[03:42] <savid> I have a file in /etc/cron.d that does not seem to be running. I put a dummy job at the top, and the dummy job does not execute. The same line in a separate file works fine. How can I debug this? Can't seem to find a logfile for cron...
[03:52] <raub> savid: I think you should see an entry in syslog when crontab finds that job
[03:53] <raub> savid: can you pastebin your file?
[03:57] <savid> raub: ok, I'll check syslog. The crontab file is rather long, lots of maintenance jobs in there.
[03:58] <savid> Ah, wrong file owner :-)
[03:59] <raub> savid: another mystery solved ;)
[03:59] <raub> savid: I meant your file in /etc/cron.d, not in crontab
[04:15] <erixNICK> hey all
[04:15] <sheptard> HI
[04:15] <erixNICK>  having trouble getting samba to run on a 12.04 LTS server.  I see that there are bugs posted,  but havent seen fix.  Can some one please shed some light on this for me
[04:16] <sheptard> what error is samba giving
[04:16] <erixNICK> one sec will pull up
[04:17] <erixNICK> smbd/server.c:1107(main)
[04:17] <erixNICK>   standard input is not a socket, assuming -D option
[04:17] <sheptard> thats normal
[04:18] <erixNICK> it seems like I cant set up even a basic guest share
[04:20] <erixNICK> sheptard,  even if I ignore error I cannot get a guest share up
[04:21] <sheptard> so samba is listening, and you can connect to it, you just can't see any shares?
[04:21] <erixNICK> from windows I get an access denied
[04:22] <erixNICK> http://pastebin.com/nbnndKuE is what I have for the share
[04:24] <sheptard> can you ls -al /shared
[04:27] <sarnold> hey sheptard :)
[04:28] <erixNICK> yeh
[04:28] <erixNICK> i have shared 777'd for testing purpose
[04:29] <erixNICK> "/shared" ^
[04:29] <sheptard> hi sarnold
[04:30] <sheptard> erixNICK: Hum
[04:30] <erixNICK> it looks like out of the box it wont allow guest shares... im also on a win8 box (unfortunatly)
[04:31] <sheptard> if you do something like
[04:31] <sheptard> smbpasswd -a test
[04:31] <sheptard> then try to connect as test, does that work?
[04:32] <erixNICK> k
[04:32] <sarnold> erixNICK: your ls -al /shared  just returns "/shared"? no . or .. directory entries? o_O
[04:33] <erixNICK> it returns a directory list
[04:33] <erixNICK> also i am able to change user "test" password
[04:35] <sheptard> erixNICK: can you connect to the samba share as test?
[04:36] <sheptard> if you have test as an allowed user
[04:37] <erixNICK> not sure how I would do that
[04:39] <erixNICK> im trying to just do a net view \\liuux_box from a dos window
[04:40] <erixNICK> even set user.group to nobody for /shared
[04:41] <erixNICK> looks more like it doesnt no what to do with windows auth
[04:43] <erixNICK> unix password sync = Yes <- would that affect a guest share
[04:46] <erixNICK> looks like alot of people are having this problem when they changed from 10.04
[05:32] <SunkSullen> anyone alive? having trouble
[05:32] <SunkSullen> with LAMP and UBuntu
[05:32] <SunkSullen> server
[05:33] <cfhowlett> !lamp
[05:33] <SunkSullen> okay but...
[05:34] <SunkSullen> the server was setup with root
[05:34] <SunkSullen> I added an account
[05:34] <SunkSullen> added to sudoers file
[05:34] <SunkSullen> and now I want to change document root
[05:34] <SunkSullen> ugh
[05:37] <sarnold> yes?
[05:39] <SunkSullen> sorry so how do I make the account I made which has sudo permisission ~/public_html my document root instead of /var/www
[05:39] <SunkSullen> ?
[05:40] <sarnold> edit your apache configuration files to set the DocumentRoot to your new location
[05:40] <sarnold> it'll be somewhere in /etc/apache2/ if I'm remembering it correctly
[05:41] <SunkSullen> k
[05:41] <SunkSullen> and also
[05:41] <SunkSullen> THANKS
[05:41] <SunkSullen> first off
[05:41] <SunkSullen> nevermind
[05:41] <SunkSullen> I can find that out my own
[05:42] <SunkSullen> the correct permissions of the fodler for /home/cameron/public_html
[05:43] <sarnold> the directory will need to be readable and executable by the apache process; so you'll need to pick permissions that grant the www-data user the permissions it needs while still meeting the security needs of the rest of the system
[05:44] <sarnold> SunkSullen: that might setting the directory to e.g. 755, so that everyone can read and execute the directory, or it might mean you need to set group ownership of the directory to www-data and then use e.g 750
[05:48] <SunkSullen> sarnold: ah confusing lol sorry
[05:49] <sarnold> SunkSullen: unix permissions are amazingly concise and sharp. it takes a few years familiarity with them to appreciate how simultaneously flexible and concise they are
[05:49] <sarnold> before then of course it's just baffling and confusing
[05:50] <SunkSullen> hehe yea
[05:55] <soren> rbasak: uvirtbot?
