[03:00] <sarkis> hey all - does upstart send the kill signal to parent+child procs?
[03:03] <sarnold> sarkis: it's a bit complicated; see step 8 at http://upstart.ubuntu.com/cookbook/#stopping-a-job
[03:05] <sarkis> shit that sucks
[03:05] <sarkis> can i override that behavior with any of the stanzas?
[03:06] <sarkis> i don't think so :(
[03:07] <sarnold> what are you trying to accomplish?
[03:11] <sarkis> i'm managing a ruby script via upstart... the script forks off and runs other scripts, when i stop upstart i need it to only send a signal to the parent script and not the ones that it's forked
[03:12] <sarkis> specifically it's managing resque (if you all know what that is)
[03:12] <sarkis> somehow the child procs that get forked are part of the same proc group ... i've confirmed this as the child procs are dying with the killsignal i give upstart
[03:13] <sarkis> this could all be by design - but that is what the end goal is for trying to modify the quit behavior ^
[03:16] <sarnold> sarkis: you -could- make your script fiddle with the process groups of its children; something like pid = fork(); if pid == -1: FAIL ; if pid == 0: setpgrp(); ... sort of thing
[03:17] <sarnold> sarkis: if ruby doesn't expose the setpgrp() syscall (and given how ugly it is I wouldn't be surprised) you might be able to achieve the same thing with a call to setsid() in a child process, too
[03:19] <sarnold> sarkis: another possibility, maybe you could have the child processes block or ignore SIGTERM; some programs just leave signal dispositions alone when they start (e.g. that's how the nohup program works)
[11:34] <rbasak> nacc: "usd merge tag" seems to fail with my current lint script, breaking it. I think we need to have a discussion about namespaces. Additionally remote names like "lpusd" created by "usd clone" are getting quite inconsistent.
[11:34] <rbasak> smoser: ^
[11:37] <rbasak> nacc, smoser: maybe time to update SPECIFICATION with our current understanding and additionally define all expected names used (namespaces, tags, branches, all of it), with rationale, and define what commands require what?
[11:58] <sb_9> http://pastebin.com/QYHjgWfq
[11:58] <sb_9> struggling with apache reverse proxy issue.
[12:32] <Walex> sb_9: maybe #httpd but your query is a bit vague.
[12:35] <sb_9> Walex: please check the http://pastebin.com/QYHjgWfq
[12:35] <Walex> sb_9: I already looked at it.
[12:42] <Walex> sb_9: have you looked at https://httpd.apache.org/docs/current/mod/mod_proxy.html#page-header
[12:42] <Walex> https://httpd.apache.org/docs/current/mod/mod_proxy.html#examples
[13:27] <coreycb> zul, ddellav: glance and trove b1 are pushed/uploaded
[13:27] <zul> coreycb: ok
[13:27] <coreycb> zul, ddellav: i think i have the ocata backports fixed up
[13:28] <coreycb> zul, ddellav: we're hitting more and more packages that have debhelper >= 10.  so i've been patching it to >= 9 in ca-patches for the ocata uca.
[13:29] <zul> coreycb: ok im trying to figure out this now https://launchpadlibrarian.net/293970474/buildlog_ubuntu-zesty-amd64.nova_2%3A15.0.0~b1-1ubuntu2_BUILDING.txt.gz
[13:33] <coreycb> zul, hmm those deps should be ok.  os-brick is stuck in proposed though.
[13:33] <zul> coreycb: yeah i think i figured it out
[13:33] <coreycb> zul, what was the issue?
[13:34] <zul> os-brick dep issues
[13:35] <coreycb> zul, i just hit something similar with my glance upload - https://launchpadlibrarian.net/293975451/buildlog_ubuntu-zesty-amd64.glance_2%3A14.0.0~b1-0ubuntu1_BUILDING.txt.gz
[13:36] <zul> coreycb: hmm...
[13:36] <zul> coreycb: this is weird
[13:38]  * zul scratches his head
[13:48] <smoser> rbasak, sure, i agree we want consistency there.
[13:51] <rbasak> smoser, nacc: I suppose the areas are: .gitconfig remote insteadOf aliases for documentation, remote names, local names for things for "usd clone", "usd merge tag" and similar and also documentation, and then separately the namespaces inside refs/ that the importer produces.
