[00:09] <Kutakizukari> Is there a way to block port 22 SSH till I need to log in? I know dome9.com does this but looking for the free way if can do.
[00:10] <KM0201> hmm, not sure
[00:12] <maxb> Kutakizukari: If you feel it's necessary, you could go with a port knocking approach
[00:13] <maxb> Alternatively, accept only pubkey-based logins, denying passwords completely, and trust that to be secure enough
[00:13] <Kutakizukari> was using fail2ban but the bots just grab another ip address and try again.
[00:14] <maxb> Possibly backed up with a fail2ban approach
[00:14] <Kutakizukari> is there docs on pubkey-based logins for ubuntu server 10.04 LTS?
[00:15] <patdk-lap> Kutakizukari, you want a knock solution probably
[00:15] <patdk-lap> where you hit up a special port y, and it enables access to other port x for that ip for x min
[00:15] <Kutakizukari> patdk-lap: never heard of it, will look it up. Thanks
[00:16] <patdk-lap> http://www.shorewall.net/PortKnocking.html
[00:16] <Kutakizukari> can port y be sniffed out to lead to port x?
[00:17] <Kutakizukari> will look into it, again thank you
[00:17] <maxb> Pubkey based SSH is a standard part of OpenSSH. I know it's described in the man pages, no idea if anyone's bothered to write customized documentation for Ubuntu; really there's little to nothing distro-specific
[00:17] <patdk-lap> not without someone intercepting your traffic, while you do it
[00:17] <maxb> !info knockd
[00:18] <patdk-lap> odd, no need for a userbased program, as iptables can do it all itself
[00:18] <maxb> it can?
[00:18] <patdk-lap> sure
[00:18] <patdk-lap> the link above is almost raw iptables commands
[00:19] <maxb> I've never come across one iptables rule mutating the ruleset itself before
[00:19] <patdk-lap> I never said 1 iptables rule :)
[00:19] <patdk-lap> but a collection of 4 will do it, and protect against portscans
[00:20] <patdk-lap> if you don't care about portscans, 2 rules, the normal accept 22 and the portknock port
[00:21] <maxb> Interesting. I was not aware of the 'recent' module
[00:22] <patdk-lap> part of normal ubuntu iptables
[00:22] <patdk-lap> no xtables needed
[01:47] <romulobr> hi, i want my server to be able to send emails. I followed ubuntu server guide and successfully (i think) installed postfix with dovecot, but i don't know what to do next, can you help me?
[04:23]  * pehden is away: I'm busy
[04:23]  * pehden is back (gone 00:00:01)
[06:29] <journeeman> Hi all. Trying to delete a node from the MAAS web interface gives an `Internal Server Error'. Anyone else facing this?
[06:52] <journeeman> cobbler.log shows the "At least one interface needs to be defined" exception being raised. checking my cobbler config
[08:02] <mmmfungo> hello! i was referred here in an attempt to get my usb ports running at 2.0 instead of 1.1
[08:02] <mmmfungo> good morning! im having an issue with my usb ports and was hoping for some help...i recently installed a firewire/usb/audio front panel and the ports are only recognized as usb 1.1, even though the board and panel both support 2.0..the output of lsusb and lspci as well as the motherboard and front panel info can be seen here : http://paste.ubuntu.com/921401/ .. lsmod here : http://paste.kde.org/454208/ .. the chipset, according to
[08:02] <mmmfungo> the manual is an Intel 82801GH 1/O Controller Hub (ICH7DH) and im using 11.10..any help in getting the ports up to speed and running as 2.0 would be greatly appreciated
[09:25] <wmp> hello, i have problem with 12.04 and repositores
[09:26] <wmp> i want to install php5.4 from ppa and i have error:
[09:26] <wmp> Depends: libonig2 (>= 5.2.0) which is a virtual package.
[09:26] <wmp>             Depends: libqdbm14 (>= 1.8.74) which is a virtual package.
[09:26] <wmp> where i can found this package?