[11:03] <jdo_dk> I have a postfix, which logs maillog to syslog. How to split the log into mail.log ?
[11:03] <jdo_dk> Ubuntu 13.10 and running rsyslog
[11:22] <ikonia> jdo_dk: look at the syslog.conf and you should see the priorities/levels which determain what goes to what file
[11:34] <jdo_dk> ikonia: http://privatepaste.com/c8c65eba26
[11:35] <jdo_dk> ikonia: As i can see in the /etc/rsyslog/50-default.conf, mail.* already goes to: /var/log/mail.log
[11:37] <ikonia> jdo_dk: what's the actual line from your syslog that you want to into mail
[11:37] <ikonia> it may not actually be a mail alert
[11:37] <ikonia> (but look like a mail alert)
[11:39] <jdo_dk> ikonia: http://privatepaste.com/40a7c0622f
[11:39] <ikonia> that's probably not going to be classed as "mail"
[11:39] <ikonia> thats an application/daemon reporting
[11:39] <jdo_dk> Ok
[11:40] <jdo_dk> But i would "expect" to see those "normal" log entries in mail.log
[11:40] <jdo_dk> Should i try to install syslog instead of rsyslog ?
[11:40] <ikonia> which normal ones ?
[11:40] <ikonia> can you give me an example of a normal one ?
[11:40] <jdo_dk> Will find another ubuntu server.
[11:40] <jdo_dk> just a sec
[11:43] <jdo_dk> Ikonia: mail.log is empty on all "my" servers...
[11:43] <ikonia> jdo_dk: yes, I'd expect that
[11:43] <ikonia> unless you have mail events
[11:45] <jdo_dk> ikonia: http://privatepaste.com/af6174fe9c
[11:45] <jdo_dk> I would expect mail.log to look like that file...
[11:45] <ikonia> nah,
[11:45] <ikonia> that's postfix
[11:45] <ikonia> not mail
[11:45] <jdo_dk> I have postfix installed.
[11:45] <ikonia> postfix is an "application" from the point of view of the syslog
[11:46] <jdo_dk> So i need to configure "postfix" in the syslog conf ?
[11:46] <ikonia> you need to configure postfix to log to the correct places
[11:46] <ikonia> or find out the "event" that syslog will monitor from postfix, I did this years ago with exim and ksyslogd (so it will be a bit different now) but it works out easier to configure the application to log correctly, rather than configure the syslog to try to manage it's events
[11:47] <jdo_dk> Ok
[11:47] <jdo_dk> I will try. Thanks mate.
[12:01] <jdo_dk> ikonia: http://ubuntuforums.org/archive/index.php/t-1893739.html
[12:02] <jdo_dk> stop rsyslog, chown / chmod start syslog solved the logging...
[12:02] <ikonia> really ?
[12:02] <ikonia> it picks up those postfix events as "mail" ?
[12:27] <marcoceppi> hey rbasak, thanks for the info for the charm-tools thing. I have no idea why the packaging branch is a head of what's in precise I just need to update the bits about recommends, etc. I'll get a public bug filed from the private one in a few though.
[12:27] <marcoceppi> How would i go about getting the right packaging branch for precise?
[12:28] <rbasak> marcoceppi: I'm confused about that too. Yeah just updating the recommends is fine. With just an SRU bug I'm happy to upload that.
[12:29] <rbasak> marcoceppi: I'm not sure how you can submit a UDD merge proposal based off the version that's in precise, but don't worry about that too much. I'll happily upload from a debdiff, or if you just update that MP without worry what it's based on, I'll fix it before upload.
[12:29] <marcoceppi> rbasak: the entire charm-tools packaging is a giant bag of two headed snakes
[12:29] <marcoceppi> so this somehow doesn't surprise me
[12:31] <rbasak> marcoceppi: BTW, is this fixed in Trusty? We need that before uploading an SRU, but I presume it is?
