[00:36] <adam_g> hallyn: around?
[00:37] <twb> Sigh.  So I backported xtables-addons from natty to lucid, so I would have a DKMSized version.  This installed PERFECTLY on the local test machine.
[00:37] <twb> I send my boy out to the airgapped prison, and it's reporting "I need kernel sources", despite not needing them here (only the headers)
[00:38] <twb> With any luck he is now sneakernetting the dkms build output & similar diagnostics to somewhere that can get to me.
[00:40] <Patrickdk> hmm?
[00:40] <Patrickdk> why didn't you send him the compiled package?
[00:44] <SpamapS> Patrickdk: too easy ;)
[00:44] <twb> Uh, because when the kernel gets a minor update, the .kos will need to be recompiled
[00:45] <twb> e.g. 2.6.32-33 to 2.6.32-34
[00:45] <SpamapS> dkms always builds though so I don't think this is a "just send him the binary" issue
[00:45] <twb> I did send him the xtables-addons-dkms .deb, of course
[00:45] <Patrickdk> ya, dkms is nice for auto-upgrades
[00:45] <twb> That was the whole point of using dkms instead of m-a
[00:45] <Patrickdk> but I build my dkms on a build machine, and push them out
[00:45] <twb> Patrickdk: I suppose I would've done that if I had anticipated this not working
[00:46] <Patrickdk> 33 to 34? you living life like 3 months ago?
[00:46] <Patrickdk> we are on -40 :)
[00:46] <twb> Patrickdk: whatever, I made thhose numebrs up
[00:47]  * Patrickdk is pissed at his isp
[00:47] <Patrickdk> their providers provider is broken
[00:47] <Patrickdk> and they *dunno* how to work around the issue
[00:47] <Patrickdk> yay for 1000ms latency
[00:59] <hallyn> adam_g: half
[01:01] <twb> at least your site is on the internet
[01:01] <twb> stupid airgap
[01:02] <adam_g> hallyn: do you know of any circumstances where libvirt would take the liberty of setting a domain to 'paused'?
[01:02] <hallyn> (thinking)
[01:03] <hallyn> i don't.  it'd be neat if it did that based on host load :)
[01:03] <adam_g> seems to be related to load, but trying to determine if its libvirt or nova thats pausing
[01:03] <hallyn> interesting
[01:03] <hallyn> i haven't heard of such a thing
[01:03] <adam_g> ok, ill look closer at nova in that case, thanks
[01:04] <hallyn> i'd look at nova
[01:04] <hallyn> yeah
[01:04] <hallyn> np - ttyl
[01:04] <adam_g> night
[01:57] <Rallias> Since my installation of nginx is no longer tasked with listening on a port below 1024, how do I set it so it never is root when it starts up?
[02:01] <twb> How is it started presently?  sysvinit or upstart?
[02:10] <aarossig> Question regarding my network.. I just did a hardware refresh and upgraded to 11.10 Server. While copying on my gigabit network, the network transfer will stop for periods of time and then start back up. Using sshfs, the cpu will go down to 0 for ~10 or 20s and then the transfer resumes. Any ideas?
[02:10] <twb> aarossig: you... aren't the guy who wants me to configure his UPS, are you?
[02:10] <aarossig> am I?
[02:11] <twb> Dunno, his name was Andrew Rossi
[02:11] <twb> of Designer Networks
[02:11] <aarossig> twb: I don't think so, I am capable of configuring my own UPS (as well as those of my clients)
[02:11] <aarossig> ah, no I don't own a business
[02:11] <twb> OK, no worries :-)
[02:11] <aarossig> :P
[02:11] <aarossig> so do you happen to have any suggestions for my situation?
[02:12] <twb> Well I would definitely encourage you to go 12.04 not 11.10, since 12.04 is LTS
[02:12] <aarossig> this happpens on both sshfs and samba, but with samba it is much more catastrophic and the transfer actually itimes out
[02:12] <aarossig> twb: yeah, I plan to do a dist upgrade to it
[02:12] <twb> Not sure about your issue; could be I/O or something
[02:13] <aarossig> moral of the story.. I can't wait anymore. I've been putting up with old packages and ppas with 10.04 for far too long
[02:13] <twb> dump ps auxf to a file every second or so and diff them around the time an issue occurs, looking for D state procs, or watch "iostat 5"
[02:13] <aarossig> thanks, i'll look into it
[02:13] <twb> if you're copying with e.g. scp, it could be the buffering when it actually hits the disk
[02:14] <aarossig> I believe it is something IO related, I am running 8 hard drives and some are on the PCI/PCI-e busses.
[02:14] <twb> so e.g. you do "scp example.com:foo.img ." and it will block when it runs out of buffer writing to the local disk
[02:14] <aarossig> yeah, that sounds accurate. I will try
[02:14] <twb> aarossig: 8 HDDs in what array, RAID6?
[02:14] <aarossig> it is actually several RAID1 arrays
[02:14] <aarossig> one for root, one for home, and then several for data
[02:15] <twb> Track down which ones have the data in question, watch their iostats
[02:15] <aarossig> okay, thanks for the suggestions
[02:15] <aarossig> i'll give it a shot
[02:15] <twb> Also look at the flush stuff in /proc see if its interval corresponds to your blocking interval
[02:15] <twb> e.g. if your ext4 filesystem is set to flush every five minutes and you see issues every five minute
[02:16] <aarossig> the frequency is much higher, but I see your point
[02:19] <aarossig> twb: yeah, iostat shows fairly constant writing
[02:19] <aarossig> you're probably right about the caching, it might be filling up the caches and then flushes them to the disk, rinses and repeats.