[13:52] <zul> rbasak: hey me and coreycb are looking at some weirdness that has us stumped https://launchpadlibrarian.net/293975451/buildlog_ubuntu-zesty-amd64.glance_2%3A14.0.0~b1-0ubuntu1_BUILDING.txt.gz
[13:53] <rbasak> zul: looks like a build-depends problem? Try using a local schroot and install all the build deps Launchpad installed manually, and apt should tell you why python-xattr >= 0.4 can't be installed.
[13:56] <zul> coreycb/rbasak: it looks like python-cffi is not installable?
[13:57] <ddellav> coreycb zul neutron-dynamic-routing done: lp:~ddellav/ubuntu/+source/neutron-dynamic-routing
[13:58] <ddellav> coreycb zul I'm gonna take networking-ova
[13:58] <coreycb> zul, rbasak I think this is the problem http://paste.ubuntu.com/23495526/
[13:58] <ddellav> *ovn
[13:58] <zul> coreycb: yeah i got there as well :)
[14:00] <coreycb> zul, maybe we just need to sync those deps
[14:01] <zul> coreycb: nah...ply was synced from debian like 2 hours ago so maybe we been caught in a transition or something
[14:01] <zul> coreycb: https://launchpad.net/ubuntu/+source/ply
[14:01] <coreycb> zul, ok
[14:02] <coreycb> thanks rbasak
[14:09] <coreycb> ddellav, n-d-routing pushed/uploaded.  i updated the version.
[14:10] <ddellav> coreycb why did you add an ubuntu delta?
[14:13] <zul> coreycb: did you upload neutron?
[14:15] <coreycb> zul, yes
[14:16] <coreycb> ddellav, because we uploaded a new version straight to ubuntu
[14:16] <coreycb> ddellav, if we were uploading to debian and then syncing from debian then we could just use  -1 instead of -0ubuntu1
[14:19] <zul> someone did neutron-fwaas and neutron-lbaas?
[14:23] <sb_9> http://pastebin.com/QYHjgWfq
[14:55] <coreycb> ddellav, did you figure out why neutron is failing on xenial by any chance?  having a hard time figuring it out.
[14:55] <coreycb> ddellav, xenial-ocata that is
[14:58] <rbasak> nacc, smoser: I'm not sure about "usd clone" pulling from both usd-import-team and ubuntu-server-dev either. That gets confusing quickly.
[14:58] <rbasak> I was using ubuntu-server-dev to store trees from uploads before we had an importer.
[14:59] <rbasak> Now that we have an importer and all usd-import-team members are core devs, can we just trust usd-import-team for now, and forget about ubuntu-server-dev?
[14:59] <rbasak> Then when we're ready we just switch everything over.
[14:59] <rbasak> Then we don't have to have logic in our tooling (eg. my lint) that deals with differences between the two, which we won't need in the end anyway.
[15:00] <rbasak> I suggest that "usd clone" doesn't clone ubuntu-server-dev for now, and that if exceptionally you need it because something legacy is in there, you pull it manually.
[15:02] <smoser> rbasak, well... i very much like a source that cannot be broken
[15:02] <rbasak> smoser: that would be nice but we don't have that luxury right now. If we try, we just get split trees and no idea which to use.
[15:03] <smoser> well, to me there are 3 things
[15:03] <smoser> a.) import pristine stuff
[15:03] <rbasak> smoser: unless you want an import-only tree (no pushing upload tags). We could maintain an effectively read-only set of package repos that could be rebuilt as necessary then.
[15:04] <smoser> b.) group maintained split delta
[15:04] <smoser> c.) users branches
[15:05] <smoser> i dont think i really feel strongly on pushing upload tags or not.
[15:05] <smoser> in the end i dont care, as long as said tags are never wrong
[15:07] <smoser> increasing the number of people who push to 'a' increases likely stuff to get screwed up
[15:07] <smoser> i do agree, though, that the 3 things is confusing.
[15:30] <coreycb> beisner, hello, can you promote python-glance-store 0.18.0-0ubuntu1.1~cloud0 to newton-updates and neutron 2:8.3.0-0ubuntu1.1~cloud0 to mitaka-updates?
