16:01 <gaughen> #startmeeting ubuntu-server-team
16:01 <meetingology> Meeting started Tue Sep 29 16:01:36 2015 UTC.  The chair is gaughen. Information about MeetBot at http://wiki.ubuntu.com/meetingology.
16:01 <meetingology> 
16:01 <meetingology> Available commands: action commands idea info link nick
16:01 <gaughen> #topic Review ACTION points from previous meeting
16:02 <gaughen> I don't see any action points from last time.
16:02 <rharper> yay, we finally moved on from asking rharper about numad
16:02 <gaughen> #topic Wily Development
16:02 <gaughen> #link https://wiki.ubuntu.com/WilyWerewolf/ReleaseSchedule
16:02 <gaughen> ha
16:02 <gaughen> is thedac on the list to run this meeting?
16:03 <coreycb> gaughen, no but he probably should be
16:03 <thedac> so far I have hidden away
16:03 <gaughen> final beta has passed
16:03 <gaughen> thedac, I'm putting you on the list. I'm kind that way.
16:03 <gaughen> I'll put you on the end... but before me ;-)
16:03 <thedac> I am reconsidering our friendship :p
16:03 <gaughen> final release is in less than a month - AAAAH!
16:04 <gaughen> final freeze is Oct 15
16:04 <gaughen> #subtopic Release Bugs
16:04 <gaughen> #link http://reqorts.qa.ubuntu.com/reports/rls-mgr/rls-w-tracking-bug-tasks.html#ubuntu-server
16:05 <gaughen> don't think there's anything worth mentioning  on the list
16:05 <gaughen> #topic Server & Cloud Bugs (caribou)
16:05 <gaughen> caribou, you have anything to say?
16:05 <gaughen> beuller
16:05 <caribou> hi ?
16:05 <gaughen> hello
16:05 <gaughen> any bugs you want to mention
16:05 <caribou> yes, bug #1496317 is up for review
16:05 <ubottu> bug 1496317 in kexec-tools (Ubuntu) "kexec fails with OOM killer with the current crashkernel=128 value" [High,In progress] https://launchpad.net/bugs/1496317
16:05 <caribou> apw is supposed to look at it soon
16:05 <smoser> oh nice.
16:06 <caribou> tl;dr create smaller initrd.img in /boot/kdump
16:06 <arges> caribou: if it gets uploaded before tomorrow, I'll sru-review it : )
16:06 <caribou> arges: then go & harrass apw about it :)
16:06 <caribou> arges: I'd like a few eyes on it. Adding stuff to /boot can be seen as invasive
16:07 <arges> yea
16:07 <caribou> gaughen: other than that I'm good
16:07 <gaughen> caribou, cool! thank you!
16:07 <gaughen> #topic Weekly Updates & Questions for the QA Team (matsubara)
16:07 <gaughen> matsubara, any QA topics?
16:07 <matsubara> gaughen, hi, nothing new from me
16:07 <caribou> smoser: debdiff is in the case if you want to have a look at it
16:08 <gaughen> #topic Weekly Updates & Questions for the Kernel Team (smb, sforshee, arges)
16:08 <smb> Nothing to see here, move along... Please...
16:09 <gaughen> #topic Upcoming Call For Papers
16:09 <gaughen> any papers?
16:09 <gaughen> topics?
16:09 <gaughen> moving on
16:09 <gaughen> #topic Ubuntu Server Team Events
16:10 <gaughen> Openstack Summit in Tokyo is coming up
16:10 <gaughen> there is also Linux Con EU next week
16:10 <gaughen> several folks are presenting  on LXD and cloud-init
16:10 <gaughen> anything else?
16:10 <gaughen> #topic Open Discussion
16:11 <gaughen> anything?
16:11 <rharper> crickets
16:11 <gaughen> #topic Announce next meeting date, time and chair
16:11 <gaughen> next week, same time, same place
16:11 <smoser> oh.
16:11 <gaughen> chair will be
16:11 <smoser> hold on.
16:11 <smoser> lets go back to open discussion
16:12 <gaughen> arosales
16:12 <smoser> i want to float one thing
16:12 <gaughen> okay, go back
16:12 <gaughen> go for it
16:12 <smoser> https://blueprints.launchpad.net/ubuntu/+spec/foundations-w-networkd-vs-ifupdown
16:12 <gaughen> do I need to switch topics officially?
16:12 <smoser> nah
16:12 <smoser> server has been asked what we think of a switch to networkd in 15.10
16:12 <smoser> "by default"
16:12 <rharper> how can we do that now ?
16:12 <smoser> i wish rbasak was here, but interested in anyone's thoughts.
16:13 <smoser> :)
16:13 <gaughen> rcj, does this impact cloud-images?
16:13 <smoser> that is the expected response.
16:13 <rharper> maybe at the start of a cycle
16:13 <gaughen> smoser, isn't it a wee bit late
16:13 <smoser> well, the argument for it goes like this: i'd rather fail horribly now
16:13 <gaughen> rcj, I expect the answer is yes
16:13 <smoser> it does affect cloud images, yes.
