[00:46] <smoser> adam_g, https://launchpad.net/uec-testing-scripts
[01:24] <adam_g> SpamapS: ping
[01:31] <SpamapS> adam_g: pong!
[01:41] <adam_g> SpamapS: working thru some nova database issues again.. maybe you can shed some light on this to a total lamer (me). if i run the same set of db migrations against a MyISAM backed mysql server and a InnoDB backed server, i end up with two totally different schemas wrt foreign key constraints. myisam ends up the wild west, and innodb creates a schema that is tightly locked down.
[01:41] <adam_g> see: http://people.canonical.com/~agandelman/schema/myisam/relationships.html
[01:41] <adam_g> vs: http://people.canonical.com/~agandelman/schema/innodb/relationships.html
[01:44] <SpamapS> adam_g: MyISAM has no such thing as FK constraints..
[01:45] <SpamapS> adam_g: so its my guess that the code that does the migrating is just being weird.
[01:49] <adam_g> SpamapS: hmm, okay this sounds more obvious that i thought :)
[01:52] <SpamapS> adam_g: right
[01:53] <SpamapS> adam_g: perhaps the migratino code is assuming that because the table is already there, that its relationships will just be there?
[01:55] <adam_g> SpamapS: well, the big problem is that the migrations are being developed against sqlite, or at best, mysql+myisam (on oneiric).  there are assumptions made, yeah..
[01:56] <adam_g> SpamapS: curious, did the switch to innodb by default come from debian, or us?
[02:00] <SpamapS> adam_g: it came from MySQL 5.1 -> 5.5, which was done in Debian.. by me. ;)
[02:01] <SpamapS> adam_g: and its truly the *ONLY* sane path to ever take. MyISAM is *crack*
[02:02] <adam_g> SpamapS: i dont doubt that, just wonder how many others will have similar problems
[02:03] <SpamapS> adam_g: perhaps the right thing to do is to open a bug against openstack that it should refuse to create tables in anything except InnoDB.
[02:03] <adam_g> SpamapS: heh
[02:03] <SpamapS> adam_g: and in migrating.. it should detect myisam tables and recreate *all* relationahips
[02:03] <SpamapS> relationships too
[02:03] <adam_g> actually, at some point in its migration it converts all tables to innodb
[02:04] <adam_g> but im not sure how sqlalchemy handles that, though
[02:05] <SpamapS> adam_g: right, so when it is doing that, it needs to say "oh crap, I'm converting from a table engine that has no relationships definable.. I need to re-create them"
[02:06] <adam_g> yup
[04:03] <axisys> how do I check if a dir is mounted .. would be nice if we can count on a exit code or exit codes
[04:04] <axisys> exit 32 seems to be when fails or already mounted .. from what I understand
[04:04] <axisys> so hard to differentiate what really the mount problem is
[04:06] <axisys> mount /foo; case $? in 0|32) run_bar;; esac did ran `run_bar' even though /foo failed to mount
[06:21] <Stef2> 'morning. any1 on and willing to answer noob questions?
[06:21] <Stef2> :)
[06:22] <Stef2> ~o) included
[06:22] <twb> !anyone
[06:30] <Stef2> it doesn't say in title if i can ask noob q and it is better to ask first. sorry for this. after i installed successfuly ubuntu 11.10 server without GUI i tryed separatly to install xubuntu gui. i don't know/guess it installed properly becouse some files in /lib/something didn't get "registered" properly. now i get an error an ubuntu start "The disk drive for /dev/mapper/cryptswap1 is not...
[06:30] <Stef2> ...ready yet or not present". looked in here https://bugs.launchpad.net/ubuntu/+source/mountall/+bug/527666 but it seems to be an old bug. the start process stops on Checking battery state... [OK]. what should i do to check what's wrong?
[06:32] <twb> Stef2: is this a fresh install?
[06:33] <Stef2> yes
[06:33] <twb> Reinstall from scratch.  Either install xubuntu to begin with, or install server and don't install a GUI.
[06:34] <twb> We recommend learning to manage servers from the CLI; we discourage installing a GUI on servers.
[06:34] <Stef2> installed server. restart. installed xubuntu. got error woth /lib/something. restart. server stops at Checking battery state... [OK]. ubove couple lines i get /dev/.. error
[06:36] <Stef2> i somehow know that, but i guessed it will help me in learning process of CLI
[06:36] <twb> I understand what happened but I can't diagnose it, let alone fix it, over IRC.  There is no point bothering since a clean install has no investment that will be lost (e.g. apps already set up, user data)
[06:42] <Stef2> i understand and agree with u, but (no offense intended at any kind) i am the problem solver somehow. i try to understand what i did/miss so i can try nto to repeat the mistake. i understand and agree the GUI problem. i hoped some1 will point me to an place so i can look at the problem. hope i'm not missunderstood. thnx for the answer :)
[06:44] <twb> The problem is probably just that mountall is crap, and when you tried to install xubuntu you missed some helper package that makes it work properly.
[06:44] <twb> cryptsetup is normally used for block-level encryption (LUKS), but you are probably either using no encryption or per-file (ecryptfs) encryption, so I don't see how you're running into that.
[06:45] <twb> upstart in general and mounting issues with upstart are a huge pain to debug
[06:45] <Stef2> ai first install , installer asked me if i want to encrypt user files (my user files). and i choose yes
[06:45] <Stef2> *at first install
[06:46] <SpamapS> twb: should be easier with newer releases
[06:46] <SpamapS> twb: but really, mountall isn't that bad.