[02:19] <twb> I hear there is also iotop; I haven't tried it
[02:19] <aarossig> i was using iotop
[02:20] <aarossig> iostat is more device lebel
[02:20] <aarossig> level*
[02:20] <twb> The other thing to read up on is "write amplification"
[02:20] <aarossig> iotop is process level
[02:22] <tarvid> boot setup temperature dropped from 76 C to 36 C after installing a new Corsair A50 fan, how would I go about monitoring remotely?
[02:23] <twb> tarvid: how are you measuring it now?
[02:23] <tarvid> the setup screen reports tempereature
[02:23] <tarvid> BIOS
[02:24] <twb> OK
[02:24] <twb> The raw data is in /sys somewhere; there is a package that tries to understand how to interpret it...
[02:24] <twb> Can't remember the name
[02:25] <tarvid> /proc/acpi/... on some boxes, not this one
[02:25] <twb> lm-sensors
[02:26] <tarvid> installed but hard disk is the only it reports
[02:26] <twb> You probably have to rice it up
[02:26] <twb> IME it's a pain and usually not worth the hassle
[02:27] <tarvid> This box was a gift from the local library - free as in free cats
[02:27] <aarossig> because the library doesn't serve beer.
[02:27] <twb> I hope they don't serve cats either
[02:27] <tarvid> alas
[02:27] <tarvid> they have one
[02:28] <aarossig> twb: I ended up here https://bugs.launchpad.net/ubuntu/+source/linux/+bug/427210, rather interesting stuff. Thanks for the general direction pointing.
[02:28] <tarvid> !lm-sensors
[02:28] <twb> aarossig: you have a server, CFQ is probably best for your case IIRC, but feel free to experiment
[02:29] <aarossig> deadline seems to have improved it slightly, the network transfer seems to be more continuous
[02:29] <aarossig> I will be doing some more debugging
[02:29] <twb> aarossig: at least in my case, the problem wasn't the scheduler, it was that I was doing so damn much I/O.  Which was mostly due to RRDs (collectd) plus a lot of write amplification.
[02:29] <twb> I fixed it by telling collectd to STFU and buffer an hour of writes and flush them all at once, instead of trying to do random-access writes into the RRDs every ten seconds
[02:30] <twb> It was interesting to see how badly it raped the system, though -- e.g. dpkg would issue an fsync for every file (or dir, depending on version) and bring the whole system crashing down.
[02:31] <aarossig> lol
[02:31] <aarossig> very good, these are good starting points for me to google form
[02:31] <twb> Incidentally, if you are into BDSM you can look at libeatmydata
[02:32] <twb> Which is an LD_PRELOAD wrapper that discards all attempts by the app to explicitly flush data.
[02:34] <twb> Patrickdk: fyi, the problem at the prison was he just had the wrong header package installed
[02:38] <aarossig> I am quite proud of my little server. I can boot up from cold 8 drives and the voltage only falls from 12.2 to 11.9V. Quite pleased.
[02:46] <tarvid> !dns
[02:58] <tarvid> where do I setup my isp's nameservers
[02:58] <tarvid> !nameservers
[02:59] <tarvid> !nameserver
[02:59] <tarvid> !name server
[03:00] <tarvid> uswed to be resolv.conf but that gets overwritten
[03:01] <twb> tarvid: resolvconf package
[03:01] <twb> IIUC if you remove that symlink you can just make it an ordinary file and it won't fuck with it
[03:01] <erichammond> tarvid: If you're using DHCP, then you might set up the nameservers in the DHCP server.
[03:02] <tarvid> static assignment
[03:02] <qman__> remove network-manager if you have it, then disable resolvconf, then set resolv.conf manually
[03:02] <qman__> or if you want quick and dirty, set resolv.conf then chattr +i
[03:04] <tarvid> thanks, that seems like the hard way, does Ubuntu havge a reason?
[03:04] <qman__> well, this stuff exists to make networking on desktops and laptops easy
[03:04] <qman__> why it gets thrown in on servers too, I have no clue
[03:04] <qman__> caused me plenty of trouble before
[03:06] <tarvid> avahi with two nics gets things wrong half the time
[03:13] <twb> qman__: resolvconf is hard dep in precise (and 11.10?) so would need to disable rather than remove... oh, you said that
[03:19] <tarvid> cfd /etc/root
[03:25] <tarvid> so how am I supposed to set up static name servers in 12.04?
[03:26] <twb> tarvid: you learn to use resolvconf, or you bypass it
[03:26] <twb> When I grumped about the same thing recently, I was told 14:01 <twb> IIUC if you remove that symlink you can just make it an ordinary file and it won't fuck with it
[03:27] <tarvid> I looked at thye man page, it is inscrutable
[03:27] <tarvid>         dns-nameservers 68.105.28.16, 68.105.29.16
[03:27] <tarvid> 	dns-search ls.net
[03:27] <tarvid> 	
[03:28] <tarvid> chattr: Operation not supported while reading flags on resolv.conf
[03:39] <tarvid> by adding nameservers to /etc/network/interfaces I now have nameservers in roslv.conf but dns does not work
[03:52] <twb> tarvid: is /etc/resolv.conf a symlink?
[03:53] <twb> Define "does not work"
[03:54] <twb> $ dig PTR 8.8.8.8.in-addr.arpa @8.8.8.8 +short ==> google-public-dns-a.google.com.
[03:56] <tarvid> can't get out, can ping peers but nothing beyond the gateway
[03:57] <twb> tarvid: so you did "mtr 8.8.8.8" and got nothing?
[03:57] <tarvid> nothing
[03:57] <twb> Then your network is fucked; nothing to do with DNS
[03:58] <tarvid> you are correct
[04:23] <tarvid> giving up for the night, thanks for the help
[04:33] <twb> Hmm, any gotchas in online growing of an XFS filesystem?
