[01:04] <darkf34r> I'd like to register my nickname on freenode, could I?
[01:05] <dax> darkf34r: http://freenode.net/faq.shtml#nicksetup
[01:05] <dax> (btw, #freenode is the usual channel for network support questions)
[02:05] <BuenGenio> hi there
[02:05] <BuenGenio> any idea why this has started happening after the last postfix upgrade?
[02:05] <BuenGenio> http://pastebin.com/VPR9wXRP
[02:05] <BuenGenio> some users - not all, and not always, also started complaining about their mail clients asking for their passwords when sending mail
[02:05] <BuenGenio> personally, I've never had this occur to me on the same server, and there's nothing obvious in logs that would suggest it's a postfix/auth issue
[02:05] <BuenGenio> however, the small proportion of users with the problem only started complaining after the upgrade
[02:12] <qman__> if they're using MS outlook, it just does that, tell them to reboot
[02:13] <qman__> as for your pastebin, looks like you're missing whatever library or module is used for quotas
[02:20] <BuenGenio> qman__, didn't used to do that "password incorrect". in any case, you know what their answer is going to be - it works with my other email account - so it's YOUR problem
[02:21] <BuenGenio> and they would be right...
[02:21] <qman__> I deal with it all the time, MS Outlook just does that sometimes, usually when the server goes offline while the computer is connected
[02:21] <qman__> the fix is to reboot, or in bad cases, recreate the profile
[02:21] <ScottK> BuenGenio: Those errors are from a postfix that was built with the unofficial VDA patch.
[02:22] <qman__> but if people besides Outlook users see it, then you may have an issue
[02:22] <ScottK> The official Ubuntu packages don't have that.
[02:22] <ScottK> So if you switched from a custom build with the VDA patch to the standard Ubuntu packages, you'd get errors like that.
[02:23] <BuenGenio> how to get rid of that?
[02:23] <ScottK> Actually there's a few there that aren't VDA related.
[02:23] <ScottK> What version of Ubuntu are you running?
[02:23] <BuenGenio> 12.04
[02:23] <BuenGenio> upgraded from 11.04 -> 11.10
[02:23] <BuenGenio> -> 12.04
[02:24] <qman__> well, you either need to get a same-patched custom version for 12.04, or deconfigure the custom bits
[02:24] <ScottK> What's the  output of dpkg -s postfix?
[02:26] <ScottK> Also what do you get from grep spf /etc/postfix?
[02:27] <ScottK> BuenGenio: ^^^
[02:27] <BuenGenio> 1 sec
[02:27] <ScottK> Also please pastebin the output of postconf -n.
[02:27] <BuenGenio> nothing
[02:29] <ScottK> Can you pastebin /etc/postfix/master.cf?
[02:29] <ScottK> Also main.cf.
[02:29] <BuenGenio> ScottK, http://pastebin.com/rW1c2Ckz
[02:29] <BuenGenio> postconf -n ^^
[02:30] <BuenGenio> ScottK: master.cf: http://pastebin.com/av5q2syH
[02:30] <ScottK> OK.  What was it you got no output for?
[02:31] <BuenGenio> ScottK, for grep -i spf
[02:31] <ScottK> OK.
[02:35] <ScottK> How about main.cf.
[02:35] <ScottK> And I'm also still looking for dpkg -s postfix.
[02:39] <ScottK> BuenGenio: Have you vanished?
[02:39] <BuenGenio> no, i'm here
[02:39] <ScottK> BuenGenio: Did you pastbin main.cf?
[02:40] <ScottK> I need to see that.
[02:40] <BuenGenio> ScottK, dpkg -s postfix: http://pastebin.com/3d7PLQ8t
[02:41] <ScottK> Good.  That's the correct one for 12.04 with all updates applied (which is what you want).
[02:41] <ScottK> Now main.cf.
[02:41] <BuenGenio> cat /etc/postfix/main.cf: http://pastebin.com/V536vh4t
[02:41] <ScottK> Thanks.
[02:41] <BuenGenio> cheers
[02:43] <ScottK> So there are a few problems here.
[02:45] <ScottK> First is the SPF one.  To use the SPF policy server you've got set up, you need to do two things:
[02:45] <ScottK> 1.  Add check_policy_service unix:private/policyd-spf to your smtpd_recipient_restrictions, right before permit_auth_destination.
[02:46] <BuenGenio> ok
[02:46] <ScottK> 2.  In line 123 of your master.cf add the missing 'd' to policy-spf, i.e. policyd-spf.
[02:47] <ScottK> That will make those warnings go away.
[02:47] <BuenGenio> yeah, so I did something wrong grepping /etc/postfix for spf the first time =)
[02:47] <BuenGenio> there's quite a few spf references there
[02:49] <ScottK> Yes.
[02:49] <ScottK> Not a problem.
[02:49] <BuenGenio> added spf restriction
[02:50] <BuenGenio> but that doesn't address the "unused parameter" issues
[02:50] <ScottK> Remove all reference to virtual_mailbox_limit_override, virtual_maildir_extended, virtual_overquota_bounce, virtual_create_maildirsize, and virtual_maildir_limit_message from main.cf.
[02:51] <ScottK> Those are VDA patch functions that aren't supported by postfix upstream or Ubuntu.
[02:51] <ScottK> Alternately, you can update your package to apply the VDA patch, but you are totally and completely on your own there.
[02:52] <ScottK> http://vda.sourceforge.net/
[02:52] <ScottK> They have their own mailing lists.
[02:52] <BuenGenio> ok
[02:53] <BuenGenio> is there a launchpad project with binaries for that?
[02:53] <ScottK> No idea.  That's part of where completely on your own kicks in.
[02:54] <BuenGenio> heh
[02:54] <BuenGenio> cool, thanks for the help
[02:55] <BuenGenio> ScottK, one error left
[02:55] <BuenGenio> /usr/sbin/postconf: warning: /etc/postfix/main.cf: unused parameter: spf-policyd_time_limit=3600s
[02:56] <ScottK> What's your service in master.cf called?
[02:56] <ScottK> Is it policyd-spf?
[02:56] <ScottK> If so, that needs to match.
[02:57] <ScottK> policyd-spf_time_limit=3600s
[02:57] <ScottK> I see it was policy-spf.
[02:58] <ScottK> So make that parameter in main.cf match the service name in master.cf plus _time_limit.
[02:58] <ScottK> BuenGenio: ^^^
[02:58] <BuenGenio> ah
[02:58] <BuenGenio> oic
[02:59] <BuenGenio> sweet
[03:02] <ScottK> One of the changes in more recent postfix releases is it started warning about these things.  Those were possibly broken all along, but no one knew because there were no warnings.
[03:30] <hilarie> okay, so my isp got mad at me for using both an static and dynamic ip (long story) so I went out and bought a 2nd nic for my ubuntu server, and its showing up as connected, but I can't figure out how to make it play nice, right now, I just have my xubuntu hooked directly up to it trying to get it to connect, so I can then put my router in place of the xubuntu netbook and everything can be connected
[03:30] <hilarie> how can I get my ubuntu dhcp server up and running? it shows its active, but my xubuntu isn't connecting
[03:38] <hallyn> stgraber: all right i don't expect this to be reliable yet, but lp:~serge-hallyn/ubuntu/quantal/lxc/lxc-api-getconfig/ .  it leaves open the question of how you want the get_config_item(lxc.network) to look
[03:39] <hilarie> how to tell if your nic is "modern" and will handle the crossover stuff without you?
[03:41] <stgraber> hallyn: I guess the easiest is to do as you said on Friday, have get_config_item("lxc.network...") return the key for all interfaces, one on each line
[03:41] <stgraber> hallyn: I can pretty easily then write some python to query all the .network keys and generate nice interface objects (or similar) that people can understand and update easily
[03:42] <hallyn> stgraber: it gives full 'lxc.network.*' output.  (none of the others do)
[03:44] <stgraber> hallyn: yeah, not sure I like the special casing that much... I think it's best to treat the network stuff just like any other key and let the higher level stuff deal with it (in this case, the python overlay).
