[00:20] <adam_g> smoser: ping
[00:22] <adam_g> smoser: if you see this, do these traces look like anything obvious to you? http://paste.ubuntu.com/809214/
[00:24] <adam_g> smoser: ah, nvm.
[01:27] <bjf> roaksoax: koan is still busted for me  http://pastebin.ubuntu.com/809254/
[02:44] <roaksoax> bjf[afk]: will have a look
[03:39] <b930913> What's the best way to set up a email daemon on a server I have to receive emails sent to a domain pointing at it?
[03:59] <Titomen> anybody can help me here: rgotten@myplasticare:~$ sudo apt-get -y --force-yes -f install libgd-perl
[03:59] <Titomen> [sudo] password for rgotten:
[03:59] <Titomen> Reading package lists... Done
[03:59] <Titomen> Building dependency tree
[03:59] <Titomen> Reading state information... Done
[03:59] <Titomen> Package libgd-perl is not available, but is referred to by another package.
[03:59] <Titomen> This may mean that the package is missing, has been obsoleted, or
[03:59] <Titomen> is only available from another source
[03:59] <Titomen> E: Package libgd-perl has no installation candidate
[03:59] <twb> What a horrible hostname
[04:01] <Titomen> can you help
[04:01] <twb> Titomen: some people are beyond help
[04:02] <Titomen> please elaborate
[05:27] <roaksoax> bjf[afk]: found the issue! bug #918538 will fix tomorrow
[07:29] <goddard> if i wanna use SSL do i need to buy a cert?
[07:42] <greppy> goddard: depends on how "official" you want to be.
[08:17] <eagles0513875> hey ikonia found the root cause of virt-manager now not being able to create a pool
[10:04] <lynxman> morning o/
[10:22] <Alan> when you regenerate ubuntu's snakeoil SSL certificate, where is the CA key it uses?
[10:22] <Alan> I want to import that into my browser so i don't get "invalid SSL certificate" warnings
[10:25] <_ruben> isn't it self-signed and thus no CA involved?
[10:26] <Alan> _ruben: does that make it impossible to add as trusted?
[10:26] <Alan> I'm not actually sure i entirely understand the SSL certificate signing... is it just signed with another SSL certificate?
[10:26] <Alan> which might in turn be signed by another SSL certificate, which might be one that is trusted?
[10:27] <Alan> and it just follows the chain to find out if the root is trusted?
[10:27] <_ruben> roughly, yes
[10:27] <_ruben> the types of those ssl certs differ tho
[10:28] <Alan> ok
[10:28] <Alan> so a self-signed cert is actually an unsigned cert?
[10:28] <_ruben> it's signed by itself :)
[10:28] <Alan> ok
[10:28] <Alan> so that makes it its own CA?
[10:29] <_ruben> sort of :)
[10:29] <Alan> but does that mean there is a certificate somewhere in the process that i could use to say "trust this certificate" ?
[10:29] <_ruben> you could create your own personal CA, and use that to sign certs for your ssl services. then you could add that ca to your trusted ca store
[10:29] <Alan> but you can't add the cert itselfd to the trusted CA store
[10:30] <Alan> because it's not a CA
[10:30] <Alan> but it kinda is
[10:30] <_ruben> that's why i said "create your own personal CA, ...." ;)
[10:30] <_ruben> then you can
[10:30] <_ruben> check some openssl howtos, there's plenty of those around
[10:30] <_ruben> or use a tool like tinyca2
[10:30] <Alan> i was hoping to trust ubuntu's snakeoil
[10:31] <_ruben> there's really not much point in that
[10:31] <Alan> jumping through a load of hoops kinda defeats the point of how "simple" the snakeoil cert setup is...
[10:31] <_ruben> it being simple also makes it limited :)
[10:33] <Alan> seems like it's possible to add it to the OS's certificate store though...
[10:33] <Alan> hmmm
[10:33] <allenap> rvba: Can you give me a +1 in https://code.launchpad.net/~allenap/maas/test-bling/+merge/89078?
[10:37] <_ruben> Alan: i must admit that my experience with the client side (the trusting of certs/cas) on linux is rather limited, most my client experience is with windows, for the server side i have experience with both :)
[10:37] <_ruben> so whether or not it's possible to trust a single cert (not a ca) on ubuntu, i don't dare to tell ;)
[10:38] <jamespage> negronjl, I'm going to scrub the hive package from the PPA (we need to use 0.7.1)
[10:38] <Alan> _ruben: it's actually pretty easy, it seems, to add a cert to NSS (which both firefox and chrome use)
[10:38] <Alan> http://blog.avirtualhome.com/2010/02/02/adding-ssl-certificates-to-google-chrome-linux-ubuntu/
[10:38] <jamespage> med, negronjl: please give me a ping when you are around
[10:38] <jamespage> need to discuss standards for packages :-)
[10:39] <_ruben> Alan: getting a browser to trust a single cert tends to be trivial indeed, i was thinking systemwide stuff
[10:39] <Alan> _ruben: not sure about systemwide, but who cares about systemwide :P
[10:54] <_ruben> Alan: those who don't want do it for each app/user individually ;)
[11:39] <allenap> rvba, jtv: Fancy a review? https://code.launchpad.net/~allenap/maas/top-of-file-boilerplate/+merge/89224
[11:41] <rvba> allenap: sure
[11:41] <rvba> eurj
[11:41] <rvba> eurk even
[11:41] <allenap> rvba: eurk?
[11:42] <rvba> fr for yuck ;)
[11:42] <allenap> rvba: Oh :( That bad?
[11:43] <rvba> just kidding ;)
[11:45] <Daviey> lynxman: What is the status of https://blueprints.launchpad.net/ubuntu/+spec/servercloud-p-cfjuju ?
[11:46] <lynxman> Daviey: in progress but I have lower priority for it coming from upstairs so... I reckon I'll try to hunt it in a couple weeks
[11:46] <Daviey> lynxman: is it actually going to get done?
[11:47] <Daviey> If not, it would be better to unassign or postpone.