[04:39] <twb> Too late, I did it anyway
[04:40] <KM0201> lol
[04:55] <andol> twb: Any fun suprises? :)
[04:57] <twb> Not yet
[04:58] <twb> But $stupid_customer has no barriers, it's an 8-disk array w/hardware raid and then LVM on top, and the ENTIRE PV is allocated
[04:58] <twb> So the "solution" to give them another 6TB of storage, was to put in another 8 disks in a second array, and pvcreate /dev/sdb and vgextend vg1 /dev/sdb and then fill /home over the entire second PV as well
[04:58] <twb> "whee"
[04:58] <twb> Also this is lenny
[04:59] <andol> Ah
[05:11] <koolhead11> hi all
[05:12] <koolhead11> Daviey: :)
[06:09] <SpamapS> twb: I've done a lot of online growing of XFS partitions in exactly the manner you described. Works fine
[06:09] <SpamapS> twb: and when i did it, it was kernel 2.6.18 (old ass redhat)
[06:19] <Maleko> why i dont see if labels in 'ip a'? i assigned few additional ips under labels/aliases
[09:01] <lynxman> morning o/
[09:13] <yolanda2> hi, good morning
[09:13] <yolanda2> i have a question about dbconfig
[09:13] <yolanda2> is there any way to skip database creation? i only want to create postgres user, but no database
[09:14] <bencer> jamespage: SpamapS would you have a look at https://wiki.ubuntu.com/JorgeSalamero/PerPackageDeveloperApplication and maybe support my application?
[10:00] <cas> Hi, I got some problems with formatting my (hardware)RAID disk
[10:00] <cas> I get the error it is in use while it is not mounted
[10:02] <cas> Any idea what I can do to "release" this disk for mkfs?
[10:13] <lynxman> jamespage: Question, I've seen a package I'd like to create for Ubuntu (and maintain) which would possibly land in Universe (or multiverse), due to this the restrictions of the freeze for precise do not exactly apply, right?
[10:17] <jamespage> lynxman, they still apply - i.e. you will need a FFe IMHO - but if its a new package they I think the release team would ack that
[10:17] <jamespage> bencer, on my list
[10:17] <bencer> jamespage: ok thanks!
[10:18] <bencer> btw, i've already sent some blueprints on things we want to work now
[10:18] <lynxman> jamespage: its a mysql load balancer (ala pgbouncer), I reckon pretty useful for ops people
[10:19] <bencer> https://blueprints.launchpad.net/~zentyal
[10:21] <iclebyte> has any one ever seen 'rt6_redirect: source isn't a valid nexthop for redirect target' in syslog? I can only seem to find references to the kernel source where the message is and other people asking the same question. I have no way of working out which node is causing this.
[10:29] <yolanda2> hi, just another question about dbconfig: i need to create an user, but this user needs to have rights to create database. Currently the user created by dbconfig cannot do it, how can i do it?
[10:37] <yolanda2> please, any help with dbconfig?
[10:47] <yolanda> please, any one can help with dbconfig?
[10:57] <soren> yolanda: Why would the user need to create the database?
[10:57] <yolanda> hi soren, because openerp works like that...
[10:57] <soren> yolanda: dbconfig is supposed to have created it ahead of time.
[10:58] <yolanda> there is a "manage databases" screen, that allows to create new databases, so the user needs to have createdb permissions
[10:58] <soren> yolanda: And it gets upset if the db is already there?
[10:58] <yolanda> well, we don't use that db that is created by dbconfig, if we could just skip the db creation will be fine. But the main problem is that the openerpuser must be able to create new db
[10:59] <yolanda> the way that openerp works, it allows to create different databases for different companies for example
[10:59] <yolanda> we cannot mess with it, it's how openerp is built
[10:59] <soren> Oh.
[11:00] <soren> I'm not sure dbconfig will let you do that. Giving admin creds to an app sounds like a recipe for disaster.
[11:00] <yolanda> soren, not admin rights, only createdb
[11:00] <yolanda> i'm trying to use install-dbadmin files to add an "alter user openerp createdb", am i right?
[11:01] <soren> yolanda: For postgres? No idea.
[11:01] <yolanda> it's a mandatory condition to run openerp...
[11:01]  * soren is a MySQL sort of person
[11:01] <yolanda> soren, the alter has right syntax, no advice for that :) ... but for the place to put it
[11:02] <soren> Then yes.
[11:02] <soren> At least, that's what the docs say. I've never used that functionality.
[11:02] <iclebyte> soren, could you please take a quick look at this for me and see if it rings a bell? http://ubuntuforums.org/showthread.php?t=1947743
[11:02] <yolanda> ok, will try
[11:02] <Daviey> yolanda: Hola!
[11:02] <Daviey> How goes it?
[11:03] <yolanda> bad...
[11:03] <Daviey> O_O
[11:03] <yolanda> Daviey, currently having some problems with user permissions
[11:03] <lynxman> Daviey: by bad she means she's having troubles with dbconfig and setting up the openerp database :)
[11:03]  * lynxman helps sometimes as a cultural translator from Spanish to English
[11:03] <Daviey> yolanda: just run everything as root?
[11:03] <Daviey> (not being serious)
[11:04] <yolanda> Daviey, i will be happy just embedding a doc in the package and telling the users to do all by themselves :)
[11:04] <soren> iclebyte: No idea. I'm remarkably lost when it comes to ipv6.
[11:04] <iclebyte> soren, as am I! no problem, thanks for looking.
[11:05] <yolanda> the problem i have, is that opnerp user is created by dbconfig, but doesn't have permissions to create new databases, which is completely mandatory for openerp
[11:05] <yolanda> i'm thinking in use install-dbadmin scripts, and paste some alteruser there
[11:07] <Daviey> yolanda: GRANT access?
[11:07] <yolanda> Daviey "ALTER USER openerp CREATEDB"
[11:07] <yolanda> this is the clue
[11:07] <Daviey> Ah no
[11:07] <yolanda> how can i do it?
