[01:21] <dmbQs> Need some help reconfiguring Grub after restoring to a new VM. Can anyone help me with the instructions here: I restored a Tar backup to a new VM. Now I need to reconfigure Grub. Need some help with the commands found here: https://help.ubuntu.com/community/BackupYourSystem/TAR. Would anyone mind helping??
[02:47] <delinquentme> sshfs is supposed to create the mount dir on execution right?
[02:47] <delinquentme> im getting a complaint from bash: fuse: bad mount point `/home/thrive/www': No such file or directory
[02:48] <sarnold> delinquentme: no, a directory needs to exist somewhere before you can mount a filesystem on it
[02:51] <delinquentme> sarnold, much apreesh
[04:35] <babak_>  question mod_security comes on 12.04 as 2.6.3 and the current version is 2.7.5 is it ok to stay with 2.6.3 ? I though we where supposed to use LTS for things like this.
[07:49] <rbasak> Daviey: I'm looking at http://reqorts.qa.ubuntu.com/reports/ubuntu-server/merges.html. It looked truncated; then I realised that it's being updated as I read it. What do you think about changing the cronjob to write to a temporary file and then renaming in, so we get an atomic update? Not worth it for the one time I hit the problem? :)
[08:30] <Daviey> rbasak: Yeah, i can do that.
[08:31] <Daviey> rbasak: if you saw this crontab... you'd be wishing we had wrappers :)
[08:31] <Daviey> (busy)
[08:31] <rbasak> :)
[08:36] <Daviey> rbasak: should be good now
[08:36] <rbasak> Daviey: thanks!
[09:01] <Daviey> zul: hey, can you check if kombu can be sync'd? kombu 2.5.12-0ubuntu2 -> 2.5.12-1
[09:02] <Daviey> jamespage: looks like gdisk can be sync'd?
[09:12] <jamespage> Daviey, probably
[11:37] <yolanda> zul, trying to build ceilometer for the tests, but having a lot of RuntimeError: No 'ceilometer.storage' driver found, looking for ''
[11:43] <zul> yolanda:  yeah ill have a look at it today
[11:43] <yolanda> i was trying to build ceilometer to run the tests, do you prefer that i wait for it?
[11:47] <zul> yolanda:  yeah
[11:47] <yolanda> ok
[11:47] <yolanda> zul, please let me know when it's fixed so i can add my tests
[11:52] <zul> jamespage/roaksoax: https://code.launchpad.net/~zulcss/python-novaclient/babel/+merge/182367
[12:04] <wsk233> hello i need to setup a ftp server for some internal use the only thing i need to be able to configure is chroot which is the most easy ftp server to set this up
[12:05] <greppy> wsk233: I've always liked proftpd if I *really* had to setup a plain ftp server, I much prefer using sftp.
[12:05] <wsk233> ofcrouse i understand
[12:08] <zul> jamespage: https://code.launchpad.net/~zulcss/python-neutronclient/deps-refresh/+merge/182372
[12:16] <zul> Daviey:  should be ok to sync
[12:40] <crass> I'd like to run the server iso from an iso on a filesystem on a usb stick. Is this possible?
[12:41] <crass> this can be done using the loopback.cfg with the desktop iso, but server's loopback.cfg just has an entry for testing the iso, not actually running it
[12:57] <Monotoko> what should I be allowing through the firewall to stop a DNS lookup failure when doing a proxypass?
[12:58] <Monotoko> outgoing is allowed... I don't see the problem
[14:25] <hallyn_> zul: if/when my laptop(s) overheat can you step in for a few mins in the virt stack talk (in 1.5 hrs)
[14:26] <zul> hallyn_:  sure
[14:26] <hallyn_> thx
[14:28] <zul> hallyn_:  if my network connection doesnt crap out on me :)
[14:28] <hallyn_> sbux?