[06:47] <twb> It drove me batshit insane trying to boot off NFS on lucid
[06:47] <SpamapS> twb: thats a use case that still doesn't really have a stakeholder in Ubuntu
[06:48] <twb> IOW nobody is paying canonical to not fuck it up?
[06:48] <SpamapS> We do lots of stuff that nobody pays us for
[06:48] <twb> Because of course it Just Works under sysv init
[06:49] <twb> And it was pretty painless under hardy for that matter.
[06:49] <SpamapS> But yeah, its been pushed to the bottom of the pile a few times.
[06:50] <SpamapS> twb: this would be something to bring up at UDS.. you just need a few interested users who are willing to be annoying about the bugs that need fixing and it will get done.
[06:51] <twb> I do not deal well with meatspace
[06:52] <SpamapS> Sad to say.. but I'm quite susceptible to the 'squeaky wheel' method of bug prioritization.. and so are many others in ubuntu-core-dev
[06:52] <SpamapS> twb: just attend virtually :)
[06:52] <twb> Wasn't UDS a in last November or so anyway?
[06:52] <SpamapS> for precise yeah
[06:52] <twb> Somebody invited me to florida and I was "noooo, I do not go near .us"
[06:53] <SpamapS> twb: next one is in May also in .us .. in Oakland, CA
[06:53] <SpamapS> then back to EU/US rotation
[06:56] <SpamapS> twb: anyway, UDS is just a thought. Really.. if you want those bugs fixed you just need to ping me/slangasek/jodh .. we're quite responsive to the vocal minority when it comes to boot bugs
[06:56] <twb> Righto
[06:57] <twb> Part of the problem is that *my* machines run Debian, so while I deal with Debian problems as they arise, I only encounter ubuntu-specific issues after they hit LTS
[06:57] <twb> (LTS being what my customers ask for, after I tell them they can't have RHEL.)
[06:57] <SpamapS> twb: we fix the LTS from time to time. :-P
[06:58] <twb> Yes, well, I don't turn backports on because I don't trust not to introduce new and exciting issues :-P
[06:58] <SpamapS> https://bugs.launchpad.net/ubuntu/lucid ... only 116 open High importance bugs
[06:59] <twb> RC status doesn't matter after the release amirite ? :P
[07:00] <SpamapS> no such thing as "release critical" in Ubuntu. Critical means "the first things we deal with"
[07:01] <SpamapS> unless our CDs eat your data or just won't boot.. we release on time (dapper, for example, did not ship on time, because it basically didn't work and needed 2 more months)
[07:01] <twb> Hehe "release when its ready" vs "release when it's not ready"
[07:01] <SpamapS> We release when *we* are ready... which is generally known to be about every 6 months
[07:03] <SpamapS> twb: for the LTS we even cut new CDs after we fix the critical bugs that we couldn't get to by "release day"
[07:03] <SpamapS> so it kind of works a little more like a Debian release
[07:05] <tarvid> I have a neglected server on karmic, can't update. Should I attempt an upgrade to lucid?
[07:05] <SpamapS> tarvid: yes
[07:05] <SpamapS> tarvid: backup first
[07:06] <twb> SpamapS: eh, I do my installs from netboot d-i anyway
[07:07] <twb> !eol
[07:07] <twb> tarvid: second link there
[07:23] <Stef2> started reinstall server. got at the point where installer ask me if i want to encrypt home directory. can i presume that my previous error started from the encryption part and now i should choose option to not encrypt my Home directory?
[07:26] <twb> Recommend choose no, at least for now
[07:26] <twb> If everything seems to be working you could try another reinstall and pick yes
[07:27] <Stef2> thnx for the answer. i'll do that. first time i choose yes and it worked. got errors after i installed xubuntu
[07:30] <SpamapS> Stef2: why aren't you installing *xubuntu* if you want xubuntu?
[07:30] <twb> SpamapS: he wants a server with a GUI on it :-(
[07:31] <twb> 17:33 <twb> Reinstall from scratch.  Either install xubuntu to begin with, or install server and don't install a GUI.
[07:31] <SpamapS> server with gui.. hah
[07:31] <SpamapS> saving yourself 1 day of frustration for years of slow, manual labor.
[07:32] <Stef2> i didn't installed GUI from the first time. i got the error when i installed GUI separatly. but now i wont even try to install GUI anymore.
[07:32]  * SpamapS feels the other way.. he wonders why his desktop has a GUI
[07:37] <Syria> Hi, does this mena that guest additions are now installed? http://paste.ubuntu.com/841373/
[07:37] <twb> SpamapS: mine doesn't
[07:38] <twb> My local LUG asked me to give a talk on how to not cry like a little girl when using fbcon for everything, but I didn't want to go outside
[07:38] <twb> Syria: dkms is just infrastructure
[07:39] <twb> Syria: normally you would "apt-get install foo-dkms" to get a DKMS-based third-party kernel module called "foo"
[07:39] <Stef2> would be better if i choose to install software like dns server, lamp server and so on separately or i should install them from start and configure after?
[07:39] <Syria> twb:  I want to install guest additions so I can run CD Rom from the Windows virtual machine.
