[01:44] <goddard> having an issue where my wordpress site is unable to connect
[01:44] <goddard> but i can connect with phpmyadmin
[02:00] <Gorian> say, anyone around in here?
[02:01] <Gorian> @coreycb
[06:51] <t4nk050> Hi
[06:53] <t4nk050> I need help with an issue. I have a Dell R610 box with 4 ethernet ports. eth0 & eth1 are bonded and connected to 2 Dell stacked switches. eth2 and eth3 are outbound to a pair of Dell 5424 iSCSI SAN switches.
[06:53] <t4nk050> The SAN switches are in Active-Standby mode.
[06:54] <t4nk050> i.e. only eth2 is in use for the SAN switch connectivity.
[06:55] <t4nk050> Whenever I restart this server, the ping towards the SAN group management IP fails
[06:55] <t4nk050> And it happens intermittently - not at every restart. Only during some restarts
[06:55] <t4nk050> This happened last night as well
[06:55] <t4nk050> Running ifdown eth2 && ifup eth2 fixed the issue.
[06:56] <t4nk050> Any pointers how can I debug the problem?
[07:09] <t4nk050> The box is running 14.04.4 LTS
[07:09] <t4nk050> Ubuntu
[07:21] <sarnold> t4nk050: have you been able to capture e.g. ip link or ethtool output with both 'good' boots and 'bad' boots?
[07:48] <sveinse> What is the easiest approach to disable mdadm and lvmetad from a 16.04 server? The obvious choice is to uninstall them, but I'd really like to have ubuntu-server install, which pulls in both of these tools
[08:09] <Zardoz84> Some one can help me ? I have a really scary situation with a Ubuntu server, with a lot of process that are stuck on "uninterruptible sleep", including a postgress database
[08:10] <sarnold> is it an NFS mount?
[08:11] <Zardoz84> Nope
[08:12] <Zardoz84> I have a EXT4 + BTRFS system over a RAID
[08:13] <Zardoz84> many of the process that are on "uniterruptible sleep" state, don't are touching the BTRFS fs, so I think that isn't a bug of btrfs
[08:13] <Zardoz84> It's a Ubuntu 14.04.4 LTS, that we just did a apt-get upgrade yesterday
[08:14] <sarnold> you may not have options beyond rebooting
[08:14] <sarnold> (I asked about nfs because it's possible to bring back missing mounts through some ugly tricks, which, once you know them, are handy to know...)
[08:14] <Zardoz84> uname -a -> Linux XXXX 3.16.0-70-generic #90~14.04.1-Ubuntu SMP Wed Apr 6 22:56:34 UTC 2016 x86_64 x86_64 x86_64 GNU/Linux
[08:15] <Gorian> don't suppose anyone online can help with making keystone+uwsgi+nginx work?
[08:15] <Zardoz84> I was afraid of reboting, because there two postgres process hanged with the D flag
[08:15] <sarnold> maybe do an 'ubuntu-bug linux' first, to get osme details filed, but there's every chance that -those- processes will also hang D ..
[08:15] <Zardoz84> Ds   jul13   0:00 postgres: checkpointer process
[08:15] <Zardoz84> and the mos worrysome process is this : " D    abr25   7:58 [jbd2/dm-4-8]"
[08:16] <t4nk050> #sarnold I was unable since it occurs in a production envionment and during this time we have to quickly do everything under the sun to bring the interface up.
[08:16] <sarnold> t4nk050: makes sense. maybe prepare a little "ohcrap" script you can run to get the info before you start fixing..
[08:17] <sarnold> Zardoz84: actually, try to grab the /proc/<pid>/stack files from that thing before you reboot
[08:17] <sarnold> Zardoz84: .. and before running the ubuntu-bug command
[08:18] <t4nk050> #sarnold Thinking of preparing a script that automates the restarting the interface using ifupdown and executing the script on post-up in ifconfig?
[08:18] <sarnold> t4nk050: I was just thinking of the ip link and ethtool commands..
[08:20] <Zardoz84> thanks sarnold , trying to reboot now
[08:20] <t4nk050> #sarnold - Sure, I can do it next time. Anything else I can check? kern logs don't hint much, they correctly register ADDRCONF(NETDEV_UP): eth2: link is not ready and  ADDRCONF(NETDEV_CHANGE): eth2: link becomes ready
[08:20] <sarnold> good luck
[08:21] <sarnold> t4nk050: nothing else comes to mind, but ask again in another hour or so, there ought to be more europeans around by then, maybe someone else will have good ideas
[08:21] <t4nk050> #sarnold ok
[08:22] <Gorian> anyone encounter a sigsegv when running uwsgi?
[08:22] <sarnold> Gorian: a sigsegv, that's almost nice, you've got something concrete to work with there :)
[08:23] <Gorian> lol
[08:23] <sarnold> Gorian: if it were me, I'd try strace on the thing and try to find out what it's doing
[08:23] <Gorian> ugh. It's one in the morning, I don't want to debug, I want a quick fix :(
[08:23] <sarnold> that's my generic go-to tool. it's not great fun but it often gives me a decent idea where to start...
[08:23] <sarnold> Gorian: ugh :( sorry to hear it
[08:23] <Gorian> trying to run keystone behind nginx
[08:24] <Gorian> follow the guide... and it breaks horrible
[08:24] <sarnold> Gorian: was it working a few hours ago? maybe look for recently-applied updates, perhaps one went funny.
[08:24] <Gorian> no
[08:24] <Gorian> just installed keystone
[08:24] <Gorian> then followed the guide to setup keystone with nginx
[08:24] <sarnold> ah
[08:25] <Gorian> I could try it with apache and mod_wsgi, per the official ubuntu documentation, just tired hoping to fix it
[08:31] <Gorian> meh
[08:32] <Gorian> runs fine under apache, it's a weird nginx/uwsgi/nginx+uwsgi error
[08:32] <Gorian> idk
[08:52] <ivoks> rbasak: bacula in your ppa fixes the problem with mysql
[08:52] <ivoks> rbasak: bconsole problem can be fixed with adding maximum console connections in bacula-dir.conf
[09:11] <sveinse> Will 16.04 break horribly if I uninstall lvm2 and mdadm from it?