[03:44] <hallyn> stgraber: but that doesn't really work bc the network stuff needs context
[03:45] <hallyn> i.e. one nic might be macvlan with vlan_id to print, one might be veth with another endpoint ifname to print
[03:46] <hallyn> gateways and bcast etc are also optional
[03:46] <hallyn> so i could drop 'lxc.network.' but keep the rest of the key names, but that just makes it all the more work for your layer since you'll have to convert between that and the config file
[03:47] <hallyn> so maybe i need a network-specific query api after all with things like 'network_get_link()'
[03:48] <stgraber> gah, lxc's network config is really a mess... let me quickly write a bit of python that looks like what I was hoping to get from get_config_item/set_config_item/clear_config_item, then we can see if that's even doable ;)
[03:48] <hallyn> stgraber: let's talk tomorrow (or after you've taken a look and can tell me which parts you hate the most)
[03:48] <hallyn> sounds good
[03:48] <hallyn> good night
[03:56] <stgraber> hallyn: http://paste.ubuntu.com/1083899/ that's rougly how I was planning to use get_config_item. Essentially iterating all possible keys, assuming I'll always get one line per interface with the line empty if it's not set
[03:57] <stgraber> hallyn: which then lets me build a list of interfaces and properties that the user can play with. When changed, I'd simply iterate through all the keys again with clear_config_item first, then do set_config_item calls to set the keys
[03:58] <stgraber> so it'd involve quite a bunch of calls to get_config_item, clear_config_item and set_config_item
[04:00] <stgraber> hallyn: which only works as long as none of these keys can exist more than once per interface (I'm not sure about that bit, if these some of these keys can be defined multiple times, it'll just fail...)
[04:00] <stgraber> (we really should really plan to move to lxc.network.<interface>.* ... that'd make things much easier to deal with ;))
[04:01] <stgraber> hallyn: anyway, I'm off, I'll take a long at the branch tomorrow (for now I was just speculating based on our previous discussion and IRC comments, haven't looked at the implementation). good night
[04:03] <hallyn> stgraber: lxc.network.ipv{4,6} are both lists (meaning multiple values for addresses)
[04:03] <hallyn> night
[04:06] <stgraber> hallyn: argh...
[04:10] <ping__> hy
[04:10] <ping__> help my ubuntu server installing lusca can't cache youtube :(
[04:38] <ping__> help :(
[05:16] <glosoli> How do I check if mail service is installed in my VDS
[05:16] <glosoli>  ?
[05:18] <andol> glosoli: Well, did you install one? :) Also, how do you define a mail service? A basic (sending) SMTP server, or more?
[05:19] <glosoli> andol: basic sending smtp server, I didin't installed one, but it might have been installed when I buyed a Cloud Server with Ubuntu, I kinda see in Parallels Panel something like sendmail mentioned
[05:22] <andol> glosoli: For starters, you can check if you have anything listening on port 25
[05:22] <andol> sudo netstat -tlnp | grep ":25"
[05:23] <glosoli> yeah send mail
[05:23] <glosoli> is being listening on 25
[05:25] <andol> glosoli: Ok, sounds like you have something install, but no idea whatever it is properly configured or not.
[05:25] <glosoli> ok will check that out as there is some php file on the website being hosted by that server
[05:25] <andol> (Also, sendmail is something I haven't touched in a while.)
[05:25] <glosoli> so if it won't popup any error while using send_mail I assume I can think that everything is o
[05:25] <glosoli>  ?
[05:25] <glosoli> is ok"
[05:26] <andol> No idea.
[05:31] <ping__> hy sir, plz help, i instal squid on ubuntu server, but i cn't cache youtube ?
[05:47] <glosoli> any ideas how to clear DNS Cache
[05:47] <glosoli>  ?
[05:50] <ping__> s
[05:51] <ping__> glosoli = sudo aptitude install nscd
[05:51] <ping__> Flush DNS Cache in Ubuntu Using the following command sudo /etc/init.d/nscd restart
[05:53] <glosoli> so likely if I updated some domains ip, if I flush DNS Cache I will get to it as the new IP already
[05:53] <glosoli>  ?
[06:03] <greppy> glosoli: are you using your own local dns server, or using an ISP or other external DNS server?
[06:03] <glosoli> greppy: ISP, so that basically means I  can't do anything yes
[06:03] <glosoli>  ?
[06:04] <greppy> glosoli: probably not
[06:29] <glosoli> what's the best email panel
[06:29] <glosoli>  ?
[06:29] <glosoli> or the most used one, for having email control center or something inside ubuntu server as some kind of web front end
[06:30] <greppy> glosoli: I have been using and liking the froxlor panel from http://froxlor.org
[06:31] <greppy> glosoli: but I also do some domain hosting for friends and family, giving them the power to manage the email for thier domains makes life easier.
[06:31] <glosoli> greppy: does it also provide something like frontend for singing with accounts and reading the mails ?
[06:32] <greppy> glosoli: nope, for that you can use something like squirrelmail or roundcube.
[06:34] <SteveRiley> glosol: for dns flush, see http://ubuntuguide.net/flush-clean-dns-cache-ubuntu-12-04-precise
[06:35] <glosoli> SteveRiley: too late, isp already updated hmm
[06:35] <glosoli> greppy: I found roundcube something pretty
[06:35] <SteveRiley> glosoli: ah, okie. sorry, just noticed your msg.
[06:35] <glosoli> SteveRiley: no problem
[06:41] <glosoli> greppy: 10GB for hosting something like email service for few mails and some other websites, is far too less for a server yes ?
[06:43] <greppy> glosoli: depends on how much space they want to use.
[06:44] <glosoli> greppy: Might be ~5GB
[06:44] <greppy> most websites are not that large unless you have lots of images, movie or music files.
[06:44] <glosoli> well I think if I add everything it might be 90percent of space wasted
[06:44] <greppy> all of my email fits in about 2.3gb, but that is archiving several years and also being on several mailing lists.
[09:42] <Anomie21> Is there any firewall for Ubuntu 11.10 that has a nice web gui?
[09:44] <jMCg> Anomie21: fwbuilder is a nice GUI and can be exported to pretty much everything.
[09:44] <jMCg> We use it at $work for everything (ipfw, iptables, etc…)
[09:45] <jMCg> Personally, I just use ufw on Ubuntu
[10:06] <jMCg> To repeat my fancy pants question from yestereve:
[10:06] <jMCg> 18:58:37 < jMCg> I'm trying to run my own repo (with reprepro), but I run into the following problem: I'm rebuilding packages and they end up with the  same version number in different distros (lucid, precise, oneiric), so when I run includedeb on new packages, I get:
[10:06] <jMCg> 19:00:41 < jMCg> http://dpaste.com/768679/
[10:08] <jMCg> The progress so far: I know that what I want is not possible, because the repo uses the same pool to serve all files from it. If I want it any other way, I'd need a different config for each distro, but HEY HEY, the config is different for each distro ANYWAY. (emphasis mine)
[10:24] <jMCg> putting Multiple in conf/incoming would allow me to put the package in multiple distributions, but of course not when the same version has different.. stuff. Is different.
[10:32] <jMCg> I'm talking to myself, I know, but that's okay, really.
[10:32] <jMCg> Ponies.
[11:01] <glosoli> http://www.webmin.com/deb.html is this ok for doing webmin installatin on Ubuntu Server
[11:01] <glosoli>  ?
[11:11] <bazhang> !webmin | glosoli
[11:12] <glosoli> bazhang: what would you recommend to change it
[11:12] <glosoli>  ?
[11:12] <bazhang> https://help.ubuntu.com/community/Zentyal
[11:16] <glosoli> bazhang: is it free
[11:16] <glosoli>  ?