[12:33] <marcoceppi> rbasak: yes, it was fixed with charm-tools 1.0.0
[12:34] <marcoceppi> but 1.0.0 broke a package which is why there isn't a backport request
[12:34] <rbasak> OK
[12:48] <marcoceppi> rbasak: here's the public bug, lp:1182905
[12:54] <rbasak> marcoceppi: great! Are you doing the SRU justification (are you familiar with that?)
[12:54] <marcoceppi> rbasak: I'm not, this is a completely new process! I'm completely interested in learning though
[12:55]  * marcoceppi hunts down a wiki page
[12:55] <rbasak> marcoceppi: https://wiki.ubuntu.com/StableReleaseUpdates#Procedure
[12:56] <rbasak> marcoceppi: it's for the SRU team, who will need to review before accepting any upload. And also to leave a trail for anybody who receives the update and wants to know why.
[12:56] <rbasak> Or wants to know why everything broke, etc :)
[13:09] <jdo_dk> ikonia: Yeah. Mail.log just works... So no need to configure syslog or postfix further...
[13:29] <ikonia> jdo_dk: that is very surprising, but well done
[13:30] <jdo_dk> ikonia: ;)
[14:07] <marcoceppi> rbasak: Okay, I updated #1182905 with SRU template. It mentions uploading the package to release-proposed, but I'm not sure how or if I should do that
[14:07] <rbasak> marcoceppi: OK, thanks. I'll take a look in a bit - a few too many balls in the air right now :-/
[14:08] <marcoceppi> rbasak: np np! I get that completely, thanks for the help thus far! :D
[14:40] <stetho> I have a hard disk on a server that's filling at quite a rate. I can see with iotop that it's a remote user doing something over nfs but I can't figure out which user. How can I figure out which files are growing?
[14:41] <rbasak> marcoceppi: juju-core doesn't exist in precise. How would you feel about just dropping the Recommends: juju to Suggests: juju?
[14:41] <marcoceppi> rbasak: uh, yeah that works
[14:42] <marcoceppi> rbasak: I realized that while writing the SRU
[14:43] <rbasak> marcoceppi: charm-helper-sh can still suggest juju, right? I don't see why that would break anything.
[14:44] <rbasak> Sorry I'm confusing myself. Let me pastebin you a debdiff in a moment.
[14:45] <marcoceppi> rbasak: charm-helper-sh is installed solely server side, and doesn't need juju at all, in fact it's not even really associated with juju
[14:45] <moparisthebest> is anything in /var/log/ important? and if so, can I mount it on tmpfs and setup logrotate to just delete things when they get so big?
[14:45] <rbasak> marcoceppi: my proposed diff: http://paste.ubuntu.com/7073884/; full new control file: http://paste.ubuntu.com/7073885/
[14:46] <marcoceppi> rbasak: I suppose suggests is fine for charm-helpers
[14:46] <marcoceppi> lgtm
[14:46] <marcoceppi> I should just stick with mvp in the future
[15:03] <jhobbs> hallyn: should containers be able to mount ext2/3/4 filesystems in loopback mode in trusty?
[15:06] <jhobbs> hallyn: looks like i still need to add some aa_profile/cgroup config to my lxc conf to get it to work
[15:07] <smoser> jhobbs, no. not by default.
[15:07] <smoser> i think there is a config that allows it though eplicitly
[15:08] <smoser> jhobbs, /etc/apparmor.d/lxc/lxc-default-with-mounting
[15:10] <jhobbs> smoser: ok - cool; i'm not really familiar with app-armor; how do i make that apply to new containers created via juju-local?
[15:11] <smoser> you can set some value in /etc/lxc/default.conf
[15:11] <smoser> to set the default app armor profile
[15:11] <smoser> but i dont have that handy at the moment
[15:11] <jhobbs> ah ok cool, i'll chase it down; thanks
[15:11] <smoser>  lxc.aa_profile
[15:11] <smoser> (see man lxc.container.conf)
[15:12] <jhobbs> so i have some settings in there already that apply to containers that i create manually; i set the aa_profile to unconfied and change some cgroup settings
[15:12] <jhobbs> but those don't seem to apply to the containers that juju-local created
[15:12] <smoser> they wont.
[15:12] <smoser> only new ones.
[15:12] <smoser> they get copied.
[15:12] <smoser> i got to run.