[10:34] <sw> !info libqdbm14 | wmp
[12:28]  * koolhead11 thinks everyone is having Easter Fun time
[14:01] <tash> Is it easy to mount external storage locally to use as your, say, mysql data drive?  I am not sure if fdisk -l will show external storage
[14:14] <rurufufuss> greetings, is there a command line gui (like that of aptitude) for updating alternatives?
[14:14] <rurufufuss> (e.g trying to set valgrind to point to the newer version)
[15:30] <wmp> sw: thx
[15:45] <aeol> hey, does anyone here work for Canonical?
[15:49] <aeol> I have been trying to convince my boss (who knows nothing about linux) that we should use Ubuntu in prod instead of Redhat and it seems no one at Canonical can be bothered to respond to my request to purchase their enterprise services
[15:54] <rockets> Is there a way to apt-get install security updates only
[15:55] <rockets> aeol, I've contacted canonical sales several times and always gotten an immediate reply sir.
[15:56] <Pici> rockets: I don't think I understand your question, apt-get should always pull the highest version package available.
[15:57] <aeol> @rockets how did you contact them? there is no phone number, only a web page to submit a contact request
[15:57] <romulobr> can you help me to set up an email server to send emails only, i followed the tutorial on ubuntu guide, and things seems to be working, but my client can't send emails.
[15:59]  * koolhead11 digs url to pass aeol 
[16:00] <koolhead11> aeol, https://forms.canonical.com/sales/  this might help
[16:00] <rockets> Pici, I only want to install security updates not general bugfixes or new versions
[16:01] <rockets> e.g. i only want to install things from hardy-security
[16:01] <Pici> rockets: Then comment out the updates and backports lines in /etc/apt/sources.list
[16:02] <rockets> Pici, how is it that unattended updates can install only security updatse
[16:02] <rockets> but theres no way to do that manually from apt
[16:14] <hallyn> rbasak: hey, did you get around to running the lxc testsuite on arm 2 weeks ago?
[16:15] <hallyn> if not, could you, this week? :)
[16:22] <hggdh> jamespage: do you perform remote Jenkins job submission in the server lab? I am finding that if I pass ?token=... I get a 403
[16:22] <hggdh> jamespage: if I take '?token=...' out, and go for a basic auth, it works
[16:32] <wbullock8> can anyone help me with an issue I'm having regarding and IP Address change and using apt-get update
[16:33] <wbullock8> I changed the IP address and now I can't get apt-get update to connect... but I'm able to see everything on the internet
[16:33] <wbullock8> the moment I change the IP address back to what it was apt-get update works fine
[16:34] <wbullock8> any help would be super awesome!
[16:39] <bluefrog> you change from what to what?
[16:39] <bluefrog> and how do you change it, what is your dns server and so on. we need ino
[16:39] <bluefrog> infp
[17:09] <kirkland> SpamapS: howdy!
[17:09] <kirkland> SpamapS: curious, still no apache upstart script, eh?
[17:09] <SpamapS> kirkland: I see no reason for there ever to be one.
[17:09] <SpamapS> shocking I know
[17:10] <kirkland> SpamapS: erm?
[17:10] <SpamapS> but really, apache has a rich init.d script that they maintain perfectly.
[17:10] <kirkland> SpamapS: okay
[17:10] <SpamapS> and I don't see apache ever being part of the system plumbing
[17:10] <kirkland> SpamapS: so, let's say I have a script that I want to run before apache starts
[17:10] <kirkland> SpamapS: what's the best way for me to do that?  modify the apache init script itself to call me?
[17:11] <kirkland> SpamapS: or put myself just before them in the init.d list?
[17:11] <SpamapS> kirkland: start on starting rc RUNLEVEL=[2345] .. will run your script before *all* sysvinit.
[17:11] <kirkland> SpamapS: hmm
[17:11] <SpamapS> kirkland: (needs to be a 'task' too)
[17:12] <kirkland> SpamapS: and I need networking to be up
[17:12] <SpamapS> kirkland: its up
[17:12] <kirkland> k
[17:12] <SpamapS> as of 11.10 anyway
[17:12] <SpamapS> before that.. you never knew when networking was "up"
[17:13] <SpamapS> kirkland: apache is less important now anyway. Seriously, I think its day has passed. nginx will *blow your mind* how much better it is.