[11:50] <lynxman> Daviey: yes it'll get done :)
[11:51] <Daviey> lynxman: rocking
[11:51] <Daviey> rbasak: should i just sync openmpi, and then we solve the ftbfs?
[11:52] <lynxman> Daviey: waiting on your merge review for ipxe btw
[11:52] <rbasak> Daviey: can you sync off experimental? I don't mind how we do it. I haven't yet investigated the ftbfs apart from that it fails if I try a build on ubuntu
[11:53] <Daviey> lynxman: ah, groovy.. Will do that myself, or find another budding reviewer today
[11:53]  * rbasak is hoping to bring more pandas online today
[11:53] <Daviey> rbasak: yeah, i'll sync that now
[11:54] <lynxman> Daviey: excellent
[11:54] <Daviey> rbasak: ah, i wonder if the ftbfs was https://bugs.launchpad.net/ubuntu/+source/openmpi/+bug/813767
[11:55] <Daviey> or, http://launchpadlibrarian.net/86689726/openmpi_1.4.3-2.1ubuntu1_1.4.3-2.1ubuntu2.diff.gz
[11:55] <Daviey> if it is those issues, it'll want a sync :/
[11:55] <Daviey> err, merge
[11:55] <rbasak> Yeah I didn't get to a conclusion about that. I didn't understand how the porters worked before christmas when I looked. Now I do, I'll be using my own panda :)
[11:57] <Daviey> rbasak: would you be able to give it some attention soonly.. seems like an easy win. :)
[11:58] <rbasak> as soon as I can. I'm sorting out getting my pandas online first so I can task switch more easily
[11:58] <Daviey> rbasak: how many do you have now?
[11:58] <rbasak> three
[11:58] <Daviey> rbasak: you know, you could buy more sdcards :)
[11:58] <rbasak> I've been using hard drives for speed
[11:58] <rbasak> but yeah I didn't think of sdcards :)
[12:03] <Daviey> utlemming: Around?
[12:03] <Daviey> jamespage & utlemming: Happen to know status of, [utlemming] update image promotion process to integrate with Jenkins automated testing: TODO ?
[12:04] <Daviey> jamespage: https://blueprints.launchpad.net/ubuntu/+spec/servercloud-p-cloud-image-testing ?
[12:06] <Daviey> jamespage: is "Publish daily testing results to jenkins.qa.ubuntu.com: TODO" actually DONE, on https://blueprints.launchpad.net/ubuntu/+spec/servercloud-p-complex-deployment-testing
[12:06] <Daviey> ?
[12:06] <jamespage> Daviey: no - not yet
[12:06] <Daviey> k
[12:07] <allenap> rvba, jtv: What do you think of using from __future__ import unicode_literals everywhere? It means that all strings would be unicode, and you get a str() with b"..." syntax.
[12:13]  * Daviey ^5's allenap 
[12:14]  * allenap ^5's Daviey, but wonders what prompted it :)
[12:18] <jtv> allenap: I'm all for it.  Maybe even absolute imports, and what the hey, I'd go for the print function as well.
[12:18]  * jtv had been quietly wishing for this but dared not mention it
[12:18] <allenap> jtv: In 2.7 we get absolute imports by default I think.
[12:18] <allenap> I shall get on it.
[12:19] <jtv> I did a bunch of cross-version development (2.6/2.7/3.2) over the holidays.  Barry's tips helped immensely.
[12:19] <jtv> In fact I keep an email of his in my mailbox about stuff I ought to blog about.
[12:20] <allenap> I too have become an avid Barry porting missive reader.
[12:20] <jtv> In a nutshell: better give up on getting your newline conversions right before you start on python 3.  :)
[12:20] <rvba> allenap: sounds good
[12:33] <jtv> rvba: overriding the runserver command works a treat.  Thanks.
[12:33] <soren> jtv: Newline conversions?
[12:34] <jtv> soren: yes… my impression is that that particular problem wasn't sufficiently re-thought in the new I/O model.  And so you can ask for newline conversion when reading a compressed file, but you can't specify the encoding!
[12:35] <jtv> And so if you get a UTF-16 character, say, where one of the bytes happens to be equal to 13...
[12:35] <soren> jtv: Oh.
[12:35]  * soren hasn't dived into Py3k yet
[12:36] <jtv> I'd been writing a module for python 2.6/2.7/3.x that would be best described as lib_open_my_god_damn_file.py
[12:36] <jtv> Transparent decompression is one of the things it did, if needed.
[12:36] <jtv> So you get a conceptual layering of:
[12:36] <jtv> unicode
[12:36] <jtv> bytes
[12:37] <jtv> compressed file
[12:37] <jtv> blocks etc.
[12:37] <jtv> Where does newline conversion happen?
[12:37] <jtv> Currently, at the "compressed file" layer.
[12:37] <jtv> Which isn't aware that there might be b'\n' bytes in the data stream that aren't u'\n' characters.
[12:38] <jtv> A-hah!  I hear you cry: just use something or other in codecs that handles newlines.
[12:38] <jtv> But no, in python 3, codecs must always convert between bytes and unicode.
[12:38] <jtv> Which is cleaner, but less malleable when it comes to this kind of thing.
[12:39] <jtv> Newline conversion has to happen at (or above) the unicode layer.
[12:39] <jtv> So either it must be part of the codec, or it needs a separate layer that AFAICS does not exist yet.
[12:39] <onre> this sounds rather painful
[12:40] <onre> possibly even like a design mistake
[12:40] <jtv> In the case of compression, it can be handled by giving the gzip and bzip2 libraries the same open() API as the built-in one.
[12:40] <jtv> I like to think of it as design skew: the standard library isn't all on the same page yet
[12:41] <jtv> —so yes, a possible design mistake but with a slightly more optimistic view.  :)
[12:41] <RoyK> py3k - is that like http://tools.ietf.org/html/rfc2550 ?
[12:41] <jtv> Don't depress me.  I've got test code in that module that will start failing in a few centuries.