[11:09] <Daviey> yolanda: I think you need to add GRANT CREATE
[11:09] <yolanda> Daviey, only need that openerp user can create new dbs
[11:09] <yolanda> so doing that in install-dbadmin is right?
[11:11] <Daviey> yolanda: So, you don't want the package to create the database, or create the schema ?
[11:11] <yolanda> no, only the user
[11:12] <yolanda> if it creates an empty db, it isn't a big problem, but the user will have to do it by the UI
[11:12] <Daviey> yolanda: In this case, there is actually little benefit to using dbconfig :(
[11:12] <yolanda> Daviey, to configure the user
[11:12] <yolanda> that's the only thing
[11:13] <Daviey> yolanda: You are going to hate me for this, but if you want the database to be managed by the app.. I wouldn't use dbconfig.
[11:13] <Daviey> especially as you only care about pg
[11:13] <yolanda> well, what i need is just to create that openerp user, and modify config file with that settings. Obviously can be done in rules, as it was done before
[11:14] <yolanda> do you think that is better?
[11:15] <yolanda> i mean, in postinst, but with shell, sorry
[11:15] <yolanda>         # Register "openerp" as a postgres superuser
[11:15] <yolanda>         su - postgres -c "createuser -s openerp" 2> /dev/null || true
[11:15] <yolanda> that is the old way
[11:17] <Daviey> yolanda: if that is ALL you want to do for database handling at package level, i think i'd probably stick with that.  I'd probably put a debconf question of medium, asking if i should do it or not.
[11:17] <Daviey> You must hate me now :)
[11:17] <yolanda> Daviey, i put you in my blacklist :)
[11:18] <yolanda> i've learned a new thing, that's good
[11:20] <Daviey> yolanda: sorry!  When i suggested dbconfig, it was when i thought you'd be handling the database at packaging level
[11:21] <Daviey> you don't care about the database at that stage, it seems
[11:21] <Daviey> i'm suhc an ass.
[11:21] <yolanda> Daviey, doesn't matter
[11:21] <yolanda> i've learned an useful thing, and fighted with packaging a bit, that is good
[11:22] <yolanda> i will do some modifications to that, because we need random password for that user, and modify config file, but this can be done by script
[11:22] <Daviey> yolanda: You can generate a random password if wanted.
[11:23] <yolanda> i know
[11:23] <yolanda> i was doing it
[11:23] <yolanda> with dbconfig :)
[11:24] <Daviey> heh
[11:26] <Daviey> jamespage: Do we care to have: jsvc, libcommons-daemon-java and tomcat6-user in main?
[11:27] <jamespage> Daviey, any specific reason you ask that question?
[11:27] <Daviey> jamespage: i removed them from the CD, and want to know if i should do anything to keep them in main
[11:29] <jamespage> Daviey, I don't think they really need to be in main TBH
[11:30] <jamespage> esp tomcat6-user
[11:30] <jamespage> its source package still will be BTW
[11:30] <jamespage> tomcat6
[11:30] <jamespage> assume you did not take tomcat6 off the ISO?
[11:31] <Daviey> jamespage: no, tomcat is still on the cd
[11:31] <jamespage> Daviey, great
[11:31] <Daviey> just not -user
[11:31] <Daviey> jamespage: happy for -user to drop to universe?
[11:31] <jamespage> Daviey, good with me - I can't see why jsvc, libcommons-daemon-java where in main anyway?  maybe euca?
[11:32] <Daviey> maybe
[11:32] <RoyK> hrmf - seems my server is booting, but after kernel bootup, jumping to initramfs
[11:32] <RoyK> any idea how to remedy that?
[11:32] <Daviey> RoyK: try the prior kernel?
[11:32] <RoyK> grub config looks good, and it must be, otherwise it shouldn't be booting kernel
[11:49] <derknecht> i want to set up a apache with php. I normally used mod-php5, but since ubuntu 11.x there are dependency issuses with apache worker and mod-php5. how is the recommended way of using php with apache? suphp? cgi/fastCGI? mod_php5? Thanks for your opinions.
[11:49] <ikonia> derknecht: there shouldn't be dependency issues
[11:49] <derknecht> problem with mod_php5 is that prefork worker is needed
[11:50] <ikonia> derknecht: apache/php modules are pretty strongly tested each release
[11:54] <derknecht> how stable is suphp? find the user/group setting attractive
[11:57] <ikonia> never used suphp
[11:57] <ikonia> never really even researched it
[12:00] <Daviey> derknecht: I've used apache2-mpm-itk when i want apache running as a user/group
[12:23] <derknecht> Daviey: itk looks very promising, thanks a lot
[12:23] <derknecht> does that work for php and html with the same user? i mean is both affected when i assign a user/group in a vhost?
[12:29] <Daviey> derknecht: yes
[12:31] <macnix> is there a way to create & control upstart jobs as a regular system user?
[12:31] <macnix> I keep getting the dbus error message whenever I use start/stop as anyone else but root
[12:31] <macnix> ideally, each app has will have its own regular user, it will create upstart jobs as that user and it will be able to start/stop them, no admin privileges required
[12:32] <Daviey> jodh: ^^
[12:32] <jodh> macnix: you can create jobs as regular users but this feature is not enabled in Ubuntu by default. See http://upstart.ubuntu.com/cookbook/#user-job
[12:33] <jodh> macnix: and presumably this is the dbus error you get: http://upstart.ubuntu.com/cookbook/#strange-error-when-running-start-stop-restart-or-initctl-emit
[12:34] <Daviey> thanks jodh
[12:34] <jodh> Daviey: np
[12:35] <ikonia> jodh: that's very useful ! I didn't know you could do that
[12:36] <ikonia> thank you
[12:36] <jodh> ikonia: np.
[12:37] <jodh> there is a restriction on user jobs currently: they do not have their output logged. That feature is essentially complete but won't be available in Precise.