[14:28] <zul> hallyn_:  rural internet
[14:46] <disposable> i have a vlan interface on top of bond(active-backup) interface on top of 2 physical interfaces. I recently upgraded from 10.04 to 12.04 and suddenly I see a lot of dropped packets on my bond0 and the inactive underlying eth interface. same problem on 3.2, 3.5, 3.8, 3.9 kernels. it seems about a 1/5 of received packets are dropped. my cards use igb module (intel 82576). has anybody come across this?
[14:47] <disposable> i've tried juniper and procurve switches
[14:54] <zul> roaksoax: mind +1ing https://code.launchpad.net/~zulcss/python-novaclient/babel/+merge/182367 and https://code.launchpad.net/~zulcss/python-neutronclient/deps-refresh/+merge/182372 please
[14:54] <zul> yolanda:  ceilometer should be fine now
[14:54] <roaksoax> zul: done
[14:55] <yolanda> zul, thx
[14:55] <zul> thanks
[14:55] <patdk-wk> disposable, works great for me
[15:03] <disposable> patdk-wk: if you type ifconfig bond0, do you not see any dropped packets?
[15:08] <patdk-wk> oh, that, your suppost to see that
[15:08] <patdk-wk> every single multicast/broadcast packet it sends out one nic, will be dropped when it's received on the other nic
[15:10] <disposable> patdk-wk: i'd understand seeing dropped packets on the ethX device that isn't the current slave, but not on bond0
[15:11] <disposable> patdk-wk: on your bond0 interface is the number of dropped packets also around 1/5 of all received?
[15:20] <patdk-wk> 1/230th
[15:20] <disposable> patdk-wk: thanks.
[15:20] <patdk-wk> it exactly matchs the received and dropped count of my backup nic
[15:21] <disposable> patdk-wk: yes, same here.
[15:21] <patdk-wk> so that easily explains it
[15:21] <patdk-wk> it's a backup, anything received isn't suppost to be, and is a dup
[15:21] <patdk-wk> and is dropped
[15:21] <patdk-wk> as I said
[15:21] <disposable> patdk-wk: weird thing is that every couple of reboots i get 0 dropped packets.
[15:21] <disposable> by couple i mean >20
[15:28] <Shadowandlight> anyone have ideas on how to deploy a cloned copy of Ubuntu faster?  I need to change passwords (mysql and linux) and settings faster then doing everything by hand... if possible http://askubuntu.com/questions/337503/easiest-solution-to-modifying-linux-mysql-passwords-and-other-settings-after-c
[15:30] <disposable> Shadowandlight: i'd suggest chef/puppet/cfengine but that may be too much overhead if you're just changing a few things. in your case i'd simply start clusterssh, log into all machines at once and just click into individual windows to enter different passwords.
[15:34] <Shadowandlight> do the config tools run locally or do they need to be web based like webmind / cloudmin?
[15:52] <thebwt> are there any reccomendations for a php ppa? I'm trying to get newer version stuff for a 12.04 server
[15:55] <arosales> jamespage, smoser is there goingto be a Server IRC meeting today since there is 1308 vUDS?
[15:56] <jamespage> arosales, nope
[15:56] <jamespage> can't do sessions and irc meeting at the same time
[15:57] <arosales> jamespage, ok
[15:57] <arosales> jamespage, I thought so just wanted to confirm
[16:00] <smoser> arosales, i say no
[16:01] <arosales> smoser, ack thanks. I'll put a reminder in ubuntu-meeting since we didn't get a mail out to the list
[16:01] <arosales> or I may have missed it if it did go out :-)
[16:23] <disposable> Shadowandlight: unfortunately, chef/puppet require installation of ruby and lots of additional software. cfengine is smaller and faster but with the steepest learning curve. for clusterssh, you just need ssh.
[16:33] <rbasak> disposable: is there a real world problem with pulling in all this "overhead"? apt-get does it automatically for you, and it doesn't use a significant amount of space on an installed server node. I also don't like the ruby dependency, but I don't think it'd affect my recommendiation to look into chef/puppet over anything else.