[12:44] <jtv> Luckily for me, my program doesn't actually mind if it gets a few too many newlines, loses carriage returns, or mistakes paragraph boundaries for line breaks.  As a friend pointed out, "about right" is the new black.
[12:45] <jtv> (By the way, b'\r\n' is a valid unicode character... in big-endian utf-16)
[12:58] <jtv> allenap: I'm new to this side of buildout… any idea how I can pull in python-oops and python-oops-wsgi?
[12:58] <jtv> Write a recipe?
[13:00] <jtv> I think I'll go look for a better headset first.
[13:55] <smoser> jamespage, Daviey . so, looking at https://jenkins.qa.ubuntu.com/view/Precise%20ISO%20Testing%20Dashboard/view/Daily/job/precise-server-ec2-daily/ARCH=i386,REGION=eu-west-1,STORAGE=instance-store,label=ubuntu-server-ec2-testing/45/artifact/None/i386/m1.small/instance-store/i-929cf6db/uec2-20120119-0340-74e55fac26b44d-terminated.console.txt
[13:55] <smoser> the mount issue came after a reboot that was issued by jenkins.
[13:55] <smoser> so it would seem to me that that reboot resulted in /mnt not being unmounted cleanly
[13:56] <smoser> maybe we could force this error just by repeatedly issuing reboot-instances
[14:13] <hallyn> kirkland: shift-f2 isn't working for me in bybou-tmux anymore
[14:22] <hggdh> roaksoax: your fence_cdu script: should the names of the machines be case-sensitive (-s)?
[14:41] <adam_g> smoser: snapshot/restore working pretty good. http://paste.ubuntu.com/809683/ reusing the same snippets earlier seems to work fine for bootstrapping juju agents
[14:42] <adam_g> Daviey: ^
[14:43] <Daviey> adam_g: \o/
[14:44] <Daviey> adam_g / jamespage: what is blocking making the jenkins results public?
[14:44] <Daviey> ie, what is left to do?
[14:44] <adam_g> Daviey: a firewall?
[14:44] <Daviey> adam_g: right, but what needs to be done before we can progress to make it public?
[14:44] <jamespage> Daviey, adam_g: are we happy with the automated process in the lab?
[14:44] <adam_g> Daviey: nothing i suppose, if theres nothing wrong with making a WIP public
[14:45] <jamespage> if so then I am OK to setup the results publishing
[14:45] <jamespage> ^^ WIP ^^
[14:45] <adam_g> im working on this snapshot stuff to reduce the # of provisioning failures
[14:45] <jamespage> nice
[14:45] <jamespage> sorry uvirtbot
[14:45] <Daviey> jamespage: I think exposing WIP has little negative impact?
[14:45] <jamespage> well maybe not
[14:45] <jamespage> its up to you
[14:45] <adam_g> ive still been focusing on getting the bootstrapping an deployment working as reliably as possible
[14:46] <adam_g> its pretty reliable as-is, but we're still running into periodic archive errors during initial installation, which require a 45 min timeout to pop before the test is failed
[14:46] <jtv> allenap, bigjools, rvba: do we have any opinions on where to put MaaS oopses?  I'm thinking maybe /var/log/maas in production, ./logs in development.
[14:46] <Daviey> yeah, if it's constaly failing there is no beneift to allowing me to see it :)
[14:47] <smoser> adam_g, great. and that is going into early_command ?
[14:47] <rvba> jtv: sounds good.
[14:47] <adam_g> Daviey: if you look at it now, the precise-openstack-essex-deploy. the minority are failures, but those are all archive errors during install. after install, those errors can be worked-around
[14:47] <adam_g> smoser: the partman early_command, yeah.
[14:48] <allenap> jtv, rvba: Up for a review? https://code.launchpad.net/~allenap/maas/coding-standards/+merge/89255.
[14:48] <jtv> allenap: I can take it
[14:48] <Daviey> adam_g: you know, that could be a udeb :)
[14:48] <allenap> jtv: Thanks.
[14:49] <adam_g> Daviey: the kernel module? yeah.. but what udeb provides it? or, how to build one myself for every new kernel thats released?
[14:50] <jtv> allenap: I guess this means we get sensible fixed-width indenting, not the squeeze-in-arguments-all-the-way-on-the-right indenting.  Nice.
[14:51] <Daviey> adam_g: sorry, this lvm work is in di or cloud-init?
[14:51] <adam_g> Daviey: d-i
[14:51] <jtv> allenap: you're lying about the database cluster though.  A plain "make" no longer sets up the database, as per your recommendation.  But "make syncdb," "make run," and "./bin/test" do.
[14:51] <allenap> jtv: I hadn't thought about that, but \o/
[14:52] <allenap> jtv: Ah, okay, that was a copy-n-paste. I'll update that.
[14:52] <jtv> Also, we now have ./bin/django deletedb
[14:52] <Daviey> adam_g: yeah, i was thinking - it could be wrapped into a udeb, passing preseed for the differing values.. rathe rthan hacking it into a late_command
[14:52] <jtv> allenap: I think it was a copy-and-paste from an older version of the readme..?
[14:52] <jtv> Ah no
[14:53] <adam_g> Daviey: i need to get a udeb together for the kernel module for sure. as for the scripting, i quite like having it embedded in the snippet, because we can make use of logic around profile name, mgmt classes, etc in cobbler to determine whether or not to restore or install.
[14:55] <rvba> allenap: oh, I've duplicated what you did with the README in my api doc, sphinx branch ;)
[14:56] <jtv> allenap: mp done
[14:56] <Daviey> adam_g: ok.. i just wondered if it made a nice thing that culd be reused in the archive..
[14:56] <Daviey> but either way :)
[15:01] <smoser> roaksoax, ping
[15:02] <allenap> rvba: Cool :)
[15:03] <allenap> rvba: Is that ready for review?
[15:03] <allenap> jtv: Thanks.
[15:03] <rvba> allenap: well, I've reverted it, I'll merge your branch…
[15:03] <rvba> allenap: almost
[15:04] <allenap> rvba: I don't have Sphinx yet. I'll sort that out next I think.