[17:13] <kirkland> SpamapS: them's strong words
[17:13] <SpamapS> yep
[17:13] <SpamapS> I haven't tried apache 2.4 yet..
[17:14] <SpamapS> but nginx certainly utterly destroyed 2.2 in every way for omgubuntu.co.uk
[17:14] <kirkland> SpamapS: yeah, jcastro was raving about it too
[17:15] <SpamapS> The area where apache will win out is in mods for auth and stuff.. but many of those are making their way into nginx land as well.. so I think its only a matter of time before we see a real shift.
[17:16] <patdk-wk> I have 2.4 installed on percise to play with
[17:16] <patdk-wk> but haven't done any performance testing yet
[17:17] <SpamapS> kirkland: also I'm not as excited about moving every service into upstart. I think its more important that classes of services be moved there, but stuff like apache that just pops up on a port can happily stay in sysvinit land for as long as it makes sense.
[17:18] <patdk-wk> doesn't that run into dependancy issues?
[17:19] <patdk-wk> expecially if I need to make sure my fastcgi servers are running before apache
[17:21] <SpamapS> patdk-wk: why do you need to make sure your fastcgi servers are running before apache?
[17:21] <SpamapS> what if they are on another host?
[17:21] <SpamapS> Its a false hope to try and get all daemons started in the right order in a modern system.
[17:21] <patdk-wk> if they are on another host, apache will retry the connection
[17:22] <SpamapS> Right
[17:22] <patdk-wk> if you use a socket file, I think it just errors
[17:22] <SpamapS> and if they are local, it will retry as well. :)
[17:22] <patdk-wk> or maybe that was fixed
[17:22] <SpamapS> patdk-wk: right, fixed.. because thats a bug if it doesn't. :)
[17:22] <rockets> What is the official way to change the hostname of an ubuntu 10.04 server permanently? I'm seeing so much conflicting advice. I'm seeing edit /etc/hosts, create an /etc/hostname, just use the hostname command (doesn't do it). I'm not sure which I should do.
[17:23] <SpamapS> patdk-wk: either way, whats more important is that network clients be generous about retries and degrade themselves/report errors properly.. not that daemons are all started in the right order.
[17:23] <patdk-wk> rockets, lots of changes to do that :)
[17:23] <SpamapS> Ordering is nearly impossible in a truly distributed system.
[17:23] <rockets> patdk-lap, What is the "proper" way :P
[17:23] <patdk-wk> personally? grep -R oldname /etc/
[17:24] <patdk-wk> and change every file you locate :)
[17:24] <patdk-wk> it's in 5 places I can think of, in a normal plain install
[17:25] <rockets> lol
[17:25] <SpamapS> rockets: the modern way is to have DHCP handing out hostnames, and change it on the DHCP server. :)
[17:25] <rockets> in other words, it's unmanageable :D
[17:25] <rockets> SpamapS, yeah well, we don't have dhcp on virtual slices on linode
[17:26] <SpamapS> how do they do that? do they inject /etc/network/interfaces for you or something?
[17:26] <SpamapS> I'm genuinely curious.
[17:26] <rockets> mm
[17:26] <patdk-wk> well, for most things, just /etc/hosts, /etc/hostname will do it
[17:26] <rockets> ill pastie my interfaces file for you if you're interested
[17:26] <patdk-wk> but then you have other configs, like mail server, xmpp server, webserver, ...
[17:26] <rockets> patdk-lap, yeah but nothing is configured yet
[17:26] <rockets> its a new host
[17:26] <patdk-wk> well, then you should only have to touch those two files then
[17:27] <patdk-wk> then reboot it
[17:27] <rockets> i just dont want the randomly generated string for a hostname i have
[17:27] <SpamapS> rockets: no thats ok, I'm just wondering how virtual providers can do it any other way.
[17:27] <rockets> SpamapS, actually my /etc/network/interfaces contains NOTHING other than loopback
[17:27] <rockets> maybe they do have some sort of internal dhcp
[17:27] <SpamapS> rockets: wow.. eth0 isn't even listed?
