Thursday, 2019-03-07

logan-i'm trying to find a way to modify the config repo gerrit acl in a way that does not get overwritten by sfconfig on the next run, is that possible?00:51
logan-ah I see that it is not possible01:13
*** sbadia has quit IRC01:45
*** sbadia has joined #softwarefactory01:46
*** ChanServ sets mode: +o sbadia01:46
*** bhavikdbavishi has joined #softwarefactory02:45
sfbenderLogan V created software-factory/sf-config master: Allow customization of config gerrit ACL  https://softwarefactory-project.io/r/1516004:03
logan-greenfield letsencrypt deployment is broken currently. sf-gateway runs an apache2 reload, which fails because the lecm role has not run yet so the cert file does not exist yet.04:21
logan-in other projects I work around this by generating a self signed cert to bootstrap with, then replace it with a letsencrypt cert later on.04:21
logan-just an idea but that might work here04:22
*** nijaba has quit IRC04:26
tristanClogan-: indeed that's an issue, perhaps we could copy the default cert in /etc/letsencrypt for the first reload04:43
tristanClogan-: though, sf-lecm should be running before sf-gateway...04:44
logan-hmm04:45
logan-correction, the failure is in sf-cauth04:45
logan-although that's on a sfconfig run where sf is already deployed and I tried to flip the letsencrypt switch04:47
logan-I can't remember where it failed on greenfield04:47
tristanClogan-: 15160 looks correct04:47
tristanClogan-: iirc we used to have a test that setup letsencrypt on greenfield and it worked04:47
tristanClogan-: and switching it on later should also work. if you have some logs to share, we can have a look04:48
tristanClogan-: also, note that cauth mush be running on the gateway host too, this is now enforced by https://softwarefactory-project.io/r/#/c/15124/04:49
*** nijaba has joined #softwarefactory04:53
logan-gotcha - I will get some logs gathered up tomorrow for the letsencrypt issue. I fixed up and tested 15160 for idempotency with the old config, and its identical now.05:04
logan-thanks!05:04
sfbenderTristan de Cacqueray created DLRN master: logging: refactor basicLogging configuration  https://softwarefactory-project.io/r/1516405:48
tristanCjpena|off: jruzicka: we are struggling to figureout how to let the user configure distroinfo and the dlrn.cfg file...07:22
tristanCjpena|off: jruzicka: why isn't the dlrn.cfg part of the distroinfo? and or, should dlrn be able to generate the dlrn.cfg based on the existing distro-info?07:22
fbothe mock config07:28
*** jpena|off is now known as jpena08:14
jpenatristanC, fbo: well, we never thought about the possibility of generating the mock config, since it's always been quite specific to the environment (base repos, chroot-installed packages and such). Looking at rdoinfo, it might be possible to generate the repos part based on https://github.com/redhat-openstack/rdoinfo/blob/master/rdo.yml#L34-L42, but I see it as a quite specific use case. I mean, having an external tool generate the08:20
jpenaconfiguration based on that looks fine to me, but I wouldn't make it part of DLRN (too tight coupling)08:20
tristanCjpena: external repos would make sense, it's a duplicated conf, and dlrn already inject its own repository08:22
tristanCjpena: and for the rest of the mock config, would you use different one for the same distroinfo ?08:23
jpenatristanC: yes, we use different ones when building for rhel 7 or 8, for example08:24
jpenaalso, not every driver depends on distroinfo08:24
tristanCjpena: but rdoinfo only has el7 information...08:25
jpenatristanC: right, so that's why I don't want to couple it too tightly08:25
jpenawith the same distroinfo information, we're currently building for el7, fedora and el808:26
tristanCjpena: ok, make sense, but then , shouldn't we remove the distrepos from distroinfo ?08:27
jpenatristanC: from the DLRN point of view we could, I don't know if it is used for any other purpose08:27
tristanCjpena: oh ok, i think it would be clearer. it's confusing to set distrepo both in distroinfo and in dlrn mock.cfg08:28
tristanCe.g., from a new user perspective, it is confusing to set build target and distrepo in both places08:31
*** sshnaidm has quit IRC08:51
tristanChmm, grafiti probably needs distrepo information08:51
*** sshnaidm has joined #softwarefactory08:52
*** bhavikdbavishi has quit IRC10:30
sfbenderTristan de Cacqueray created software-factory/sfinfo master: format-patch: remove git global  https://softwarefactory-project.io/r/1516710:48
sfbenderMerged software-factory/sfinfo master: Add path exclusion and rename 'cat' by 'note'  https://softwarefactory-project.io/r/1506310:51
*** bhavikdbavishi has joined #softwarefactory11:16
*** rfolco is now known as rfolco|rucker12:08
jruzickatristanC, so you don't like duplication of information in rdoinfo and dlrn mock.cfg, correct?12:11
jruzickadistroinfo is just a simple parser/templater/fetcher, it knows nothing like certain John ;)12:12
jruzickajpena, hey! so let's assume I have all the pieces after patch_rebaser in place (which I don't :D), where should I put them? will there be a script calling individual preprocessing steps or...?12:20
jpenajruzicka: we have a script that calls patch_rebaser, so it could call the extra preprocessing step12:20
jpenaalternatively, we can configure DLRN to run several preprocessing scripts in order12:20
jruzickajpena, where does it live?12:20
jpenajruzicka: right now, it's in the osp trunk VM12:21
jpena#!/bin/bash12:21
jpenasource ~/patch_rebaser/.venv/bin/activate12:21
jpenapython ~/patch_rebaser/patch_rebaser/patch_rebaser.py12:21
jpena^^ pretty simple ATM12:21
jruzickajpena, and expected result is the updated distgit, right? does it matter at which branch does the script end?12:22
jpenajruzicka: no, doesn't matter12:22
jruzickajpena, so I say just add `git co $DISTGIT` and `rdopkg patch -l` to that script and we only have the proper changelog left and I'm reading them cards about that right now.12:25
jpenajruzicka: I guess that would work12:28
jruzickawhy shouldn't it? :)12:28
jruzickaif it won't work I'll make it work :D12:29
jruzickaworked for me, though12:29
*** jpena is now known as jpena|lunch12:46
*** rfolco|rucker is now known as rfolco|ruck12:57
*** jpena|lunch is now known as jpena13:50
sfbenderJavier Peña created DLRN master: Refactor downstream driver filtering  https://softwarefactory-project.io/r/1517014:10
sfbenderMatthieu Huin created software-factory/sfinfo master: Add python-jsonpath-rw  https://softwarefactory-project.io/r/1517217:19
*** jpena is now known as jpena|away17:30
*** sshnaidm is now known as sshnaidm|off17:47
*** irclogbot_3 has joined #softwarefactory18:03
sfbenderJon Schlueter created rdopkg master: WIP: expand test_guess unit test coverage for patches branch  https://softwarefactory-project.io/r/1517318:29
*** bhavikdbavishi has quit IRC19:21
*** jpena|away is now known as jpena|off19:32
logan-tristanC: following up on that letsencrypt stuff from yesterday. here are 2 logs, the first is a greenfield deploy w/ letsencrypt enabled. the second is a deploy without letsencrypt, then attempting to flip the letsencrypt flag on and sfconfig. both fail the same way it seems like.20:04
logan-tristanC: greenfield letsencrypt https://pasted.tech/pastes/68219754a5624efc62b16d58854965e2a57cc859.raw20:04
logan-and a separate deploy without letsencrypt, attempting to flip it on afterward and re-sfconfig: https://pasted.tech/pastes/080cc900dda228c2a844082f9af029b6cdf17ee8.raw20:04

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!