[12:01] <zul> good morning
[12:06] <jamespage> morning zul
[12:06] <jamespage> zul: hey - I uploaded a fix for openvswitch so that it will actually start with the 3.5 kernel
[12:06] <zul> jamespage: cool...sorry i couldnt get to it soonser
[12:07] <jamespage> however the DKMS package is still broken - I did try to pick some patches but it got to the point where I may as well just done a HEAD snapshot
[12:07] <jamespage> zul, also the kernel module is NOT provided by the kernel on arm*
[12:07] <zul> jamespage:  ok ill take a look at it this afternoon...i swear
[12:08] <jamespage> zul, sorry - I was not hassling - I'm through my immediate blocker with this mornings upload
[12:08] <jamespage> just wanted to impart what I had discovered
[12:09] <zul> jamespage: i know you were not hassling..i need to spend some time on it
[12:19] <feisar> Hi I have a question regarding UFW, I'd like to do the following: sudo ufw allow from any to 192.168.255.5 port 60000:65535/tcp
[12:19] <feisar> but I get the error: ERROR: Port ranges must be numeric
[12:30] <jdstrand> feisar: change /tcp to ' proto tcp'
[12:32] <Pupeno_W> What's the proper place to modify the prompt system wide?
[12:48] <feisar> jdstrand: thank you : ) 'sudo ufw allow proto tcp from any to 192.168.255.2 port 60000:65535' did the trick!
[13:18] <RoyK> any idea what kernel I should choose for an ubuntu server guest (x86_64) under vmware esx?
[13:29] <stgraber> hallyn: so, thinking about it some more (but still haven't checked your branch, sorry), I guess the easiest way of dealing with this mess would be to add an index argument to all the *_config_item() functions and an extra count_config_item() function
[13:30] <stgraber> hallyn: for most of the keys (!network basically), this would be set to 0/NULL but for network, you would be able to query the various sections independently using it
[13:31] <stgraber> hallyn: though I don't really like the count_config_item name, would have to think of something better as it wouldn't be returning the number of key with a matching name, but instead the number of sections to use with index=
[13:33] <stgraber> hallyn: I suppose it could simply be count_config_section() and based on what we currently have, the only option where it'd return >1 would be for anything under lxc.network
[13:36] <jamespage> zul, around? wanted to discuss something openvswitch-ish with you
[13:37] <zul> jamespage: yep
[13:37] <jamespage> zul, OK _ let me set the scene
[13:37]  * zul gets his popcorn
[13:38] <jamespage> I'm working on packaging a tool called mininet which makes use of the openvswitch-controller
[13:39] <jamespage> however; it uses the ovs-controller executable directly
[13:39] <zul> ok
[13:39] <jamespage> BUT the openvswitch-controller package installs an init scripts and starts it up automatically
[13:40] <jamespage> I considered doing bad things in the postinst/postrm script for mininet to stop/disable it
[13:40] <jamespage> but going on the big linitian error I don't think using update-rc.d to manipulate another packages init scripts appears to be acceptable
[13:41] <zul> ok
[13:41] <jamespage> zul: how would you feel about some restructuring of the openvswitch package so that say all binaries are provided by -common
[13:41] <zul> jamespage: i wouldnt mind
[13:41] <jamespage> and the -controller et al packages just have the init scripts
[13:42] <jamespage> zul, do you know the debian maintainer?  I think its one of the upstream guys
[13:43] <zul> jamespage: yeah he is...i would talk to the debian guy...simon horms is his name i think
[13:44] <jamespage> zul, ack
[13:44] <jamespage> I'll email them both
[14:04] <jamespage> rbasak, nice triaging BTW
[14:09] <hallyn> stgraber: I figured only network would have a c->num_networks(), and then c->get_network_config(c, "macvlan.vlan_id") which returns -1 if invalid
[14:12] <rbasak> thanks!
[14:13] <stgraber> hallyn: surely with an extra parameter to get_network_config to tell it what network to retrieve the key from?
[14:18] <jamespage> any chance someone could verify an SRU for me - bug 993291
[14:19] <jamespage> cheers
[14:20] <hallyn> stgraber: of course
[14:20] <hallyn> lemme do some busywork and then i guess i'll write it all back down and see how it looks
[14:22] <stgraber> hallyn: ok. I'm not a big fan of special casing networking like that, but at the same time my suggestion would essentially do that too (as count_config_section() would only return >1 for network)
[14:28] <hallyn> stgraber: well there is another possibility - we can use our own tweak on the config syntax, and have:
[14:28] <hallyn> get_config_item(c, "lxc.network") -> ("veth", "macvlan", "veth")
[14:29] <hallyn> get_config_item(c, "lxc.network.nic2.macvlan_id") -> "4802"
[14:29] <hallyn> uh, make that "veth\nmacvlan\nveth" for the first return value
[14:31] <stgraber> hallyn: could you make that:
[14:31] <stgraber> get_config_item(c, "lxc.network") => "eth0\neth1\neth2" (with ethX being whatever name lxc would use in the container)
[14:32] <stgraber> then have get_config_item(c, "lxc.network.eth2.macvlan_id") => 4802
[14:33] <stgraber> because I'm kind of planning to ask for a config format change for lxc.network to essentially match the above, so that way, if that change ever gets upstream, we wouldn't need any API change
[14:33] <stgraber> (or rather, API consumers wouldn't need to change)
[14:37] <hallyn> stgraber: but they don't have to have a name
[14:37] <hallyn> i suppose for unnammed ones i could use 'nic%d', and use the name if it exists...   complicates (and makes more fragile) by a bit, but gives you what you want in most normal cases
[14:38] <stgraber> hmm, good point...
[14:38] <stgraber> ok, let's just go with lxc.network.<index>.<key> => lxc.network.0.macvlan_id
[14:39] <stgraber> or nic0 if you prefer but I'm not sure it really makes it any clearer :)
[14:39] <hallyn> agreed
[14:39] <hallyn> i suspect from a user pov this will mean we'll want a get_keys() fn, i.e. get_keys("lxc.network.0")
[14:42] <stgraber> hallyn: yeah, that'd make my life much easier when writing the python code to deal with networks. Otherwise I'd have to include an hardcoded list of keys and go through them all, not difficult to do but seems to be a bit of a waste of CPU time ;)
[15:12] <glosoli> Hey folks, I have some question, so here it goes: I have made backup via Direct Admin from some shared hosting, is there any way to restore that backup in some Dedicated Server I own with something other than Direct Admin ? cheers
[15:16] <jamespage> Ursinha, do you have the mootbot meeting log for last week?
[15:16] <jamespage> I seem unable to find it :-)
[15:16] <Ursinha> oops, sure, just a moment
[15:18] <jamespage> Ursinha, thanks for running the meeting BTW - much appreciated
[15:19] <Ursinha> jamespage, no problem at all, sorry I haven't updated the page as I should
[15:20] <Ursinha> jamespage, here's is the output generated by the bot: http://ubottu.com/meetingology/logs/ubuntu-meeting/2012/ubuntu-meeting.2012-07-03-16.00.moin.txt
[15:23] <jamespage> Ursinha, ta
[15:31]  * jamespage nudges Daviey about the thrift packages in the NEW queue
[15:32] <Daviey> jamespage: undoubtedly
[15:44] <jamespage> Daviey, ta
[15:46] <Daviey> jamespage: It's all accepted, but can you confirm python-thrift needs Arch: any ?
[15:46] <jamespage> Daviey, yep - it builds a .so
[15:46] <Daviey> ah, super
[16:00] <Daviey> o/
[16:00] <Daviey> gah
[16:04] <bencer> there were some ongoing efforts to package ovirt, what's the status of this on 12.10?
[16:09] <smb> zul, Hm, I just realized (by failure) that between my xen version and yours from the later source there seems to be a tiny but evil difference that it does not create the /usr/lib/xen-default link. Something that libvirt does not like much...
[16:09] <Daviey> !!!
[16:09] <zul> smb: ergh...ok
[16:10] <smb> zul, Not sure why. To me the update-alternatives --remove in the xen-utils postinst looks a bit weird...
[16:27] <zul> smoser: i have a euca2ools patch if you want to review it later this week
[16:28] <smoser> zul, have you/can you do a pull request to upstream?
[16:28] <smoser> https://github.com/eucalyptus/euca2ools
[16:29] <zul> smoser:  no/yes, it adds armhf as a valid architecture
[16:29] <zul> so i dont know if its suitable for upstream inclusion
[16:29] <smoser> i would suspect it is.