[17:27] <rockets> SpamapS, nope
[17:27] <rockets> I have no idea how they do it
[17:27] <rockets> its xen though
[17:27] <SpamapS> weird
[17:27] <patdk-wk> what kernel?
[17:28] <rockets> oh wait
[17:28] <SpamapS> perhaps they jiggered their own startup scripts in
[17:28] <rockets> im looking at the wrong server
[17:28] <rockets> lulz
[17:28] <rockets> hold on
[17:28] <rockets> yeah it looks like it is dhcp, "auto eth0"
[17:28] <rockets> iface eth0 inet dhcp
[17:28]  * SpamapS really hopes virtual host providers fall in line and start using the official Ubuntu cloud images instead of all this weirdness.
[17:28] <rockets> i dont have ctonrol over the dhcp though
[17:28] <SpamapS> rockets: gotchya
[17:28] <rockets> SpamapS, well, this is 10.04 when they had to make their own modified kernel to make it work
[17:29] <rockets> because it wasnt xenified or something
[17:29] <SpamapS> rockets: we have 10.04 EC2 images.. EC2 == xen.. so... ?
[17:29] <rockets> or maybe that was for kvm
[17:29] <patdk-wk> ya, I use the ec2 kernel on xen
[17:29] <rockets> this is what slicehost told me, but im using linode now
[17:29] <SpamapS> meh, well anyway.. no worries. So with dhcp You may have to tell it to not overwrite your hostname
[17:29] <rockets> SpamapS, nah, /etc/hostname works
[17:30] <SpamapS> rockets: hopefully that sticks. :)
[17:30] <rockets> SpamapS, I just tested it.
[17:30] <rockets> well i suppose my lease hasn't been renewed
[17:30] <rockets> we'll see hehe
[17:30] <SpamapS> rockets: reboot would test it
[17:30] <rockets> SpamapS, i've rebooted several times
[17:31] <patdk-wk> if it doesn't, go edit /etc/dhcp3/dhclient.conf
[17:36] <SpamapS> rockets: alright, could you please go delete all the confusing docs you read from the intarwebs kthxbai
[17:37] <rockets> lol
[17:49] <thesheff17> I currently use apt-mirror and manage my own mirror.  I wonder if it is was possible to create a date based mirror where I could just do http://ubuntuMirror/ubuntu/20120904/ and it would only contain those files from that date off the mirror...so basically a rolling snapshot of a mirror.
[17:49] <adam_g> win 10
[18:03] <hallyn> utlemming: haven't checked your patch yet, but if you haven't yet, could you roll in the minor change that stgraber has staged at ubuntu:lxc ?
[18:05] <hallyn> (lunch, biab)
[18:17] <zul> adam_g: im thinking we should do some more database smarts in the packaging
[18:20] <adam_g> zul: like?
[18:20] <zul> adam_g: backup a copy of the database if the schema changes
[18:21] <adam_g> zul: how does that happen for a populated databse with millions of rows?
[18:21] <adam_g> or for an external database
[18:21] <zul> adam_g: i dont know yet
[18:21] <adam_g> zul: i dont see that happening for 12.04, tbh
[18:22] <zul> adam_g: well yes not for 12.04
[18:23] <zul> adam_g: its just a gleam in my eye right now
[18:24] <thesheff17> I would love a way to use source.list with a date of a snapshot of mirror.  I wrote a program that symlinks all of the mirror to a folder in apache but I'm sure certain text files change...it prob should also have db behind it.
[18:27] <SpamapS> zul: smarts in packaging == fail usually. ;)
[18:27] <SpamapS> zul: that sounds like something openstack should support.
[18:27] <zul> SpamapS: *cough* bacula
[18:28] <SpamapS> as in, if its going to do a schema change, it should have a '--backup' option that does the backup, we shouldn't do it all in the preinst or something.
