| *** Guest44289 has quit IRC | 02:45 | |
| *** Guest44289 has joined #softwarefactory | 02:51 | |
| *** nchakrab has joined #softwarefactory | 06:07 | |
| *** Guest44289 has quit IRC | 06:09 | |
| *** Guest44289 has joined #softwarefactory | 06:24 | |
| *** jangutter has quit IRC | 06:31 | |
| *** jangutter has joined #softwarefactory | 07:05 | |
| *** jangutter_ has joined #softwarefactory | 07:31 | |
| *** jangutter has quit IRC | 07:32 | |
| *** Guest44289 has quit IRC | 07:45 | |
| *** yassine has joined #softwarefactory | 08:06 | |
| *** yassine is now known as Guest90681 | 08:06 | |
| *** Guest90681 has quit IRC | 08:09 | |
| *** yassine_ has joined #softwarefactory | 08:10 | |
| sfbender | Tristan de Cacqueray created software-factory/sf-release master: Re-add depends https://softwarefactory-project.io/r/13001 | 08:12 |
|---|---|---|
| sfbender | Merged software-factory/sf-docs master: github setup doc: multiple improvements https://softwarefactory-project.io/r/12982 | 08:16 |
| sfbender | Fabien Boucher created software-factory/sf-ci master: deactivate test_pages temporarily (flacky test) https://softwarefactory-project.io/r/13002 | 08:17 |
| sfbender | Fabien Boucher created software-factory/repoxplorer-distgit master: Bump to last master https://softwarefactory-project.io/r/13003 | 08:55 |
| *** jangutter_ has quit IRC | 09:10 | |
| *** jangutter has joined #softwarefactory | 09:17 | |
| sfbender | Tristan de Cacqueray created www.softwarefactory-project.io master: create_release: add missing untag-build sf-release-9999 instruction https://softwarefactory-project.io/r/13006 | 09:37 |
| rcarrillocruz | pabelanger: i think we went thru this a while ago. The stock images are behind paywalls, we need to reach out PMs and people to get access to them after $time, therefore cloud images works better for us as we can't have a place to put the stock images, so dib can kick in and upload to nodepool | 09:57 |
| tristanC | rcarrillocruz: actually, i'd like to integrate cloud-image management in sf-config directly, e.g. have a playbook to run, create/update cloud-image and update the imageid in nodepool.yaml | 09:58 |
| rcarrillocruz | yah, that'd be good | 09:59 |
| rcarrillocruz | like | 09:59 |
| rcarrillocruz | when a new stock image gets out | 09:59 |
| rcarrillocruz | and we sort out access | 09:59 |
| rcarrillocruz | we build on our side | 09:59 |
| rcarrillocruz | we drop it somewhere in nodepool side | 09:59 |
| rcarrillocruz | and that gets uploaded | 09:59 |
| rcarrillocruz | i guess that's what you have in mind, right? | 09:59 |
| tristanC | yes, exactly | 09:59 |
| rcarrillocruz | ++ | 09:59 |
| tristanC | we also want to stop using dib for sf-ci, it would be better to run our test on the centos cloud image, since it's what users are using | 10:00 |
| tristanC | but we would like to pre-install mock and other things to speed up test, so we need a way to automate that | 10:01 |
| tristanC | too bad nodepool dropped snapshot support... | 10:01 |
| rcarrillocruz | that'd be good yeah...initially i was thinking to use nested images (i.e. a fedora/centos running network appliance on boot with qemu), as that would simplify jobs (the node is actually what you rnu the test on) and we wouldn't need to open ports on the secgroup as we were on the same provider as others in RDO | 10:02 |
| rcarrillocruz | but now that we are in vexxhost | 10:02 |
| rcarrillocruz | im less concerned | 10:02 |
| rcarrillocruz | so the net appliances will be the noes | 10:02 |
| rcarrillocruz | and the jobs will run the ansible-playbook based tests from either executor or an OCI node | 10:03 |
| rcarrillocruz | tristanC: inorite? it was so useful back in the day :/ | 10:03 |
| tristanC | rcarrillocruz: same thing with swift log upload, very useful too back in the day :-) | 10:04 |
| rcarrillocruz | heh | 10:04 |
| rcarrillocruz | tristanC: alternative, instead of doing that cloud-image uploader, could be writing a dummy dib element that doesn't create any image, just the post install does 'openstack image create BLAH' | 10:05 |
| rcarrillocruz | i toyed with something along those lines a while ago | 10:05 |
| rcarrillocruz | tristanC , pabelanger : https://github.com/rcarrillocruz/networking-dib-elements/blob/master/packer/element-deps | 10:08 |
| rcarrillocruz | the thing to do that is no-final-image | 10:09 |
| rcarrillocruz | that tells dib to not create an image | 10:09 |
| rcarrillocruz | so you could hook logic in the steps to do just 'check $folder for image, upload to vexxhost' kind of thing | 10:09 |
| rcarrillocruz | now the thing is to put logic on the element that doesn't fail if we don't have an image put in the folder | 10:10 |
| rcarrillocruz | i think it can be done | 10:10 |