[16:29] <smoser> or surely we can make it such that it is
[16:37] <glosoli> I am installing Zentyal
[16:37] <glosoli> anyone can say what does that mean "   │ Please enter the password for the admin entry in your LDAP directory.  │  "
[16:37] <glosoli>  ?
[16:37] <glosoli> LDAP Directory what exactly is it for
[16:40] <xranby> glosoli: LDAP can be used to keep track on manu users in larger organizations. like centralized login authentication for multiple systems
[16:40] <xranby> many
[16:40] <glosoli> so that password should be hard to ques ?
[16:40] <glosoli> guess"
[16:41] <Daviey> zul: so.. should we do a pep8 1.1 upload?
[16:42] <zul> Daviey: wouldnt it be a bit backwards since we already have 1.2 in quantal
[16:42] <Daviey> zul: i mean, 1.1 to quanta
[16:43] <Daviey> quantal
[16:43] <Daviey> As in, i don't think openstack is going to 1.2 anytime soon
[16:43] <xranby> glosoli: sorry i have not setup any Zentyal system so i cant guide you..
[16:43] <glosoli> xranby: what do you use
[16:43] <glosoli>  ?
[16:43] <zul> Daviey: yeah i started getting the pep8 tests past with 1.2, its not going to be easy
[16:44] <zul> so yeah
[16:44] <Daviey> zul: i think we should just go back :(
[16:44] <Daviey> i'll do that now
[16:44] <zul> Daviey: i dont want to....but yes
[16:46] <zul> Daviey: pep8 will need to be blacklisted for syncs then
[16:46] <zul> jamespage: i dont understand it...according to the changelog for openvswich you patched to work with 3.5 but dkms building still fails for me
[16:47] <jamespage> zul, I did not - "use of openvswitch without openvswitch-datapath-dkms"
[16:47] <jamespage> I could not get the dkms build to work
[16:47] <smb> zul, So yes, I think that xen-utils-4.1.postinst should actually have a "update-alternatives --install /usr/lib/xen-default xen-default /usr/lib/xen-4.1 50" (or whatever prio is normal) instead of --remove-
[16:48] <zul> jamespage: ok ill start there then :)
[16:48] <jamespage> zul, it probably needs a 1.7.x git snapshot with some patches I think
[16:48] <zul> smb: 4.1.3~rc1+hg-20120614.a9c0a89c08f2-3 removed the /usr/lib/xen-default according to the changelog
[16:49] <zul> smb: so libvirt needs to be fixed
[16:49] <zul> jamespage:  or patch the snot out of the dkms module
[16:50] <jamespage> zul, I got to 5 patchsets; none of which applied cleanly and then gave up
[16:50] <zul> jamespage: *sigh* :)
[16:50] <smb> zul Just read that... Though I wonder how well that works if xen suddenly becomes xen-4.2... But probably not an immediate problem...
[16:51] <zul> smb: i guess we will find out eventually :)
[16:51] <smb> Pretty much very likely. :)
[16:55] <zul> jamespage: crud... with the openvswitch git tree http://paste.ubuntu.com/1084805/
[16:56] <jamespage> zul: yeah - git tree + patch to allow 3.5 kernel
[16:56] <zul> jamespage: right
[16:56] <jamespage> zul, there is only one explicit kernel check in datapath/datapath.c
[16:57] <jamespage> and note that the only think that this provides over the kernel openvswitch provided module is the brcompat module
[16:57] <jamespage> which is not advised....
[16:57] <zul> yeah
[16:59] <zul> jamespage: i wonder why we are still using the 1.4 branch
[16:59] <jamespage> zul, not sure; its been pretty much untouched from Debian AFAICK
[16:59] <jamespage> AFAIK
[16:59]  * jamespage <- should learn to type
[16:59] <zul> maybe stick 1.6.1 in a ppa so we can test it out
[17:00] <zul> jamespage: there is a program called hooked on phonics :)
[17:00] <jamespage> zul, it still lacks the 3.5 support tho
[17:00] <jamespage> I think thats only in the 1.7 branch
[17:00] <jamespage> sorry - not even there....
[17:00]  * jamespage sighs
[17:01] <jamespage> zul: worth a discussion with upstream - bpl just pointed out an obvious way for me to fix my mininet problem...
[17:01] <jamespage> without restructuring the package....
[17:01] <jamespage> blp that is
[17:02] <jamespage> not bpl - he's someone else....
[17:20] <glosoli> http://dpaste.org/PmoiE/ how do I fix this ?
[17:51] <piercedwater> I am trying ton convert FLAC to ALAC using avconv, but i get this error and avconv stops: [flac @ 0x824baa0] Format detected only with low score of 25, misdetection possible!
[17:51] <piercedwater> any ideas?
[17:53] <r3dLunchb0x> looking for a good server reporting tool. Just need basics, top/network/disk/cpu/mem
[17:53] <r3dLunchb0x> something I can send to management
[17:54] <tonyyarusso> Hi, I just copied rsyslog config files over from an older Ubuntu system (either 10.04 or 8.04) to a new Ubuntu 12.04 system, and something's different with our logging behavior.  This is a central logging box, and it's supposed to write to separate files per source host.  Problem is, it's doing that AND writing EVERYTHING to the normal /var/log/syslog.  Did something change in how that's handled in the config file?  How do I ...
[17:54] <tonyyarusso> ... tell it not to duplicate the remote stuff, and only put local messages in there?
[17:54] <tonyyarusso> Here's the config we're using:  http://pastebin.com/exZtpiSK
[17:55] <tonyyarusso> r3dLunchb0x: What exactly are you trying to report?  They're current status?  Capacity trends?  Outage history?
[17:56] <r3dLunchb0x> trends over 5-7 days. I tried nagios and zabbix but i would like simple...bash scripts even
[18:00] <glosoli> Hmm wha can be the problem
[18:00] <glosoli> apache2 is isntalled
[18:00] <glosoli> there is no init script for it
[18:03] <adam_g> win 9
[18:07] <koolhead17> hoi all
[18:45] <r3dLunchb0x> tonyyarusso: trends over 5-7 days. I tried nagios and zabbix but i would like simple...bash scripts even
[18:56] <tonyyarusso> r3dLunchb0x: For things like disk space, a little bash script would be fine.  Network traffic would be a little trickier, depending on what level of granularity (how frequent of checks) you need.  You could use the Nagios plugins and just run them from cron, or use snmpget for everything, again from cron.
[18:56] <tonyyarusso> If you wanted to actually calculate trend lines or anything there are ways to do it, but it'd be a bit more complicated.
[19:06] <r3dLunchb0x> tonyyarusso: snmpget....i tried that on ubuntu server 10.04 running on HP proliant hw, I get nothing worth while...Is there not a pkg for something like this?
[19:27] <rbasak> du2rrd (http://oss.oetiker.ch/optools/wiki/du2rrd) is awesome for monitoring disk space
[19:50] <zastaph> can't decide if I should setup RAID 0 or 1 from BIOS which is a fake RAID or from Ubuntu https://help.ubuntu.com/community/Installation/SoftwareRAID on my HP Microserver
[19:55] <RoyK> zastaph: I'd drop raid config and rather use software raid
[19:55] <RoyK> just my opinion...
[19:55] <zastaph> right
[19:55]  * RoyK has setup dozens of servers on software raid...
[19:56] <zastaph> I got one 160GB and 2 250GB in the server.. could probably do RAID 5, but then all 3 would be 160GB i guess
[19:57] <RoyK> you have a wee bit more control with software raid when something goes wrong
[19:57] <RoyK> two small partitions for the root and the rest for raid-5 should do
[19:57] <RoyK> although, personally, I never mix those, but I guess there aren't too much space in that thing
[19:58] <RoyK> what about two tiny SSDs for the root mirror and three largish drives for the data?
[19:58] <zastaph> i got 2 small servers.. this one will be for KVM on a USB stick, and then 160GB space for 3-4 VM's .. then on the other server I'll install a samba/NFS for actual storage
[19:59] <zastaph> i figured that would be the right way to do things
[19:59] <RoyK> you could use two USB sticks for the root and the drives in raid-5
[20:00] <RoyK> you'll hit the wall once it starts swapping, though
[20:00] <zastaph> if the USB stick breaks I'd just lose the RAID master and KVM server... that could be restored I guess
[20:01] <RoyK> zastaph: keep in mind swap if you want to use a usb stick for the root - those sticks are usually slow
[20:01] <zastaph> right..