[09:12] <Zardoz84> sarnold: Well, looks that is a RAID problem. I have a "resyncing" message when I reboot on safe mode and I open the system sumary
[09:12] <Gorian> I would guess not, if you aren't using them? Lol, i have no clue. I never use mdadm or lvm
[09:12] <Zardoz84> I thought that I had installed smartmontools to monitoring the health of the hard disks....
[09:13] <sveinse> Gorian: No I'm not. Point is ubuntu-server (which IMHO is a good package to start from on a server) pulls them in
[09:14] <Gorian> http://askubuntu.com/questions/99774/exclude-packages-from-apt-get-upgrade?
[09:14] <Gorian> I've never actually used that, so idk
[09:16] <Gorian> huh. I've never used that package before
[09:16] <Gorian> any reason that you have to use the metapackage instead of just downloading what you want individually?
[09:40] <ivoks> rbasak: in addition to everything, new MySQL doesn't like bacula's default values for datetime
[09:40] <ivoks> rbasak: i figured out that replacing '0000-00-00 00:00:00' with NULL in mysql's install script solves the problem
[09:41] <ivoks> rbasak: make that 'in bacula's mysql install script'
[09:54] <ivoks> rbasak: pretty much, it seams that there should be an upgrade script for mysql 5.6, which would alter all '0000-00-00 00:00:00' with NULL
[10:06] <sveinse> Hmm, when I build my recipe from scratch it fails with "Could not find qmake configuration file linux-oe-g++.", but I find it in ./sysroots/x86_64-linux/usr/lib/qt5/mkspecs/linux-oe-g++, almost next to qmake itself.
[10:24] <rbasak> nacc: ^^
[10:24] <rbasak> ivoks: thanks. nacc was looking at the bacula situation. I think he's got something for me to sponsor.
[10:25] <ivoks> honestly, mysql should do that, because mysql changed
[10:25] <ivoks> but... i'll leave it to you guys :)
[10:30] <rbasak> ivoks: I'm not sure what you mean. Major release bumps break reverse depends all the time. Usually the dependencies keep up, or distro engineers send patches.
[10:30] <ivoks> rbasak: yeah, sorry. so, mysql 5.6 changed and previously valid values for datetime and timestamp are not valid anymore
[10:30] <ivoks> rbasak: one of such values is 0000-00-00 00:00:00
[10:31] <rbasak> Right - because that was a non-sensical value anyway. SQL uses NULL for that.
[10:31] <ivoks> rbasak: that was a value bacula used
[10:31] <ivoks> right
[10:31] <rbasak> MySQL upstream have been cleaning up with a ton of things on the "MySQL is stupid" hitlist :)
[10:31] <ivoks> :)
[10:31] <rbasak> Bacula upstream are active on Launchpad bacula bugs BTW
[10:32] <ivoks> but what i wanted to say is that, once mysql is upgraded, mysql package should check all databases and fix that bs
[10:32] <ivoks> instead of bacula
[10:32] <ivoks> imho
[10:32] <ivoks> i know, i've seen kern commenting
[10:32] <rbasak> I disagree. The bacula package needs to do it. It makes no sense for the mysql package to have knowledge of all reverse depends.
[10:33] <ivoks> that's fine, as i said, i'll leave it to you to do it in the proper place :)
[10:33] <rbasak> OK. I understand what you mean now anyway, thanks :)
[10:33] <ivoks> heh, context switching all day :)
[10:33] <ivoks> sorry for confusion
[10:33] <rbasak> np
[11:21] <t4nk050> I need help with an issue. I have a Dell R610 box running Ubuntu 14.04.4 LTS with 4 ethernet ports. em1 & em4 are bonded and connected to 2 Dell stacked switches. em2 and em3 are outbound to a pair of Dell 5424 iSCSI SAN switches. The SAN switches are in Active-Standby mode. route -n shows http://pastebin.com/YNCNkvNR Whenever I restart this server, the ping towards the SAN group management IP fails. And it happens intermitten
[11:22] <antonispgs> hi guyes, fresh install on a dedi server and after apt-get upgrade this comes up
[11:22] <antonispgs> http://imgur.com/OIlnpKw
[11:22] <antonispgs> what should i do?
[11:22] <t4nk050> ..Only during some restarts. This happened last night as well. Running ifdown em2 && ifup em2 fixed the issue. Any pointers how can I debug the problem? I could not capture ip link and ethtool output since it occurs in a production envionment and during this time we have to quickly do everything under the sun to bring the interface up.
[11:30] <vbotka> antonispgs, "show the differences ..." and try to figure out if you need the local changes
[11:44] <antonispgs> ok thanks
[11:48] <Madhu__> hi
[12:03] <junaidali> Hi everyone, should there be any issues if an lxc is running in parallel to and lxd?
[12:03] <junaidali> I have an lxc which should start a few services when it is created. If I create lxc when there are no lxds on the machine, it works as expected but the services inside lxc doesn't get started when that lxc is created while an lxd is also on the same machine
[12:04] <junaidali> sorry, somehow the message got cut.. :)
[12:04] <junaidali> I have an lxc which should start a few services when it is created. If I create lxc when there are no lxds on the machine, it works as expected but the services inside lxc doesn't get started when that lxc is created while an lxd is also on the same machine
[12:13] <junaidali> I'm new to LXDs, any idea what might be the issue?
[12:16] <meekrat> Running 14.04.4 LTS and a "do-release-upgrade" says "no new release found"
[12:18] <t4nk050> I need help with an issue. I have a Dell R610 box running Ubuntu 14.04.4 LTS with 4 ethernet ports. em1 & em4 are bonded and connected to 2 Dell stacked switches. em2 and em3 are outbound to a pair of Dell 5424 iSCSI SAN switches. The SAN switches are in Active-Standby mode. route -n shows http://pastebin.com/YNCNkvNR Whenever I restart this server, the ping towards the SAN group management IP fails. And it happens intermitten
[12:18] <t4nk050> Only during some restarts. This happened last night as well. Running ifdown em2 && ifup em2 fixed the issue. Any pointers how can I debug the problem? I could not capture ip link and ethtool output since it occurs in a production envionment and during this time we have to quickly do everything under the sun to bring the interface up.