| rcarrillocruz | not superautomated as the trove nested image pattern tho, as i said in networking we don't have the privilege to have all things opened like 'curl this public http site for image', due to licensing etc | 10:11 |
| rcarrillocruz | so probably | 10:12 |
| rcarrillocruz | we have a ssh key on our side | 10:13 |
| rcarrillocruz | so we can scp images to nodepool-builder | 10:13 |
| rcarrillocruz | and the dib element with no-final-image thingy | 10:13 |
| rcarrillocruz | (btw, disregard the packer thing from the link i put earlier, that was me experimenting with packer to build network images, thing is the concept of having an element that doesn't build an image, it just 'runs something' in the dib context) | 10:14 |
| rcarrillocruz | tristanC: iirc, i chatted with monty about the issue of cloud-images caching UUID, i believe he pushed a patch where it tries to use UUID, but if it fails then it falls back to cloud-image name | 10:45 |
| rcarrillocruz | re: sf being able to upload clouad images *and* updating image id in nodepool.yaml | 10:45 |
| rcarrillocruz | me thinks it may not be needed if cloud images names are used | 10:45 |
| tristanC | rcarrillocruz: yes, but we still want to support customization | 10:46 |
| tristanC | ideally, we restore the snapshot feature in nodepool and it's managed by the service | 10:46 |
| rcarrillocruz | fair enough, curious how would that work as usually nodepool.yaml is kept in source control | 10:46 |
| tristanC | but it might be easier to support custom playbook to do such things | 10:46 |
| tristanC | rcarrillocruz: in sf, the config-update already merge the nodepool/*.yaml content, it could take care of resolving the cloud-images names | 10:47 |
| rcarrillocruz | ah neat | 10:47 |
| sfbender | Fabien Boucher created software-factory/managesf master: configurations/repoxplorer: prevent failure if default cnx does not exists in resources cnx https://softwarefactory-project.io/r/13009 | 11:09 |
| sfbender | Merged software-factory/repoxplorer-distgit master: Bump to last master https://softwarefactory-project.io/r/13003 | 11:30 |
| sfbender | Tristan de Cacqueray created software-factory/sf-config master: tenant: fix tenant deployment when config-location is on github https://softwarefactory-project.io/r/13012 | 11:39 |
| tristanC | rcarrillocruz: an example of github tenant deployment using sf-3.1, the tenant is called ansible-test, it's deployed here: https://ansible-test.softwarefactory-project.io/zuul/status.html | 11:43 |
| tristanC | and the zuul tenant configuration is on github: https://github.com/morucci/an-sf-config/blob/master/resources/_internal.yaml | 11:43 |
| tristanC | then if new projects are added to github.com/morucci/an-sf-config, the tenant-update job will add the projects to the scheduler of sf-project.io | 11:44 |
| rcarrillocruz | nice, when vendors come along then everything will come from gh the bootstrapping | 11:45 |
| *** yassine_ has quit IRC | 12:23 | |
| *** yassine_ has joined #softwarefactory | 12:24 | |
| *** nchakrab_ has joined #softwarefactory | 12:30 | |
| *** nchakrab has quit IRC | 12:33 | |
| *** nchakrab_ has quit IRC | 13:15 | |
| *** nchakrab has joined #softwarefactory | 13:16 | |
| *** nchakrab has quit IRC | 13:37 | |
| *** nchakrab has joined #softwarefactory | 13:38 | |
| *** nchakrab has quit IRC | 13:43 | |
| sfbender | Merged software-factory/sf-ci master: deactivate test_pages temporarily (flacky test) https://softwarefactory-project.io/r/13002 | 13:56 |
| *** nchakrab has joined #softwarefactory | 14:16 | |
| *** nchakrab_ has joined #softwarefactory | 14:22 | |
| *** nchakrab has quit IRC | 14:26 | |
| *** nchakrab_ has quit IRC | 14:44 | |
| sfbender | Merged software-factory/sf-config master: jobs-base: protect config-check and config-update job https://softwarefactory-project.io/r/12969 | 15:28 |
| sfbender | Logan V created software-factory/sf-config master: Allow anonymous github API usage https://softwarefactory-project.io/r/13023 | 16:04 |
| *** nchakrab has joined #softwarefactory | 16:05 | |
| *** nchakrab has quit IRC | 16:22 | |
| *** nchakrab has joined #softwarefactory | 16:31 | |
| *** yassine_ has quit IRC | 16:46 | |
| *** yassine_ has joined #softwarefactory | 16:47 | |
| *** nchakrab has quit IRC | 17:34 | |
| sfbender | Matthieu Huin created www.softwarefactory-project.io master: 2018-28 summary https://softwarefactory-project.io/r/13030 | 20:19 |
| *** apevec has joined #softwarefactory | 20:35 | |
| *** apevec has joined #softwarefactory | 20:35 | |
| *** apevec has quit IRC | 21:57 | |
| *** yassine_ has quit IRC | 23:01 | |
| *** yassine_ has joined #softwarefactory | 23:13 | |
| *** yassine_ has quit IRC | 23:51 | |
| *** yassine_ has joined #softwarefactory | 23:57 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!