[12:38] <koolhead17> hi all
[13:02] <zul> good morning
[13:16] <lynxman> zul: morning
[13:28] <macnix> jodh: yes, that's the error
[13:28] <macnix> jodh: my LTS is running upstart 0.6.5
[13:28] <macnix> is it safe to force an upgrade to 1.3?
[13:29] <wiherek> Hi
[13:29] <jodh> macnix: no
[13:30] <jodh> macnix: well, you can do it, but you'd need to recompile a few packages your self including nih and dbus.
[13:30] <wiherek> should I go with Ubuntu 10.10 or 11.04 and 32 or 64 bit on a 4GHz 4GB RAM VPS?
[13:30] <Angtagapagligt> 11.04 32 Bit
[13:30] <macnix> jodh: don't really want to open that can of worms
[13:31] <wiherek> k, thanks.
[13:31] <Myrtti> 10.04 :-P
[13:31] <macnix> jodh: can I accomplish my goal in a different way?
[13:31] <jodh> macnix: quite - I wouldn't recommend it.
[13:31] <wiherek> cant do 10.04 :(
[13:31] <wiherek> its not in the options.
[13:31] <Myrtti> :-(
[13:31] <wiherek> :-(
[13:31] <Angtagapagligt> :-P
[13:32] <macnix> jodh: I mean, others must have tackled this
[13:32] <macnix> system user per app with upstart jobs, without sudo privileges, can't be THAT difficult, surely
[13:32] <jodh> macnix: you could play elaborate tricks with sudo+initctl+user directories below /etc/init/, but again, I really wouldn't recommend it. What types of jobs do you want your users to be able to create? When do you want these jobs to start?
[13:34] <jodh> macnix: I really don't know I'm afraid. But what is the user case here? You can run system jobs as a different user if required.
[13:35] <macnix> jodh: this is what I'm doing now 1.1 - 1.3 https://github.com/gerhard/deliver/tree/master/strategies
[13:36] <macnix> jodh: that requires anyone who deploys to have sudo privileges, so that upstart jobs can be created properly, and the app start via sudo start
[13:37] <macnix> jodh: so we have developers which have sudo privileges just because they can't deploy new versions of an app without it.
[13:38] <jodh> macnix: I don't understand - why does all this need to run through upstart and need root privs? Can't this app server run on non-priv ports?
[13:39] <macnix> jodh: yes, it can, but how do you re-generate new upstart files for that app (new services get added, old ones removed) without sudo privileges?
[13:40] <macnix> jodh: it's not the ports, it's writing those upstart files in /etc/init & starting/stopping the service which requires sudo
[13:41] <jodh> Why do you need to regenerate new upstart jobs? Why not have 1 central (root) job that uses "instance" and can somehow "look" for other instances of this app-server thing in non-priv users home directories?
[13:42] <macnix> jodh: how would that work? I know the instance job, but not sure how it could "look" for other jobs part of this app
[13:43] <macnix> let me give you a live example
[13:43] <jodh> macnix: look here: http://upstart.ubuntu.com/cookbook/#instance
[13:44] <jodh> macnix: that first example is looking for instances of some service in /var/lib/queues. Tweak to suit.
[13:46] <macnix> jodh: tweaking away. will let you know how it goes. cheers!
[13:46] <jodh> macnix: np. good luck.
[14:26] <hallyn> Daviey: hey
[14:28] <hallyn> Daviey: https://bugs.launchpad.net/ubuntu/+source/vde2/+bug/776818/comments/13   I'd missed that comment.  I don't know how to answer that :)
[14:32] <hallyn> Daviey: specifically wrt bug 776650
[14:32] <hallyn> shoudl this be added to today's mtg agenda?
[14:32] <hallyn> I'll add it
[14:37] <wiherek> can I build a views search form that would work by just adding /search argument to the end of a taxonomy term url?
[14:37] <wiherek> so like this: mysite.com/taxonomy/term/search
[14:38] <Pici> wiherek: Are you sure you're in the right channel for your question, this is #ubuntu-server
[14:38] <wiherek> lol
[14:38] <wiherek> thought it was drupal
[14:38] <wiherek> sry :D
[14:38] <Pici> tis okay ;)
[14:45] <rbasak> smoser: in cloud images, I can sudo without specifying a password, but the sudo configuration does not look like it's changed and the password is just disabled by prefixing a "!". When I do this in a netinst, I still need a sudo password (or NOPASSWD adding to /etc/sudoers). Any hints as to how the cloud images manage to work this way? I want to do the same in a netinst.
[14:45] <smoser> sudoers.d
[14:46] <smoser> $ sudo cat /etc/sudoers.d/90-cloudimg-ubuntu
[14:46] <smoser> # ubuntu user is default user in cloud-images.
[14:46] <smoser> # It needs passwordless sudo functionality.
[14:46] <smoser> ubuntu ALL=(ALL) NOPASSWD:ALL
[14:46] <Daviey> passwords are overrated
[14:46] <rbasak> #includedir /etc/sudoers.d
[14:47] <rbasak> and there I was thinking that the "includedir" directive was commented out, so I didn't examine sudoers.d.
[14:47] <rbasak> thanks :)
[14:50] <hallyn> stgraber: do you have a minute?
[14:51] <hallyn> stgraber: let's say that user namespaces came with a limitation: you have to map userids in container to userids on host a priori (prior to starting container at least).  How bad is that in your opinion?