[15:05] <roaksoax> smoser pong
[15:07] <rvba> allenap: I need to merge your branch but otherwise, it's up for review: https://code.launchpad.net/~rvb/maas/maas-api-doc/+merge/89263
[15:07] <roaksoax> hggdh: uhmmm I thought it was case sensitive
[15:07]  * allenap looks
[15:08] <hggdh> roaksoax: it is, I just wanted to confirm it is as planned
[15:12] <allenap> rvba: Conflicts!
[15:12]  * rvba looks
[15:17] <roaksoax> hggdh: yeah I'm not doing anything to not consider it as case sensitive :).
[15:17] <hggdh> heh
[15:17] <allenap> bigjools: How are investigations of cobbler going?
[15:18] <hggdh> roaksoax: I guess my question was more "is it as designed, or just because" -- but you already answered it ;-)
[15:22] <jtv> bigjools: review basic oops integration?  I'd like your feedback on how much more we want to add on top: https://code.launchpad.net/~jtv/maas/oops/+merge/89267
[15:23] <xorred> has anyone had any experience with installing memcache on ubuntu server here?
[15:23] <xorred> it seems it's installed correctly, it's running, but I can't seem to be able to see it enabled in php
[15:23] <xorred> php -r 'phpinfo();' | grep 'memcache' returns nothing
[15:29] <roaksoax> hggdh: ehehehe ;)
[15:30] <patdk-wk> xorred, why would it?
[15:30] <patdk-wk> you need to *access* it from php somehow
[15:30] <patdk-wk> either by making your own socket calls, installing a php memcache lib, or whatever
[15:33] <onre> xorred, you need to separately install either one of the memcache extensions for php.
[15:34] <onre> xorred, php5-memcache and/or php5-memcached, latter is "better" from programmer point of view
[15:35] <SpamapS> jamespage: re zookeeper+start-stop-daemon.. no, I never pushed it anywhere, but I think I can now..
[15:35] <jamespage> SpamapS, sweet - thanks
[15:37] <kirkland> hallyn: are any other keybindings broken?
[15:37] <kirkland> hallyn: what version of byobu?
[15:38] <SpamapS> jamespage: testing built debs now.. seems to work
[15:39] <xorred> onre: I just ran apt-get install php5-memcached, and my site went DOWN
[15:39] <xorred> holy ****
[15:39] <xorred> please help, onre
[15:39] <xorred> what happened?
[15:39] <onre> xorred, well, look in apache error log to see what's failing.
[15:40] <xorred> apache runs just fine, apache2ctl restart shows no error
[15:40] <onre> xorred, /var/log/apache2/error.log
[15:40] <onre> haven't had that happen to myself
[15:40] <xorred> 13)Permission denied: make_sock: could not bind to address 0.0.0.0:80
[15:40] <xorred> 13)Permission denied: make_sock: could not bind to address 0.0.0.0:80
[15:40] <xorred> no listening sockets available, shutting down
[15:41] <xorred> Unable to open logs
[15:42] <xorred> ok
[15:42] <xorred> onre: now I see A LOT of segmentation faults
[15:42] <xorred> on the log
[15:43] <xorred> ls
[15:43] <jamespage> SpamapS, great
[15:44] <onre> well, either remove the module or find the config file that loads it. can't remember where it is, somewhere under /etc/php5/ iirc
[15:44] <rvba> allenap: conflicts fixed (diff updating).
[15:45] <allenap> rvba: Cool.
[15:48] <bigjools> allenap: coming along!
[15:48] <bigjools> jtv: sure
[15:50] <xorred> guys, my site is down
[15:50] <xorred> onre: man, please help me out
[15:50] <xorred> this happened after I followed your suggestion to install memcached
[15:50] <xorred> now, I have ran the php5-memcached remove, but nothing happens
[15:51] <allenap> rvba: Is there meant to be a prerequisite branch? There's a lot of code changing in there.
[15:51] <xorred> the log shows segmentation faults all over
[15:51] <rvba> allenap: no rereq, but most of it is boiler plate code added by the initialisation of sphinx.
[15:52] <onre> xorred, do you have any non-standard stuff installed or configured in there, and what exactly is segfaulting?
[15:52] <rvba> allenap: docs/Makefile and docs/
[15:52] <rvba> …docs/conf.py
[15:52] <bigjools> jtv: you want me to look at that MP as well as allenap's review?
[15:52] <xorred> onre: I found the conflictiing module
[15:52] <xorred> now... just if I were able to un-install xcache :D
[15:53] <onre> xorred, alternatively you can just disable it from being loaded.
[15:53] <onre> xorred, and be done with it.
[15:55] <jtv> bigjools: I'd just like to know what else you want done on top of this.
[15:55] <bigjools> jtv: ok, looking at it now
[15:55] <jtv> thx
[15:59] <roaksoax> bjf: howdy!! I just uploaded a patch for what seems to be your bug. So should be built soon let me know how it goes
[16:01] <bjf> roaksoax: cool, look forward to giving it a spin
[16:02] <kirkland> roaksoax: have you seen https://wiki.ubuntu.com/Kernel/Reference/Orchestra ?
[16:02] <kirkland> Daviey: ^
[16:04] <roaksoax> kirkland: cool
[16:04] <Daviey> kirkland: yeah, they were digging it last week.
[16:08] <kirkland> Daviey: good stuff, man
[16:09] <allenap> rvba: There's quite a lot of new code, like the whole of NodesHandler.
[16:09] <rvba> allenap: that's right :)
[16:10] <allenap> rvba: Okay...
[16:10] <rvba> allenap: but most of it is simply refactoring
[16:10] <rvba> allenap: and documentation
[16:12] <allenap> rvba: Okay, if you promise :) It looks like lots of new untested code, but then I can imagine that it's something to do with what Piston expects.
[16:12] <hallyn> soren: i'm going to add kvm nesting as an option through /etc/default/kvm.  I'm leaning toward making it enabledy by default.  Test shows no impact on guest perf.  Do you object to enabling by default?