[20:01] <RoyK> zastaph: if you have sufficient memory, decrease vm.swappiness to something very low, like 0
[20:02] <zastaph> so could use the 160GB for KVM, and buy 3 new for RAID-5 for VM's
[20:02] <RoyK> kvm/libvirt without the VMs is just config files
[20:02] <Lcawte> Anyone around to help me with some firewall problems
[20:02] <RoyK> so two 16GB plugs in a mirror for the root should suffice
[20:02] <zastaph> yes but you said if I use usb stick I'll get a bottleneck?
[20:03] <zastaph> don't have 2 plugs, only 1
[20:03] <RoyK> a usb hub? :)
[20:03] <RoyK> another single point of failure, so maybe not
[20:03] <zastaph> :)
[20:04] <RoyK> Lcawte: !ask
[20:04] <RoyK> erm
[20:04] <RoyK> !ask
[20:04] <RoyK> zastaph: are you sure there aren't more USB outlets on the mobo?
[20:04] <RoyK> there usually are
[20:04] <zastaph> RoyK, internally no.. but sure outside.. but hmm?
[20:05] <zastaph> and raid on USB sticks?
[20:06] <RoyK> why not? :)
[20:06] <Lcawte> Ok, so I'm running Ubuntu Server, and I have iptables installed, anyway... the firewall I assume, it doesn't seem to be the router, as other people were getting through, used to "randomly" block certain people, like one of my friends used to get blocked, but now everyone but my IP seems to be blocked. Although, on ports other connections are getting through from the same server... whats going on?
[20:07] <Lcawte> domain is lewiscawte.info
[20:07] <zastaph> and that's a stripe raid i guess?
[20:07] <RoyK> zastaph: raid-1 aka mirror
[20:07] <zastaph> but then I don't get any speed from it.. so what's the point? if the KVM itself breaking isn't any critical?
[20:07] <RoyK> zastaph: so that when (not if) one of them dies or is unplugged, the other can do the job
[20:08] <zastaph> the VM's could still be restored i guess
[20:08] <RoyK> zastaph: well, if the root dies, your VMs will probably have a rather hard time
[20:08] <zastaph> hmm, right
[20:08] <zastaph> and installing RAID on the root is as easily as doing it on other disks?
[20:09] <RoyK> but then, if you don't need uptime, just go with a single drive
[20:09] <zastaph> no I don't need uptime
[20:09] <RoyK> zastaph: with ubuntu server, yes, just create partitions for boot and root and swap as "physical partitions for md" (equal size on both drives) and configure md
[20:09] <RoyK> that's during install
[20:10] <RoyK> you can't do that with ubuntu desktop install
[20:10] <zastaph> right, I read that
[20:10] <zastaph> what if I choose minimum install? (for virtual machine)
[20:10] <RoyK> but then, running ubuntu desktop off a usb plug won't be much fun :P
[20:10] <zastaph> should I choose normal
[20:10] <RoyK> erm
[20:10] <RoyK> the VMs shouldn't do much thinking about md
[20:11] <RoyK> just setup your server with two USB sticks, make a mirror on  those, or three, one for boot, one for root, one for swap
[20:11] <RoyK> once installed, create a raid-5 on the three spinning drives (or do it during install, up to you)
[20:12] <zastaph> i also had this design in mind: 1 USB for root (KVM on Ubuntu) .. 160GB for VM's, and RAID 1 on x2 250GB's, doing daily/weekly backup of the VM's to that. Anyway if I put my VM's on a RAID it's not a backup
[20:12] <RoyK> mount that somewhere like /usr/lib/libvirt (if you do it after installation, make sure to rsync -avPHAS the data from the old dir to the newly mounted)
[20:12] <zastaph> and then all data on my secondary server, RAID 10 setup or something
[20:12] <RoyK> zastaph: didn't you say you had a secondary server? if so, use that for backup
[20:13] <zastaph> someone told me it's always a good idea to use a decoupled NFS for data, to avoid having data in VM's, keep them tidy
[20:13] <RoyK> I don't think that would be of much need
[20:14] <RoyK> I've been running VMs off NFS for some time
[20:14] <zastaph> I like the idea of decoupling logic and data
[20:15] <RoyK> zastaph: sure, the data is under /var/lib/libvirt/images, the logic is mostly under /etc/libvirt
[20:16] <RoyK> zastaph: but then, decoupling sufficiently, and you lose control over what's really needed, to keep good backups until the day you need them
[20:16] <zastaph> yes and I would just mount the other server folders on the KVM server
[20:17] <RoyK> zastaph: well, not much changes in kvm land, except the images, that is, unless you change the VMs
[20:17] <zastaph> lets say I have my git repositories and my sql databases taking up 1 TB .. Would you store those inside the actual VM image? or on a secondary image on the same server?
[20:17] <RoyK> zastaph: if I were you, I'd setup the server with two small, spinning drives, and perhaps a spare, and just use the storage server for the VMs, over NFS
[20:19] <RoyK> zastaph: you *can*, but it's not necessarily a good idea to run databases in a VM. Better use the fileserver for that. For Git, I don't think it matters much, but perhaps an NFS share on the fileserver for storage will look better the day you need to extract it and the VM has somehow died
[20:19] <zastaph> would you make small images for VM's? 30GB something, and then secondary images for the server, and everything on the storage server?
[20:20] <zastaph> duh, secondary images for the data
[20:21] <RoyK> just tell kvm not to allocate the VM images, and let them grow dynamically, then you can make them as big as you like (or whatever is healthy)
[20:21] <zastaph> then you would store the actual data (git repositories, etc.) inside the VM's ?
[20:21] <RoyK> for data, I'd say use the fileserver with NFS from the VMs
[20:21] <zastaph> heh i'm lost
[20:22] <zastaph> i thought the mantra about VM's is that they are "encapsulated" and can be easily moved
[20:23] <RoyK> control question: You're setting up a fileserver with a bunch of space, right? And, you're setting up a smallish server to serve VMs
[20:24] <zastaph> fileserver has 4 1TB's .. I was thinking about setting up vanilla Ubuntu server on that, with manual samba or NFS sharing..
[20:25] <zastaph> but yes to your question
[20:26] <RoyK> zastaph: in my world, I'd use VMs for small things or application specific setups, like a Zimbra server, or some webserver for my brother, or this irc login host
[20:26] <RoyK> for things demanding I/O specifically, I'd use iron
[20:26] <zastaph> what is Iron
[20:27] <RoyK> hardware
[20:27] <RoyK> like your fileserver
[20:27] <zastaph> performance is not my main goal.. i'll be the only user of these servers.. maintainability, stability and not losing my data is important
[20:27] <RoyK> zastaph: if a VM needs to serve large amounts of data, I would do that via NFS from the server to avoid that data locked up in a VM image
[20:28] <zastaph> i'd like to keep things simple though
[20:28] <RoyK> if it's smallish, it's no problem
[20:28] <RoyK> if it's a database, I'd use the fileserver, if the database is significant in size
[20:29] <RoyK> if it's something small, like a wordpress db, just use it locally
[20:29] <zastaph> install mysql on the fileserver?
[20:29] <zastaph> not install mysql in a VM, and then the actual data on the fileserver?