[07:39] <twb> Syria: I don't know what that means
[07:40] <twb> Stef2: I never pick anything like that until I know the server can actually boot
[07:40] <Stef2> thnx for tip
[07:40] <twb> Stef2: I don't like that UI anyway but you can rerun it later, it's called "tasksel"
[07:40] <twb> Stef2: I prefer to just use apt-get/aptitude
[07:46] <Syria> twb: I have a virtual windows machine and the host is ubuntu server 10.04.3 ... I want to read the CD from the virtual machine.. What shoud I do please?
[07:47] <twb> No idea, I don't support Windows.
[07:48] <Syria> Thank you.
[08:07] <SpamapS> wow erlang takes a long time to build. :-P
[08:07] <SpamapS> I always forget that :-P
[08:21] <Syria> Somebody help me please... I have a windows virtual machine and it is hosted bu ubuntu server.. how can I read cd's thorugh the virtual machine?
[08:28] <Jeeves_> Syria: That depends on your management tool
[08:32] <Syria> Jeeves_:  Virtual Box not the ose edition.
[08:36] <Jeeves_> Oh, no clue then :)
[08:40] <Syria> Jeeves_:  Thanks.
[08:44] <_ruben> most virtualization tools make it fairly trivial to provide cds to its vms, then again, i never used vbox either :)
[08:46] <Stef2> !details
[08:47] <Stef2> !gq
[08:47] <Stef2> !poll
[10:00] <lynxman> morning o/
[10:05] <jamespage> morning lynxman
[10:06] <lynxman> jamespage: good morning sir!
[10:23] <linocisco> hi every one
[10:23] <linocisco> my ISP internet needs us to set proxy in browser to use internet. I would like to bypass this settings using ubuntu server.
[10:35] <linocisco> how to accomplish?
[11:07] <ludo89> Hi
[11:08] <ludo89> i've got a question about tcpdump output on one of the interfaces of my firewall.
[11:08] <ludo89> i've got this :
[11:08] <ludo89>  11:59:46.827529 00:1c:2e:84:7e:00 > 09:00:09:00:00:67 Unknown DSAP 0xf8 Unnumbered, ui, Flags [Command], length 97
[11:09] <ludo89> any idea of what "Unknown DSAP 0xf8 Unnumbered" means ?
[11:14] <lynxman> ludo89: http://serverfault.com/questions/202777/what-do-unknown-ssap-and-unknown-dsap-mean-in-tcpdump
[11:32] <ludo89> lynxman: Thanks a lot for your link !
[11:56] <CasmoNL> Hi, even after removing sendmail (with apt-get, running Ubuntu 10.04), it's still running, is there any way to completely disable it? Googling didn't help me that much..
[12:08] <koolhead11> jamespage: here
[12:09] <jamespage> koolhead11, yep
[12:09] <koolhead11> https://bugs.launchpad.net/ubuntu/+source/cobbler/+bug/841726
[12:09] <koolhead11> status says fixed released
[12:09] <koolhead11> i have wasted almost whole day on thios
[12:09] <koolhead11> and then found about this bug
[12:10] <koolhead11> cobbler profile loads without any problem
[12:10] <koolhead11> its cobbler system which got screwed up
[12:11] <koolhead11> after i did cobbler sync
[12:12] <koolhead11> jamespage: gimme few minute i will paste my cobbler config and dhcp template file
[12:14] <koolhead11> poor virtulbox
[12:16] <jamespage> koolhead11, ack
[12:18] <koolhead11> so manage_dhcp: 0 inside /etc/cobbler/settings
[12:18] <koolhead11> now when i added a system and later did cobbler sync
[12:18] <koolhead11> i see something interesting
[12:19] <koolhead11> inside my dhcpd.conf file :P
[12:21] <koolhead11> jamespage: http://paste.ubuntu.com/841616/
[12:22] <jamespage> koolhead11, can I see a log of the sync please
[12:22] <jamespage> also I don't think cobbler will actively stop managing you dhcp file
[12:22] <koolhead11> jamespage: sure
[12:23] <jamespage> i.e. it will leave stuff around that it created
[12:23] <jamespage> koolhead11, are you running on precise or oneiric?
[12:24] <koolhead11> oneiric
[12:25] <koolhead11> jamespage: http://paste.ubuntu.com/841623/  cobbler sync
[12:26] <jamespage> koolhead11, and what does you dhcpd.conf file have in it now?  I specialize mine quite alot for my local network topology/setup
[12:27] <jamespage> koolhead11, I seem to have done:
[12:27] <jamespage>         #filename "$iface.filename";
[12:27] <jamespage>         filename "pxelinux.0";
[12:27] <jamespage>         ## Cobbler defaults to $next_s
[12:27] <koolhead11> jamespage: nopes its plain DHCP file with my dhcp network info and this "system" infos
[12:27] <koolhead11> jamespage: yeah these options
[12:28] <jamespage> koolhead11, so what exactly is not working for you?
[12:29] <koolhead11> jamespage: gpxe/menu.gpxe
[12:30] <jamespage> koolhead11, thats not going to work - thats the bug
[12:30] <jamespage> you need to change:  filename "pxelinux.0";
[12:31] <koolhead11> jamespage: yeah i did that, then did cobbler sync
[12:31] <koolhead11> still no luck i will try again
[12:31] <jamespage> koolhead11, so manage_dhcp: 0 inside /etc/cobbler/settings?
