=== mgagne is now known as Guest20946 === r-daneel_ is now known as r-daneel === kukacz_ is now known as kukacz === r-daneel_ is now known as r-daneel [17:42] Thanks again to rharper and blackboxsw for their guidance yesterday. Here is how things panned out. https://gist.github.com/RichardBronosky/fa7d4db13bab3fbb8d9e0fff7ea88aa2#result [17:43] I will take that /var/lib/cloud/scripts/per-boot/00_run-parts.sh concept to the Raspbian community and see how they feel about it. https://gist.github.com/RichardBronosky/fa7d4db13bab3fbb8d9e0fff7ea88aa2#file-cloud-init-setup-sh-L46 === brunobron-afk is now known as brunobronosky [17:45] I am worried about how to package it seeing as Debian is pegged to cloud-init version 0.7.9 for the foreseeable future. [18:18] brunobronosky: hey; we can probably help you work with 0.7.9; that's not that old despite the jump in versioning [18:20] rharper that would be great. My concern about 0.7.9 is that it seems to insist on a volume labeled cidata. [18:20] it also supports the /var/lib/cloud/seed/nocloud-net; so you could switch to that; [18:20] That kernel cmdline flag doesn't seem to be introduced until 17.1 [18:21] the seedfrom should still work on 0.7.9 though; lemme see [18:22] yeah, you can definitely seed from older than 17.1 [18:23] just may need to write a config file to /etc/cloud/cloud.cfg.d/datasource.cfg which specifices NoCloud and sets seedfrom: /boot/ etc [18:24] Oh. That's not too bad. Actually that may be more acceptable (upstream) than modifying the kernel cmdline. === r-daneel_ is now known as r-daneel [18:39] smoser is that completely missing from the old documentation, or am I just not finding it? http://cloudinit.readthedocs.io/en/0.7.9/topics/datasources/nocloud.html [18:44] probably missing [18:47] Okay, I guess I have to learn to grok https://github.com/cloud-init/cloud-init/blob/7fb6f78177b5ece10ca7c54ba3958010a9987f06/cloudinit/sources/DataSourceNoCloud.py [20:03] blackboxsw: rharper [20:03] http://paste.ubuntu.com/p/2VxttpZFjK/ [20:03] looking [20:03] that is what i was describing with IFS and shell. [20:04] smoser: I was somewhat confused, I didn't realize it was set -- which was dropping the duplicates [20:05] wow [20:08] it is documented. [20:08] see https://linux.die.net/man/1/bash [20:08] 'word splitting' [20:12] Any character in IFS that is not IFS whitespace, along with any adjacent IFS whitespace characters, delimits a field [20:12] ie, no empty words [20:12] clear as mud [20:12] what does *that* mean? That if you set IFS to non-default whitespace characters it'll also observe standard whitespace chars in addition to provided IFS value? [20:13] no, just that, you have to have a sequence of non-IFS chars between IFS [20:13] bash and sh are at least consistent [20:13] no empty words [20:13] rharper: well no. see paste [20:13] empty words are allowed with non whitespace as delim [20:13] yeah that paste looks inconstent based on what IFS you provide [20:14] pipe and comma look like bugs [20:15] I suspect something other than ifs wordsplitting is happening due the | being for setting up new fs and execs [20:15] " Any character in IFS that is not IFS whitespace, along with any adjacent IFS whitespace characters, delimits a field" [20:15] not sure what comma means to bash [20:15] *that is not IFS whitespace* [20:16] hrm, so it should read, don't use a subset of the default IFS [20:16] that's pretty strange thing to do [21:29] blackboxsw or rharper https://code.launchpad.net/~smoser/cloud-init/+git/cloud-init/+merge/338470 [21:29] i got to run, will check back in later. and i will review stanguturi and blackboxsw mps then [21:29] later [22:59] rharper or blackboxsw pretty please ^ ? stanguturi would like it too . [22:59] will grab it smoser [22:59] sorry, was unsuccessfully heads down on vsphere [23:50] smoser: approved your branch, stanguturi has a comment in there that I'll let you fix/merge [23:51] @smoser. Thanks Scott. I tested your patch in my test environment and it worked.