[16:54] <wedgwood> Is there a good way to set environment variables when calling lxc-attach? I need something like --clear-env plus a few extra values.
[16:58] <wedgwood> hallyn_: If I might trouble you, any ideas? ^
[16:58] <rbasak> wedgwood: I call env after lxc-attach. lxc-attach --clear-env ... -- env foo=bar my_command
[16:59] <wedgwood> rbasak: ah, I'll give that a shot. there are ~a dozen things to set in some cases, but I still think that could be manageable
[17:25] <smoser> ok. i have a stupid question.
[17:26] <smoser> i wnat to make 'cloud-utils' package basically only depend on another package now.
[17:26] <smoser> transitional if you will.
[17:28] <hallyn_> wedgwood: had to look back at the changelog;  right now rbasak's is the best option.  There is code for 'extra_keep_env', but it's not yet hooked up to the cli
[17:29] <hallyn_> smoser: what is your q?  (does it belong in -devel? :)
[17:29] <smoser> i think its never mind.
[17:30] <hallyn_> ok :)
[17:30] <roaksoax> smoser: yeah that would work
[17:30] <roaksoax> smoser: cloud-utils would become a transitional package indeed
[17:31] <roaksoax> smoser: or you could make it a meta-package
[17:31] <smoser> whats a meta-package specifically ? a task?
[17:31] <roaksoax> smoser: no, a metapackage are packages that simply depend on others and do not install anything
[17:31] <smoser> example?
[17:32] <roaksoax> smoser: for example, 'maas' is a metapackage that installs 'maas-region-controller', 'maas-cluster-controller'
[17:32] <smoser> perfect. thank you.
[17:33] <roaksoax> smoser: now, transitional packages are usually needed for upgrades
[18:11] <gartral> hey all, I've been looking into Landscape to help manage a private server and I was wondering if I install the landscape-client package, and decide later that I don't want it, will it break the server if I remove it?
[18:12] <sarnold> gartral: I had no trouble on my machine removing the landscape client after playing with it for a few weeks
[18:13] <gartral> sarnold: and exactly how much IS the Landscape service, I'm quite dismayed and put off by Cononical's secrecy on that subject
[18:14] <sarnold> gartral: heh, I'm sorry I can't actually fix that :) I had a free account on account of being an employee...
[18:17] <sarnold> gartral: hey! this looks like it. :) http://www.canonical.com/enterprise-services/ubuntu-advantage/server
[18:18] <gartral> sarnold: 320 USD *PER SERVER*?! holey crap! never mind that!
[18:19] <sarnold> gartral: yes that seems a bit steep if landscape is the only part of the plan you want
[18:21] <gartral> sarnold: yea.. I was thinking it would be like $10-20 US/year for a single server.. $320 is far, far *far* too rich for my blood
[18:23] <sarnold> gartral: yes, it looks like it was priced to be competitive against rhel and sles: https://www.redhat.com/apps/store/server/ https://www.suse.com/products/server/how-to-buy/
[18:25] <gartral> sarnold: yea.. I'd rather just run my single server through an SSH console..
[18:25] <sarnold> gartral: and perhaps the "unattented-upgrades" package would provide a lot of what you'd want from landscape anyway
[18:38] <mgw> Is btrfs recommended for general use on server machines? I'm interested to use in conjunction with lxc. As I understand it, this will help reduce disk usage from duplicated container file systems.
[18:41] <RoyK> mgw: btrfs is not recommended for production systems
[18:41] <mgw> RoyK, thanks
[18:41] <xnox> mgw: you can equally use overlayfs or lvm snapshots for that.
[18:41] <RoyK> mgw: afaik btrfs doesn't have any dedup stuff yet
[18:42] <xnox> mgw: which are more stable than btrfs.