[20:30] <RoyK> I have a single server at an ISP running a few VMs. A few of the VMs need databases, and the host runs those. It makes better use of available memory than spreading that buffering out to the VMs, allowing me to shrink memory use on the VMs
[20:30] <RoyK> I don't run databases in VMs unless it's something very isolated
[20:30] <RoyK> or unless someone is sufficiently geeky to demand full admin rights to that db :P
[20:30] <zastaph> make sense.. and indeed it would be a good idea to just install mysql to the fileserver running vanilla ubuntu
[20:31] <zastaph> it should be able to handle that and samba/NFS at the same time
[20:31] <RoyK> yes
[20:31] <RoyK> even a very, very, very cheap machine would be able to handle that
[20:31] <RoyK> even a cheap cellphone :P
[20:31]  * RoyK reminds everyone that 1999 is over
[20:32] <zastaph> actually.. I could also install git-core on the fileserver, rather than letting a secondary server/VM handle that
[20:32] <zastaph> i'm starting to rule out the VM server's purpose now :p
[20:32] <RoyK> zastaph: so long noone needs direct access to the git root, yes
[20:32] <RoyK> zastaph: but if you want someone to access the git file store directly, that's better done within a VM
[20:33] <zastaph> i.. don't think I need that
[20:33] <RoyK> zastaph: but then, perhaps the git file store may be on the fileserver, shared with nfs
[20:33] <resno> my machine booted to busybox, ibelieve my hdd is failing, how can i get it running temporarliry?
[20:33] <zastaph> just a generic git user, behind ssh
[20:33] <resno> i tried fsck
[20:33] <RoyK> resno: boot on a live cd and start troubleshooting from there
[20:33] <zastaph> and when I think about it.. KVM only complicates things :)
[20:33] <RoyK> :)
[20:34] <zastaph> and if I can run git+mysql+samba/NFS on one server, I have a whole server leftover for playing around
[20:34] <RoyK> zastaph: you can probably run those VMs on the fileserver as well, normally samba+mysql+git won't use that much cpu
[20:34] <zastaph> and install KVM where?
[20:35] <RoyK> that's what I ended up doing, just two 1,5TB drives in a mirror, cpu is just a core2duo, 6 gigs of RAM, currently 5 VMs in production
[20:35] <RoyK> on the fileserver
[20:35] <resno> RoyK: production wise, what hypervisor do you use?
[20:36] <RoyK> just make sure you have (a) mirrored root on separate drives and (b) a small RAID set, RAID-5 or better, for the data, also on separate drives
[20:36] <RoyK> !kvm
[20:36] <resno> oh you use kvm everywhere?
[20:36] <RoyK> resno: I just use kvm, some people prefer Xen, since it's slightly cooler, but it's harder to work with (IMHO)
[20:36] <zastaph> RoyK, fileserver has 4 1TB's .. can't I just RAID10 those for one main ubunut server? and then install git+mysql+samba/NFS ?
[20:37] <RoyK> resno: at work, we use vmware in a rather large setup - kvm can't do that IMO
[20:37] <resno> RoyK: i was considering going esxi
[20:37] <RoyK> zastaph: do you have room for another two oldish 160GB drives for the root?
[20:37] <zastaph> resno, I tried all of them before I settled with KVM.. they are only headaches :)
[20:37] <RoyK> resno: it's better, I guess, but KVM works well for me
[20:38] <zastaph> RoyK, not without some modifications.. but.. if I'm not going to KVM, why do I need a root?
[20:38] <resno> RoyK: my question has been how do you setup drives on a baremetal hypervisor?
[20:38] <resno> esxi on usb, and then os drive and then data drives?
[20:39] <zastaph> esxi lost me when I tried to clone and had to pay cash :)
[20:39] <RoyK> resno: depends on the hardware, obviously, but my typical setup would be two small spinning drives for the root, and a largish raid-5 or raid-6 for the data
[20:39] <resno> RoyK: "root" being? oses stuff?
[20:39] <RoyK> or raid-1+0 if you're worried about IOPS
[20:40] <RoyK> resno: root being your root filesystem with its programs, /boot and swap
[20:40] <resno> gotcha
[20:41] <RoyK> zastaph: erm - you need a root filesystem on the server ;)
[20:41] <zastaph> RoyK, then we're back to scratch.. root on one server, and data on fileserver
[20:41] <RoyK> zastaph: and you can't boot off a raid-[56]
[20:41] <zastaph> oh
[20:42] <zastaph> I only have 4 HDD slots in each server
[20:42] <zastaph> rest have to be mounted in special ways
[20:42]  * RoyK hands zastaph a roll of duct tape
[20:42] <zastaph> decisions decisions
[20:42] <RoyK> or perhaps a new chassis? ;)
[20:42] <zastaph> ouch
[20:43] <zastaph> all the time wasted to customize these 2 :)
[20:43] <RoyK> most of us have a few around, and they should be usable for most mobos
[20:44] <zastaph> so.. even if I wanted to dedicate one for solely fileserving, I would still need 1-2 extra disks for its root? if I wanted the 4 HDD slots for RAID-10 ?
[20:45] <RoyK> yes, you need more than four drive slots for a good fileserver
[20:45] <RoyK> you can, however, run the root off a couple of usb drives, but it will be slowish to boot, and not much fun if it starts swapping
[20:45] <resno> heh, a fileserver with limited drive expandability?
[20:45] <RoyK> but then, swapping is rarely fun
[20:45] <RoyK> resno++
[20:45] <zastaph> resno, but a very energy efficient one ;)
[20:46] <zastaph> they both have eSATA :p
[20:46] <resno> is using sas not an option?
[20:47] <RoyK> zastaph: how many esata ports?
[20:48] <soren> RoyK: Actually, grub2 should support booting from raid5.
[20:48] <RoyK> zastaph: a port multiplier won't be much good for performance
[20:48] <zastaph> 1 in each
[20:48] <RoyK> soren: oh - when did that happen?
[20:48] <RoyK> !grub2
[20:48] <zastaph> soren, RAID-10 too?
[20:48] <soren> RoyK: I have mailing list posts from at least 2009 suggesting that it works.
[20:48] <soren> RoyK: http://www.mail-archive.com/grub-devel@gnu.org/msg08994.html
[20:49] <zul> hallyn: ping
[20:49] <RoyK> soren: I thought I tried that within a vm with precise just recently and found it didn't work at all
[20:49] <zastaph> and does that work for software raid too?
[20:49] <zul> soren: grub2 works?
[20:49] <RoyK> still
[20:50] <RoyK> even if it *is* supposed to work, I stand my ground: I would recommend two smallish drives for the root, and a small bunch of largish drives for a RAID-5 or -6 set for the data
[20:51] <RoyK> that will separate system and data, and make recovery far easier the day it all goes wrong
[20:51] <zastaph> hmm right..
[20:51] <zul> hallyn: so libvirt-lxc segfaults for me
[20:52] <hallyn> zul:		/win 20
[20:53] <soren> zul: It's been known to not be entirely broken at times.
[20:53] <hallyn> (*&(* lag
[20:53] <zastaph> RoyK, then I could get a dual-disk device -> eSATA and boot from that, and use 4 HDD's in each for RAID-10
[20:53] <hallyn> zul: where does it segfault?
[20:53] <zul> hallyn: when its trying to load veth0
[20:54] <zul> hallyn: hold on
[20:54] <RoyK> zastaph: sounds like a good idea ;)
[20:54] <zul> hallyn: this is the dmesg
[20:54] <zul> hallyn: http://paste.ubuntu.com/1085179/
[20:54] <RoyK> zastaph: or even better, you said you weren't worried about performance, if so, rather use raid-6, same net data, better safety
[20:55] <zul> hallyn: http://paste.ubuntu.com/1085180/ thats the xml domain
[20:55] <zul> hallyn: thats the log file: http://pastebin.ubuntu.com/1085182/
[20:56] <zul> hallyn: and yes veth is loaded
[20:56] <zastaph> RoyK, thanks for all the ideas
[20:57] <RoyK> zastaph: out of interest, how many VMs do you plan to run concurrently?
[20:57] <RoyK> if it's just a few, and perhaps a local mysql or postgresql db on the host, and a git thing, then raid-6 or even raid-5 will probably suffice
[20:57] <zastaph> yeah 3-4
[20:57] <RoyK> linux' i/o scheduling and buffering does a fairly good job for those things
[20:58] <zastaph> but don't forget it's atom based servers, and software raid
[20:58] <RoyK> which atom?
[20:58] <RoyK> H?
[20:58]  * RoyK sniggers
[20:59] <RoyK> sorry, just, which atom cpu?
[20:59] <RoyK> and how much memory?