[12:20] <meekrat> etc/update-manager/release-upgrades has prompt=lts also - so everything looks ok
[12:20] <t4nk050> #meercat 14.04 to 16.04 will be offered on July 21st when 16.04.1 is released
[12:20] <meekrat> t4nk050: ah - that explains it.  Thanks
[12:32] <mowthegrass> is there anyway to preseed grub install to specific device / install grub on the device where the installation took place
[12:32] <mowthegrass> by default installer looks at sda
[13:14] <coreycb> ddellav, jamespage: all of our core packages are uploaded for b2 (minus aodh/ceilometer which haven't released yet).  horizon is failing to install but I think I have that figured out, just need to test it.
[13:16] <coreycb> ddellav, jamespage: working through ca backport issues now.  I patched python-cryptography in ca-patches for newton to get past it's backport issue.
[13:18] <jamespage> coreycb, \o/
[13:51] <LaserAllan> hi there, I have 2 NFS mount commands i want ot be done at a startup, where should i put the mount -t nfs server:/path/to/files /mnt/folder
[13:55] <t4nk050> #LaserAllan You can try placing them in /etc/network/interfaces below the interface configuration information. See http://pastebin.com/KBNxCtuS
[13:55] <LaserAllan> t4nk050: With or without sudo?
[13:55] <t4nk050> Without
[13:56] <t4nk050> #LaserAllan just ensure you are writing below the correct interface through which the network share will actually be available
[13:57] <LaserAllan> source /etc/network/interfaces.d/*
[13:57] <LaserAllan> # The loopback network interface
[13:57] <LaserAllan> auto lo
[13:57] <LaserAllan> iface lo inet loopback
[13:57] <LaserAllan> Thats the only thing i hav ein the file so i am writing those two commands under those lines
[14:17] <jayjo> I'm trying to create an ssh tunnel and forward my local port 5555 through a server i have ssh access to. I tried ssh -nNR -L 5555:localhost:5432 jayjo@<server address> and upon trying to connect to this instance I get a "Connection refused"
[14:17] <jayjo> can I get more verbose output, or is there something clearly wrong that I'm trying to do
[14:22] <ddellav> coreycb o/ awesome
[14:30] <setuid> I'm having a bear of a time getting my preseed + kernel append line to get past the initial Language prompt right when the ISO boots.. .once I interactively hit enter on English, and then Enter again on the "Install Ubuntu Server", the rest of the automation works.
[14:31] <setuid> I've tried every option in the preseed file and append lines to force en, en_US, etc. but it's not working
[14:41] <nate_> I'm trying to setup a server using software raid 1. When booting, I get: "error: attempt to read or write outside of disk `hd0'."
[14:42] <sky> I setup a long running cron script... supposed to be fired once a day
[14:42] <nate_> I have 2 disks, used automatic partitioning. Then made two raid devices, then set them to be used as ext4 with a root mountpoint, and the other as swap.
[14:42] <sky> when that time comes, its fired every 60 seconds
[14:42] <sky> I guess this is anacron or something watching the process, terminating and respawning every 60?
[14:44] <nate_> sky: You're using 0 0 * * * ?
[14:44] <sky> I'm using * 2 * * *
[14:44] <sky> it only fires once per day. but then when it starts, it keeps retrying every 60 seconds
[14:45] <nate_> That would execute every minute on the 2nd hour.
[14:45] <nate_> You probably want 0 2 * * *
[14:45] <sky> oh
[14:45] <sky> doh
[14:45] <sky> thanks
[14:45] <nate_> sky: no problem.
[14:53] <caribou> rbasak: does "Unable to read /etc/mysql/my.cnf.migrated" upon upgrade of mysql-server-5.7 rings a bell ?
[14:53] <caribou> rbasak:  I know you were hacking at it recently
[14:53] <caribou> rbasak: just want to know if  I should open a bug about it
[15:03] <AndyWojo> Is there a Ubuntu specific cloud channel? For OpenStack etc?
[15:03] <rbasak> caribou: that's bug 1602963
[15:04] <rbasak> caribou: the fix has landed. I've set the bug to Fix Released.
[15:04] <caribou> rbasak: thanks for the info; thought it'd be better to check with you first
[15:05] <caribou> rbasak: most probably not hit my mirror yet
[15:07] <setuid> Can someone set eyes on this? http://paste.ubuntu.com/19502064/
[15:08] <setuid> I'm still getting prompted right before the boot selection menu, to choose a language
[15:08] <setuid> When I choose "English", then choose "Install Ubuntu Server", the rest of the unattended/automated install works as expected
[15:09] <setuid> My append line looks like this:
[15:09] <setuid> file=/cdrom/preseed/unattended.seed auto=true priority=critical debian-installer/language=en debian-installer/locale=en_US kbd-chooser/method=en localechooser/preferred-locale=en_US.UTF-8 console-setup/ask_detect=false console-setup/layoutcode=us netcfg/get_hostname=maas-test netcfg/get_domainname=maas initrd=/install/initrd.gz --
[15:11] <nacc> setuid: i'm here too
[15:11] <nacc> setuid: let me read scrollback
[15:12] <setuid> I'm 99.8% of the way there, just the first two interactive prompts
[15:12] <setuid> But oddly, this works on 14.04's version of virt-install, but not on 16.04's version (1.3.2)
[15:12] <nacc> ivoks: could you test the version in a PPA?
[15:13] <setuid> I just plucked the relevant entries out of questions.dat after doing an interactive install, they didn't work when put into the preseed and append line
[15:13] <nacc> setuid: heh, was just helping someone else yesterday with random virt-install stuff (not playing nice with cobbler); let me look
[15:13] <setuid> nacc, the version of virt-install? Sure, one sec
[15:13] <setuid> cobbler and chef are next on my list, actually ;)
[15:13] <setuid> I need to be able to rebuild the whole openstack stack, then maas + juju using chef + cobbler
[15:14] <setuid> There doesn't appear to be a ppa for virtinst
[15:14] <nacc> setuid: cobbler doesn't play great with virt-install (and ubuntu) afaict, just fyi
[15:15] <nacc> setuid: seems to be assumptions about what kind of --location that virt-install gets which cobbler doesn't satisfy
[15:15] <nacc> setuid: you say it works on 14.04's virt-install but not 16.04; installing the same ubuntu in each (e.g., 14.04 as the victim OS?)