[15:31] <coreycb> beisner, also newton-staging can be promoted to newton-proposed while you're in there.
[15:34] <beisner> howdy coreycb
[15:36] <coreycb> beisner, howdy
[15:36] <huwjr> hi, can someone tell me how i can add kernel modules during a preseed?
[15:37] <huwjr> problem is i’m trying to boot KVM machines with virtio modules, but i have to boot as IDE, then add the modules and edit the guests xml then reboot
[15:38] <ddellav> coreycb you talking about b1 or ci?
[15:38] <ddellav> re:neutron
[15:38] <coreycb> ddellav, both i guess, it's basically the same
[15:39] <ddellav> coreycb i asked because i built neutron in xenial-ocata before i pushed it up and it built fine
[15:39] <ddellav> i haven't messed with CI today since we are doing b1 releases atm
[15:40] <coreycb> ddellav, here's the xenial-ocata build that's failing for b1: http://10.245.168.2:8080/job/backport_package/1579/consoleFull
[15:41] <coreycb> ddellav, i'm trying a build that runs tests serially
[15:42] <ddellav> coreycb i built it locally with the sbuild-ocata helper
[15:46] <smoser> huwjr, can you explain more whats the goal ?
[15:46] <huwjr> currently KVM sets up host with preseed and virt-install script
[15:47] <huwjr> if i want to deploy the hosts with VIRTIO devices (instead of IDE for disk and realtek for NIC) it will preseed, but the machine won’t boot because it doesn’t recognise the virtio devices without kernel modules being loaded
[15:47] <huwjr> so is my only option wget | bash a script to fix up initramfs and update?
[15:48] <huwjr> or can i do it cleaner
[16:09] <nacc> rbasak: yes, i can adjust that, similar to the default for `usd import` having been changed for now
[16:11] <nacc> rbasak: and yes, re: consistency, let's talk about that
[16:12] <smoser> huwjr, you did an ubuntu installation and you do not have virtio devices in the image that came out ? is that what i'm understanidng ?
[16:12] <huwjr> fixed it :)
[16:22] <rbasak> nacc, smoser: got time to chat about it in the next 90 minutes or so? Or do you want to leave it today? I'm not being particularly productive today. Everything I tried to do recently got sidetracked by some development bug or other :-/
[16:23] <smoser> i can chat for some time, but not too long.
[16:26] <coreycb> ddellav, running tests serially didn't help. i'm guessing there's a dependency difference.
[16:28] <ddellav> coreycb i don't understand why it's failing for you but not on my local schroot. It's fully updated.
[16:28] <nacc> rbasak: i don't particularly care about rbasak yes
[16:29] <nacc> rbasak: bah! typed two lines :/
[16:29] <nacc> rbasak: i have time! :)
[16:29] <rbasak> smoser, nacc: https://hangouts.google.com/hangouts/_/canonical.com/git?hl=en-GB&authuser=0
[16:29] <rbasak> :-)
[16:35] <zul> coreycb: ply is still running atuopkgtest
[16:35] <coreycb> zul, ok
[16:40] <foo> Any idea why I'm seeing this? https://bpaste.net/show/fd925ff31e6d - I can't run apt-get update, it always dies with Killedg package lists... 87%
[16:53] <tarpman> foo: check dmesg? wonder if you have enough memory
[16:59] <foo> tarpman: actually, that's very likely what it is - good idea.
[16:59] <foo> This time it said this: FATAL -> Failed to fork.
[16:59] <foo> Yup. Nov 18 16:59:04 apertureps kernel: [2154877.507478] Out of memory: Kill process 18287 (apt-check) score 106 or sacrifice child
[16:59] <foo> Nov 18 16:59:04 apertureps kernel: [2154877.511963] Killed process 18287 (apt-check) total-vm:114360kB, anon-rss:53392kB, file-rss:1204kB
[17:00] <foo> tarpman: Here's free -m : Nov 18 16:59:04 apertureps kernel: [2154877.507478] Out of memory: Kill process 18287 (apt-check) score 106 or sacrifice child
[17:00] <foo> Err, whoops.