[12:31] <jamespage> won't have any effect :-)
[12:32] <koolhead11> gosh. lemme make it 1
[12:32] <koolhead11> but in that case wont i have to add the same configs inside that file as well
[12:35] <koolhead11> i meant dhcp.template
[12:37] <jamespage> koolhead11, so if you tell cobbler to manage your dhcp configuration it will overwrite the dhcpd.conf file
[12:37] <jamespage> so ONLY makes changes in dhcp.template
[12:38] <koolhead11> jamespage: so i should remove all the configuration i added on dhcpd.conf ?
[12:38] <koolhead11> and paste the same inside the dhcp.template file
[12:39] <jamespage> koolhead11, you need to ensure that the cobbler dhcp.template contains all of the right configuration for your setup
[12:39] <jamespage> + the fix for filename "pxelinux.0";
[12:42] <koolhead11> jamespage: could there be some cache or something? am still getting same error
[12:43] <koolhead11> shall i add a new system by removing this
[12:43] <koolhead11> ?
[12:43] <koolhead11> and test if it works
[13:00] <melvincv> what does the -C option in tar do?
[13:10] <TeTeT> melvincv: man tar,      -C, --directory DIR
[13:10] <TeTeT>            change to directory DIR
[13:32] <koolhead11> hola TeTeT :)
[14:10] <TeTeT> hi koolhead11, how are you?
[14:11] <koolhead11> am doing good how about you?
[15:21] <Ursinha> I thought I would chair the meeting today, but that's not what the wiki page says...?
[15:22] <Aerodonkey> What is this meeting about?  Sorry that I'm new here.
[15:24] <Ursinha> Aerodonkey, it's the server team weekly meeting, where we discuss server related stuff :) and welcome :)
[15:24] <koolhead11> Ursinha: how much time we have before it starts :)
[15:25] <Ursinha> koolhead11, in 35 minutes, if calendar isn't lying to me
[15:25]  * koolhead11 will grab his dinner in meantime
[15:27] <Aerodonkey> I have problems when running Ubuntu instance on Amazon EC2.  May I ask questions here?
[15:37] <Tixos> someone please please help me with page load times
[15:37] <Tixos> getting some results of 20,000ms/request
[15:37] <Tixos> cant figure it out
[15:38] <Aerodonkey> what page? memory page, or web page?
[15:39] <Tixos> its a web page, its defaintely apache causing it, the server response is very fast
[15:39] <Tixos> i used 'httperf'
[15:39] <Tixos> nothing to do with the web page btw, its default apache
[15:39] <patdk-wk> what is the web page?
[15:39] <patdk-wk> a static html file?
[15:39] <Tixos> it was fine on centos, i feel its ubuntus default apache configs
[15:39] <Tixos> yes it is static html
[15:39] <Tixos> port 443 is fine
[15:40] <patdk-wk> is it on the internet? can I test?
[15:40] <Tixos> been testing for 2 days now lol
[15:40] <Tixos> sec
[15:40] <Tixos> ill PM
[15:40] <Tixos> i also tried 'ab' apache benchmarking, horrific results lol
[15:41] <Tixos> i would blame DNS as i have just changed servers, but using IP is just the same
[15:42] <patdk-wk> na, it's physically on that system somewhere
[15:43] <Tixos> try   ab -c 10 -n 100 IP   patdk-wk
[15:43] <Tixos> server like 110ms, processing 10,000ms lol
[15:43] <patdk-wk> why should I?
[15:43] <patdk-wk> that doesn't prove or help anything
[15:43] <Tixos> ok?
[15:43] <Tixos> to me it proves its taking too long to respond?
[15:44] <Tixos> no point in using a browser right? with caching etc, these two apps tell me better
[15:44] <patdk-wk> what apache modules are enabled?
[15:44] <Tixos> sec
[15:44] <Tixos> 'default'
[15:44] <Tixos> but let me get list
[15:44] <patdk-wk> I dunno what default is
[15:45] <Tixos> http://pastebin.com/TencnaS0
[15:45] <Tixos> i disabled 'mod_reqtimeout, because it was causing lots of 408s because of this issue
[15:46] <patdk-wk> what kind of server is this?
[15:46] <Tixos> basically, its a brand new server, ubuntu 10.04LTS 'default' install
[15:46] <Tixos> dedicated
[15:46] <Tixos> its getting alot of traffic
[15:46] <patdk-wk> did you try vmstat?
[15:46] <patdk-wk> or iostat?
[15:46] <Tixos> so i guess ubuntus default apache.conf isnt good enough
[15:47] <Tixos> no, never heard of them
[15:47] <patdk-wk> I dunno about good enough, I haven't had any issues with it
[15:47] <patdk-wk> well, how do you ever locate an issue?
[15:47] <Tixos> ive never had a problem in 2 years on centos
[15:47] <patdk-wk> for all we know, your harddrive is dead
[15:48] <Tixos> no man, it was fine before it received traffic
[15:48] <Tixos> its happened on 2 ubuntu servers
[15:48] <patdk-wk> harddrive overloaded then :)
[15:48] <Tixos> i moved from one provider because i thought it was 'them'
[15:48] <Tixos> turns out its ubuntu
[15:48] <patdk-wk> do "vmstat 1"
[15:48] <Tixos> or a config rather
[15:48] <Tixos> ok sec
[15:48] <patdk-wk> let it run for like 20seconds, and past it
[15:48] <patdk-wk> pastebin it
[15:49] <Tixos> http://pastebin.com/1MMjDYjy
[15:49] <patdk-wk> hmm, now I can't connect
[15:49] <patdk-wk> how about a, netstat -atn
[15:50] <patdk-wk> that is going dump a lot of lines though
[15:50] <Tixos>  should i | less it ?