[18:28] <SpamapS> that stuff always comes back in floods of apport bugs :-P
[18:44] <RoyK> something tells me this value is a bit above "normal"   7 Seek_Error_Rate         0x000f   045   045   030    Pre-fail  Always       -       11282972468059
[18:45] <SpamapS> RoyK: thats a bit above normal, if the saucer section of the Enterprise is only considered "a bit" of the whole ship.
[18:47] <RoyK> something in the terms of "To say that Richard Mayhew was not very good at heights would be ... It would be like describing the planet Jupiter as bigger than a duck."
[18:47] <RoyK> drive still works, though
[18:48] <RoyK> but I would guess replacing it soon, as in tomorrow, might be a good idea ;)
[19:18] <railsraider> hi guys, my HAproxy went down completely unresponsive and i had to hard boot it, i am trying to figure out why
[19:18] <railsraider> i have fail2ban, arno-iptables, and chkrootkit installed
[19:19] <railsraider> all i can see from the logs is that eth0 went promiscuous and i think that accepted all traffic
[19:20] <sebokie> hello
[19:21] <sebokie> I have a ubuntu server with nfs running with only 127.0.0.1 allowed for my share
[19:21] <sebokie> then I create a ssh tunnel between the client and the server as described here: http://nfs.sourceforge.net/nfs-howto/ar01s06.html
[19:22] <sebokie> however I am not able to mount the nfs share, and I don't get any error message
[19:23] <sebokie> is it supposed to be logged by the server when someone attempts to mount a nfs share?
[19:25] <hallyn> jdstrand: http://paste.ubuntu.com/922334/  I assume given the comments in the code, I'm on my own with this failure?
[19:28] <rockets> Here's a silly question: is there a way to make all subdirectories of a directory inherit the parent's permissions when created? I tried sgid, no dice.
[19:36] <jdstrand> hallyn: yeah-- why qemu isn't output anything from 'info block' seems like a regression
[19:36] <hallyn> jdstrand: it does output for info block
[19:36] <jdstrand> hallyn: I'm confused
[19:37] <hallyn> I assumed the 'kill monitor; run nc by hand' bit was just not working
[19:37] <hallyn> jdstrand: if i just run qemu by hand and say 'info block', i get results
[19:39] <jdstrand> hallyn: but that isn't what the script is doing. it is launching it via livirt, then trying to connect to libvirt's monitor file
[19:40] <hallyn> jdstrand: i see, i was thinking it failed getting any output at all.  but still, if i do 'info block' by hand, it works.
[19:40] <jdstrand> hallyn: in fact, based on the weird formatting of the error output, it looks like 'nc -U <libvirt monitor file>' failed
[19:40] <hallyn> on the bright side, it's the last failing case i think \o/

[19:41] <jdstrand> hallyn: there is a timeout there-- I know I can't run test-qemu.py on a heavily loaded system. maybe the your system was under high load?
[19:41] <hallyn> jdstrand: hm, ok, i'll try again then, thanks.
[19:42] <jdstrand> (as in, maybe you are seeing with list-libvirt.py the types of things I would see with test-qemu.py)
[19:47] <rockets> An especially dumb question - I can't seem to find a regular git client package in the lucid repos. Is it not there? I see a lot of things that refer to git but no actual git
[19:47] <rockets> ah, git-core
[20:27] <gary_poster> zul or adam_g, I've got an openstack question if one of you have a moment.  For Canonistack, I want the publicDnsName and privateDnsName to have a *.canonistack domain suffix rather than a *.novalocal domain suffix.  That way, with a bit more configuration on chinstrap, devs can set up a .ssh/config stanza that makes everything just work (particularly nice for juju).  AFAICT, that's controlled on the Openstack side
[20:27] <gary_poster> by nova.network.linux_net's dhcp_domain value, so I asked IS to make that change, along with the resolution bits on chinstrap.  They've done so, and I've tested, and things are not quite working as I hoped yet.  The change does make the vm correctly report its hostname as, for instance, server-8275.canonistack, but publicDnsName and privateDnsName now have no suffix at all, for instance reported as simply "server-82
[20:27] <gary_poster> 75" by euca-describe-instances.  Is there another openstack setting we can make to get those values to report the *.canonistack names, or is this an openstack bug, or can you suggest some other path to try to find out the answer?