[15:13] <jhobbs> yes, i had the settings there prior to creating the juju-local containers
[15:13] <jhobbs> ok
[15:13] <jhobbs> thanks for the lp
[15:13] <jhobbs> help
[15:20] <hallyn> jhobbs: if you need it you need it, but of course it is disallowed bc there *is* danger in allowing it,
[15:21] <hallyn> so i recommend against enabling by default
[15:21] <jhobbs> hallyn: ok
[15:21] <hallyn> (sorry, had some hubbub around these parts)
[15:22] <jhobbs> hallyn: that's cool; i just wanted to know if the behavior i was getting matched what was expected
[16:01] <jamespage> rbasak, are you covering the HWE eol meeting?
[16:02] <rbasak> jamespage: yes
[16:02] <jamespage> great
[16:02] <jamespage> listening in whilst I wait for mysql-xx to build
[16:03] <zzxc> ahhhhh so this is why I'm always getting mentions from this channel.
[16:14] <irv> howdy, i installed 13.10 to an hp dl320e g8, when the setup ran, (it setup off a bootable USB), grub installed to /dev/sda (which was the USB drive). I want to install grub to the HD. I've since rebooted the machine (it will only boot to ubuntu with the USB in there) and did sudu grub-install /dev/sda (/dev/sda is the HD once the setup rebooted), says no errors. reboot and the same issue.
[16:14] <irv> any ideas on how i can verify that grub is indeed installed to the MBR on the HD
[16:14] <irv> it's a SATA drive single volume on a b120i raid controller
[16:14] <irv> when i boot without the USB in it says non system disk or disk error for C
[16:19] <markthomas> irv, it sounds as though it still is not installed.  Are you certain that once the system is booted that your SATA drive is /dev/sda?
[16:20] <irv> when i run df -h when it's booted, i see 457 gb or something on /dev/sda1
[16:20] <irv> there's only the one drive and the USB is 8gb
[16:20] <irv> so that's how i figured it was sda at that point
[16:21] <markthomas> irv, okay.  Try update-grub and then grub-install /dev/sda again.
[16:21] <irv> k sec
[16:22] <irv> rebooting now
[16:22] <markthomas> If it doesn't work, then we may have to dissect it a bit.
[16:23] <irv> sounds good. i did install a lowlatency kernel, but it must've run update-grub since that was the kernel that booted off the USB after i installed it
[16:23] <irv> but who knows :p
[16:23] <irv> just booting up now
[16:24] <irv> hrm, still no joy
[16:24] <irv> could it be something to do with the raid controller?
[16:25] <irv> booting off hte USB now
[16:30] <irv> okay, sudo blkid (after removing the USB that booted it) shows /dev/sda1 and /dev/sda5 as swap
[16:30] <irv> sda1 is ext4 and is the partition all the files are on heh
[16:31] <markthomas> irv, Just to be thorough, can you confirm whether your /boot is a separate partition?
[16:34] <irv> it is not, i did a guided entire disk partioining
[16:37] <irv> tried just running sudo grub-install /dev/sda and then sudo update-grub /dev/sda as per someone in #ubuntu
[16:37] <irv> also didn't work
[16:38] <irv> "non system disk error"
[16:41] <irv> maybe an issue with my fakeraid? i can put it into legacy mode or AHCI mode as well
[16:48] <markthomas> OH, fakeraid.
[16:48] <markthomas> irv, Your best bet with fakeraid is to turn it off in the BIOS and use software RAID.
[16:48] <irv> ya it's only a single drive, so i'm not actually doing any RAID
[16:49] <irv> legacy mode?
[17:04] <RoyK> irv: ahci it be
[17:07] <genii> Is there some reason I would not see a process running if I ssh-ed in and ran it? ( not inside screen)
[17:07] <irv> i set it to AHCI and then re-ran grub-update and install-grub but no joy, gonna do a full reinstall on the AHCI'd drive
[17:07] <irv> genii: wehre are you looking to see if it's running?
[17:08] <irv> different userspace?
[17:08] <RoyK> irv: probably different mapping
[17:08] <genii> irv: ps aux doesn't shor it, either on server or client machine running
[17:10] <irv> RoyK: i'm surprised it even booted at all even off hte USB
[17:10] <irv> i figured the files would be gone after i changed the controller's 'mode'
[17:10] <irv> just writing a Ubuntu server USB again
[17:10] <irv> since the one i used to install now has GRUB On it
[17:10] <irv> minor side-lesson, select 'no' for grub install when installing from a USB :P
[17:46] <irv> ok so after changing the sata config to AHCI, reinstalling ubuntu server and in the setup i picked "no" for the grub install which exposed a new screen that I was able to pick the destination of GRUB. i put it to /dev/sdb (which was the HD as the USB install key shows up as /dev/sda during hte install)
[17:46] <irv> now it boots straight off the HD :)
[17:46] <irv> thanks for the help all
[18:03] <keithzg> Well that's mysterious. Trying to use cygwin+rsync to periodically pull some Doxygen-generated documentation off of a Windows autobuild machine and put it up on our internal website, and the rsync share is including cygdrive, dev and proc? WTF?