[15:50] <patdk-wk> na
[15:50] <Tixos> or file output
[15:50] <patdk-wk> netstat -atn | pastebinit
[15:50] <Tixos> netstat -atn > file.txt
[15:50] <Tixos> ?
[15:50] <patdk-wk> pastebinit is lazy way :)
[15:51] <Tixos> but thats what you mean right? lol
[15:51] <patdk-wk> ya
[15:51] <patdk-wk> I'm thinking you maxed out your apache connections
[15:51] <patdk-wk> default is 256
[15:51] <patdk-wk> and with a 15second keepalive I think
[15:52] <patdk-wk> normally changing keepalive down to 1 second is good
[15:52] <patdk-wk> it looks like you have enough ram to increase that 256limit also, if needed
[15:52] <Tixos> yea i thought this, because i compared my centos httpd.conf, and the onyl difference really was the MPM settings for prefork
[15:52] <Tixos> and also keepalive is off on that server
[15:53] <Tixos> ill PM you link
[15:53] <patdk-wk> ya, you need generic tuning
[15:53] <Tixos> so keepalive = 1 is better than off?
[15:53] <patdk-wk> keepalive off or 1sec will help it
[15:53] <Tixos> could you guide me?
[15:53] <patdk-wk> I would perfer 1sec
[15:53] <Tixos> ok let me give it a whirl
[15:54] <Tixos> should i up anything in perfork ?
[15:54] <patdk-wk> cause it only wastes it for 1 sec, but then they get all the pictures and stuff quick
[15:54] <patdk-wk> you have craploads of free ram
[15:54] <patdk-wk> so probably safe to double the max
[15:54] <Tixos> yes i know
[15:54] <patdk-wk> maybe more
[15:54] <patdk-wk> but I wouldn't put it higher than needed
[15:54] <Tixos> the server is decent enough, so thats why i was lost, this is all a bit beyond me but i done alot of reading :P
[15:55] <Tixos> i read 256 is the higher?
[15:55] <Tixos> or is that for worker or something ?
[15:55] <patdk-wk> I dunno, what one you have installed?
[15:55] <patdk-wk> apache2-prefork, apache2-worker, ...
[15:55] <Tixos> well, again lol by 'default' i didnt install either
[15:55] <Tixos> but it has both sections in the config
[15:55] <Tixos> same as centos does
[15:56] <patdk-wk> how many apache2 programs do you see?
[15:56] <patdk-wk> worker should have like 2-4
[15:56] <Tixos> programs?
[15:56] <patdk-wk> prefork will have like 200 or so in your server
[15:56] <patdk-wk> ps ax
[15:56] <Tixos> sec
[15:57] <Tixos> /usr/sbin/apache2 -k start    there is about 100 of these?
[15:57] <Tixos> maybe more
[15:57] <patdk-wk> then your using prefork
[15:57] <Tixos> ok
[15:57] <Tixos> these are my centos settings, so i give these a go?
[15:57] <Tixos> http://pastebin.com/i8Rz3ind
[15:58] <Tixos> if you get the httperf time/request down i will love you forever :)
[15:58] <Tixos> so far 4 people couldnt help :)
[15:59] <Tixos> sidenote: what is the correct way to reload configs when changing apache.conf.   reload, restart or stop/start?
[15:59] <patdk-wk> depends
[15:59] <EvilResistance> Tixos, i'm unfamiliar with Apache, but if its anything line nginx's nginx.conf (which controls the instance of nginx and not the sites), i'd assume restart
[16:00] <patdk-wk> I normally just restart
[16:00] <EvilResistance> even when i was dabbling in Apache, i just used 'restart'
[16:00] <patdk-wk> but doing a reload is alittle more friendly
[16:00] <Tixos> yea i normally do
[16:00] <Tixos> im not feeling friendly :)
[16:00] <Tixos> so should i match those prefork settings patdk-wk ?
[16:00] <patdk-wk> hmm?
[16:00] <Tixos> let me pastebin both
[16:01] <Tixos> these are 'defaults'
[16:01] <Tixos> http://pastebin.com/HQRMPF2E
[16:02] <andrew667> test
[16:02] <EvilResistance> i very very rarely feel friendly towards my systems, so i just use 'restart' for misbehaving things and rehashes of the config (so to speak)
[16:02] <patdk-wk> tixos, I would do this
[16:02] <patdk-wk> StartServers 50
[16:03] <patdk-wk> MinSpareServers 20
[16:03] <patdk-wk> MaxSpareServers 50
[16:03] <patdk-wk> ServerLimit 300
[16:03] <patdk-wk> MaxClients 300
[16:03] <patdk-wk> and keepalive at like 1sec
[16:03] <patdk-wk> maybe increase the 300 to something larger if needed
[16:04] <Tixos> patdk-wk:  your keepalive = 1 fixed it..
[16:04] <Tixos> what are the pitfalls of this?