[18:42] <RoyK> xnox: 'cept lvm snapshots are dead slow :P
[18:42] <mgw> xnox: thanks, looking at overlayfs
[18:42] <xnox> RoyK: well dedup is being packaged by me, soon ;-)
[18:42] <RoyK> mgw: try zfs
[18:42] <xnox> RoyK: not in my testing.
[18:42] <sarnold> RoyK: is the zfs support entirely via fuse?
[18:42] <RoyK> mgw: zfs has a decent dedup implementation if you have *lots* of memory
[18:43] <RoyK> sarnold: there's zfsonlinux - separate ppa
[18:43]  * xnox wouldn't recommend zfs on linux to anyone. Solaris or Freebsd maybe.
[18:43] <RoyK> sarnold: works well
[18:43] <xnox> sarnold: zfsonlinux is not legally clean.
[18:43] <sarnold> RoyK: hrm. how is that legal?
[18:43] <sarnold> aha :) hehe
[18:43] <RoyK> xnox: heh - I've worked with zfs on rather large systems, as in quater of a petabyte, works well
[18:43] <sarnold> RoyK: linux?
[18:43] <RoyK> sarnold: it's legal, it's legal to download and install non-gpl software on your own
[18:44] <RoyK> sarnold: zfsonlinux does just that
[18:44] <mgw> ubuntu-zfs is zfsonlinux?
[18:44] <RoyK> sarnold: it's like graphics drives that compile on install
[18:44] <RoyK> mgw: no, ubuntu-zfs is fuse-based, dead slow
[18:47] <mgw> RoyK: ok… so I'll need to download the package from http://zfsonlinux.org/
[18:47] <RoyK> mhm
[18:48] <sarnold> RoyK: neat. thanks :D
[18:48] <RoyK> mgw: keep in mind that zfs doesn't support things like expanding a VDEV (that is, a RAIDz1 or RAIDz2 aka RAID-5 or RAID-6)
[18:49] <mgw> RoyK: I *think* that would be ok for us.
[18:49] <keithzg> xonx: It's only not legally clean to distribute the code compiled into the kernel; but for example a binary module for the kernel is just fine, same as the proprietary NVIDIA blobs are.
[18:50] <RoyK> mgw: takes a wee bit more of planning
[18:50] <keithzg> ^xnox, I mean.
[18:51] <sarnold> I think the zfs modules work in part because the code clearly originated elsewhere, rather than works only on linux.
[18:51] <sarnold> it's hard to claim it's a derived work when the primary platform was something else entirely :)
[18:52] <mgw> RoyK, xnox, sarnold: thanks for the assistance
[18:52] <sarnold> mgw: thanks for asking the question at the right time :) I'd not seen zfsonlinux before. it looks cool. :)
[18:54] <mgw> RoyK: is it stable?
[18:54] <RoyK> mgw: works for me (tm)
[18:54] <RoyK> mgw: and it's used by some large storage providers
[18:54] <mgw> ok, great
[19:04] <RoyK> mgw: what sort of storage are you planning?
[19:07] <mgw> RoyK: you mean the use case, or the type of hardware?
[19:22] <RoyK> mgw: both
[19:34] <RoyK> mgw: ping?
[19:34] <mgw> RoyK: most of my machines have 2x1TB with HW RAID
[19:34] <mgw> SATA
[19:34] <RoyK> ok
[19:35] <RoyK> for that, zfs should work well, that is, for even larger stuff like multihundred terabytes, zfs should work well, if properly planned
[19:36] <mgw> I'm wanting to back lxc containers in such a way as to not have to replicate the 500MB ubuntu files for every container…. does zfs have a versioning system that would allow that?
[19:36] <lifeless> mgw: have a look at lxc-start-ephemeral
[19:36] <lifeless> mgw: which does exactly that
[19:36] <RoyK> mgw: somehow, yes, you can clone a filesystem which will make it dedup what's there, but not the new stuff
[19:37] <RoyK> mgw: you can turn on dedup on the dataset, but it'll require large amounts of RAM to run efficiently
[19:37] <mgw> lifeless: lxc-start-ephemeral… so the / filesystem would be read only/shared?