[14:52] <hallyn> so for instance, installing mysql without planning in advance could cause setuid(mysql) to fail until you stop the container, reconfigure, and restart
[14:52] <hallyn> (of course if you just map everything in container from 0..500 and 1000..1100 you might be pretty safe)
[14:53] <stgraber> hallyn: sounds pretty bad as your regular user won't be able to create users on the host
[14:53] <hallyn> well in any case the mapping will need to be created on the host, so that part i'm not sure is a problem
[14:54] <hallyn> basically, i think you'd find a way to say host ids 50000-55000 are usable by hallyn, for whatever
[14:54] <hallyn> then uid 0 in my container maps to 50000 on host, etc
[14:54] <hallyn> the part that worries me is that the mapping may not be dynamically changeable
[14:55] <hallyn> it may become so, depending on performance impact
[14:57] <hallyn> stgraber: I'm going to cc you on my next email regarding this
[14:59] <stgraber> hallyn: hmm, right, thinking some more, if we assign 5k uids per container it should be fine for most cases
[14:59] <stgraber> hallyn: that's as long as they don't use LDAP or any other large user directory
[15:00] <hallyn> right, but the q is is it reasonable to assume the container can predict the contaienr uids needed, so they can map them
[15:00] <hallyn> let's move to email :)  thx
[15:03] <antihero> how do I make it do an automatic security update each night?
[15:03] <antihero> by "it" I mean a clean ubuntu server distro
[15:46] <SpamapS> antihero: do you want an ISO each night, or do you want to just download and install upgraded packages every night?
[15:57] <lynxman> roaksoax: ping
[16:02] <hallyn> smoser: regardign kvm perf regression, feb 20 I uploaded a pkg based on upstream git head to ubuntu-virt ppa, might be worth trying with that
[16:02] <hallyn> (i'm waitnig for lucid vm to finish building)
[16:03] <smoser> hallyn, well, you can remove variables (specifically variables i dont care about) by using the cloud images
[16:04] <hallyn> drat.  yeah
[16:07] <roaksoax> lynxman: pong
[16:10] <lynxman> roaksoax: I'm having some issues deploying precise with orchestra (sitting on an oneiric machine), I'd like to pick your brain for some mins whenever its okay for you :)
[16:10] <roaksoax> lynxman: sure,
[16:11] <roaksoax> lynxman: sure, what are your issues
[16:11] <lynxman> roaksoax: It looks like when importing the isos if the profile already exists it won't be updated, I'm having an issue with precise where it complains about different kernel versions between the archive and the pxe boot
[16:12] <roaksoax> lynxman: cobbler-ubuntu-import -u precise-i386
[16:12] <lynxman> roaksoax: let me try
[16:14] <lynxman> roaksoax: yeah did that already, facing the same issue :/
[16:15] <roaksoax> lynxman: is cobbler-ubuntu-import updating the ISO?
[16:18] <lynxman> roaksoax: http://pastebin.ubuntu.com/902393/
[16:18] <lynxman> roaksoax: it says "skipping import", that's the issue, I'd pretty much like to force it :)
[16:19] <roaksoax> lynxman: cobbler-ubuntu-import -U
[16:20] <lynxman> roaksoax: no -U option on the oneiric version
[16:20] <roaksoax> lynxman: ah then you need to remove the prefoile and import one
[16:20] <lynxman> roaksoax: cool! will do that then :)
[16:20] <roaksoax> lynxman: unless you wanna install the precise version of cobbler-ubuntu-import
[16:21] <lynxman> roaksoax: nah it's fine, will go that way
[16:21] <roaksoax> ok
[16:21] <lynxman> roaksoax: the only problem is that the oneiric orchestra version does -c which throws the same error, doesn't update the profile
[16:22] <roaksoax> lynxman: -c = check
[16:22] <roaksoax> -u should update
[16:22] <roaksoax> lynxman: that's all fixed in precise btw
[16:22] <lynxman> roaksoax: yeah that's good news, but it'd be also interesting to do a SRU for oneiric
[16:22] <lynxman> roaksoax: don't you think?
[16:41] <Adri2000> I've just realized I proposed a merge request into lp:ubuntu/horizon instead of lp:~ubuntu-server-dev/horizon/essex ...
[16:41] <Adri2000> is it going to be handled anyway?
[16:46] <arosales> smoser: utlemming: Do you think you be able to get to  ubuntu-cloudimg-query and cloud-publish-ubuntu to query2 along wit hthe cloud-image-sync this cycle?
[16:47] <arosales> utlemming: smoser: per https://blueprints.launchpad.net/ubuntu/+spec/servercloud-p-cloudutils
[16:47] <smoser> i think both of those qualify as Feature, and I really think we're past that point in the cycle.
[16:47] <smoser> both suck that we dont have them.
[16:47] <smoser> but..
[16:49] <utlemming> smoser, I agree with your assesment here
[16:49] <Daviey> smoser: if we didn't have control of the API, and the API was deprecated mid-cycle, we'd consider it for an SRU.
[16:49] <utlemming> I would like to ship it, but we're late.
[16:49] <Daviey> therefore, we can still consider it for release if the risk benefit is met.
[16:50] <arosales> sounds like something we can still push for ?
[16:52] <utlemming> arosales: yes...but I need to do a bit of work to get it ready. The bits are done, the documentation is so-so and the packaging needs doing.
[16:53] <smoser> i'll let utlemming comment on it, and i can review that. having cloudimg-query read /query2 woudl be good, but its not terribly useful.
[16:54] <smoser> it would remove 1 hack, which is that you can't really download (or get data for) the disk images with it.
[16:54] <smoser> you just have to get the url for the tarball ("uec style") image and then take off .tar.gz and add .img
[16:57] <smoser> so my personal feeling would be to not bother with cloudimg-query, but if we had sync, that would be good.
[16:57] <arosales> utlemming: do you have cycles to work on sync?
[16:57] <utlemming> smoser: I concur...
[16:58] <utlemming> arosales: I may or may not...it really depends on whether another project interfers. Right now I have the cycles since I'm blocked. So I can give a noble effort.
[16:59] <Daviey> utlemming: Have you solved the LANG PITA?
[16:59] <smb> hallyn, smoser If some hard evidence comes out of the kvm perf issue, it would be best to have a bug report for collecting data and working on it. If you then could mail me (and I thinkg tgardner is interested too) that number it will have a lesser chance to get lost in noise.