[15:15] <setuid> I wonder if jumping to uvtool would obviate those gaps
[15:16] <setuid> Yes, installing trusty using virt-install on native 14.04, and installing trusty using 1.3.2 virt-intall on native 16.04
[15:16] <setuid> Trying to change as little as possible, before I start playing with different guest vms
[15:17] <nacc> setuid: hrm, that's an unexpected issue, i'd guess
[15:17] <setuid> 14.04 ships 0.600.4, 16.04 ships 1.3.2
[15:17] <nacc> setuid: *if* the preseed works with 14.04 and doesn't work with 16.04
[15:17] <nacc> as the preseed shouldn't be being parsed at all by virt-install, only by the installer, aiui
[15:17] <nacc> anyone else, cmiiw
[15:17] <setuid> Ok, so it's the append lines then?
[15:18] <nacc> setuid: are they identical between the two cases?
[15:18] <setuid> I've stripped out all the irrelevant pieces (netcfg/* values for example) to try to simplify the debugging
[15:18] <ivoks> nacc: which ppa?
[15:18] <nacc> yeah
[15:18] <nacc> ivoks: ppa:nacc/bacula
[15:18] <ivoks> nacc: i'd like to avoid major upgrades
[15:18] <nacc> ivoks: it's just the fixes right now
[15:19] <nacc> ivoks: to the existing xenial & yakkety packages
[15:19] <ivoks> 7.4.1~dfsg-1.1~ppa8.is.actually.7.0.5+dfsg-4.1~ppa3 ?
[15:20] <ivoks> if i install that, i'll have issues going back to real 7.0.x
[15:20] <nacc> ivoks: let me build you nicer versioned one
[15:20] <nacc> one sec
[15:20] <ivoks> but i've already altered my database
[15:20] <nacc> ah
[15:20] <nacc> ok, nm
[15:21] <nacc> i'll try and reproduce it locally
[15:21] <ivoks> install bacula on 14.04 and upgrade
[15:21] <nacc> yep, understood
[15:21] <ivoks> if you just install it on 16.04, i'm not sure you'll hit the problem with database
[15:22] <ivoks> on first sight, everything runs, but problem appears only when you start running a backup process
[15:22] <nacc> ivoks: you do, actually, but we've fixed that in my ppa build
[15:22] <nacc> ivoks: which is a testbed for the sru
[15:22] <ivoks> i don't recall if this machine was 14.04->16.04, or 12.04->14.04->16.04
[15:23] <ivoks> i could find out thou :)
[15:24] <ivoks> 2.6.24 kernel
[15:24] <ivoks> sounds like 12.04
[15:24] <nacc> yeah
[15:24] <nacc> although 12.04 right now is on 3.2.0.106.122
[15:24] <nacc> so that might actually be an un-updated 12.04.0 ?
[15:25] <ivoks> ha?
[15:25] <ivoks> now
[15:25] <ivoks> it was installed as 12.04
[15:25] <ivoks> upgraded to 14.04
[15:25] <ivoks> and yesterday upgraded to 16.04
[15:25] <nacc> 2.6.24 isn't in the archives at all
[15:25] <nacc> afaict
[15:25] <ivoks> maybe it was 10.04
[15:25] <nacc> which hasn't been supported for ... 2 years? :)
[15:25] <ivoks> that's irrelevant
[15:26] <ivoks> 2.6.24 was 8.04
[15:26] <nacc> it's relevant for what upgrade paths i need to test
[15:26] <ivoks> so, it was installed as 8.04 and was upgraded over time
[15:26] <ivoks> yesterday it was upgraded from 14.04 to 16.04
[15:27] <setuid> I wonder if this is a 1.3.2 parsing bug, ignoring values passed in
[15:27] <ivoks> it's pretty cool nothing more than that broke :D
[15:27] <nacc> setuid: it could be (i'd check upstream to see if already reported?)
[15:27] <nacc> setuid: in my experience the only value i've needed to preseed is debian-installer/locale
[15:27] <t4nk050> I need help with an issue. I have a Dell R610 box running Ubuntu 14.04.4 LTS with 4 ethernet ports. em1 & em4 are bonded and connected to 2 Dell stacked switches. em2 and em3 are outbound to a pair of Dell 5424 iSCSI SAN switches. The SAN switches are in Active-Standby mode. route -n shows http://pastebin.com/YNCNkvNR Whenever I restart this server, the ping towards the SAN group management IP fails. And it happens intermitten
[15:27] <nacc> setuid: if you set more than that, it will start prompting
[15:27] <t4nk050>  Only during some restarts. This happened last night as well. Running ifdown em2 && ifup em2 fixed the issue. Any pointers how can I debug the problem? I could not capture ip link and ethtool output since it occurs in a production envionment and during this time we have to quickly do everything under the sun to bring the interface up.
[15:28] <nacc> setuid: but just setting locale sets language, country & locale, iirc
[15:29] <ivoks> t4nk050: you have two routers for same network over different nics with the same metric?
[15:29] <ivoks> s/routers/routes/
[15:29] <setuid> nacc, The rest of the automated install works without prompting, it's just those first two intial values... the first language input and the boot selection
[15:30] <ivoks> t4nk050: if i understood everything correctly, i'm actually surprised it works at all
[15:30] <ivoks> t4nk050: sounds like you should have an active-passive bond on your ubuntu box
[15:31] <ivoks> so bond0, i presume lacp, and bond1 active-passive
[15:31] <nacc> setuid: right, understood that part
[15:33] <setuid> Something isn't right... because questions.dat's values are used to successfully install interactively, but those same values don't work unattended.
[15:33]  * setuid boggles
[15:39] <nacc> setuid: what is questions.dat ?
[15:40] <setuid> nacc, When you install a machine, automated or otherwise, the answers to every prompt are put into /var/log/installer/cdebtool/questions.dat
[15:40] <setuid> They're almost identical to preseed format
[15:48] <setuid> everything I've read says that langauge=en should be sufficient in an append line to pass the intial dialog
[15:49] <nacc> setuid: sorry, internet hiccup at home
[15:50] <nacc> setuid: have you tried just passing debian-install/locale=en_US ?