[19:38] <lifeless> mgw: have a look at the script; it should answer your questions :)
[19:38] <mgw> (looking at man now)
[20:23] <stuxlof> I'm running ubuntu 11.10 x86_64 with a dell h700 raid controller containing megaraid sas firmware. I'm not sure if I either need to install megasasctl or megacli. I've got megacli64 running but it only gives me an 'Exit Code: 0x00'. Now megasasctl needs some 32 bit libraries (https://github.com/gnif/ARMT/tree/master/utils/megactl) to function, and this is where I'm stuck at atm
[20:26] <stuxlof> the megaraid_sas driver is loaded in the kernel
[20:27] <mgw> lifeless: do I understand that lxc-start-ephemeral still needs overlayfs?
[20:28] <lifeless> or aufs yeah
[20:28] <lifeless> you could do a similar script though
[20:29] <lifeless> or manually setup your lxc mount tables
[20:29] <lifeless> so that you have a readonly fs and a /var/state/myproject that is readwrite
[20:29] <mgw> yeah, i'm thinking i'd need to use that script as a ref
[20:29] <mgw> but create my own
[20:45] <delinquentme> I've got a file I've used for sshfs mounting remote files
[20:45] <delinquentme> I want to delete the local version of this file
[20:45] <delinquentme> sudo rm -rf localFile/
[20:46] <delinquentme> will do that without removing the content of the remote dir
[20:46] <delinquentme> correct?
[20:48] <sarnold> delinquentme: does the output of mount show the mount currently active?
[20:51] <delinquentme> mount: can't find dopamine_live/ in /etc/fstab or /etc/mtab  << looks like im good
[22:17] <Darkstar1> I can't seem login via ssh. I'm on 12.04
[22:22] <RoyK> Darkstar1: what error message?
[22:23] <Darkstar1> None. I think the server was hacked
[22:23] <Darkstar1> just looking through the  bash history
[22:23] <RoyK> lots of nice stuff? ;)
[22:24] <Darkstar1> it seems someone got in yesterday and unpacked something called portuser.
[22:24] <Darkstar1> Just trying to get ssh back
[22:25] <Darkstar1> ssh logs to system normally right?
[22:25] <RoyK> Darkstar1: try rkhunter and chkrootkit
[22:26] <Darkstar1> RoyK: It's a remote vm
[22:26] <Darkstar1> but I'll look at those toools
[22:26] <RoyK> they attempt to do the same job, but may overlap a bit
[22:26] <RoyK> Darkstar1: yiu can run those on a remote machine
[22:30] <Darkstar1> Think I need to boot into single user mode
[22:32] <sarnold> Darkstar1: best would be taking the machine entirely offline and inspecting the filesystem from a known-good system. preferably a known-good system you can throw away when you're done.
[22:35] <Darkstar1> sarnold: thing is I don't know what to look for tbh. I looked at the history of bash and managed to find the one thing that was installed. What that thing did I do not know, but I removed the compromised user and inspecting the bash history has allowed me to remove the unpacked stuff
[22:36] <sarnold> Darkstar1: check especially for unexpected dot files in all directories, debsums mismatches, unexpected kernel modules, etc.
[22:36] <Darkstar1> debsums?
[22:37] <sarnold> Darkstar1: debsums checks the sizes / hashes of dpkg-managed files, it's a useful way to get a quick overview of what might have changed if an installed rootkit is pretty amateurish
[22:39] <Darkstar1> ok.
[23:52] <crass> something has changed wrt to how ubuntu handles inactive lvm vgs within the last few releases
[23:52] <crass> any one have ideas on how to really release all resources using a vg so that I can luksClose the underlying device?