[20:27] <gary_poster> eh, that was a novel, sorry
[20:38] <smoser> SpamapS, bug 820699
[20:38] <smoser> i was about to open "tgt does not start on installation"
[20:38] <smoser> is that a dupe of this bug ?
[20:42] <hggdh> jcastro: there?
[20:43] <jcastro> yo
[20:43] <hggdh> jcastro: OK. I already fscked my maas install -- forgot the useruser password...
[20:43] <hggdh> jcastro: anyway of recovering apart from full uninstall & reinstall?
[20:44] <jcastro> no clue, I've only done rudimentary set up with it
[20:44] <hggdh> heh
[20:44]  * hggdh goes for the reinstall
[20:44] <jcastro> wait!
[20:44] <jcastro> we should figure that out
[20:44]  * hggdh waits
[20:45] <hggdh> jcastro: another thing -- I installed maas-dhcp, and nothing changed in the dnsmasq config
[20:45] <hggdh> at least visibly
[20:46] <jcastro> hggdh: no clue that one either, but here's the first one: http://askubuntu.com/questions/120436/how-do-i-reset-my-maas-username-password
[20:46] <jcastro> matsubara: any idea? ^^
[20:47] <hggdh> jcastro: will follow the Q, thanks
[20:47] <jcastro> "maas changepassword" seems to exist
[20:48] <hggdh> jcastro: and... it works! -- it does change the password without requiring to enter a current one
[20:48] <matsubara> hggdh, jcastro: you can flush the db and create a new password
[20:49] <hggdh> (pretty much like sudo passwd xyz)
[20:49] <matsubara> well, I guess that's even easier :-)
[20:49] <jcastro> ah but how do you list users?
[20:49] <jcastro> without dumping the whole db I think?
[20:51] <hggdh> yeah, adding an answer to askubuntu now
[20:51] <hggdh> just a sec
[20:51]  * jcastro nods
[20:52] <matsubara> jcastro, not very easy but you can do something like this: http://pastebin.ubuntu.com/922456/
[20:53] <jcastro> hggdh: ^^ good enough for an answer for now, heh
[20:53] <hggdh> matsubara: better than 'sudo dumpdata |grep username'
[20:53] <hggdh> wich is what I used
[20:54] <hggdh> jcastro: I will edit the answer to add super-matsubara's way of finding an user
[20:54] <Daviey> hola
[20:54] <Daviey> How is everyone?
[20:57] <guntbert> Daviey: welcome - do have a support question?
[20:57] <guntbert> *do you have ...
[20:58] <Daviey> guntbert: I hadn't, no.. do you?
[20:58] <smoser> utlemming, any ideas here: http://paste.ubuntu.com/922462/
[20:59] <smoser> that is from a us-east-1 instance
[20:59] <smoser> freaking slow
[20:59] <guntbert> Daviey: no, are you aware that this channel is not intended for chatter? If yes - my apologies
[20:59] <zul> lol
[20:59] <hggdh> Daviey: you got no support question??
[21:00] <jcastro> he should be kicked out!
[21:00] <hggdh> LOL
[21:00] <utlemming> smoser: cloud-front?
[21:00] <zul> Daviey: you are the weakest link good bye
[21:00]  * sw hands Daviey the AK47
[21:01] <Daviey> guntbert: Are you sure this channel isn't for chatter?
[21:02] <smoser> utlemming, just a download from cloud-images.ubutnuc.om
[21:02]  * guntbert tries hard to get his foot out of his big mouth :-)
[21:02] <utlemming> smoser: I'm checking from eu-west-1 right now
[21:03] <hggdh> oooohhh Daviey gains superpowers, and kills self
[21:04] <utlemming> smoser: I get pretty decent internet (15Mb/s) and I pull the images this morning at around 200k, so I am wondering if its crossing that small puddle in between New York and the UK
[21:05] <smoser> well, normally i think its better than that.
