| *** mhen_ is now known as mhen | 01:17 | |
| *** elodilles is now known as elodilles_pto | 06:14 | |
| *** iurygregory__ is now known as iurygregory | 16:58 | |
| stephenfin | fungi: does the following pbr test fail for you in your local env? tox -e py38 -- 'pbr.tests.test_packaging.TestVersions.test_valid_tag_honoured' | 17:47 |
|---|---|---|
| stephenfin | I see it failing locally. I've traced it down to GNUPGHOME needing to be set to the value of HOME, so that it uses the temporary home dir we're creating for that test. I'd like to know if this is Fedora-specific behavior or not | 17:48 |
| fungi | stephenfin: is GNUPGHOME set to something in your environment normally? | 17:48 |
| stephenfin | not that I can | 17:49 |
| stephenfin | tell | 17:49 |
| stephenfin | however, by not setting that, gpg seems to _not_ create the '~/.gnupg' in the temporary home directory, presumably because it already exists in the "real" home directory | 17:50 |
| fungi | gimme a sec and i'll have your test answer | 17:50 |
| stephenfin | ...which causes file not found errors later on | 17:50 |
| fungi | {0} pbr.tests.test_packaging.TestVersions.test_valid_tag_honoured(preversioned) [0.793693s] ... ok | 17:51 |
| fungi | {0} pbr.tests.test_packaging.TestVersions.test_valid_tag_honoured(postversioned) [1.071473s] ... ok | 17:51 |
| fungi | i do have a ~/.gnupg dir though | 17:52 |
| stephenfin | as do I | 17:52 |
| fungi | i can try moving it aside for a moment | 17:52 |
| fungi | passes even if i don't have a ~/.gnupg | 17:53 |
| stephenfin | hmm, and what version of gnupg? | 17:53 |
| stephenfin | I'm seeing gpg (GnuPG) 2.4.7 and libgcrypt 1.11.0-unknown | 17:53 |
| fungi | gpg (GnuPG) 2.4.7 libgcrypt 1.11.0 | 17:57 |
| fungi | so same as yours | 17:57 |
| fungi | though i'm using debian sid/unstable | 17:57 |
| stephenfin | weeeeird | 17:58 |
| stephenfin | fungi: last test: what happens if you run these command? | 18:02 |
| * fungi is standing by waiting for the commands to run | 18:04 | |
| stephenfin | https://paste.openstack.org/show/bNz6FXx0xt0LqTKL0QUy/ | 18:04 |
| fungi | was that cat supposed to be redirecting stdout to something? | 18:05 |
| stephenfin | sorry, to `key-config` | 18:05 |
| stephenfin | `$PWD/key-config` (== `$HOME/key-config`) for the final command | 18:06 |
| stephenfin | I'm trying to manually reproduce the failing step from those test | 18:06 |
| stephenfin | s | 18:06 |
| *** iurygregory_ is now known as iurygregory | 18:07 | |
| fungi | stephenfin: https://paste.openstack.org/show/bnGij0XFL8cQPy61Iw6z/ | 18:09 |
| stephenfin | Okay, I get this | 18:10 |
| stephenfin | https://paste.opendev.org/show/bwRARTBa2asqPeYeMGLx/ | 18:10 |
| stephenfin | and it passes if I set GNUPGHOME | 18:11 |
| stephenfin | https://paste.opendev.org/show/bvYk2plryaEQjEgZniWx/ | 18:11 |
| fungi | i do have console/cli pinentry plugins installed since i run these things remotely over an ssh session most of the time | 18:12 |
| fungi | maybe it's something to do with trying to identify the pinentry mechanism, even if it ends up not being called | 18:13 |
| fungi | stephenfin: try exporting env DISPLAY="" GPG_TTY=$(tty) | 18:14 |
| fungi | see if that helps at all | 18:14 |
| fungi | as for packages, i have pinentry-curses and pinentry-tty installed specifically (those are their names in debian, dunno about fedora) | 18:15 |
| fungi | i also have pinentry-gtk2 and pinentry-gnome3 though i don't ever use them | 18:16 |
| stephenfin | no dice https://paste.opendev.org/show/baC2vwFAg8B5BkvX7ZDn/ | 18:16 |
| fungi | hrm... | 18:17 |
| fungi | i think it defaults to trying to use pinentry-tty unless overridden in e.g. ~/.gnupg/gpg-agent.conf with a pinentry-program line | 18:18 |
| stephenfin | Neither the Fedora package Git server nor the Debian one seem to be loading for me (HTTP 503 for both), so I can't look for obvious things like patches to change the behavior | 18:20 |
| stephenfin | I also don't have anything in /etc/gnupg nor do I have a ~/.gnupg/common.conf file, so I should be using standard configuration | 18:21 |
| stephenfin | setting selinux to permissive mode doesn't help... (I didn't think it would: I'd have seen avc denials otherwise) | 18:23 |
| stephenfin | so I guess my commit message is "this fixes things on Fedora but I don't know why" 😅 | 18:23 |
| stephenfin | fungi++ thanks for checking those for me | 18:24 |
| fungi | stephenfin: were you able to replicate the issue on another fedora machine, ruling out some local oddity on yours? | 18:26 |
| stephenfin | fair point, let me spin one up to see | 18:27 |
| fungi | my machines always have plenty of local oddities, which is why that's usually one of the first things i do ;) | 18:29 |
| stephenfin | yup works just fine on a Fedora 42 Cloud VM | 18:44 |
| fungi | at least that narrows it to "this fixes things on stephenfin's computer" ;) | 18:47 |
| fungi | (and maybe other systems with an as-of-yet unidentified commonality) | 18:48 |
| fungi | but yeah, most of the discussions i found related to that error pattern seem to stem from pinentry and agent daemon problems | 18:48 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Add openstack-tox-py313 job https://review.opendev.org/c/openstack/pbr/+/940117 | 21:24 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: tests: Skip tests that break on newer setuptools https://review.opendev.org/c/openstack/pbr/+/953839 | 21:24 |
| opendevreview | Stephen Finucane proposed openstack/pbr master: tests: Set GNUPGHOME in tests https://review.opendev.org/c/openstack/pbr/+/953840 | 21:24 |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!