[18:19] <jrwren> gah, just realized its UDS this week.
[18:29] <keithzg> UDS doesn't have the same pomp and circumstance it used to, eh?
[18:29] <keithzg> Moving entirely online will do that, I suppose.
[18:31] <ogra_> it is a lot more focused
[18:31] <ogra_> ... but you have to drink alone afterwards
[18:48] <jamespage> hallyn, if you are around can you take a look at https://bugs.launchpad.net/ubuntu/+source/libvirt/+bug/1254872
[18:48] <jamespage> I'm trying to get to the bottom of it but this appears to be impacting openstack-gate in some way
[18:51] <hallyn> jamespage: I can't keep up with the cutesy cloud archive names :)  which ubuntu release is that libvirt from?
[18:51] <jamespage> hallyn, so that bug appears to relate to 12.04 itself - 0.9.8 release
[18:52] <jamespage> I have another
[18:52] <jamespage> for 1.1.1 (saucy)
[18:52] <hallyn> meaning you have the same error on both?
[18:52] <jamespage> hallyn, oh - I see you already on https://bugs.launchpad.net/nova/+bug/1228977
[18:52] <jamespage> hallyn, good-oh
[18:53] <jamespage> that was only yesterday
[18:53] <hallyn> jamespage: right that one is in saucy only and no way am i backporting those patches to precise :)
[18:54] <hallyn> but certainly the commit mentionedin comment #13 is worth trying
[18:55] <jamespage> hallyn, no indeed
[18:55] <jamespage> hallyn, is the 1.2.2 we have in trusty going to suffer from bug 1228977 as well?
[18:56] <hallyn> jamespage: 1.2.2 was *just* released so i don't think so
[18:56] <hallyn> oh, but, yes
[18:56] <jamespage> hallyn, can you join me in #openstack-infra - doing piggy in the middle right now :-)
[18:57] <hallyn> zul: ^ we'll probably need the patch in comment #40 of bug 1228977 added to trusty's libvirt
[18:57] <jamespage> hallyn, zuls in florida on a beach
[18:57] <hallyn> curses
[19:13] <adam_g> hallyn, jamespage FWIW https://bugs.launchpad.net/ubuntu/+source/libvirt/+bug/1248025 will eventually be a big issue for the upstream gate, at least the portions of it that run on rackspace.
[19:13] <adam_g> stgraber, im happy to give you an instance on rackspace if you want to poke at that yourself ^
[19:13] <jamespage> adam_g, you mean smb
[19:14] <adam_g> doh
[19:14] <adam_g> smb, ^ :)
[19:14] <jamespage> adam_g, :-)
[19:15] <smb> adam_g, yeah, thanks. If you can have something running for me in the morning and email me the details for looking there tomoroow
[19:15] <smb> *tomorrow
[19:15] <jamespage> thanks smb
[19:15] <jamespage> and adam_g
[19:15] <jamespage> :-)
[19:15] <adam_g> smb, sure. rackspace gave out a bunch of free credit at an ODS a while ago--might as well put it to good use
[19:17] <smb> adam_g, Cool, yeah it really mysterious what is different there since all seems well on my test systems even with a cloud-image based guest
[19:17] <smb> I suspect its something in the setup but hopefully one can find traces of that inside the guest
[19:20] <smb> adam_g, Just as a note that the day is mostly over, so to make best use of any credits you can bring up the guest late your day (assuming you are on the other side of the Atlantic) :)
[19:22] <adam_g> smb, just spun a precise and saucy up. should be fine to keep running as long as you need. ill email you the details
[19:22] <smb> adam_g, Ok, thanks
[19:26] <Lord_Set> What is the config file which has all installed packages in it? I need to force remove a package that wont't remove via apt-get purge properly or reinstall properly... I just forgot the name of the file that I have to remove the entries of the package so that Ubuntu doesn't even think it is installed.