[15:50] <setuid> Yes
[15:50] <nacc> *debian-installer
[15:51] <setuid> append   file=/cdrom/preseed/unattended.seed debian-installer/locale=en_US console-setup/ask_detect=false netcfg/get_hostname=maas-test netcfg/get_domainname=maas initrd=/install/initrd.gz --
[15:51] <setuid> that's what I'm using now
[15:51] <setuid> with all the appropriate same options turned on in the preseed
[15:51] <setuid> 1 second after boot, I get the language selection menu, before the "Install ubuntu Server" option
[15:51] <nacc> setuid: can you pastebin your preseed file?
[15:51] <setuid> It's obviously not the language, locale, country, etc.
[15:52] <setuid> yep, oen sec
[15:52] <nacc> setuid: i've never cdrom installed with preseed, only pxe; and i've never seen the 'install ubuntu server' dialog :)
[15:52] <setuid> http://paste.ubuntu.com/19506925/
[15:53] <setuid> Ok, let me try using --location and my local mirror path to the install tree, vs. using the iso
[15:53] <kidn3ys> Hello, I have a dvr package that is installed that is supposed to pull the MAC of the server for licensing purposes. Anyone have any suggestions on how I might be able to determine HOW it's doing that? Even fresh installs of the package result in a mac of all zeros.
[15:55] <patdk-wk> kidn3ys, there could be hundreds of ways
[15:55] <patdk-wk> no idea what this mythical dvr package is using
[15:55] <setuid> Using --location and --extra-args, passing in the appropriate append values, works. No initial prompt for language.
[15:56] <nacc> setuid: interesting
[15:56] <setuid> But when using --cdrom (omitting --location and --extra-args), it prompts before the boot selection
[15:56] <nacc> setuid: so it does seem to be some weird interaction between virt-install ad the args?
[15:56] <nacc> setuid: if you do --cdrom + --extra-args?
[15:56] <nacc> *what if
[15:56] <setuid> you can't use --cdrom with --extra-args, it exits because they're incompatible options
[15:56] <nacc> ah
[15:56] <nacc> setuid: sorry, i've not used it much :)
[15:56] <setuid> I had to remaster the iso file to inject the append lines into isolinux/txt.cfg
[15:57] <jonah> Hi just wondered if anyone can help. I'm running a few ubuntu servers but keep having to manually change file permissions to get things to work. When I ftp a CMS it won't load, so I manually change all the files to 0644 and directories to 0755 and it works great, the installer runs. But then from within the CMS backend if I try install a plugin or addon etc it often sets the permissions wrong for the addon and again I have to come out and
[15:57] <jonah> manually set permissions... Does anyone have any tips on this so that default perms are correct when writtn by a normal (not root) user from apache or CMS etc?
[15:57] <jonah> Thanks for any help
[15:57] <setuid> jonah, Did you ask the relevant CMS channel?
[15:57] <setuid> jonah, Probably ownership, not permissions
[15:58] <setuid> Also, 0755 will probably open you up to remote exploits, so be careful (esp. using Drupal or Wordpress)
[15:58] <kidn3ys> patdk-wk: that's what I figured. I tried stracing it on startup but didn't have much luck. =/
[15:58] <jonah> setuid: it's not one particular CMS, it's pretty much any of them. So it is more somethign with my ubuntu server setup rather than down to a CMS. FTP also acts strangely with perms too, doesn't set them right on uploading stuff...
[15:59] <setuid> jonah, ownership, probably... not permissions
[15:59] <setuid> If you're serving this over http, the files have too be readable by your webserver user, probably www-data, NOT the user uploading the files
[16:03] <setuid> nacc, I'd rather do this with an iso, because it's 50% faster than doing it over http (even to my local Ubuntu mirror on the same LAN segment)
[16:03] <jonah> setuid: it seems to be perms though, as when I ftp in after a cms installs the ownership is correct, but then I have to change all the perms...
[16:04] <setuid> jonah, "have to change all the perms"?
[16:04] <setuid> jonah, Where are the files coming from? Windows? Or Linux -> Linux?
[16:05] <setuid> 5m to build with an iso, 10m to build over http
[16:06] <nacc> setuid: i guess that makes sense, esp. for virt installs potentiall
[16:06] <setuid> nacc, I wonder... if there's a boot menu that happens with an iso that doesn't happen when pulling a remote initrd
[16:06] <setuid> If I'm remastering, I can rip that menu out... or find the right append hook to pass through it
[16:07] <setuid> Maybe using a kickstart file + preseed?
[16:07] <setuid> can virt-install use a kickstart file?
[16:08] <nacc> setuid: well, it's not wehther virt-install can or not; it's whether ubuntu can, and it can, sort of
[16:08] <setuid> True
[16:08] <nacc> https://help.ubuntu.com/community/KickstartCompatibility
[16:08] <setuid> kickstart is a dead-end, that's RH specific
[16:08] <nacc> well, there's a compat layer
[16:08] <nacc> for some things
[16:08] <nacc> and you can preseed what isn't in the kickstart-compat
[16:09] <setuid> I shouldn't have to jump through these hoops to get a completely unattended install
[16:09] <nacc> setuid: right, i'm thinking it's an iso specific thing
[16:09] <nacc> setuid: i think very few people preseed iso installs :)
[16:09] <setuid> Remastering all of my ISOs to support unattended is also a hassle
[16:09] <setuid> How does anyone using a cloud to stand up new images, do it? If I'm building 500 machines, that's 500 copies of the same packages coming through the lan
[16:10] <setuid> It's logical to do it from the same, local source iso
[16:10] <nacc> setuid: using a caching proxy?
[16:10] <setuid> The preseed works with the iso everywhere except those first 2 prompts
[16:10] <setuid> hrmph!
[16:11] <nacc> https://help.ubuntu.com/community/Installation/UnattendedCD ??
[16:11] <nacc> hrm
[16:11] <nacc> heh, it says locale should work too
[16:12] <setuid> yep, but it doesn't :(
[16:12] <nacc> setuid: have you tried bumping priority?