[20:59] <zastaph> Acer easyStore H340 -> Intel Atom 230 ... HP ProLiant Microserver -> AMD N36L
[20:59] <zul> hallyn: and im at a lost
[20:59] <zastaph> ahh memory.. only 2 GB in the easyStore but 8 GB in the microserver
[21:00] <zastaph> which calls for the easyStore being just a dumb fileserver
[21:00] <zastaph> and Atom 230 has no hardware extensions
[21:00] <zastaph> for virtualization
[21:00] <RoyK> then better separate them
[21:01] <resno> atoms virtualizing..
[21:01] <hallyn> zul: trying...
[21:01] <resno> this is just sounding worse to me
[21:01] <zastaph> HP microserver is quite good for virtualization with a minimum energy use
[21:01] <zul> hallyn: 0.9.13 does the same thing i think
[21:01]  * RoyK thought his fileserver was fried today and checked pricing for a new mobo+cpu+8GB RAM and it cost about NOK 1300 with a two (or four?) core amd thing
[21:02] <RoyK> I don't really see a good reason for using atoms for servers...
[21:02] <hallyn> zul: you get this with what's stock in quantal?
[21:02] <zul> hallyn: yep
[21:02] <RoyK> zastaph: how much did those machines cost?
[21:02] <zastaph> RoyK, yeah perhaps I live up to the saying when you buy cheap you pay double :)
[21:03] <RoyK> zastaph: and btw, that was from a local computer shop, not something from ebay ;)
[21:03] <zastaph> RoyK, don't recall.. easyStore I bought for windows home server (which broke, obviously) and HP Microserver was quite cheap
[21:03] <RoyK> NOK 1300 is ~USD 200, btw
[21:04] <zul> hallyn: if i removed the nwfilter stuff it works fine
[21:05] <zastaph> looking for a eSATA device for 2 HDD's
[21:06] <RoyK> zastaph: really, get an old chassis with room for some drives and start over - it'll be easier and probably just as cheap ;)
[21:06] <zastaph> RoyK, and I have a wet dream to get ZFS working :p http://sstahlman.blogspot.dk/2011/02/acer-easystore-h340-using-eon-to-fool.html
[21:06] <RoyK> zastaph: and quite possibly very much better
[21:06] <RoyK> zastaph: with linux?
[21:06] <zastaph> mmm.. wanna buy 2 small servers? :)
[21:07] <zastaph> nah solaris
[21:07] <RoyK> I've used openindiana in production for some time
[21:07] <RoyK> but I guess smartos or omnios may be better supported these days
[21:07] <zastaph> I looked at the install docs and just gave up :)
[21:07] <zastaph> i tried smartos, same feeling
[21:07] <resno> i was once excited about zfs, but now im against it
[21:08] <zastaph> if they make ZFS in something as easy to use as ubuntu server, I'm on
[21:08] <resno> non-expandablity of raid was a kill
[21:08] <RoyK> resno: I've used zfs for some 400TB of storage and it rocks, but I'm using linux md for my home server for the reason you mention
[21:09] <resno> yea, for my home server being stuck willnever work
[21:09] <RoyK> resno: the only way to expand a redundant zpool correctly, is to replace every drive with larger ones, which is somewhat a showstopper for a homeserver
[21:10] <RoyK> linux md rocks that way :D
[21:10] <RoyK> make a mirror, convert to raid-5, add a drive, add another, convert to raid-6, oops need more space, convert to raid-5 ...
[21:11] <zastaph> someone needs to simplify RAID :)
[21:12] <RoyK> zastaph: it's a bit hard for a complex system like zfs
[21:13] <RoyK> since raidz is more like raid-3 than anything else - there are no block-level stripes, just write stripes
[21:14] <RoyK> and then there's a snapshot, or a clone, and another dedup, so to change that to something else while running, you need to rewrite the whole system, which isn't very easy
[21:14] <zastaph> everything can be simplified.. system shouldn't ask you silly questions like pools and stripes.. instead it should show you a GUI with your harddisks, and ask if you want reliability or speed, and then set it up behind the scenes :)
[21:14] <resno> RoyK: im just curious if i have two disks free and a 3rd with data... can i do that?
[21:14] <resno> zastaph: do you run a gui on your server?
[21:14] <RoyK> resno: detail that, please
[21:14] <zastaph> nah, just saying that RAID could be simplified a lot
[21:15] <RoyK> zastaph: it really can't
[21:15] <RoyK> zastaph: RAID is rather complex
[21:15] <hallyn> zul: bridging to virbr0 worked fine, bridging to br0 seems to have killed my instance
[21:15] <resno> RoyK: ok. 3x2TB drives. 2 of the drives are empty. 1 drive has stuff on it. so, i want to start raid but dont have 3 disks to do it (raid 5)
[21:15] <RoyK> zastaph: it'd be like saying "scientists, can you please simplify this physics thing?"
[21:15] <zastaph> so was programming in the 80es.. but it's gotten easier :)
[21:15] <hallyn> or just killed its network
[21:15] <RoyK> resno: just create a RAID-5 on those two drives
[21:16] <resno> RoyK: ah, you can do that... ool
[21:16] <RoyK> resno: then create a filesystem on that one, or perhaps an lvm system first if you want more flexibility, then create a filesystem on the lvm
[21:16] <resno> swell
[21:16] <resno> RoyK: gotta run andpikc up the kid :) bbl
[21:16] <RoyK> resno: then move the data, then expand the raid to the third drive, it'll take half a day or so, but the system will be online during that time
[21:17] <stgraber> hallyn: hmm, I have a weird one for you, maybe you have a clue of what's going on here :)
[21:17] <stgraber> ~ # ls /proc/sys/net/ipv4/conf/
[21:17] <stgraber> all      default  eth0     lo       sit0
[21:17] <stgraber> ~ # ls /proc/sys/net/ipv6/conf/
[21:17] <stgraber> all      default  lo       sit0
[21:17] <stgraber> hallyn: found that one while testing my automated ipv6 tester with quantal containers
[21:18] <stgraber> hallyn: basically eth0 vanished from ipv6/conf but not from ipv4/conf (I'd expect it to be present in both)
[21:18] <stgraber> I didn't even know that was possible (and not quite convinced it should be ;))
[21:18] <RoyK> stgraber: can you pastebin 'ifconfig -a' ?
[21:18] <RoyK> !pastebin
[21:19] <hallyn> stgraber: interesting.  i don't get that in precise at least
[21:19] <stgraber> RoyK: no ifconfig, I'm debugging that in an installer initrd ;) and don't worry, I know pastebinit, I wrote it.
[21:19] <stgraber> hallyn: that's with a quantal container running on a precise host
[21:19] <stgraber> hallyn: running the exact same test with precise container on the same quantal host works fine
[21:20] <hallyn> quantal container on quantal host also works for me.
[21:21] <hallyn> lemme create a quantal container on my precise laptop
[21:21] <stgraber> hallyn: yeah, seems to be specific to whatever d-i is doing, the router container works great on quantal
[21:21] <hallyn> funky
[21:21] <stgraber> yeah, also based on the logs it worked for a while as it managed to run rdisc6, get an IP and everything, then dhclient started and no more v6 after that :)
[21:22] <hallyn> and so you don't expect me to see this in a quantal container on precise host?  you were doing something 'special'?
[21:23] <stgraber> hallyn: well, starting debian-installer in a container, that's all ;) (which you may consider as "special")
[21:24] <hallyn> i do
[21:24] <hallyn> is it possible for you to have it straced?
[21:24] <zastaph> RoyK, hey.. I thnk both servers support PXE boot.. can't I just use the 1 server for root then, and boot second server from first, thus dedicate all 4 disks to RAID-6 ?
[21:26] <stgraber> hallyn: stracing the whole installer is a bit tricky as it's downloading quite a few megs of stuff from the network, I'm trying to figure out exactly what bit triggers that bug so I can strace it
[21:28] <RoyK> zastaph: you can, but I still think it's a jolly good idea to use an old chassis with some cheap mobo with enough SATA controllers and room for drives and more SATA controllers as the drives fill up
[21:28] <hallyn> zul: jinkeys!  this is a kernel bug
[21:28] <hallyn> smb: ^
[21:28] <hallyn> well, ok, the kernel bug doesnt' really look like what zul is reporting, but crikey
[21:28] <stgraber> hallyn: though besides half of the libraries changing, a clear change is isc-dhcp that got bumped from 4.1 to 4.2, I'll diff the dhclient script quickly to see if it wouldn't be that simple :)
[21:28] <RoyK> erm - what kernel bug?