[19:32] <rbasak> Lord_Set: I usually edit the maintainer script in /var/lib/dpkg/info to fix it so it does remove (eg. with an exit 0 at the top), and then do whatever cleanup it was trying to do manually instead.
[19:33] <rbasak> I've always felt that this is cleaner. Then at least the package manager can maintain its own state, so no chance of corrupting it.
[19:34] <Lord_Set> Thanks rbasak
[19:36] <Lord_Set> But just in the other case what is the file that has a list of all installed packages?
[19:44] <smb> adam_g, Ah, hm. So there is xe-guest-utilities which causes a xenfs mount to /proc/xen and that is causing the failure. That package probably comes from managing the guests with XenServer. There is also a nova-agent using some file exported in the. Not sure where that comes from, it does not seem to have a package installed.
[19:45] <adam_g> smb, so essentially libvirt conflicts with  xe-guest-utilities?
[19:45] <smb> adam_g, So I think that is enough for first round of data gathering. I can check that locally tomorrow. And will put that info into the bug
[19:45] <smb> Not really conflicts but it leads to some unexpected situation. Maybe the error of getting capabilities needs to be made a soft failure
[19:46] <smb> Or it needs another check to avoid thinking the proc mount alone is hinting a dom0
[19:47] <smb> But it should be enough info for me to reproduce and then discuss this upstream
[19:48] <smb> adam_g, So I am off the guests and you can shut them down to save credits
[19:49] <adam_g> smb, great. thanks a bunch
[20:09] <raub> How do you configre openldap to use a comodo-bought cert? i.e. what should I feed TLSCACertificateFile with?
[20:09] <Lord_Set> rbasak: But just in the other case what is the file that has a list of all installed packages?
[21:51] <thumper> hallyn: if I snapshot a dir based lxc image with aufs, and the base container gets updated (update/upgrade), is it expected that the cloned image will works?
[21:51] <thumper> s/snapshot/clone/
[21:53] <hallyn> thumper: no.  snapshots of dir backed containers should not be done
[21:53] <hallyn> and lxc should be defaulting to a copy-clone
[21:53] <hallyn> you should create an aufs clone first, then snapshotting that
[21:53] <hallyn> unfortunately
[21:53] <thumper> I think I used snapshot wrong
[21:54] <thumper> let's step back
[21:54] <thumper> with juju, we want to improve the local testing process
[21:54] <thumper> the idea is to create containers with clone
[21:54] <thumper> so we can skip the apt update/upgrade step mostly
[21:55] <thumper> btrfs works wonderfully
[21:55] <thumper> but we want a stable solution that works for people without btrfs
[21:55] <thumper> that still gets them good speed, good disk combo
[21:55] <thumper> hallyn: do you think that cloned containers using aufs will give us this?
[21:56] <thumper> one thought was btrfs on loopback
[21:56] <thumper> which I had working locally, but there are issues around keeping track of the size, mounting in the right place, resizing etc
[21:56] <thumper> I could go down this track if it will give us the best user experience
[21:56] <hallyn> thumper: btrfs *is* the ideal solution,
[21:56] <hallyn> thumper: overlayfs has its own bugs, and aufs has its own (stgraber can tell you about them)
[21:57] <hallyn> oh that's right, aufs doesn't do xattrs?
[21:57] <thumper> so in your opinion, if they don't have btrfs natively, offer btrfs with loopback devices?
[21:57] <hallyn> anyway, if you want to use aufs, it shoudl mostly work
[21:57] <hallyn> you just have to create a unchanging container base, then create clones from those
[21:57] <thumper> one problem with aufs that I have been told about, is that wordpress charm doesn't work
[21:57] <hallyn> and not update the base
[21:57] <thumper> and given that is our CI, and used in all the examples
[21:57] <thumper> not ideal
[21:57] <hallyn> heh
[21:57] <hallyn> wonder why
[21:58] <thumper> wordpress installs nsf kernel modules
[21:58] <thumper> nfs
[21:58] <thumper> not nsf
[21:58] <hallyn> that should fail with btrfs too
[21:58] <hallyn> wordpress should not do that
[21:58] <thumper> hmm...
[21:58] <thumper> I should really go try
[21:59] <thumper> but would that also fail in plain containers?
[21:59] <hallyn> yes
[21:59] <hallyn> containers cannot install modules
[21:59] <thumper> ok, in which case it must be something different
[21:59] <thumper> because wordpress works in dir backed containers
[22:00] <thumper> backing up a little
[22:00] <thumper> if I wanted to test aufs backed clones
[22:00] <thumper> I should make sure that the base image is not updated for the lifetime of the clone?