[16:59] <utlemming> Daviey: yes, but I want to talk about it with you and smoser during our meeting shortly
[16:59] <Daviey> utlemming: If you have capacity, kernel upgrades still suck.. and infinity doesn't seem to be touching it as promised.
[16:59] <arosales> utlemming: ok, lets see where you are able to make progress on sync. Perhaps leave sync as a todo and postpone cloudimage-query?
[17:00] <utlemming> arosales: that sounds like a plan
[17:00] <utlemming> Daviey: are you referencing the grub error messages?
[17:01] <arosales> SpamapS: thanks for charing todays meeting
[17:01] <SpamapS> always a pleasure :)
[17:05] <hallyn> smb: was just checking to see if smoser had filed one, looks like no.  I'll file one against qemu-kvm and linux (later today)
[17:06] <smoser> hallyn, i have not.
[17:06] <smoser> i really just don't have amore than a finger in the wind
[17:06] <smb> hallyn, sounds good
[17:06] <smoser> and i'm embarrarisingly ignorant about what even *is* good
[17:06] <hallyn> ditto, really.  but i'm pretty sure 9m/s is bad :)
[17:06] <smoser> regressions are easy to spot, though, given 2 setups to compare.
[17:13] <tarvid> can I reload without setting up the dm again?
[17:14] <tarvid> or how can I get dns resolution working again?
[17:15] <tarvid> since switching to static ip address, I've lost dns
[17:16] <e_t_> tarvid: Did you switch to static IP in NetworkManager? Did you specify a DNS server to use there?
[17:16] <tarvid> I don't havge network namager, this is a basic server install
[17:16] <tarvid> manger
[17:16] <tarvid> without desktop
[17:17] <e_t_> What's the output of cat /etc/resolv.conf ?
[17:17] <tarvid> it had nameserver 192.168.1.1
[17:18] <tarvid> even after adding dns-nameservers to /etc/network/interfaces
[17:19] <e_t_> Do you get anything from "dig @192.168.1.1 google.com" ?
[17:19] <tarvid> no name resolution works
[17:21] <e_t_> That's what the @ is for, it directs dig to point to a specified server, regardless of defaults. Also try it with @8.8.8.8
[17:22] <tarvid> no name servers could be reached
[17:22] <e_t_> can you ping other hosts?
[17:22] <tarvid> only this machine
[17:23] <e_t_> The your real problem is that networking is busted.
[17:23] <tarvid> true
[17:23] <tarvid> but it worked before attempting a static ip and nameservers
[17:24] <e_t_> Can you pastebin your /etc/network/interfaces ?
[17:25] <tarvid> http://paste.ubuntu.com/902512/
[17:27] <e_t_> What address do you get from DHCP?
[17:28] <tarvid> I'll swith from static to auto and see what happens
[17:30] <Aison> on my ubuntu server, the networking is completly fucked up :(((
[17:30] <tarvid> I'm downloading the daily build, can I reinstall without setting up the dm again?
[17:30] <Aison> I didn't change anything in the configuration, but now, after reboot I have to configure the network devices all by hand
[17:31] <e_t_> tarvid: what do you mean by "dm" ?
[17:31] <Aison>  /etc/network/interfaces is somethow not working right
[17:31] <tarvid> RAID1
[17:31] <tarvid> Aison, at least you have company
[17:32] <tarvid> resolvconf is supposed to take dns-nameservers from that file
[17:33] <e_t_> tarvid: I think the installer will detect an existing Linux RAID.
[17:33] <tarvid> encoueraging
[17:33] <tarvid> took overnight the last time to erase it
[17:33] <Adri2000> zul: is it a problem that my merge request is for lp:ubuntu/horizon instead of lp:~ubuntu-server-dev/horizon/essex ?
[17:34] <Adri2000> (realized that today :/)
[17:34] <e_t_> tarvid: At that point in the installer, nothing has been changed, so you can just reboot if the raid isn't detected.
[17:34] <zul> Adri2000: yeah can you merge it against lp:~ubuntu-server-dev/horizon/essex please
[17:35] <Aison> tarvid, since ubuntu 11.04, I've got also this strange "Waiting for network connection" message at boot up, on ALL my desktop machines, it's just crap
[17:35] <tarvid> does not make one cheery
[17:43] <tarvid> dhcp fails on install, time to fix that first
[17:45] <Adri2000> zul: done https://code.launchpad.net/~adri2000/horizon/rename-apache-config-file/+merge/99574
[17:46] <zul> Adri2000: cool thanks
[17:49] <tarvid> how am i supposed to use network-manager to add aliases?
[17:55] <bluefrog> tarvid, aliases? oh thought you had your own dns server
[17:56] <tarvid> I do but want to add ethernet aliases to a host
[17:56] <bluefrog> tarvid, is your resolv.conf still a symlink?
[17:57] <bluefrog> tarvid, deit /etc/hosts for aliases
[17:57] <bluefrog> edit
[17:58] <tarvid> about to reload, testing static IP addresses issued by ISP, that is why I want to add aliases on another machine to make sure the routes work
[18:03] <koolhead17> popey, around?
[18:05] <koolhead17> popey, just wanted to tell you we should not feed those trolls!! :)
[18:07] <tarvid> how come aliases in network-manager do not show up in ifconfig?
[18:08] <rbasak> Daviey: can we talk about my service orchestration / system management blueprints tomorrow?
[18:09] <Daviey> rbasak: ping me in the morning?
[18:10] <Daviey> rbasak: well have a voice chat
[18:22] <roaksoax> SpamapS: how can I start an upstart job after a init script has started a daemon
[18:40] <SpamapS> roaksoax: two options. Either a) change the init.d script to emit an event when it is started, or b) convert the init.d script to an upstart job
[18:41] <roaksoax> SpamapS: cool thanks and yes I had to patch the init script
[18:45] <rbasak> Daviey: will do, thanks
[19:23] <Daviey> rbasak: still around?