[16:12] <hallyn> jamespage: smoser: zul: ^
[16:13] <smoser> i do not object.
[16:13] <rvba> allenap: really, no new code was added!  Only the create method was splitted into two.
[16:14] <rvba> allenap: well, resource_uri is new, but that's it.
[16:14] <smoser> smb, how did you do that ascii art ?
[16:14] <allenap> rvba: 770 lines diff?
[16:14] <smb> smoser, the keyboard? ;)
[16:15] <smoser> once again, smb shows he's a better man than i.
[16:15] <hallyn> smoser: as for your question about 2nd level nesting perf, it's about what you'd expect.  compiling a tiny program jumps from .8 seconds to 2.8 seconds without kvm in second level
[16:15] <smb> smoser, MAybe just a bit more anal... :-P
[16:15] <rvba> allenap: 769 yes, but again, most of it is docs/Makefile and docs/conf.py.
[16:16] <smoser> hallyn, i dont follow.
[16:16] <smb> smoser, I thought it takes me probably longer to find out how that nicer moin things work again
[16:16] <smoser> smb, i figured you used some modeling tool that could output ascii art
[16:17] <smb> smoser, Yeah, but no. Just alot of patience
[16:18] <hallyn> smoser: run kvm on the host.  inside that kvm guest, run kvm, and compile a small program (suckless' ii).  If you run the kvm guest without '-cpu host' (to provide nested kvm), it takes 2.8 seconds.  With '-cpu host', it takes .8 seconds
[16:19] <smoser> wow.
[16:19] <smoser> thats not what i would have expected at all.
[16:19] <smoser> much better improvement than i would have thought.
[16:20] <SpamapS> jamespage: https://code.launchpad.net/~clint-fewbar/ubuntu/precise/zookeeper/drop-daemon-dep/+merge/89284
[16:20] <hallyn> well, aiui, the cpu should be host speed, you just have (a) slower disk, and (b) issues with shadow paging in memory.
[16:20] <hallyn> anyway, i'll go enable it.  thanks.
[16:22] <SpamapS> hallyn: why doesn't kvm default to nesting if its available?
[16:26] <hallyn> SpamapS: they don't say.  relevant docs are http://www.usenix.org/events/osdi10/tech/full_papers/Ben-Yehuda.pdf   and Documentation/virtual/kvm/nested-vmx.txt in the kernel source.
[16:26] <hallyn> presumably bc i'ts new
[16:32] <SpamapS> Yeah that would be nice if it just happened
[16:33] <hallyn> of course otoh i don't want mysterious new failures due to a bug in the nesting support...
[16:33] <hallyn> paper says 6-8% perf impact on second level guest vs. first level.  not bad.
[16:34] <SpamapS> hallyn: in the past I'd have agreed, but IMO, proper test coverage (both across software and hardware types) should alleviate the need for such caution.
[16:35] <SpamapS> But yeah, maybe a release or two with it not on by default is prudent. :-)
[16:38] <hallyn> SpamapS: yes, i'm figuring if there are bugs, we have 2 months to run into them
[16:39] <hallyn> so right now i'm going to enable it.  it worked great for me yesterday and today under pretty heavy use.
[16:43] <bjf> roaksoax: looks like the same error - http://pastebin.ubuntu.com/809812/
[16:47] <roaksoax> bjf: uhm ok... so http://192.168.1.6/cblr/ks_mirror/precise-x86_64/ is empty then?
[16:47] <bjf> roaksoax: The requested URL /cblr/ks_mirror/precise-x86_64/ was not found on this server.
[16:48] <roaksoax> bjf: can you pastebin /var/www/cobbler/ks_mirror and /var/www/cobbler/ks_mirror/precise-x86_64
[16:51] <bjf> roaksoax: ok, the directory name is different as seen here: http://pastebin.ubuntu.com/809816/
[16:51] <hallyn> SpamapS: hey upstart guru, do you have any objections to the way I'm doing this?  http://people.canonical.com/~serge/debdiff
[16:52] <roaksoax> bjf: ah that's the error, did orchestra imported those ISO's directly?
[16:52] <roaksoax> adam_g: are we updating the ISO's by a cronjob?
[16:52] <bjf> roaksoax: yes, i've done nothing manually/myself other than install orchestra
[16:53] <roaksoax> s/by a/with a/
[16:54] <roaksoax> bjf: interesting issue, I'll look into that
[16:54] <bjf> roaksoax: any way i can work around it (symlink i suppose)
[16:54] <roaksoax> bjf: do this then: sudo cobbler profile edit --name precise-x86_64 --ks-meta tree=http://@@http_server@@/cblr/repo_mirror/precise-x86_64-x86_64
[16:55] <bjf> roaksoax: note, a netboot of bare metal worked
[16:57] <roaksoax> bjf: yeah, because koan uses the "tree" variable from ks_meta in order to determine whre the installation tree is located
[16:57] <bjf> roaksoax: --ks-meta is a bad option
[16:57] <adam_g> roaksoax: where?
[16:58] <bjf> roaksoax: ah, --ksmeta (no hyphen)
[16:58] <roaksoax> bjf: --ksmeta sorry
[16:58] <roaksoax> yeah
[16:58] <roaksoax> adam_g: in the import scripts we use for orchestra/cobbler? are they been run by a cronjob to update the ISO's?
[16:59] <adam_g> roaksoax: ya. as of precise, if a distro exists, orchestra-import-isos calls: cobbler-ubuntu-import -c $r-$a && cobbler-ubuntu-import -u $r-$a to update it if necessary
[17:00] <bjf> roaksoax: running koan after that produce: http://pastebin.ubuntu.com/809826/
[17:10] <roaksoax> adam_g: ok I think there might be something wrong with that as it creates directories with distro-arch-arch
[17:11] <roaksoax> bjf: libvir: error : Cannot get interface MTU on 'virbr0': No such device
[17:11] <roaksoax> bjf: specify the virt interface you are using with virt-bridge i think
[17:16] <bjf> roaksoax: so i need to define a bridge device ?