[16:12] <nacc> i wonder though if this is all before d-i is really running
[16:12] <setuid> priority=critical, so far
[16:12] <setuid> hrm, their preseed syntax is different
[16:12] <nacc> for which?
[16:12] <setuid> Mine comes from the actual iso (I injected it)
[16:12] <setuid> theirs is on a remote http
[16:12] <nacc> yeah
[16:13] <nacc> it shouldn't matter, but i wonder if it puts it in a different mode when it knows it's getting stuff off the network
[16:13] <setuid> I can change that, my preseed is identical in both places
[16:14] <nacc> setuid: in the installed system have you tried runing `debconf-get-selections --installer` to see if there's any relevant entries that maybe aren't being set?
[16:15] <jonah> setuid: well i upload a cms from linux, but it doesn't matter where it comes from. The problem is more when the CMS is working and running and then from within it I install an addon in the backend, this then has the wrong permissions and I have to go in via ftp or other means and change them...
[16:15] <setuid> jonah, It absolutely matters where it comes from, because file permissions from Windows -> Linux will not be what you expect them to be
[16:15] <setuid> Transferring from Linux -> Linux will retain source permissions, Windows -> Linux will not
[16:16] <setuid> nacc, Let me try that...
[16:16] <ivoks> jonah: right
[16:16] <jonah> setuid: but I'm talking about after you have transferred the files and set them all up right - once then logged in to a website apache is then setting them wrong...
[16:16] <ivoks> jonah: this depends on how apache+php is configured
[16:17] <jonah> ivoks: hi thanks
[16:17] <ivoks> if it's using php as a module in apache
[16:17] <ivoks> then all files created by your web page will have www-data uid/gid
[16:17] <jonah> ivoks: currently i believe I use fastcgi running as the domain owner
[16:17] <setuid> jonah, The CMS may be reconfiguring them to ensure consistency/security
[16:17] <ivoks> fastcgi gives more options
[16:18] <ivoks> when running as fcgid, all php files are executed with your uid
[16:18] <ivoks> and therefore create files with your uid as owner
[16:19] <ivoks> what kind of fixes do you have to do over ftp?
[16:19] <jonah> ivoks: yeah using fcgid, and it is setting the files as the right owner so it is running as domain owner correctly, just for some reason the perms are always 0777 rather than 0755 for example on directories
[16:20] <ivoks> umask?
[16:20] <jonah> ivoks: i have to just ftp in and change all files to 0644 and directories to 0755
[16:20] <jonah> ivoks: yeah i thought that, but I don't want to change anythign that will break security or root perms etc so just wasn't sure
[16:20] <ivoks> when you create a new file over FTP, what mask does it have?
[16:21] <jonah> ivoks: when I create a test file with ftp it has no owner or perms! in filezilla it just says xxx for the numerica value and owner.group is blank...
[16:21] <ivoks> that's just impossible. :)
[16:21] <setuid> Sounds like a filezilla preference is incorrect
[16:22] <jonah> ivoks: i read a bit about unmask but haven't delved into anything as I don't want to start breaking system processes etc
[16:22] <setuid> 0111 perms?
[16:22] <setuid> x-x-x?
[16:22] <jonah> ivoks: ah hang on, I clicked refresh on the view and it does have default perms!! the owner is right and group as the domain user
[16:23] <jonah> ivoks: but the perms of the file are 664
[16:23] <ivoks> and mask of the file?
[16:23] <ivoks> ok
[16:23] <ivoks> that means your umask is 002
[16:23] <ivoks> which is ok
[16:23] <jonah> ivoks: when i do a folder the perm is 775
[16:23] <ivoks> are you sure files created by webapp are 777?
[16:24] <jonah> ivoks: well for example I just installed prestashop, the installer ran half way but then must of been creating new files and failed. When I ftped in I had to change all files to 0644 and dirs to 0755, then reran the installer and it completed.
[16:25] <jonah> ivoks: when I then go in to the backend and try install an addon, same issue, that addon then doesn't work or the page for it doesn't work until ftp in and update perms
[16:26] <jonah> ivoks: but that's just an example, it's the same with drupal, modx and most cms I've tried. i think wordpress was ok but most stuff seems to break itself with the 664 files and 775 folders.
[16:26] <jonah> ivoks: but my server seems to like 0644 and 0755!! haha
[16:26] <ivoks> that doesn't make sense
[16:26] <jonah> ivoks: crazy huh!?
[16:26] <ivoks> and files created by cms are owned by your user?
[16:27] <ivoks> not by www-data or something like that?
[16:27] <jonah> ivoks: yeah
[16:27] <jonah> ivoks: let me double check that...
[16:27] <coreycb> beisner, jamespage: can you tell what's wrong with the promotion of the packages in staging here? http://reqorts.qa.ubuntu.com/reports/ubuntu-server/cloud-archive/liberty_versions.html
[16:27] <ivoks> the behavior you are describing would make sense if you use mod-php
[16:28] <jonah> ivoks: yeah looks like all the files are the right user
[16:29] <coreycb> beisner, jamespage: neutron-lbaas 1:2015.1.4-0ubuntu2 is ready to promote to kilo-proposed when you get a chance
[16:29] <nacc> ivoks: curious; without the PPA fixes for bacula, bacula-dir doesn't even run with mysql (afaict)
[16:29] <ivoks> nacc: correct
[16:30] <nacc> ivoks: so you just noticed manually the tables were wrong?