[16:04] <Tixos> its fine... totally fine
[16:05] <Tixos> Request rate: 4.3 req/s (234.3 ms/req)
[16:05] <patdk-wk> it's not fine
[16:05] <patdk-wk> it's *better*
[16:05] <Tixos> same as google for me :P
[16:05] <Tixos> yes i will fix prefork up like you say also
[16:05] <patdk-wk> keepalive is how long a client can *tie up resources* on your server, while waiting for the next page request
[16:06] <patdk-wk> before you had it off, so every thing needed a new connection, slow
[16:06] <Tixos> no it was on, but set @ 15
[16:06] <patdk-wk> making it 1sec, basically means, they can download the page, and images, and js,css,.. all in one connection
[16:06] <Tixos> it was off on my centos server yes, but it was fast here
[16:06] <patdk-wk> then a new page will take a new connection
[16:06] <patdk-wk> so having it on, makes a page faster, but makes OTHER people wait
[16:07] <patdk-wk> since you maxed out how many connections you could have at once (150), all waiting around for 15sec, no one could connect
[16:08] <Tixos> yea, great man
[16:08] <Tixos> thanks so much :)
[16:08] <patdk-wk> your tests didn't see that, since you tests didn't use keepalive
[16:08] <Tixos> so if there is noneed for keepalive at all, it can be turned off, but 1-2 seconds if always better in your opinion?
[16:09] <patdk-wk> if you have webpages that are more than one item, (includes javascript, css, images, ...) I would keep it on
[16:09] <patdk-wk> it makes them display faster for the user
[16:09] <antihero> Hey, I can't install php5-memcached
[16:09] <patdk-wk> plus, the more distance the user is from you, the longer it would keep your server tied up, without keepalive
[16:09] <antihero> any ideas? (lucid)
[16:10] <antihero> it's here http://packages.ubuntu.com/lucid/php5-memcached
[16:10] <antihero> but apt-get says it has no installation canddiate
[16:12] <Tixos> patdk-wk: would you use mod_reqtimeout, i disabled it, but it was installed by default, now this is fixed should i re-enable?
[16:12] <Tixos> it was giving tons of 408 errors
[16:12] <patdk-wk> I can't remember what it does
[16:16] <brendan0powers> jamespage: Hi, I have some questions about the debian/watch file
[16:16] <brendan0powers> jamespage: Aside from the man pages, I've fixed all of your comments
[16:17] <jamespage> brendan0powers, hi!
[16:17] <brendan0powers> jamespage: So, the watch file should point to a directory containing source releases?
[16:18] <jamespage> brendan0powers, typically a http location for a tar.gz of the source release
[16:18] <Tixos> ok, thanks patdk-wk  much love :D
[16:18] <brendan0powers> I tried putting the exact file name I wanted, and got an error from uscan beacuse there was no pattern
[16:19] <brendan0powers> jamespage: and bitbucket does not support listing tags, so i can't use a wildcard for the URL
[16:20] <jamespage> brendan0powers, OK so that might not be practical then
[16:20] <brendan0powers> I could host the source somewhere else, I guess
[16:21] <brendan0powers> I mean, the .tar releases
[16:22] <jamespage> brendan0powers, if it looking hard ATM then I suggest you document in a README.source file how to create the orig.tar.gz file manually.
[16:23] <jamespage> that should be acceptable
[16:23]  * jamespage is not an archive admin
[16:29] <jamespage> brendan0powers, that could involve documenting how to clone from bitbucket and create the orig.tar.gz
[16:30] <jamespage> which could be put in debian/rules
[16:45] <mdeslaur> *sigh* looks like I have to push mysql 5.5.20 and 5.1.61 as security updates to our stable releases, as Oracle won't disclose what the security updates are :(
[16:46] <andol> mdeslaur: Making MariaDB seem even more tempting?
[16:46] <mdeslaur> andol: definitely
[17:01] <SpamapS> mdeslaur: have they confirmed that 5.5.20 and 5.1.61 *actually fix the problem* ?
[17:01] <mdeslaur> SpamapS: yes, those are the versions listed in the advisory
[17:02] <SpamapS> mdeslaur: Ok, I have 5.5.20 ready to go in Debian, so I can push that into precise just so precise isn't vulnerable if you like.
[17:02] <mdeslaur> SpamapS: yes, please, that would be awesome
[17:02] <SpamapS> mdeslaur: Ok I'll upload 5.5.20 later today... cool?
[17:02] <mdeslaur> SpamapS: yes, thanks!
[17:06] <koolhead17> hi all
[17:23] <hallyn> jjohansen: I'm going to go and review some userns patches (some = >100 of them :), please ping me when I should be testing a apparmor+mounts patchset
[17:24] <jamespage> iamfuzz, hey - sorry but jenkins does not use gwt :-(
[17:25] <iamfuzz> jamespage, I didn't think so, but Andy told me that so I got excited :-)
[17:25] <iamfuzz> no worries, was just a stab
[17:25] <jjohansen> hallyn: hrmm, well I am packaging up a first run of the ppa now, but there will be some revisions throughout the day.  I have full option matching that I am going to be testing and will roll out some time
[17:26] <jjohansen> hallyn: so if you want to play it will be available in an hour or so but changes, but later might be better
[17:26] <hallyn> jjohansen: ok, no problem - I"ll wait a bit, thanks.  just ping me when i should look :)
[17:26] <jjohansen> hallyn: okay, btw full option matching means none of the limitations/concerns from the other day
[17:27] <hallyn> \o/
[17:42] <smoser> adam_g and Daviey , roaksoax looked at the big/small buckets thing.
[17:42] <smoser> he would know more. but he asserted upstream bug on that.