[22:01] <hallyn> yes
[22:01] <thumper> how badly will it break if it does happen?
[22:01] <hallyn> depends on what is changed
[22:01] <thumper> still able to destroy the containers?
[22:01] <hallyn> yeah
[22:02] <hallyn> mainly, if file /a/b was overwritten in the clone, then updates in the main container to /a/b won't show up,
[22:02]  * thumper nods
[22:02] <thumper> also if a is updated and b is running
[22:02] <thumper> missing inotify for things that aren't overwritten?
[22:02] <thumper> stuff like that?
[22:03] <hallyn> for things that are overwritten
[22:03] <hallyn> remember, we're not sure (afaik) whether aufs will end up in final trusty kernel or not
[22:04] <thumper> ok
[22:04] <thumper> so, back to juju awesome user experience
[22:04] <thumper> for trusty, we are going with clone for creating containers
[22:05] <thumper> if you are on btrfs, it is awesome
[22:05] <thumper> if not, lots of I/O and disk usage
[22:05] <thumper> have a plugin to support btrfs on loopback
[22:05] <thumper> sound like the best option?
[22:12] <hallyn> thumper: I dn't know how to prioritize the downsides.  I think using aufs and never upgrading the base is good.
[22:12] <hallyn> this also might be a use-case for a qcow backend,
[22:12] <hallyn> (but that doesn't exist yet :)
[22:12] <thumper> qcow>
[22:12] <thumper> ah
[22:13] <hallyn> and really it'd have the same expectations as aufs
[22:13] <thumper> is there a way to ask the container (aufs clone) what it is cloned from?
[22:13] <hallyn> the base rootfs is in the config file
[22:14] <hallyn> the rootfs for the clone will be "aufs:<rootfs_base>:delta_dir"
[22:14] <thumper> so, could easily parse the config files for all the current containers
[22:14] <hallyn> yeah,
[22:14] <thumper> hmm...
[22:14] <hallyn> or juju could just call the canonical container "juju-precise-20140311"
[22:14] <thumper> well, the canonical one now is 'juju-precise-template'
[22:14] <hallyn> then the next time it creates a base, 'juju-precise-20140315'
[22:15] <thumper> what do you mean 'next time it creates a base'?
[22:15] <thumper> make a dir based clone of the base template?
[22:16] <hallyn> if juju ever updates teh template
[22:18] <thumper> hallyn: ok, so if I want to use clone with aufs, all I need is a base container to start with (dir backed), and go "lxc-clone -B aufs -o juju-precise-template -n new-name" right?
[22:19] <hallyn> yeah (plus -s)
[22:21] <thumper> oh, need the --snapshot too?
[22:22] <thumper> the help says "The new container's rootfs should be a LVM or btrfs snapshot of the original."
[22:22] <thumper> but ok, will add it
[22:22] <thumper> i just thought that if I specified the backingstore, it would just know
[22:24] <hallyn> thumper: there may be cases where someone doesn't include "-s" and really means "full copy".  in that case aufs should retufn failure, asit can't do that
[22:24]  * thumper nods
[22:24] <thumper> ok
[22:25] <hallyn> maybe the clearest thing to do would be to add a --nosnapshot option and have the default be "whatever is best"
[22:25] <hallyn> but i don't nwat to change that right now
[22:27] <thumper> sure
[22:27]  * thumper goes to experiment with aufs
[23:05] <apb1963> Could someone run dhcpdump on their machine and tell me if you're sending out continual bootp requests (about every 15 seconds or )?  That's what my machine does and I'd like to know if anyone else is seeing this as well (and doesn't know it).
[23:11] <sarnold> apb1963: nothing with 'sudo tcpdump -n -i wlan0 'udp and (port bootpc or bootps)'' when waiting a minute-ish here
[23:13] <apb1963> sarnold: thank you... what version of ubuntu are you running?
[23:13] <sarnold> apb1963: 13.10
[23:14] <apb1963> hmm... 12.04 here.
[23:14] <apb1963> plus i'm wired with a cable provider
[23:15] <apb1963> I need more of an apples to apples comparison I suppose
[23:15] <sarnold> could be :) dhcp is finnicky.
[23:15] <apb1963> is it?
[23:16] <sarnold> the never-ending stream of bug reports of folks running in vms or containers when the kernel doesn't fill in the udp checksums because it thinks that's being offloaded to a NIC.. there must have been months of those bugs. :) heh
[23:16] <apb1963> hmm
[23:18] <apb1963> i'm glad I missed out on that spot o' fun
[23:19] <apb1963> it was purely accidental that I found this issue....  I've added a whitelist to my firewall... and was logging dropped packets.... that's when I saw this.