[21:06] <smoser> i'mi using lftp and pget now
[21:06] <guntbert> Daviey: to answer your last question: according to the /topic I should be sure - listening to some of the regulars here I get doubts :)
[21:06] <patdk-wk> utlemming, the speed is kind of iffy
[21:06] <patdk-wk> some days I can get 2-8MB/sec, others I'm happy to even get 100k
[21:06] <utlemming> smoser: 7.0MB/s in Eeu-west-1
[21:07] <smoser> utlemming, well...
[21:07] <smoser> i have strong suspicion that you can't really trust that.
[21:07] <smoser> that things get cached.
[21:07] <smoser> ie, if you had run that 2 times, magically the second would be faster.
[21:09] <utlemming> So we need a European to test my theory
[21:10]  * patdk-wk builds one
[21:11] <Daviey> utlemming: I am a European !
[21:11] <hggdh> no, you are English...
[21:11] <patdk-wk> swallow?
[21:12] <utlemming> Daviey: Can you do "wget http://cloud-images.ubuntu.com/server/precise/20120409/precise-server-cloudimg-amd64.tar.gz -O /dev/null" and report the time that it took?
[21:12] <Daviey> utlemming: European datacentre or homeish adsl?
[21:13] <utlemming> Daviey: which ever one is less-likely to implement caching and if the DC isn't Canonical or Amazon, that might be the best benchmark
[21:14] <Daviey> utlemming: Non-Canonial or Amazon, Downloaded: 2 files, 211M in 19s (11.1 MB/s)
[21:15] <patdk-wk> that all?
[21:15] <patdk-wk> 2012-04-09 17:15:09 (11.8 MB/s) ashburn,va
[21:17] <patdk-wk> 2012-04-09 17:16:43 (6.04 MB/s) comcast (maryland) home
[21:18] <utlemming> smoser: my home slowness is due to a 62% packet loss thanks to L3
[21:22] <RoyK> L3?
[21:22] <sw> Daviey: utilised /kick over my AK47! :<
[21:23] <hallyn> jdstrand: ok, it's simpler than i thought
[21:23] <hallyn> jdstrand: the _stop_daemon() just does killall -9 libvirt-bin
[21:23] <hallyn> upstart respawns it :)
[21:23] <hallyn> hm, no
[21:24] <hallyn> ok, i need to figure out why it's not going away,b ut second part was,
[21:24] <hallyn> the monitor isnow json
[21:24] <hallyn> so 'info block' will give back errors regardless
[21:25] <jdstrand> oh, that would do it :\
[21:25] <jdstrand> annoying
[21:28] <hallyn> yes
[21:30] <hallyn> jdstrand, but wait... courtesy of our dberrange,
[21:30] <hallyn> virsh qemu-monitor-command --hmp qatest-i386  'info block'
[21:30] <SpamapS> smoser: back from long lunch.. looking now
[21:31] <jdstrand> hallyn: oh, easy enough-- then on release >= 12.04 pass '--hmp'
[21:32] <hallyn> yup :)
[21:57] <RoyK> jdstrand: I first read --hmp as something like --hrmpff!
[22:08] <jdstrand> RoyK: hehe
[22:22] <ideaman> Anyone: I'm scripting an install in Ubuntu, and having some issues matching a string with sed. Any takers?
[22:24] <SpamapS> ideaman: just ask, and hang out for a while. Some people read backscroll :)
[22:24] <ideaman> k
[22:24] <SpamapS> ideaman: also askubuntu.com and serverfault.com are good places as well
[22:24] <ideaman> cool thanks
[22:25] <Patrickdk> the question, we will never know
[22:26] <ideaman> I'm trying to replace lines 7 and 11 with new text via sed, so I have sed -i 's/old text/new text/7' but anything I try never works...
[22:26] <ideaman> the best I can get is a global replace /g, which I don't want
[22:27] <Patrickdk> seems odd way to do it
[22:27] <ideaman> I'm up for any suggestions
[22:28] <Patrickdk> '7 c \New line here'
[22:32] <ideaman> Patrickdk: thanks a million, that's so much easier
[22:32] <smoser> SpamapS, tgt ?