[21:29] <hallyn> the one where when i do 'brctl addif br0 veth0' my networking hangs
[21:29] <zastaph> RoyK, these 2 might have more SATA controllers inside, but then I need to figure what to do with power aswell
[21:30] <RoyK> zastaph: most drives don't use too much power these days, at least not the "green" ones
[21:31] <zastaph> and why is PXE a bad option?
[21:31] <RoyK> zastaph: just please listen - using a traditional pc with very cheap mobo+cpu+memory will make your life a bit less miserable once you need to upgrade
[21:31] <hallyn> zul: does your whole compute node become inaccessible after that segfault or not?
[21:31] <RoyK> zastaph: it's just another thing that can fail
[21:31] <zul> hallyn: no dont think so
[21:31] <hallyn> odd
[21:31] <zastaph> RoyK, sure.. but it's hard to accept that I have to give up 2 servers :)
[21:32] <hallyn> all right let me go grab my quantal laptop and test on bare metal
[21:32] <zul> hallyn: no it doesnt
[21:32] <RoyK> zastaph: sell them on ebay :D
[21:32] <RoyK> zastaph: or keep one for playing around, use it for something cool
[21:32] <zastaph> might try that
[21:32] <zastaph> I use VM's for playing around
[21:33] <RoyK> zastaph: I've been working with data storage as my prime job for about three years, with mostly open systems, and I'm just giving you a bit of advice
[21:33] <zastaph> but I don't want to build a server myself.. then rather buy something ready
[21:34] <RoyK> zastaph: it's just a motherboard, a cpu and some memory and drives
[21:34] <hallyn> zul: which kernel are you on?
[21:34] <RoyK> and a havoc of cables
[21:34] <zul> 3.5
[21:34] <zul> 3.5.0-3-generic
[21:34] <zastaph> with me, buying hardware is never just something, it's lots of options and decisions :)
[21:35] <RoyK> well, that's your choice
[21:35] <RoyK> I'm going to bed
[21:35] <zastaph> good night
[21:35] <RoyK> nite, lads and lassies
[21:36] <zul> hallyn: i noticed that CONFIG_NET_CLS_CGROUP is built as a module
[21:37] <stgraber> hallyn: http://paste.ubuntu.com/1085250/
[21:38] <stgraber> hallyn: addr flush is the problem :)
[21:39] <stgraber> hallyn: simplified version: http://paste.ubuntu.com/1085255/
[21:40] <hallyn> stgraber: .... "don't do that" ? :)
[21:40] <hallyn> man page says "this is a pretty dangerous operation" :)
[21:42] <stgraber> hallyn: well, in netcfg we need to flush all the addresses and routes everytime we call netcfg
[21:42] <stgraber> hallyn: I interpret this warning as "it's dangerous as you'll loose all your addresses and routes" but that's exactly what I want, except for the side effect of loosing the interface in proc :)
[21:50] <hallyn> stgraber: now this only happens with veth (in a container) right?  if you do it on bare metal it doesn't?
[21:51] <stgraber> hallyn: same on bare metal
[21:51] <hallyn> zul: i'd say open a bug.  i can't get a box ready to reproduce today (and technically am not supposed to this week :)
[21:51] <stgraber> hallyn: well, at least with a bridge interface. Let me try with a real interface
[21:51] <zul> hallyn: cool...will do so
[21:51] <hallyn> zul: thanks
[21:52] <stgraber> hallyn: yep, reproduced on bare metal with eth0 on my netbook
[21:53]  * stgraber patches isc-dhcp to workaround the bug for now
[21:54] <hallyn> zul: when bridging to a br1 (which has a lxc container running on it) my libvirt container starts fine, so i'm not sure what is going on when i have it clinging to br0 with eth0
[21:54] <hallyn> zul: and i haven't yet reproduced your segv :)
[21:54] <zul> yeah...fun fun
[21:55] <hallyn> haha.  shutdown my libvirt-lxc instance, 'why didn't it go away?'.  oh yeah...
[21:56] <hallyn> zul: there you go.  it's the filterref
[21:56] <zul> hallyn: thats what i supsected
[21:56] <hallyn> http://paste.ubuntu.com/1085277/
[21:57] <hallyn> bleh
[21:57] <hallyn> well installing libvirt-bin-dbg should make it simple enough to debug.  next week.
[21:58] <hallyn> stgraber: still looking through git log
[21:58] <stgraber> hallyn: so far I know it affects 3.2 to 3.5 at least. I don't have anything running an earlier kernel to test against
[22:00] <hallyn> stgraber: oh, but 'ip addr flush eth0' doesn't do that in precise userspace.  so it's not a kernel change
[22:01] <stgraber> hallyn: it does in precise too, I was just lucky enough that isc-dhcp was using "ip -4 addr flush" back then
[22:01] <stgraber> hallyn: now in 4.2 they changed it to "ip addr flush" instead which flushes both ipv4 and ipv6 and triggers the bug
[22:01] <zastaph> so, RAID 1 is the only RAID ubuntu support for boot/root out of the box?
[22:02] <hallyn> stgraber: no, i'm doing 'ip addr flush eth0' and it doesn't go away
[22:03] <hallyn> but in q container it does
[22:03] <stgraber> hallyn: definitely does here :) tried on bare metal 12.04 and in a clean 12.04 container
[22:04] <hallyn> odd
[22:04] <hallyn> in fact even if i chroot into my q container i still can't get /proc/sys/net/ipv6/conf/eth0 to go away
[22:04] <hallyn> eth0 being my real but unused nic
[22:04] <hallyn> maybe it has to be a non-init procfs?
[22:05] <hallyn> no, doesn't help
[22:06] <hallyn> all right, i need to run.  hopefully i can finish up the api stuff later tonight - but i'm not optimistic for tonight :(
[22:06] <stgraber> hallyn: do you have an ipv4 on it?
[22:06] <stgraber> hallyn: it probably needs something to flush to trigger it
[22:07] <hallyn> stgraber: yeah, just did 'ifconfig eth0 10.0.9.9 up'.
[22:07] <hallyn> (it wasnt' actually live, mind you)
[22:07] <stgraber> hallyn: strace: http://paste.ubuntu.com/1085297/ (not that it's really useful as the netlink messages aren't exactly easy to decipher)
[22:08] <hallyn> grumble :)
[22:08] <cpg> hi, i have a 12.04 system that after upgrading everything, it seems to not be willing to upgrade the kernel for some reason:
[22:08] <cpg> The following packages have been kept back:
[22:08] <cpg>   linux-generic-pae linux-headers-generic-pae linux-image-generic-pae
[22:09] <cpg> seeking advice as to why this could be
[22:11] <hallyn> stgraber: i sacrificed my wlan0, but still couldn't get it to disappear
[22:11] <hallyn> stgraber: ttyl
[22:13] <stgraber> hallyn: really weird, I had no problem reproducing it on 3 machines here, two on 12.04 (both containers and host) and one on 12.10...
[22:14] <stgraber> anyway, ttyl
[22:15] <resno> zastaph: raid has been made easy by unraid
[22:16] <resno> RoyK: have you ever looked at unraid?
[22:16] <zastaph> he left
[22:16] <resno> meh
[22:16] <zastaph> I think I'm gonna buy this for my root/boot: http://www.raidsonic.de/en/products/soho-raid.php?we_objectID=6863
[22:16] <zastaph> and then I can use the 4 internal disks on the HP Microserver for RAID-10
[22:18] <zastaph> was kinda hard to find, as it's not a hdd enclosure nor a NAS
[22:18] <zastaph> the category is sohoraid :)
[23:49] <thys> hi
[23:50] <ping__> n thys hy to
[23:53] <thys> Im confused, files say that I have wzdftp but when I type in pstree it says it vsdftp. which is it have then?
[23:59] <ping__> service vsftpd restart