[17:00] <foo> tarpman: Here's memory: http://screencast.com/t/lRmz78Ya2 - I wonder what my options now. I only have postgres and nginx on this server
[17:02] <tarpman> 66 MB free is not very much
[17:02] <tarpman> your options are - tune postgres/nginx to use less memory, or get the VB more memory
[17:02] <tarpman> I consider 66 MB free reasonable on my NAS that has 128MB memory total ;)
[17:03] <foo> tarpman: Ha. This is a VPS at digital ocean, I think it has 512 total. It's still dev, I can shut down postgres and run this - ha
[17:03] <foo> tarpman: Actually, I wonder if allocating more swap would solve this
[17:03] <mybalzitch> more swap is never a good answer
[17:05] <foo> mybalzitch: Why? Well, I suppose that doesn't solve the root of the problem.e g. if postgres is a memory hog and needs to be tuned
[17:05] <foo> (which is possible)
[17:06] <tarpman> foo: adding swap will trade your problem of processes getting killed, for a problem of processes taking minutes to accomplish anything because the disk is thrashing
[17:07] <foo> tarpman: ah, ok - thank you, I'll see about tuning postgres. Thank you
[17:10] <coreycb> ddellav, ok i'll try sbuild-ocata
[17:11] <coreycb> i think that picks up from openstack-ubuntu-testing ppa so that could be the difference
[17:18] <zul> coreycb: im seeing some weirdness with keystone as well
[17:24] <beisner> coreycb, python-glance-store promoted to newton-updates re: https://bugs.launchpad.net/bugs/1604397
[17:27] <beisner> coreycb, neutron promoted to mitaka-updates re: https://bugs.launchpad.net/ubuntu/+source/neutron/+bug/1635369 and
[17:27] <beisner> https://bugs.launchpad.net/ubuntu/+source/neutron/+bug/1606652
[17:29] <coreycb> beisner, thanks
[17:30] <beisner> coreycb, yw.  and newton-staging is on its way to newton-proposed now.
[17:52] <coreycb> beisner, just one more promotion if you can. neutron 2:7.2.0-0ubuntu1~cloud1 from liberty-staging to liberty-proposed.
[17:53] <beisner> coreycb, done
[17:54] <coreycb> beisner, appreciate it
[18:44] <coreycb> zul, was there a python-vmware-nsxlib MIR
[18:44] <coreycb> ?
[18:44] <zul> coreycb: not yet...debian/copyright needs to be fixed
[18:48] <coreycb> zul, ok
[19:16] <Josh__> I'm having some problems with diffrent hosts acquiring locks with nfs.  The server keeps saying lockd: cannot monitor <hostname>  Any ideas?
[19:21] <jge_> hey all, any of you here familiar with tomcat in ubuntu 16.04?
[19:22] <jge_> tomcat complains it cant find a system.properties file
[19:23] <sarnold> Josh__: are all hosts running the lockd or whatever the daemon is? the portmapper daemon? how about firewalls on the machines? or between them?
[19:23] <jge_> the init script points to different paths for catalina_home and catalina_base
[19:23] <jge_> which I think is the problem
[19:23] <nacc> jge_: which tomcat?
[19:23] <jge_> nacc: tomcat 8
[19:23] <nacc> jge_: let me take a look
[19:25] <Josh__> sarnold: Everything should be working on all the other machines.  Someone installed an extra hard drive and nfs stopped working because they messed up fstab.  I got that working agian, and we can all mount or nfs drives, but can't acquire locks..
[19:25] <jge_> nacc: thanks
[19:26] <nacc> jge_: symptomatically, to make sure i can reproduce, what do you run and what do you see? (pastebin is fine)
[19:27] <jge_> nacc: our app looks for a system.properties file in any of the paths definied in common.loader=... inside catalina.properties, I've added a path for {catalina.base}/conf, place my file under /var/lib/tomcat8/conf/ but it fails to find it
[19:28] <jge_> pasting trace now
[19:28] <jge_> nacc: http://hastebin.com/xuhefasafi.lua
[19:34] <nacc> jge_: hrm, the init-script, though, says catalina.base is /var/lib/tomcat8 ?