[17:17] <roaksoax> bjf: it defaults to virbr0 but if you don't have it, then you would have to specify it. 'sudo koan --server localhost --virt --profile precise-x86_64 --virt-bridge=br0'
[17:17] <roaksoax> for exmaple
[17:18] <roaksoax> bjf: or you can define it in the cobbler profile
[17:18] <bjf> roaksoax: so the error indicates that i don't have it, should it have been created during install?
[17:19] <roaksoax> sudo cobbler profile edit --name precise-x86_64 --virt-bridge br0 --> If you do this, you don't need to specify it wen using koan
[17:19] <roaksoax> bjf: I think kvm should have created it automatically on install
[17:20] <roaksoax> hallyn: ^^
[17:20] <roaksoax> should kvm create a virbr0 on install?
[17:20] <hallyn> no
[17:20] <hallyn> that's libvirt
[17:21] <roaksoax> err yeah, should libvirt create it ayutomatically then after install?
[17:21] <hallyn> yes
[17:21] <hallyn> unless you have in the past removed it ( as some people have )
[17:22] <hallyn> if it's not, pls do file a bug, as that's wrong
[17:22] <bjf> hallyn, ack
[17:22] <hallyn> thx
[17:22] <roaksoax> hallyn: thanks ;)
[17:25] <bjf> roaksoax: bug 918807
[17:29] <mdeslaur> hallyn: whoa, nested kvm!
[17:31] <hallyn> mdeslaur: yup, \o/
[17:56] <bjf> roaksoax: i'm not convinced virbr0 wasn't created. i have this in my syslog: ADDRCONF(NETDEV_UP): virbr0: link is not ready
[18:04] <SpamapS> zul: hey, can you delete this recipe? It has never built correctly https://code.launchpad.net/~zulcss/+recipe/memcached-daily-maverick and just emails me weekly with its FTBFS
[18:06] <roaksoax> bjf: uhmmm I;ll do a fresh install and see what I find, and then will be able to confirm.
[18:21] <pmatulis> mdeslaur: nested kvm?
[18:24] <mdeslaur> pmatulis: kvm working inside a kvm guest
[18:25] <pmatulis> mdeslaur: understand the concept.  so it's upcoming?  announcement?
[18:25] <mdeslaur> pmatulis: hallyn uploaded a qemu-kvm package to precise that turned it on, that's how I noticed
[18:26] <pmatulis> mdeslaur: i see.  nice
[18:26] <hallyn> i recon i need to at least blog about how to use it, but yeah it needs to go into release notes at some point
[18:26] <hallyn> to use it, you'll still need to add '-cpu host' or -cpu qemu64,+vmx in your qemu args
[18:27] <pmatulis> hallyn: yeah, i'd like to see your blog when it's done
[18:31] <hallyn> it'll be pretty short :)  it'll be on planet.u.c though, if you check that.  if not i'll let you know when it's up
[19:08] <osaka> I'm planning to use KVM and OpenVZ, which kernel shall I install server or virtual?
[19:08] <osaka> or is virtual used when its being a guest?
[19:10] <osaka> https://help.ubuntu.com/community/ServerFaq#What_are_the_differences_between_the_server_and_virtual_kernels.3F
[19:16] <cwillu_also_at_w> osaka, virtual is intended for guests, yes
[19:17] <zul> hallyn: no objections for nested kvm
[20:28] <tdn> How do I play video without X? I thought I could to mplayer video.avi  -fb /dev/fb0, however /dev/fb0 does not exist on a deafult Ubuntu Server 11.10? What to do?
[20:38] <niksoft> hello
[20:42] <niksoft> hmmm, very quiet in here... Hey anyone want to talk about the broken LACP implementation in the Ubuntu server? At least as far as i can tell it is lightly broken
[20:42] <tdn> How do I get sound working in Ubuntu Server? What packages do I install?
[20:54] <SpamapS> tdn: what program do you want to play sounds with?
[20:55] <SpamapS> tdn: I'd imagine you just need to install the program, and it will pull in the deps that enable it to play sound through any hardware you have
[20:59] <youlysses> 2 questions. 1. unlike ubuntu desktop is the server edition comprised of all FOSS? 2. Is anybody or can anybody reccomend it as a general file/print server? Seems good for the fit.
[21:01] <dvance> 1. I am not sure 2. I would, yes
[21:03] <youlysses> Ok cool, I'll guess i'll further look into it. But I just setup my "battlestattion" and it's quickly becoming a pain to try to keep these files synced... :P
[21:04] <dvance> been there, is it nessesary that it's 100% FOSS ?
[21:04] <youlysses> Morally yes, but I could manage ... i guess...
[21:13] <kirkland> smoser: hey, are you around?
[21:13] <smoser> here
[21:13] <kirkland> smoser: can you help edygarcia with a problem with ubuntu-cloudimg-query?
[21:13] <kirkland> smoser: it's working fine for me, but not at all for him
[21:13] <kirkland> smoser: we're both on 11.10
[21:15] <kirkland> smoser: i gotta run, but thanks in advance!
[21:15] <edygarcia> smoser: hi, I am trying to configure my environment, the ec2-describe-images is working fine
[21:16] <edygarcia> smoser: the error I get is faild to find ami, when I run ubuntu-cloudimg-query t1.micro oneiric
[21:16] <edygarcia> smoser: or any description for that matter
[21:17] <utlemming> edygarcia: try rm -rf ~/.cache/ubuntu-cloudimg-query
[21:17] <utlemming> and then try again
[21:17] <smoser> edygarcia, i suspect you have bad data in cache
[21:17] <smoser> but
[21:17] <smoser> wait
[21:17] <smoser> before you do that
[21:17] <smoser> please save that off
[21:17] <smoser> in case it was it.
[21:19] <Daviey> smoser / roaksoax: Fancy reviewing, and if suitable - uploading https://code.launchpad.net/~lynxman/ubuntu/precise/ipxe/newsnapshot/+merge/88329 ?
