[14:17] rharper: http://paste.ubuntu.com/23456363/ is the modifications I've just made locally to cloudinit-analyze to get it to run over `journalctl -o short-precise -u "cloud-*"` output. [14:19] Basically the line format is different (because it has the process name/PID in there), and it's a syslog-style timestamp but with microseconds. [14:19] And also I was getting a repeated "finish" event, which the second hunk ignores. [14:19] But that was just a hack, I didn't actually think about it. [14:29] Odd_Bloke: I have short-precise working [14:30] rharper: Nice! Can you point me at the code? [14:30] yes, lemme get it pushed into the repo [14:42] Odd_Bloke: ok, pushed to cloudinit_analyze trunk [14:43] and I think I already sent you the bits about extracting the systemd journal directory ? [14:43] if we have that as an artifact, then we can reconstruct the data via -o short-precise offline [14:44] You did, but I didn't actually finish doing it. [14:44] Let me add a work item for that. [14:46] ok [14:59] smoser: re: systemd unit files, the following changes would address slangasek and pitti's request to use sysinit.target; I've tested in Xenial and Yakkety Openstack instances, in Yakkety, tested both ifupdown and netplan versions, and it's all working fine. I still need to test those changes in UC16 images and ideally in an Azure instance with the early disk mount (but I don't have access to an instance) [14:59] http://paste.ubuntu.com/23456512/ [15:07] rharper, there are some others too though, local-filesystems isnt right. [15:07] rharper: Can I launch an instance for you? Or do you have some specific requirements? [15:10] smoser: I only took the suggestion of replacing Before=basic.target dbus.sock, to Before=sysinit.target; and then the networking changes to support both networking and networkd [15:10] Odd_Bloke: that'd be perfect; I just install the latest cloud-init. modify the services files reboot (and hope for the best) [15:11] smoser: this is what it looks like after the change: http://paste.ubuntu.com/23456557/ [15:17] rharper, ok. reading... i was responding to pitti and updating https://code.launchpad.net/~smoser/cloud-init/+git/cloud-init/+merge/310386 [15:17] now i'll read back [15:18] Odd_Bloke, thats nice (the journalctl stuff) [15:23] smoser: yeah, pitti showed me the -o short-precise [15:23] and I realized that I won't need changes to the logging conf file [15:23] which makes merging cloudinit-analyze less bothersome [15:25] Odd_Bloke: I'm going pull in that last bit to help drop that extra finish event that showed up (looks cleaner than what I recall I did) [15:25] rharper: What's your LP ID? [15:26] raharper [15:26] rharper is is long-lost account that he does not have creds to [15:26] :) [15:26] rharper: ubuntu@yakkety-161110-1518.cloudapp.net [15:26] smoser: Thanks. :) [15:26] rharper: (Love your hostname on that key.) [15:30] heh [15:31] rharper: And ubuntu@xenial-161110-1527.cloudapp.net [15:34] rharper, going to ask pitti some in -devel [15:36] ok [15:37] rharper: Well, except now I'm thinking about beer. [15:40] =) [15:41] (Specifically https://untappd.com/b/cloudwater-brew-co-cosweisse/1757056) [15:42] did you see the upcoming Jester King Collab with cloudwater ? [15:42] http://jesterkingbrewery.com/cloudwater-brew-co-collaboration [15:42] * rharper likes how #cloud-init is now #cloud-init-beer [15:43] Odd_Bloke: wow, that one sounds amazing [15:43] Brettier than a bretty thing on St. Brett's Day. [15:43] lol [15:56] Odd_Bloke: smoser: on the changes to cloud-init service; the After=systemd-networkd-wait-online.service is *dependent* on systemd-networkd.service dropping dbus.service from it's After= value; This only affects instances that do networkd *and* need early mount (so netplan image in Azure, for example) [15:56] rharper, https://code.launchpad.net/~smoser/cloud-init/+git/cloud-init/+merge/310547 [15:56] Odd_Bloke: the changes I proposed work fine in azure yakkety; testing xenial now [15:57] smoser: cool [15:58] rharper, so confused by the above. [15:58] you're saying systemd-networkd.service has to change to have these work, right ? [15:59] to support all scenarios, yes [15:59] but pitti knows this [15:59] it's mentioned in the one bug [15:59] I've commented it [15:59] and it only affects the "mount stuff early"; so networkd/netplan enabled instances in Azure (which needs the early mount) [16:00] so UC16 is affected [16:00] why does azure need the early mount ? [16:00] I think it was just the place where early mount/reformat of disk occurred [16:01] other clouds could also want to do that; but none do by default? [16:01] IIRC< the bug was that fstab was updated before cloud-init reformatted the drive ? [16:01] due to not running cloud-init early enough ? [16:01] something to that effect [16:01] that was why we had Before=basic.target [16:14] smartos does it, rharper [16:14] but user-data can affect it too [16:21] ah [16:27] Odd_Bloke: where do we file bugs against waagent ? [16:28] rharper: In the Ubuntu package? [16:29] in the code that runs in the instance [16:29] python stacktrace [16:29] just wanted to pass it along [16:29] rharper: Sorry, I meant: in the Ubuntu package (unless that doesn't make sense). [16:29] ah, sure [16:29] That question mark was doing a lot of work. ;) [16:29] hehe [17:39] Odd_Bloke: smoser: thanks for the azure instances; the changes in smosers branch appear to work fine for xenial and yakkety; I'm going to test out UC16 on openstack next [17:39] Odd_Bloke: I'm done with those, if you want to kill em off [17:40] rharper, did you reboot ? [17:40] yeah [17:40] after having that entry in /etc/fstab [17:41] ok. yeah [22:10] rharper, just pushed to trunk the merge for Martin, [22:10] to be clear, I can't commit this (and upload to Ubuntu) until bug 163692 is fixed in the the given release. Thats what you're saying, right? [22:10] bah [22:10] the fix for https://code.launchpad.net/~smoser/cloud-init/+git/cloud-init/+merge/310547 [22:11] smoser: it only affects UC16 [22:11] but in general, yes that's right [22:11] we should wait [22:12] well, we can definitely go to zesty [22:12] and i feel safe saying it can go to azure also [22:12] er... to xenial [22:12] as we dont really support networkd in xenial outside of core. [22:12] and that fix is comming. [22:13] rharper, the docs are up at http://cloudinit.readthedocs.io/en/latest/ [22:13] now also [22:13] http://cloudinit.readthedocs.io/en/latest/topics/boot.html [22:13] smoser: yeah [22:13] and separated out the datasource to their own files so we can document each of the search criteria in them. [22:14] and i got to go. [22:14] later