[19:34] <nacc> jge_: meaning it should find it, i mean
[19:34] <nacc> jge_: can you clarify where you think the bug is?
[19:35] <Josh__> sarnold: any more ideas?
[19:35] <sarnold> Josh__: not really :/ maybe the bringup order of daemons is important? I'd hope they'd have the necessary ordering built into their unit files but you never know..
[19:35] <sarnold> Josh__: if it's fixed across a reboot that might be the case
[19:37] <Josh__> sarnold: Maybe.. Thing is, a week ago, everything was working...  I'm not sure how that stuff would have got messed up.
[19:37] <Josh__> sarnold: I'll check though.  Maybe I'll get lucky!
[19:37] <jge_> nacc: yeah it should find it fine.. I just tested by creating a directory /etc/tomcat7 (our old stack) and it works fine now.. something within the app is hard coded to reference it at that location (my guess)
[19:38] <nacc> jge_: ok :)
[19:38] <jge_> that's lame
[19:38] <jge_> thanks anyway
[19:40] <Josh__> sarnold: it looks like statd and nfs-kernel-server are in the right order.  nfs starts first and then statd does, which i think makes sense.
[19:42] <sarnold> Josh__: that feels right to me too. huh.
[19:43] <Josh__> sarnold: it'll be painful, but maybe I'll just completely reinstall nfs..  Sucks that everyones home drives are mounted there :/
[19:48] <jge_> nacc: devs are saying they dont hard code paths to it anywhere, :(
[19:49] <nacc> jge_: i'm honestly not sure; i don't know too much about tomcat8, but afaict, it shouldn't have any such issues by default -- if you want to file a bug, i can try and debug it later
[19:52] <jge_> nacc: ok, where would I file a bug?
[19:54] <nacc> jge_: `ubuntu-bug tomcat8`
[19:54] <nacc> jge_: launchpad, basically
[19:54] <zul> coreycb: ping
[19:56] <jge_> nacc: what's really confusing me is why would it work when I place it under /etc/tomcat7/conf and not /var/lib/tomcat8/conf .. sounds like it's hard coded somewhere and not a bug :(
[19:56] <zul> coreycb: so keystone weirdness  the database creation fails and i think i know why though....there are some missing files from the database migration stuff, whoever its in the source but not in the egg
[19:56] <nacc> jge_: yes, i'm not convinced it's a bug either :)
[19:56] <nacc> jge_: i assume you menat /etc/tomcat8/conf ?
[19:57] <jge_> nacc: no, tomcat 8 conf directory is just a symlink under /var/lib/tomcat8/conf which points to /etc/tomcat8/
[19:58] <nacc> jge_: right, above you wrote /etc/tomcat7 ?
[19:58] <nacc> jge_: could it possibly be permissions?
[19:59] <coreycb> zul, hmm
[19:59] <zul> coreycb: yeah so i dunno :)
[19:59] <jge_> nacc: so the above is correct /etc/tomcat7 is what we use in production now
[20:00] <coreycb> zul, what files?
[20:00] <zul> coreycb: basically the migrate.cfg files
[20:04] <jge_> nacc: maybe this will help clarify things, currently we deploy straight from a tar.gz to /etc/tomcat7 (catalina_home and catalina_base) end up being the same. Now we deploy tomcat 8 by whatever is on repos, but conf file now ends up being at /var/lib/tomcat8/conf
[20:04] <jge_> so I put my system.properties file in there and it fails to find it..