[21:19] <edygarcia> smoser: same error after deleting the cache
[21:19] <roaksoax> Daviey: sure
[21:20] <Daviey> roaksoax: ta
[21:23] <smoser> edygarcia, can you pastebin ~/.cache/ubuntu-cloudimg-query/oneiric.server.released.current.txt ?
[21:28] <edygarcia> smoser: http://paste.ubuntu.com/810083/
[21:28] <niksoft> Let me try again, hi, i'm trying to configure an ubuntu server to serve files on a 20Gbit link, or rather a 2 x 10 Gbit bonded link, from what i see the 802.ad (LACP) implementation fails to correctly balance the TX, favoring one of the nics and not sending any data on the other. Can anyone help me look into this?
[21:29] <niksoft> 802.1ad*
[21:29] <smoser> edygarcia, hm..
[21:29] <elfurbe> So I've got two SSDs in a server I'm building. I'd like TRIM to work but I need them mirrored. I've tried creating a mirrored lv but had no luck either at install time (dm_mirror module doesn't exist) or after (can't add mirrors to an existing lv apparently)
[21:29] <elfurbe> Anyone have any thoughts?
[21:29] <elfurbe> My understanding is that md devices don't support trim which is why I'm trying it this way
[21:30] <niksoft> elfurbe raid? software or hardware
[21:30] <elfurbe> software, I can't make the pseudo-raid on this motherboard work for anything
[21:30] <niksoft> oh no mobo implementations are very incomplete most of the time
[21:31] <elfurbe> It's a Supermicro, has both Intel and Adaptec codebases, neither of them make volumes that are visible to the installer
[21:31] <elfurbe> So whatevs, I put the ssds in AHCI mode and figured I'd work something out
[21:31] <elfurbe> But so far I can't make it work
[21:32] <niksoft> elfurbe what kernel are you on?
[21:32] <elfurbe> This is during the installer for 11.10
[21:33] <elfurbe> I need the rootfs to be on the mirror
[21:33] <niksoft> can you drop into a shell and do uname -a ?
[21:34] <elfurbe> Yeah, but I'll have to relay output the old fashioned way
[21:34] <niksoft> morse code?
[21:34] <elfurbe> Linux gluon 3.0.0-12-generic #20-Ubuntu SMP <date> x86_64 GNU/Linux
[21:34] <elfurbe> No, by reading one screen and typing it out :D No copy and paste
[21:35] <elfurbe> The installer's partitioner has no support for mirrored lvs so I'm trying to do it by hand from one of the other ttys
[21:35] <niksoft> intel brought in the functionality in at 2.6.33
[21:36] <elfurbe> The trim sauce you mean?
[21:36] <elfurbe> Yeah, but it still doesn't work for mdraid devices, right?
[21:36] <elfurbe> Only dm?
[21:38] <niksoft> http://serverfault.com/questions/227918/possible-to-get-ssd-trim-discard-working-on-ext4-lvm-software-raid-in-linu
[21:39] <niksoft> i believe you are correct
[21:39] <niksoft> as of yet software raid may not support trim
[21:43] <niksoft> Hmm perhaps worth asking around in ubuntu kernel channel...
[21:44] <smoser> roaksoax, https://code.launchpad.net/~smoser/ubuntu/precise/cobbler/improvements.cui/+merge/89347
[21:44] <smoser> edygarcia,
[21:45] <edygarcia> smoser: yes
[21:45] <roaksoax> smoser: looking
[21:46] <smoser> can you pastebin: bash -x ubuntu-cloudimg-query t1.micro oneiric > out 2>&1
[21:46] <roaksoax> smoser: looks good to me.
[21:47] <roaksoax> smoser: you want me to upload those?
[21:48] <smoser> roaksoax, sure.
[21:48] <smoser> that makes updating as easy as : sudo cobbler-ubuntu-import --update-existing
[21:52] <roaksoax> smoser: awesome!! I'll upload it tonight
[21:53] <smoser> edygarcia, see pastebin request above
[21:57] <edygarcia> smoser: http://paste.ubuntu.com/810113/
[21:58] <smoser> edygarcia, remove psace between 2 and > anda 1
[21:58] <smoser> something waswrong
[21:58] <smoser> you want
[21:59] <smoser> can you pastebin: bash -x ubuntu-cloudimg-query t1.micro oneiric 2>&1 | pastebinit
[21:59] <smoser> ( if you dont have pastebinit: sudo apt-get install pastebinit )
[22:00] <edygarcia> smoser: http://paste.ubuntu.com/810118/
[22:02] <hallyn> ahs3: Can I sling you a debdiff to enable 'make check' for netcf on sid?
[22:02] <ahs3> hallyn: go for it
[22:03] <hallyn> thx - be a few mins
[22:03] <ahs3> np
[22:05] <hallyn> ahs3: http://people.canonical.com/~serge/netcf-enable-test.debdiff
[22:09] <smoser> edygarcia, i'm not sure why, but try adding us-east-1
[22:09] <smoser> (ie, specifying the region)
[22:11] <ahs3> hallyn: kewl.  i'll see if i can get that in tonight or sometime tomorrow.
[22:12] <hallyn> ahs2: thanks!
[22:12] <smoser> hallyn, https://bugs.launchpad.net/ubuntu/+source/lxc/+bug/918946
[22:13] <hallyn> smoser: yeah.  were there others you needed to run libvirt?
[22:14] <hallyn> smoser: I'm just getting around to uploading fix for bug 918343 btw
[22:15] <hallyn> smoser: if no others, i'll goa head and post that fix right now
[22:15] <hallyn> oh, i see
[22:15] <smoser> hallyn, still not tested all the way.
[22:15] <smoser> wait.
[22:16] <hallyn> smoser: actually that bug you just opened, that is fixed in the version i pushed an hour or two ago
[22:16] <smoser> oh?
[22:16] <smoser> you can run libvirt in a container now?