[16:30] <ivoks> nacc: so, i installed bacula from rbasak's ppa
[16:30] <nacc> ivoks: ah ok
[16:30] <ivoks> jonah: makes no sense
[16:30] <ivoks> jonah: you must be missing something
[16:30] <ivoks> :)
[16:31] <ivoks> jonah: fcgid will create files owned by you, in a directory owned by you
[16:31] <ivoks> jonah: there's nothing that would prevent that
[16:31] <ivoks> jonah: mod-php will require 777 on the dir, to be able to put files in it
[16:31] <ivoks> since dir is owned by you, and files are owned by www-data
[16:32] <ivoks> there must be something else
[16:32] <ivoks> i'd love to help, but it's 6:30pm
[16:32] <ivoks> and it's friday :)
[16:32] <Sling> ivoks: definitely not 777
[16:32] <ivoks> Sling: it would
[16:33] <Sling> files should not be owned by www-data
[16:33] <ivoks> mod php would require user to allow apache user to write
[16:33] <ivoks> and the only way to do it is to create rwx for all
[16:33] <ivoks> unless acl
[16:33] <Sling> so, create a group for that
[16:33] <Sling> put www-data in the group
[16:33] <Sling> user in the group
[16:33] <Sling> voila
[16:33] <patdk-wk> apache user? don't you mean www-data user
[16:33] <ivoks> he's a user, not a root on the system
[16:33] <Sling> patdk-wk: www-data is just the user configured on some distro's
[16:33] <Sling> 'apache user' is meant in general
[16:34] <patdk-wk> personally, I would just not use mod-php, unless the server is only used by one php application
[16:34] <ivoks> i love mod-php, because it's easy
[16:34] <patdk-wk> switch to like php-fpm, and run each php site as a different user
[16:34] <ivoks> but
[16:34] <Sling> mod_php is a pretty lame way of serving php these days :)
[16:34] <ivoks> fpm is the way to go
[16:35] <setuid> nacc, there's no command debconf-get-selections, thouogh there is a debconf-set-selections command
[16:35] <setuid> did you mean dpkg --get-selections?
[16:36] <maxb_> There is a debconf-get-selections, but it's in a different package to debconf-set-selections
[16:36] <setuid> Ok, not part of the default install, Ill search
[16:37] <nacc> setuid: sorry, yeah, it's in debconf-utils, iirc?
[16:38] <setuid> FOund it, what am I looking for in the output?
[16:38] <nacc> setuid: so if you run that, e.g., `debconf-get-selections --installer > preseed.cfg` it will generate an exact preseed syntax output fo every installer selection
[16:38] <nacc> now, you shouldn't use that as a preseed
[16:39] <nacc> but you can parse through it, or feel free to pastebin, and we can see if there's something else we could set
[16:42] <nacc> ivoks: reproduced, thanks
[16:44] <setuid> nacc, Seems to be non-chronological
[16:44] <nacc> setuid: yeah it's just everything (iirc)
[16:45] <setuid> I think I found it
[16:47] <nacc> setuid: what's the line?
[16:48] <setuid> d-i     localechooser/languagelist      select  en
[16:48] <setuid> but it didn't work as an append
[16:48] <setuid> localechooser/lanuagelist=en
[16:49] <setuid> debconf/language=en also fails
[16:49] <nacc> setuid: just to be sure, are you passing 'auto-install/enable=true' ?
[16:50] <setuid> auto=true priority=critical
[16:50] <nacc> k
[16:53] <setuid> http://paste.ubuntu.com/19513727/
[16:53] <nacc> setuid: i have no idea if this is still true, but: http://unix.stackexchange.com/questions/196874/prevent-language-selection-at-ubuntu-installation
[16:53] <setuid> Tried all of those, both as append and in preseed
[16:54] <nacc> setuid: yeah, so i'm thinking it's an isolinux thing
[16:54] <nacc> not a preseed thing
[16:54] <setuid> I tried this last night, but it just leaves me with a selection menu of 1 langauge, which I still have to hit Enter on
[16:55] <nacc> setuid: did you try with the timeout?
[16:55] <setuid> Right, it happens well before the preseed is parsed
[16:55] <nacc> setuid: and with isolinux/lang (not langlist)
[16:56] <nacc> e.g., from http://askubuntu.com/questions/122505/how-do-i-create-a-completely-unattended-install-of-ubuntu/122506#122506
[16:56] <setuid> AH!
[16:56] <setuid> echo en > isolinux/lang gets rid of the initial language selection
[16:56] <nacc> :)
[16:56] <setuid> now I'm stuck at the grub menu asking me to choose a boot option
[16:56] <setuid> *facepalm*
[16:56] <nacc> grub menu or isolinux menu?
[16:57] <setuid> the boot selection menu, top enter ("Install Ubuntu Server") is highlighted
[16:57] <setuid> hitting enter, proceeds with the rest of my unattended install
[16:57] <nacc> ok, that's isolinux
[16:57] <nacc> 'txt.cfg' file
[16:57] <nacc> so i *think* it's just a matter of (since it's the default0 putting a timeout or something in
[16:58] <setuid> Yep, probably a timeout
[16:58] <setuid> I'll dig into that... thanks! THis has been hours of searching
[16:58] <nacc> setuid: i think you just put TIMEOUT 1
[16:58] <nacc> or something
[16:59] <nacc> http://www.syslinux.org/wiki/index.php?title=SYSLINUX
[16:59] <setuid> No syslinux on an ubuntu installer cd
[16:59] <nacc> "A timeout of zero will disable the timeout completely. The default is 0."
[16:59] <nacc> isolinux is syslinux compatibile, afaict
[16:59] <setuid> Prbably something in isolinux/txt.cfg
[16:59] <nacc> same syntax
[16:59] <nacc> setuid: so you can either remaster the iso to only have the one option
[16:59] <nacc> or put a non-zero timeout
[17:04] <setuid> Got it... isolinux/isolinux.cfg
[17:04] <setuid> Tried a few other files in there first, based on google results
[17:04] <setuid> http://unix.stackexchange.com/questions/32243/how-do-i-configure-syslinux-to-boot-immediately
[17:04] <setuid> So it look like it's working, although it still hits my local mirror for pulling packages, not the iso it booted from
[17:05] <nacc> setuid: it will use both, presumably
[17:07] <setuid> I need to blog this up, because this is a culmination of 2 days of testing across several hundred builds to vet this out, and most of the details on google aren't complete, or are out of date
[17:07] <setuid> Now I'll try this same process on my 14.04 box and see if it continues to work, or if I have to port the commands and write up two versions
[17:08] <nacc> setuid: gl! please do blog about it, it's good to hear you're finally successful!
[17:12] <setuid> With this method, I can automate the build of a functional maas server from a local iso or a remote http repo, in 5m or 10m
[17:15] <Pici> /21/25
[17:52] <nacc> rbasak: re: bacula + mysql-server install ordering. I think the issue is the configuration step of bacula is running just after the configuration step of mysql-server, but mysql-server hasn't finised installing yet so bacula fails to find it running. Is it possible for one package's dependency to be ordered in such a way that the dep has to be fully installed first?