[17:43] <smoser> and i think that theh goal is to move to squid-deb-proxy, which wouldn't have these buckets (although maybe we could get them into s-d-p, as they are kind of nice, to have kind of one limit for ISOs and one limit for archive things).
[17:49] <ivoks> adam_g: i'm not sure how to request a merge cause LP doesn't allow me; but if you are interested, take a look at lp:~ivoks/charms/precise/nova-cloud-controler/trunk; adds public_interface to FlatDHCPManager
[18:00] <koolhead17> jamespage: you were correct :)
[18:01] <koolhead17> am still at same issue!! :)
[18:17] <smoser> utlemming, bug 932088
[18:22] <SpamapS> lynxman: hey, I'm working on updating rabbitmq-server to 2.7.1 ..
[18:22] <zul> SpamapS: pretty sure...its dependent on the rabbitmq-server
[18:22] <SpamapS> lynxman: might that affect your stomp packages?
[18:24] <zul> SpamapS: he might not be around either
[18:26] <SpamapS> zul: true, too late.. well I'll take a look at them
[18:26] <zul> SpamapS: k just wanted to give you a heads up
[18:47] <med_> I think lynxman is in flight.
[18:53] <antnash_> Hi guys. Anyone able to give me any hints as to why my server becomes hideously unresponsive after being on for about 7 hours?
[18:55] <antnash_> I have to power it off for 30s or so for it to actually come back to life, otherwise it hangs at bios while discovering disks
[19:06] <SpamapS> antnash_: anything in dmesg? perhaps something is throttling due to temperature?
[19:08] <antnash_> Hey SpamapS. How can I check it after a reboot?
[19:09] <SpamapS> antnash_: great point. :)
[19:09] <SpamapS> antnash_: perhaps hook up another box as a serial console
[19:09] <SpamapS> antnash_: and make sure you boot with 'noquiet' so you get all messages on console
[19:10] <antnash_> Hmm. Would if I had another box I could do it with, and knew how!
[19:10] <antnash_> Of course the latter there is much less of a problem than the former
[19:12] <antnash_> SpamapS, no way to check in system logs?
[19:15] <SpamapS> antnash_: /var/log/syslog would potentially have stuff too
[19:18] <antnash_> SpamapS, you were right. Throttling due to temp
[19:18] <antnash_> Time to get a better cpu fan I think!
[19:24] <zul> adam_g: im in the middle of the nova ubuntu/debian merge would you have complaints if i moved --connection=libvirt into the nova-{kvm,lxc,uml,qemu}.conf ?
[19:25] <zul> im tryin to figure out how to handle the nova-xen/nova-xcp/nova-libvirt stuff properly
[19:26] <adam_g> zul: thats where it should go, but there was a database migration that made it required that its defined on the node thats running 'nova-manage db sync'
[19:26] <adam_g> zul: i think that might have changed, though, gimme a minute ill check
[19:27] <zul> adam_g: k
[19:30] <adam_g> zul: heres the bug https://bugs.launchpad.net/nova/+bug/921294
[19:30] <zul> adam_g: k ill try it here first
[19:32] <adam_g> zul: i think
[19:32] <adam_g> zul: er, i think moving that is okay for new installs but not going to work for upgrades
[19:32] <zul> great now i have blood dribbling from my ears ;)
[19:33] <adam_g> read the bug
[19:40] <adam_g> zul: does debian keep that flag in the nova-compute-*.conf?
[19:40] <zul> no nova.conf
[19:41] <zul> but i think debian might be a bit broken right now
[19:42] <adam_g> zul: because they also support xcp?
[19:42] <zul> adam_g: yeah
[19:43] <adam_g> zul: that is why i raised the bug to begin with, i cant see how it'd be possible to support either without nova-common-{xcp, libvirt, etc} and corresponding compute packages.
[19:44] <adam_g> zul: if the flag is only set on the compute packages, and compute is running on seperate nodes, a database migration from any other node will end up migrated for xen, even if compute nodes are running libvirt.
[19:45] <zul> yeah im going to put off that problem for now and continue merging stuff here
[19:45] <adam_g> zul: if its set globally in nova.conf, it can be run from any node, but to support other hypervisors, we'll need an entire new subset of nova packages (not just compute)
[19:45] <zul> adam_g: agreed
[19:51] <gary_poster> hallyn, I have a container that I can start with lxc-start but when I try to use lxc-execute it gives me an error: "No such file or directory - failed to exec /usr/lib/lxc/lxc-init" "invalid sequence number 1. expected 2".  In the dim past (http://comments.gmane.org/gmane.linux.kernel.containers.lxc.general/1734) you recommended copying over /usr/lib/lxc/lxc-init to the guest (so /var/lib/lxc/lpdev/rootfs/usr/lib/lx
[19:51] <gary_poster> c/ for me).  That dir in the guest does not exist for me.  Do I need to start the container and install some packages there first, or do you have another suggestion?
[19:57] <Degot> Hi, Is there a way to automatically copy files to just attached USB Stick/HDD?
[19:59] <hallyn> gary_poster: stgraber's workaround is the best - install lxc in the guest
[20:00] <gary_poster> hallyn, ok cool thanks
[20:01] <stgraber> gary_poster: install it with --no-install-recommends though to avoid pulling dnsmasq, bridge-utils, ...
[20:02] <gary_poster> stgraber, ok makes sense thanks
[21:35] <chaos_zero> heya, i am looking to do something that might be impossible. anyone want to help me? =P
[21:36] <chaos_zero> well i want to run a windows app in the "background" with no X.
