[02:12] <minimal> cc_migrator was mentioned during the Summit, don't remember by who - so it is "legacy" and only required for moving from very old cloud-init releases? The docs (and source code) are not very clear about its purpose
[02:48] <blackboxsw> minimal: that comment was me about dropping cc_migrator. it was put into cloud-init in version 0.7.0 in 2012. the migrate the semaphore marker file names from versions prior to 0.7.0. I honestly don't think that we are on an upgrade path from 0.7.0 -> 23.X that we need to worry about anymore. Any VM that is on an upgrade path from 0.7.0 would be on python 2.7 or earlier and will not be able to upgrade past 19.4.
[02:51] <minimal> ok, that clarifies. So I'll take it out of cloud.cfg on images I create ;-)
[02:58] <blackboxsw> 19.4 was the last python 2.7 version if I recall. so... I think it highly unlikely any image created in the last decade will still make use of an upgrade path that would warrant rewriting the names of semaphore files.
[03:05] <minimal> removing that will shave fractions of a millisecond off c-i boot times ;-)
[03:12] <minimal> blackboxsw: related to upstreaming of patches to c-i from distros that was briefly mentioned in Summet, I noticed that the rendered cloud.cfg for "amazon" doesn't specify any packages (i.e. RPM) management module at all, so I guess AWS are tweaking this in their own packages
[03:13] <blackboxsw> think of all the savings across your thousands of VMs :)
[03:15] <blackboxsw> minimal: I'm guessing the same, I thought I recalled them mentioning that they roll their own cloud.cfg in packaging.
[03:15] <minimal> actually must add a comment to this effect to my PR #4272 (i.e. I didn't break it, it was broken already) ;-)
[03:20] <minimal> seems strange for the c-i repo's template to have "amazon" templating then if it isn't used
[13:19] <meena> let's see if this passes on Linux! https://github.com/canonical/cloud-init/pull/4348
[13:19] -ubottu:#cloud-init- Pull 4348 in canonical/cloud-init "util.get_proc*: Fix tests on non-Linux systems" [Open]
[13:24] <meena> this now brings us down to 4 failing tests.
[13:25] <meena> time for lunch.
[15:32] <meena> hrm… okay, i think i went down the wrong path
[21:16] <meena> progress!
[22:07] <meena> and another one: https://github.com/canonical/cloud-init/pull/4351
[22:07] -ubottu:#cloud-init- Pull 4351 in canonical/cloud-init "cc_growpart: fix tests to run on FreeBSD" [Open]
[22:21] <meena> https://github.com/canonical/cloud-init/pull/4352
[22:21] -ubottu:#cloud-init- Pull 4352 in canonical/cloud-init "cmd/clean: fix tests on non-Linux platforms" [Open]
[22:21] <meena> and now we're down to one failure
[22:22] <holmanb> nice
[22:33] <meena> oh great, this one passes in isolation, but not as part of the whole suite
[22:35] <meena> E       AssertionError: b'hello world' != b'/usr/local/bin/bash: warning: setlocale: [78 chars]orld'                                                                                                                                                                          
[23:05] <meena> i got a fix.
[23:23] <meena> fix: https://github.com/canonical/cloud-init/pull/4355
[23:23] -ubottu:#cloud-init- Pull 4355 in canonical/cloud-init "subp: Fix spurious test failure on FreeBSD" [Open]
[23:23] <meena> just for giggles, i tried running the test-suite without Bash… 26 failures.
[23:24] <meena> i think if we ignore the OpenNebula stuff which actually needs it, everything else could be changed to /bin/sh without anyone noticing anything.
[23:26] <meena> done. I need sleep