[18:21] <rbasak> nacc: there's Pre-Depends, but usually dpkg tries to ensure the dependency's postinst has run first anyway.
[18:21] <rbasak> nacc: do you know why mysql-server hasn't finished installing even though the postinst has finished?
[18:21] <nacc> rbasak: maybe it just hasn't been started yet?
[18:21] <nacc> rbasak: it's not clear to me
[18:22] <rbasak> It should be started and running by the time the postinst is done. If it's not, I think that's a bug.
[18:22] <nacc> rbasak: ok, i'll trya nd reproduce it again, trying to fix the underlying bug first
[18:22] <rbasak> eg. "apt-get -y install mysql-server mysql-client && mysql -u root ..." should always work
[18:33] <nacc> rbasak: i'll keep digging and let you know
[18:33] <nacc> rbasak: switching tacks, re: puppet bug in 16.04; would appreciate if you could review those, as users have been getting restless. I'll work on getting upstream fixed properly before 16.10 comes out
[19:03] <beisner> hey coreycb - looking re: liberty staging
[19:26] <nacc> rbasak: ok, so i think what's happened is that "Setting up mysql-server-5.7 (5.7.13-0ubuntu4) ..." hasn't finished yet when dbconfig-common runs for bacula-director-mysql; and it tries to connect immediately and fails
[19:30] <beisner> coreycb, pushed staging to liberty-proposed and kilo-proposed.  lmk if the report doesn't reflect that within +1 hr.  thx!
[19:31] <coreycb> beisner, thx!
[19:31] <beisner> coreycb, yw sir
[19:32] <setuid> nacc, Thanks for the help earlier, I'm building out some bash templates, so I can gen the iso with some replacement vars and spin this up quickly without having to touch the preseed or isolinux files
[19:32] <nacc> setuid: seems reasonable :)
[19:33] <setuid> The less places I have to twiddle the netcfg vars, the better
[19:33] <setuid> I've got the build and teardown automated, just not the replacement vars for internal vs. external network and iso vs. http, but that's easy
[20:16] <rattking> Hey all I am having some secure apt issues with my personal apt mirror. how can I check what key the packages were signed with?
[20:17] <sarnold> packages aren't signed; the InRelease files are signed, and those have hashes of Packages* files, that have hashes of the packages
[20:20] <rattking> thanks for the info, I am not seeing a .gpg for my mirror in /var/lib/apt/lists so thats probably the issue
[20:21] <kidn3ys> Is there a simple way in /etc/fstab to mount an iscsi volume?
[20:21] <sarnold> the detached signatures on the Release files in "old releases" leads to race conditions that, across as many users as ubuntu has, mean someone saw the race and got errors from apt, on a daily basis
[20:21] <rattking> my aptly mirror is fine, but I have been trying to use fai-mirror to make a mirror of every installed package to bring over to some offline systems
[20:22] <sarnold> so the new approach is the InRelease file and the by-hash/ directories, to avoid the races
[20:23] <rattking> nice! does this new approach apply to precise? I didnt see any InRelease files just Release and Release.gpg
[20:25] <kidn3ys> It seems that the entry I added to fstab is attempted before the initator comes up =/
[20:27] <sarnold> rattking: InRelease files are even in precise and newer; the by-hash directory is in xenial and newer
[20:28] <sarnold> kidn3ys: isn't that a common issue for networked systems, that the remote peer may be offline when booting?
[20:28] <rattking> thanks for the info, I now know whats missing!
[20:30] <kidn3ys> sarnold: its not though =/
[20:30] <kidn3ys> sarnold: it looks like whatever is in fstab executes before the iscsi initiator starts on the local system
[20:30] <sarnold> kidn3ys: sigh. I ofcourse screwed up initiator vs target.
[20:31] <sarnold> I know that those terms make sense but .. sometimes I screw it up.
[20:31] <kidn3ys> sarnold: I confuse them too. I think I found something... '_netdev' seems to be the key.
[20:32] <sarnold> kidn3ys: are you on xenial? or pre-xenial?
[20:33] <kidn3ys> sarnold: pre, I think. 14.04
[20:34] <sarnold> kidn3ys: hrm. I'm not sure what to suggest there, except perhaps changing the line to "noauto", and add a manual mount sysv init script.. and make sure it's numbered to start after the iscsi initiator, if the iscsi initiator also starts via a sysv init script
[20:36] <kidn3ys> sarnold: adding '_netdev' as an option in fstab seems to mount it on boot but the 'mounting network filesystems' task shows as 'failed' during boot now.
[20:37] <sarnold> kidn3ys: ooh. did that add any new error messages that might indicate what to work on next?
[20:41] <kidn3ys> sarnold: not seeing anything =/
[20:42] <sarnold> kidn3ys: dang
[20:42] <kidn3ys> sarnold: pretty new to linux, i was looking in dmesg -- is there a better place to look?
[20:43] <sarnold> kidn3ys: /var/log/upstart/ .. maybe look for a 'mountall' or a file for your initiator..
[20:43] <sarnold> kidn3ys: or /var/log/syslog (or maybe it's called 'messages' these days, I can't keep track...)
[20:46] <kidn3ys> sarnold: mount: special device /dev/sdb1 does not exist
[20:46] <kidn3ys> and then mountall: mount /mnt/edvr/1 [840] terminated with status 32
[20:50] <kidn3ys> I see what I did. I had '/dev/sdb1' in the device name field instead of the UUID
[20:51] <kidn3ys> sarnold: that did it, thanks for your help :)
[20:52] <sarnold> kidn3ys: excellent :D the uuid and the _netdev combined?
[20:52] <kidn3ys> yep
[20:52] <sarnold> awesome :D
[20:52] <kidn3ys> sarnold: _netdev seems to 'mark' that entry as a network device so it bring sup the network stack/iscsi before it mounts it
[20:52] <sarnold> I wonder if that still works in xenial. heh.
[20:54] <kidn3ys> hehe :)
[21:52] <terje> hi, is it possible to script 'sudo lxd init' such that I don't have to answer any questions?
[21:54] <terje> nevermind, I see that it is..