[23:21] <apb1963> nobody seems to know why it's happening.  I've stumped the entire world :/
[23:21] <sarnold> apb1963: strace the process responsible?
[23:23] <apb1963> that's actually a good idea...  I'm not sure I know what's responsible, but I'm guessing it's dhclient
[23:24] <apb1963> just killed it
[23:24] <apb1963> lets see if that kills the packet stream
[23:27] <GeekD00d> Any reason I can't get php to work with nginx? It gives me a 502 bad gateway error
[23:27] <GeekD00d> whenever I try to load a php file, that is
[23:28] <apb1963> sarnold: it seems to have stopped the outgoing, but I still have incoming Replys coming back... not sure if they're going  to run out like they're buffered... or if it will continue forever.
[23:29] <apb1963> GeekD00d: #nginx #php
[23:29] <GeekD00d> thanks apb1963
[23:30] <sarnold> apb1963: oh that's odd..
[23:30]  * teward yawnws
[23:31] <apb1963> sarnold: it's broadcasting Reply packets
[23:31] <apb1963> I presume it's my ISP since it's coming from a 10.65 address
[23:32] <sarnold> apb1963: eww. gross.
[23:32] <apb1963> IP: 10.65.192.1 (0:24:14:61:38:d9) > 255.255.255.255 (ff:ff:ff:ff:ff:ff)
[23:33] <sarnold> wait, the ethernet mac is set to ff:ff:ff:ff:ff:ff? I thought dhcp replies were sent to the specific mac that asked for the address?
[23:34] <apb1963> I THINK, that's an IP address, not a MAC
[23:34] <apb1963> it's just converting to hex
[23:35] <apb1963> wait... maybe not
[23:35] <sarnold> heh, then it'd only be four bytes long rather than six
[23:35] <apb1963> yeah, I was just about to say... that may be an IP6 address
[23:36] <sarnold> in that case there'd be more than six bytes :) hehe
[23:38] <apb1963> sarnold: you're right... so I guess it is a MAC...
[23:44] <apb1963> sarnold: interesting conversation on it here: https://learningnetwork.cisco.com/thread/32255
[23:45] <jrwren> that is eithernet broadcast mac
[23:45] <jrwren> is it CDP or something?
[23:45] <apb1963> ?
[23:45] <jrwren> cisco discovery protocol
[23:45] <apb1963> no idea
[23:45] <apb1963> I don't think so
[23:46] <apb1963> but I couldn't tell you for sure
[23:46] <apb1963> which brings me back to ... no idea :)
[23:46] <jrwren> is that tcpdump?
[23:46] <apb1963> I use ethernet to connect to my cable modem
[23:47] <apb1963> beyond that... no idea
[23:47] <apb1963> the abovfe?
[23:47] <jrwren> yes, the above.
[23:47] <apb1963> the output I posted?
[23:47] <apb1963> it's actually dhcpdump
[23:47] <jrwren> oh, i'm not familiar with that
[23:48] <apb1963> it's like tcpdump... but...  deciphers dhcp packets
[23:49] <apb1963> this is tcpdump: 16:48:51.974312 IP 10.65.192.1.67 > 255.255.255.255.68: BOOTP/DHCP, Reply, length 300
[23:49] <jrwren> that wasn't much deciphering :)
[23:49] <apb1963> oh, that's because I only took an isolated line
[23:49] <jrwren> oh, pastebin the whole thing.
[23:49] <apb1963> ok
[23:49] <apb1963> a moment
[23:50] <jrwren> tcpdump with -vv does a pretty good job of dhcp protocol decode
[23:53] <jrwren> and tcpdump with -vv -X shows hexdump payload, and dhcp responses are reasonably readable
[23:53] <apb1963> http://fpaste.org/84549/39458201/
[23:54] <apb1963> jrwren: good to know
[23:56] <sarnold> o_O bootfile name..
[23:57] <apb1963> sarnold: yes?
[23:58] <apb1963> it's what I'd expect in a bootp reply :)
[23:59] <apb1963> just not necessarily broadcast to the world.. but according to that cisco doc.. it's not wrong.
[23:59] <sarnold> apb1963: well yes and no, I'm curous what your ISP is trying to get you to boot :)
[23:59] <apb1963> haha
[23:59] <apb1963> yeah