[21:36] <chaos_zero> has this ever been done?
[21:48] <adam_g> zul: are you planning on just overwriting the orig.tar.gz for keystone with the ksl branch?
[21:48] <zul> yes
[21:50] <zul> im talking about it with anotherjesse on #openstack-dev so you might want to chime in
[21:53] <adam_g> zul: if we could put that packaging somewhere other than the archive + the CI repo (a seperate openstack-ubuntu-testing PPA), id prefer to have some time ensuring its at least mostly working. otherwise we risk breaking CI for everything else
[21:54] <zul> adam_g: yep
[22:10] <adam_g> Daviey: you around?
[22:11] <Daviey> adam_g: always
[22:14] <Aison> somehow the mysql service is not starting
[22:14] <Aison> when I do service mysql start it hangs forever
[22:18] <adam_g> Daviey: wondering if its okay to transfer these WIs over to roaksoax: add SSL support to cobbler-enlist, add auto-generation of hostname when not provided by dhcp to cobbler-enlist, Hardware detection and inventory snippet. i believe he's already got the hostname one complete
[22:35] <Daviey> adam_g: As roaksoax is owning that area of things, it probably makes sense
[22:36] <Daviey> adam_g: postpone hw detection
[23:08] <Aison> when I reinstall mysql-server-5.1 package, why is the file /etc/init/mysql.conf not reinstalled also?
[23:08] <Aison> I deleted it
[23:16] <Jeeves_> Aison: Did you purge when you removed it?
[23:18] <Aison> Jeeves_, no, but now I did and it works ;)
[23:20] <Jeeves_> Good :)
[23:25] <Daviey> adam_g: still around?
[23:30] <adam_g> Daviey: yea
[23:31] <Daviey> adam_g: heya, how are things?
[23:33] <adam_g> Daviey: decent. trying to work on bug 883988 but hitting some strange database locking issue that only seems to happen against mysql on precise
[23:34] <adam_g> that seems unrelated to the bug
[23:35] <Daviey> aww, crap.
[23:35] <Daviey> adam_g: are you likey to have capacity to 'make green' CI testing before EOD?
[23:36] <adam_g> Daviey: yes
[23:37] <Daviey> adam_g: aww, that rocks.  I might back a start in a bit, but will not be able to finish in time.
[23:37] <adam_g> Daviey: huh?
[23:47] <Daviey> adam_g: sorry, i'll make a start - but i will not finish
[23:47] <adam_g> Daviey: i see whats wrong
[23:48] <adam_g> Daviey: deployments are timing out because, during installation, they cannot reach their mirror for whatever reason
[23:48] <Daviey> free disk space? ;)
[23:48] <adam_g> seems that lab is cut off from gb.archive.ubuntu.com, whic is what they're trying to reach
[23:48] <adam_g> not sure exactly why
[23:48] <Daviey> ah, no
[23:49] <stiv2k> hmmm
[23:49] <stiv2k> i created a new user on my server
[23:49] <stiv2k> but when i see programs in ps aux that he started
[23:50] <stiv2k> it just says user 1003
[23:50] <stiv2k> not his name
[23:50] <stiv2k> why?
[23:50] <Daviey> stiv2k: Does, cat /etc/passwd , work?
[23:51] <stiv2k> Daviey: yesh
[23:51] <stiv2k> yeah*
[23:51] <stiv2k> knightsrepair:x:1003:1003:,,,:/home/knightsrepair:/bin/bash
[23:51] <stiv2k> ^
[23:51] <Daviey> adam_g: do you see that archive.ubuntu.com is pinned to a mirror?
[23:51] <Daviey> (that works)
[23:51] <adam_g> Daviey: i just did that
[23:51] <adam_g> rebooting one now
[23:52] <Daviey> adam_g: I suspect that the IP address of gb.archive.ubuntu.com changed, and the firewall is based on IP address
[23:53] <adam_g> Daviey: any idea what is setting the hosts to use gb.archive.ubuntu.com?
[23:54] <stiv2k> Daviey: ?
[23:54] <Daviey> stiv2k: check to see you can read /etc/passwd as that user.
[23:55] <Daviey> adam_g: is this on test-01, or the nodes?
[23:57] <adam_g> Daviey: the nodes seem to be attempting to install from gb.archive.ubuntu.com which is unreachable from both test-01 and the rest of the cluster
[23:57] <Daviey> adam_g: so, test nodes use test-01 for dns resolution.  If the ip for archive.ubuntu.com in /etc/hosts works, we'll add an alias to gb.archive.ubuntu.com ; and we should be GOLD>
[23:57] <stiv2k> d	ok
[23:57] <stiv2k> Daviey: ok
[23:57] <Daviey> adam_g: can you trigger a re-test
[23:58] <Daviey> i think it should work now
[23:58] <stiv2k> Daviey: i can read it
[23:58] <adam_g> Daviey: just a sec
[23:58] <bascotie> I setup an Ubuntu Desktop 11.10 Samba/FTP server. I can connect through ftp and share through samba fine but only when putting files ON the server, not pulling them off. I've tried port forwarding, playing with permissions,etc, but I still seem to get a permissions error. I am trying to connect from my Windows 7 machine to the Ubuntu Server.
[23:59] <Daviey> bascotie: you might get more success from a samba dedicated channel.