[22:17] <edygarcia> smoser: that did the trick, thank you, I have EC_REGION set to that but it is not using it :)
[22:17] <hallyn> (thinking)  maybe not
[22:18] <hallyn> no, unrelated, nm
[22:19] <smoser> edygarcia, well, even this seems to work fine for me:
[22:19] <smoser> EC2_REGION=us-east-1 ubuntu-cloudimg-query t1.micro oneiric
[22:19] <smoser> but i'm using precise.. possible that i've fixed that.
[22:20] <edygarcia> smoser: works for kirkland as well and he is on oneiric, I will look closer into it tomorrow, thanks again
[22:22] <smoser> hallyn, am i doing something wrong
[22:22] <smoser> $ sudo lxc create -t ubuntu -n lv1
[22:22] <smoser> E: -t - no such container
[22:23] <smoser> yes.
[22:23] <smoser> yes i am
[22:23] <hallyn> smoser: 'lxc-create', not 'lxc  create'
[22:23] <smoser> lxc-create
[22:23] <smoser> dagum feeble brain
[22:24] <hallyn> pushing fix about now
[22:24] <hallyn> doh!  branches diverged.  i hate that crap
[22:24] <hallyn> you know, i *try* to make SpamapS happy, but he keeps disappointing
[22:27] <smoser> hallyn, you broke lxc-list
[22:27] <smoser> $ lxc-ls
[22:27] <smoser> lv1
[22:27] <smoser> lv1
[22:27] <smoser> gr...
[22:27] <smoser> maybre not
[22:28] <smoser> you should just exit early on all those scripts if you think that they can't be run as non-root
[22:28] <hallyn> smoser: i fixed lxc-ls just now...
[22:28] <hallyn> it *can* be run as non-root, it just couldn't bc i was mistakenly mkdiring the cgroup dirs as 0700 not 0755
[22:28] <smoser> well, lots of them can't be
[22:29] <smoser> but just error randomly or not at all
[22:29] <hallyn> like i say there are more little naggly cleanups queued up on my other laptop, hopefully i push tomorrow
[22:29] <hallyn> yes, those should be fixed
[22:29] <hallyn> smoser: which ones are you thinking of?
[22:30] <ZiggyTheHamster> I've got a really stupid problem with sendmail that I am going insane trying to figure out. Anyone around that might have some suggestions? :)
[22:31] <ZiggyTheHamster> (I'm using sendmail to send from the server only, not to receive mail)
[22:32] <smoser> dont know really. one of them just got me a bit ago
[22:32] <smoser> i'll let you know
[22:35] <smoser> hallyn, annoying failure:
[22:35] <smoser> http://paste.ubuntu.com/810148/
[22:36] <hallyn> that was as unpriv user?  That's one you wanted to exit out when you're not root?
[22:37] <smoser> well a simple "permission denied" is bette rthan that.
[22:37] <smoser> or "you cant do that, jim"
[22:41] <hallyn> right - cause 'not root' is not sufficient check (if you're using lxc-setcap).  But yes, it should be checking early.
[22:47] <undriedsea> If I have two servers connected to the same chassis of SAS disks in a RAID10 (on the chassis), could I install IET (iSCSI target software) on both, export the disk array, setup iSCSI multi-pathing on my client and have an HA storage environment? I was thinking active/passive.
[22:51] <cwillu_also_at_w> undriedsea, high availability is not the result of a combination of techniques
[22:52] <cwillu_also_at_w> it's the result of the combination of understanding and process
[22:54] <cwillu_also_at_w> taken at face value, what you describe provides about the same level of availability in principle as a raid1 , but with a much higher complexity
[22:55] <cwillu_also_at_w> whether that added complexity actually results in improved service is a matter of how well you design the system, and how well the process you put in place works to prevent mistakes during both normal and degraded operation
[22:56] <hallyn> roaksoax: any chance I could get you to upload http://people.canonical.com/~serge/netcf-precise-runtests.debdiff ?
[22:58] <hallyn> i seem to lack privs
[22:59] <cwillu_also_at_w> (consider that the rate of fatal crashes for small twin-engine aircraft is nearly identical to that for single-engine aircraft:  the extra complexities almost perfectly offsets the redundancy provided at that level of operation)
[23:06] <roaksoax> hallyn will sure do
[23:07] <Daviey> roaksoax: just uploaded.
[23:08] <hallyn> roaksoax: thanks!
[23:09] <roaksoax> Daviey lol k
[23:10] <undriedsea> cwillu_also_at_w, the failure I am trying to guard against is the iSCSI target server. In real production the failure rate of everything else has been very low. I guess I am wondering if the setup described would work.
[23:15] <adam_g> undriedsea: drbd+pacemaker+ietd active/passive is a pretty straightforward and common HA clustering setup.
[23:16] <undriedsea> adam_g, I have been seeing that as well, in this case since the disks are connected to both iSCSI servers, I was hoping to avoid using DRBD.
[23:17] <undriedsea> I don't want a SW-RAID1 when I already have a HW-RAID10
[23:17] <adam_g> undriedsea: ah, misread originally
[23:18] <undriedsea> Oh np, thanks for the suggestion
[23:25] <edygarcia> smoser: I found the problem, I setup my EC2_URL per https://help.ubuntu.com/community/EC2StartersGuide in the format of ec2.<region>.amazonaws.com , but it is parsed as <region>.ec2.amazonaws.com
[23:26] <edygarcia> smoser: commenting out my EC2_URL, and leaving the EC2_REGION fixed it
[23:30] <adac> guys, is it possible to install virtualbox on ubuntu server? mean without grafical interface? does someone know a good tutorial to that?
[23:35] <niksoft> Anyone have any good suggestions on building a high tps and high throughput web box? before you answe, i am talking over 10k tps and over 10gig throughput. Currently i have a bonded 20Gig, 96 gb of ram, 24 virtual cores, nginx compiled with threaded malloc from google performance, and tons of tweaks to the network defaults
[23:36] <niksoft> serving out of a ramdisk
[23:37] <niksoft> aanyone?