16:14 <smoser> not a *ton*, but it it would.
16:14 <gaughen> smoser, would we have enough time to recover if we failed horribly now?
16:14 <rharper> smoser: hopefully the argument is: networkd fixes the following issues we have with ifupdown
16:14 <smoser> the bigger thing is anythign that writes (or reads) files in /etc/network/interfaces
16:14 <rharper> if the size of that list is not large, it's hard to argue that it's worth the trouble in the first place
16:14 <rcj> We have cloud image specifics around the interfaces files so we'll break clouds
16:15 <smoser> a 1x1 static change is easy enough.
16:15 <rcj> we have things that generate interfaces on the fly
16:15 <smoser> honestly, i wouldnt expect the cloud images to be terribly effected.
16:15 <smoser> oh?
16:15 <smoser> rcj, where do you have such things ?
16:15 <rcj> for cloud providers that have dynamic networking and multiple devices
16:16 <rcj> Joyent is one
16:16 <smoser> ok. so i'm very interested in that... can you give examples ?
16:16 <gaughen> probably not here
16:16 <smoser> (to be clear, i'm not advocating for this, but rather playing devils advocate).
16:16 <rcj> But I don't know what tools like walinuxagent and gce tools are doing for certain either.
16:16 <gaughen> what does non-devils advocate smoser think is the right thing to do here?
16:17 <smoser> rcj, can you give a link to code that does "dynamic config" ?
16:17 <rcj> smoser, I don't know that I'll find it quickly with Ben out
16:18 <smoser> gaughen, the non-devils advocate thinks this is very late in a cycle.  and also even possibly very late in an LTS cycle (ie, as in now is late for *16.04*)
16:18 <rharper> smoser: the most damning discussion I've seen is that networkd hasn't been (in the past at least) focused on server configuration, so things like parallel interface bringup or even ordered interface bring up (think bonds) have been addressed
16:18 <rharper> the ifupdown + bonding threads brought up the issue and networkd folks, no we don't do that, lemme see if I can find the link
16:18 <smoser> well, this would be a server install (and cloud image) config.
16:18 <gaughen> smoser, can we do a test build with it and see what falls out?
16:18 <rharper> but if we don't get equivalent function, and bug fixing, it's hard to argue that we should switch (and certainly harder late in the cycle)
16:19 <smoser> rharper, i think that you're mistaken. in that networkd *is* server centric.
16:20 <smoser> https://blueprints.launchpad.net/ubuntu/+spec/foundations-w-networkd-vs-ifupdown is the blueprint, and /usr/share/doc/systemd/README.Debian.gz on your wily system describes how to change
16:20 <smoser> pitti has done a good job of documenting that.
16:20 <rharper> I'll track down the emails I read;
16:21 <smoser> for easier reference, that README.Debian.gz is http://paste.ubuntu.com/12613789/
16:21 <smoser> i agree, rharper that lack of bond is certainly a stop ship issue.
16:21 <smoser> gaughen, so yeah, you could reasonably easily "see what falls out".
16:22 <smoser> so the big fallout is in anything that is reading or writing /etc/network/interfaces{,.d/*}
16:22 <smoser> which is curtin for one
16:22 <smoser> and also juju
16:22 <smoser> i'm interested i knowing anything else that people are aware of.
16:22 <smoser> largely this is a fact finding expedition
16:22 <smoser> also relevant is resolvconf -> systemd-resolvconf would be probably done
16:23 <gaughen> smoser, if we're still in fact finding mode that means to me we're not ready to pull the trigger
16:23 <gaughen> and thus too damn late for wily
16:23 <gaughen> my 2 cents
16:23 <smoser> the general reason for the interest in networkd is the same as that of systemd
16:23 <smoser> its not something you generally *chose*
16:24 <smoser> but something that is chosen for you. and you have to justify the cost of maintaining the other option (ifupdown-/etc/network/interfaces) indefinitely.
16:25 * caribou likes the fact that this discussion is under Announce next meeting date :)
16:25 <gaughen> :-)
16:25 <gaughen> so anything else to talk about on this topic. smoser what's the next step?
16:26 <smoser> thus, if it is inevitable, then dropping ENI/ifupdown (which people love to hate also) by 16.04 , means we get to drop support for ifupdown 2 years sooner.
16:26 <smoser> does anyone have other examples of readers or writers of /etc/network/interfaces ?
16:27 <smoser> the other one that I know of is cloud-initramfs-dyn-netconf (which writes /etc/network/interfaces style file based on 'ip=' arguments on the cmdline).  And also open-iscsi is probably somehow affected (as it has to not bounce network interface of the root iscsi device).
16:27 <smoser> anyone have anything else?
16:27 <gaughen> smoser, should this one be emailed about on ubuntu-server?
16:28 <gaughen> *crickets*
16:28 <gaughen> okay I'm declaring this mtg done
16:28 <smoser> done
16:28 <smoser> thanks.
16:28 <gaughen> until next week, same time, same place.
16:29 <gaughen> thank you smoser for your little topic
16:29 <caribou> thanks gaughen
16:29 <gaughen> #endmeeting