[19:23] <Daviey> rbasak: Does OpenMPI need to be in beta2?
[19:24] <rbasak> Daviey: I'm not aware of a need. I think janimo uploaded it
[19:25] <Daviey> http://launchpadlibrarian.net/98532550/openmpi_1.4.3-2.1ubuntu2_1.4.3-2.1ubuntu3.diff.gz
[19:31] <tarvid> installing 12.04 server - long pause after configuring network manually. What is it doing?
[19:45] <SpamapS> tarvid: testing your patience ;)
[19:46] <SpamapS> Ursinha: what happened to the old server team SRU report?
[19:46] <Ursinha> SpamapS, it's supposed to be there
[19:47] <SpamapS> Ursinha: where?
[19:47] <Ursinha> SpamapS, where it was before :) let me find the link
[19:47] <SpamapS> should go in the knowledge base
[19:48] <SpamapS> Ursinha: and thank you for finding it for me. :)
[19:49] <smoser> Daviey, we're expecting cloud-images for beta-2 candidates tomorrow?
[19:49] <smoser> (and the same for ISO)
[19:52] <zul> adam_g: i added convert_to_sqilte.sh and sample_data.sh to keystone.install
[19:53] <Ursinha> SpamapS, hm, it seems it's a bit broken
[19:53] <Ursinha> zul, where did you place your old sru report?
[19:53] <adam_g> zul: whats convert_to_sqlite do?
[19:53] <Ursinha> it was you?
[19:55] <Daviey> smoser: Yep, i'm not going to bed until we have a viable candidate
[19:55] <smoser> you have a bed?
[19:56] <Ursinha> lol
[19:56]  * SpamapS sleeps on a pile of old scraps and feathers
[19:58] <zul> Ursinha: should be in the same place as last time
[19:58] <Ursinha> zul, I can't remember where
[19:58] <Ursinha> hahahaha
[19:58] <Ursinha> we're all doomed
[19:58] <zul> http://people.canonical.com/~chucks/SRUTracker/
[19:58] <Ursinha> ah, it's chucks
[19:58] <Ursinha> I was trying zul
[19:59] <Ursinha> y u no use the same nick!?
[19:59] <zul> alter-ego
[19:59] <Ursinha> lol
[19:59] <Ursinha> thanks zul
[19:59] <Ursinha> bad news is that it seems equally broken
[19:59] <Ursinha> sigh
[19:59] <Ursinha> SIGH
[19:59] <Ursinha> SpamapS, I'll fix that
[20:00] <SpamapS> Ursinha: thank you!
[20:00]  * SpamapS had not looked at it in a while but was wondering how we were doing at keeping up w/ SRU's
[20:01] <Ursinha> SpamapS, it's here: http://reports.qa.ubuntu.com/reports/ubuntu-server/sru-report.html
[20:01] <Ursinha> you can bookmark it
[20:01] <Ursinha> :)
[20:03] <SpamapS> Ursinha: done, and maybe once its fixed, add it to the KnowledgeBase page as well?
[20:03] <Ursinha> SpamapS, surely
[20:04] <Ursinha> we might want to add stats.qa.ubuntu.com/reports/ubuntu-server/ as well
[20:09] <zul> SpamapS: im not keeping up with the sru report at all
[20:11] <allowoverride> whats a good program for vitualizing OS on debian based servers like ubuntu
[20:11] <allowoverride> i did a headless a while back on ubuntu 6.x. wondering what other free progs are avail... thanks!
[20:12] <guntbert> !kvm | allowoverride
[20:13] <allowoverride> uses libvrt
[20:13] <allowoverride> ok, anything else?
[20:13] <allowoverride> i like virt-manager :) and xm cmd's
[20:14] <allowoverride> oh im new to xen by the way
[20:20] <rhys> so orchestra. anyone using it yet?
[20:54] <pabelanger> Is the ubuntu release team responsible for actually creating the ISO (specifically the mini.iso)?
[20:55] <pabelanger> I'm looking for the process that is actually used (eg: scripts)
[21:11] <raubvogel> editing rsyslog config: rsyslog.conf, rsyslog.d/50-default.conf, or create a file in rsyslog.d/ ?
[21:52] <kirkland> hmm, seems that dkim-filter no longer exists in Ubuntu 12.04
[21:52] <kirkland> I was using that happily on Ubuntu 10.04
[21:52] <kirkland> anyone know if there's a replacement?
[22:24] <bemu> Anyone seen apt-get failures on EC2 akin to: https://gist.github.com/0f8b5d1be792990d1bc6
[22:24] <bemu> It's strange because it looks like the files are being pulled from the same place with --fix-missing: https://gist.github.com/2703c9b26d594d36f6a5
[22:43] <dork> hi all, i've got a few ubuntu servers connected to an ip based kvm. i've got communication with the serial port during post, grub, and if therer isn't a failure between runlevel 1 and 2 i can see that and everything going forward, if there is an issue during boot up like prompting for raid degraidation boot override or if a software array cant assemble, or any failure for that matter that results to a busybox prompt, i lose my communication with the box
[23:42] <mgw> is it possible (reasonably easily) to run sshd on separate ports, with different sshd config settings for each port?
[23:45] <raydog45000> I just recently set lamp up on my ubuntu desktop and am trying to convert it to pure server, I can access it via ssh on mylocal host, does anyone know a good guide to make it accessible from the net or know how?
[23:54] <ChmEarl> raydog45000, iptables -A INPUT -p tcp -m state --state NEW -m tcp --dport 22 -j ACCEPT
[23:55] <raydog45000> what does that do?
[23:55] <ChmEarl> raydog45000, if you have a router (hw firewall) then open port 22 there
[23:55] <raydog45000> ok