[20:04] <jge_> my head is dizzy now :\
[20:05] <jge_> conf directory *
[20:05] <zul> coreycb: im bugging the infra people
[20:05] <coreycb> zul, that's interesting, it seems to be new since 2:10.0.0-0ubuntu1, those files were installed with that version
[20:06] <zul> coreycb: yeah
[20:07] <coreycb> zul, MANIFEST.in was dropped
[20:10] <jge_> nacc: damnit, it was perms
[20:10] <nacc> jge_: :)
[20:10] <nacc> jge_: glad you figured it out, that was my best guess
[20:10] <jge_> tomcat's stack trace could be more descriptive
[20:10] <nacc> jge_: always true :)
[20:10] <nacc> jge_: and not just of tomcat :/
[20:11] <jge_> it's innacurate, not found != permission denied
[20:11] <jge_> big difference
[20:12] <jge_> waste of my entire day
[20:12] <jge_> :*(
[20:13] <nacc> jge_: no, but it's the same from unix's perspective (open failed, i'm guessing)
[20:13] <nacc> jge_: it could do more verbose stuff, probably
[20:13] <CarlFK> assuming this is u-s - minimal odroid image from  http://odroid.com/dokuwiki/doku.php?id=en:odroid-c2#software_release wget http://east.us.odroid.in/ubuntu_16.04lts/ubuntu-16.04-minimal-odroid-xu3-20160706.img.xz
[20:13] <CarlFK> how do I set a static IP?
[20:14] <zul> coreycb: hmm....i dunno its probably best for everyone if i go EoD
[20:18] <zul> coreycb: but no migrate.cfg libconfig-model-dpkg-perl
[20:18] <zul> coreycb: but no migrate.cfg http://paste.ubuntu.com/23497265/
[20:19] <coreycb> zul, did you get ahold of anyone upstream?
[20:20] <coreycb> zul, if those files are needed they should be part of the release tarball
[20:20] <zul> coreycb: yeah they said i was on crack
[20:20] <zul> and i agreed
[20:20] <coreycb> zul, well yes, but
[20:20] <zul> coreycb: :P
[20:20] <zul> asshat
[21:06] <CarlFK>  conf that works find if there is a dhcp server on the lan http://paste.ubuntu.com/23497416/
[21:06] <CarlFK> otherwise it takes 5 min to time out.  where can I drop the dhcp client timeout to ... 10 seconds ?
[21:12] <Epx998> Can someone please point me to a working ubuntu14 preseed (efi partitions) file?  having issues with getting one to work properly.
[21:13] <CarlFK> Epx998: I would ask in #ubuntu-installer
[21:23] <ikonia> CarlFK: I think there is a kernal param for it, but I'm not sure
[21:24] <CarlFK> kernal param?!  thats surprising.   unless it maps to some /etc/foo.conf option too.  that's what I am expecting
[21:24] <ikonia> certain things you can pass in
[21:25] <ikonia> I think there is a timeout param thats passed into dhclient
[21:25] <ikonia> the only way to do it is from a kernel boot arg,
[21:25] <ikonia> (you could do it from the config file too if this is not an install)
[21:25] <ikonia> I may have miss read that you where doing this as part of an install
[21:26] <ikonia> if not - just change the dhcp.conf or the args that network manager/systemd unit file pass into dhclient
[21:37] <CarlFK> dhcp.conf .. all I find is etc/samba/dhcp.conf
[21:38] <CarlFK> this is post install.  I have 5 boxes I need to move between 2 lans.  one doen't have a dhcp server
[21:40] <nacc> CarlFK: sorry, can you re-explain what your issue is?
[21:41] <CarlFK> nacc  conf that works find if there is a dhcp server on the lan http://paste.ubuntu.com/23497416/
[21:42] <CarlFK> if I plug that box into a lan (or no lan) with no dhcp server, it takes 5 min to time out
[21:42] <sarnold> if you're moving computers back and forth between networks with dhcp and without dhcp .. this is going to be an interesting thing to solve.
[21:42] <CarlFK> I want to drop that to like 10 seconds
[21:42] <nacc> i would guess /etc/dhclient/dhclient.conf ?
[21:42] <nacc> default is 'timeou3 00;'
[21:43] <nacc> *timeout 300;
[21:43] <nacc> CarlFK: drop that to 10?
[21:43] <sarnold> man 5 interfaces  shows an example of moving between networks and using a script to figure out which network you're on
[21:45] <CarlFK> dhclient.conf looks like what I was looking for
[21:46] <CarlFK> yep.
[21:52] <CarlFK> sarnold: I may look into that some day.  for now  having 2 interfaces is fine
[21:53] <CarlFK> these are more like appliances (for recording conference talks) not my laptop that I take to coffee shops and such
[21:53] <sarnold> at some point I started pining for just running the ip addr and ip route commands by hand again. :/