[22:32] <SpamapS> smoser: still reading
[22:32] <smoser> k
[22:38] <locuse> hi.  i'm putting my 1st ubuntu-servers into production.  i'd like to configure critical-power event recovery (effectively, yank the power cord in mid operation ...) so that after power recovery, the server automatically powers up, trying a maximum of "5" times if unsuccessful.  BIOS is involved, i'd guess, but what are the appropriate pieces on a headless Ubuntu server?
[22:43] <Patrickdk> locuse, you would need a bios to do that
[22:44] <Patrickdk> well really probably a bmc
[22:44] <Patrickdk> and the ones I have only let you do it 3 times, not less, and not more
[22:44] <Patrickdk> and it requires some kind of os watchdog program
[22:44] <Patrickdk> and the amount of time that os watchdog program checks in, is fixed
[22:44] <Patrickdk> I normally turned those things off, cause the bios couldn't boot fast enough, before it would timeout, let along the os starting up
[22:45] <locuse> Patrickdk: "bmc"?
[22:45] <Patrickdk> yep
[22:46] <locuse> acronymitis .... ah, that's Baseboard Management Controller (BMC)
[22:47] <locuse> hm.  one'd think this'd be a bit more straighforward .
[22:51] <SpamapS> smoser: I have not forgotten you on tgt..
[22:57] <SpamapS> smoser: that upstart job is *really* really wrong.
[22:58] <SpamapS> smoser: that said, the reason tgt is not started on install is that the postinst does not start it
[22:58] <SpamapS> smoser: 	dh_installinit --no-start -u"start 89 2 3 4 5 . stop 11 1 ."
[22:59] <SpamapS> smoser: which is intentional
[23:00] <SpamapS> smoser: remove the args to dh_installinit and that will be fixed
[23:00] <SpamapS> smoser: also the stop on is wrong.
[23:00] <SpamapS> as is the start on actually
[23:09] <miceiken_> How do I make my ubuntu server obtain the correct time and date?
[23:10] <SpamapS> miceiken_: install the 'ntp' package
[23:11] <SpamapS> miceiken_: note that ntp will refuse to change the time/date by a large amount (it will just keep the time accurate), so you  may *also* want to install 'ntpdate' and force it in while ntp is not running.
[23:11] <miceiken_> i just did an "ntpdate ntp.ubuntu.com"
[23:12] <miceiken_> what about timezones though, does it figure that out by itself?
[23:23] <SpamapS> miceiken_: you need to set your timezone during install, but yes, NTP always communicates with NTP servers in UTC
[23:25] <smoser> SpamapS, so how bad is it that that doesn't start ?
[23:25] <smoser> and how, if you dpeend on it, should you start it ?
[23:26] <SpamapS> smoser: it affects all users, but has a workaround (manually start it) so I'd call it Medium.
[23:26] <smoser> well, as you said, its by design.
[23:26] <SpamapS> smoser: for something that Depends: tgt .. they can workaround by making sure it is started. But I'd think at that point.. just make it start.
[23:26] <SpamapS> its not so much by design..
[23:26] <SpamapS> as by the way its always been
[23:27] <SpamapS> policy suggests that if there is a sane default configuration, a daemon should be started on installation
[23:27] <SpamapS> not knowing tgt, I can't answer whether that if is true or not
[23:29] <miceiken_> wow how do I not know this
[23:29] <miceiken_> how do I output the time
[23:29] <miceiken_> from the server
[23:29] <RoyK> date
[23:30] <miceiken_> great, thanks RoyK
[23:30] <miceiken_> and thank you SpamapS, my time is all syncronized now :)
[23:30] <RoyK> date +%H:%M
[23:30] <RoyK> for instance
[23:32] <smoser> SpamapS, the daemon will start, will listen on a standard port, and have no targets if queried.
[23:32] <smoser> so, for my limited knowledge, that is a sane default
[23:48] <SpamapS> smoser: agreed. Thats probably a bug worth fixing in precise.. whether we do it b4 release or not.. I dunno ;)
[23:52] <smoser> well, personally, id say if its not fixed before release, then its not fixed in precise.
[23:52] <smoser> bug 977621
[23:52] <smoser> SpamapS, ^