel-dev
19:12 kgalanov_ joined #fuel-dev
19:12 teran_ joined #fuel-dev
19:13 kozhukalov_ joined #fuel-dev
19:13 sslypushenko_ joined #fuel-dev
19:21 mescanef joined #fuel-dev
19:21 MiroslavAnashkin joined #fuel-dev
19:21 izinovik_ joined #fuel-dev
19:21 Samos123 joined #fuel-dev
19:21 rvyalov_ joined #fuel-dev
19:22 openstackstatus joined #fuel-dev
19:22 vkramskikh joined #fuel-dev
19:23 salmon_ joined #fuel-dev
19:36 dims_ joined #fuel-dev
19:36 dims_ left #fuel-dev
19:37 nadya joined #fuel-dev
19:43 dims_ joined #fuel-dev
19:56 dims joined #fuel-dev
20:17 e0ne joined #fuel-dev
20:30 mescanef joined #fuel-dev
20:35 HeOS joined #fuel-dev
20:50 angdraug joined #fuel-dev
21:21 gariveradlt joined #fuel-dev
21:52 xarses_ joined #fuel-dev
21:54 gariveradlt joined #fuel-dev
23:05 rvyalov joined #fuel-dev
23:10 meow-nofer joined #fuel-dev
23:22 openstackstatus joined #fuel-dev
23:45 angdraug joined #fuel-dev

| Channels | #fuel-dev index | Today | | Search | Google Search | Plain-Text | summary