[01:14] <twb> "WARNING!!!  The filesystem is mounted.  If you continue you ***WILL*** cause ***SEVERE*** filesystem damage.  Do you really want to continue (y/n)?"
[01:15] <twb> If e2fsck is so confident that'll cause problems, why does it even offer to continue?
[01:16] <mdeslaur> twb: some people like pain? :)
[01:33] <BuenGenio> is the PHP 5.3.10 update coming to 11.04/10 any time soon?
[02:16] <pmatulis_> no
[02:21] <twb> portmap is so fucking stupid
[02:21]  * twb rages
[03:16] <cloudgeek> host own domain
[03:52] <pmatulis_> irc spam?
[04:39] <linocisco> http://paste.ubuntu.com/849536/
[04:39] <linocisco> hi
[04:42] <twb> linocisco: I know of no way to do what you want.
[07:00] <uvirtbot`> New bug: #936756 in lxc (universe) "lxc complains about cgroup not available" [Undecided,New] https://launchpad.net/bugs/936756
[08:22] <lynxman-> morning o/
[08:23] <x2x> morning
[08:23] <x2x> I need some suggestion on my NFS planning
[08:28] <MatBoy> mornign
[08:39] <x2x> anyone here can help me?
[08:39] <x2x> I need some suggestion on servers planning
[08:40] <linocisco> x2x, what kind of?
[08:40] <x2x> I am going to setup KVM
[08:41] <linocisco> x2x, u mean u want virtual machine?
[08:41] <x2x> There are 4 yes
[08:42] <x2x> 1 server will be cluster controller+storage controller, 1 server will be a cloud controller and 4 servers will be node controller
[08:42] <x2x> these 4 servers controller will use external storage for their fs.
[08:43] <linocisco> x2x I am sorry it is beyond my knowledge. You want fault torlerant highly available system
[08:43] <x2x> these 4 servers will be connected to 2 NFS where these 2 NFS will connect to a same storage
[08:43] <x2x> yes
[08:43] <x2x> let me simplify it
[08:44] <x2x> A HA or Cluster or others tech I can use
[08:44] <x2x> 2 NFS connect to 1 external storage thru FC
[08:44] <x2x> NFS used to prevent the single point failure
[08:45] <x2x> 2NFS*
[08:45] <x2x> Is that out of your knowledge?
[08:45] <x2x> If yes, are there any else that I can ask? or which room?
[08:47] <linocisco> x2x; sorry I could not help.
[08:47] <linocisco> x2x,  you can find some in youtube
[08:49] <x2x> It's okay. nevermind.
[08:49] <x2x> Thank you for your help. :)
[08:51] <uvirtbot`> New bug: #936762 in lxc "lxc-ubuntu template sets user shell, without checking it's installed" [Undecided,New] https://launchpad.net/bugs/936762
[08:53] <lynxman-> x2x: I reckon you have two ways of doing that, one (the one I use) is to maintain the sessions (which are stored in /var/lib/nfs/), best way to do that is to have that directory itself in the storage and soft link to it
[08:53] <lynxman-> x2x: The second way (which I haven't tried and I'm not certain it supports NFS) is to use multipath support
[08:54] <x2x> that means the first method is active passive just work like a cluster?
[08:54] <lynxman-> x2x: active/passive
[08:55] <x2x> is it possible to make it work as active/active?
[08:55] <jamespage> morning all
[08:55] <x2x> the first method shouldn't be a problem.
[08:56] <x2x> you just use heartbeat to control it?
[08:56] <x2x> How do you find it? Is it stable?
[08:56] <lynxman-> x2x: hmm in order to ensure consistency you could do that by dividing your nfs shared data in several exports, assign a virtual IP per export point and make everyone under that share to connect to the same server
[08:56] <lynxman-> x2x: it's quite stable unless there's problems transitioning the IPs
[08:56] <lynxman-> jamespage: morning sir
[08:56] <x2x> jamespage, morning.
[08:57] <jamespage> rbasak, around yet?  fed some PPA's over the weekend with openmpi rbd's - looks OK
[08:57] <rbasak> jamespage: morning!
[08:57] <x2x> what do you means everyone under that share to connect to the same server?
[08:58] <rbasak> jamespage: Debian have released exp2 incorporating your openmpi patch
[08:58] <jamespage> rbasak, so I saw
[08:58] <jamespage> \o/
[08:58] <lynxman-> x2x: that everyone that mounts the same mountpoint should be mounting it from the same server in order to ensure a painless transition, otherwise problems may arise (at least in my experience)
[08:58] <jamespage> rbasak, https://launchpad.net/~james-page/+archive/openmpi/+packages
[08:59] <jamespage> and https://launchpad.net/~james-page/+archive/openmpi2/+packages
[08:59] <rbasak> How have you been bumping the dependent packages? I'm thinking of scripting that and uploading to an ARM-enabled PPA
[08:59] <jamespage> I could not fit it all in one PPA (thats a first!)
[08:59] <jamespage> rbasak: http://paste.ubuntu.com/849702/
[09:00] <x2x> lynxman, thx for your advise. I should have a test on it.
[09:00] <jamespage> rbasak, I think if we want todo this we need to get the FFe filed today
[09:00] <lynxman-> x2x: no problem :)
[09:00] <rbasak> jamespage: OK
[09:00] <x2x> thank you. :)
[09:00] <x2x> bye bye.
[09:00] <linocisco> any mail server expert?
[09:00] <jamespage> rbasak, only 4 rebuild failures - two of them already FTBFS in the archive
[09:01] <jamespage> one timed out - I can't repro that locally
[09:01] <linocisco> sorry any port changer expert?
[09:01] <jamespage> and the other looks odd - mpi4py
[09:02] <rbasak> One still in progress?
[09:03] <lynxman-> linocisco: ask away and I'm sure somebody will pick it up
[09:03] <linocisco> our office internet needs 19xx port. but our ISP allows only 8080 and 443. how to do ?
[09:03] <jamespage> rbasak, yeah - I hit the button of desparation again - thats the one that stops and then times out
[09:03] <linocisco> mail server from  our office needs 19xx port. but our ISP allows only 8080 and 443. how to do ?
[09:05] <diplo> Get a decent ISP ?
[09:05] <diplo> A business line even
[09:06] <diplo> Or do you mean you have some proxy software and need it to go via 19xx
[09:07] <linocisco> diplo, yes.
[09:07] <rbasak> mpi4py - I've had a similar failure
[09:08] <diplo> Just forward port on your router ?
[09:08] <diplo> http://portforward.com/
[09:09] <rbasak> starpu-contrib - I was also missing a dependency on libcuda1 which doesn't appear to exist in the archive
[09:11] <x2x> Hi Diplo
[09:11] <x2x> Hi Linocisco
[09:11] <linocisco> x2x, hi
[09:11] <rbasak> elmerfem - I have the same error. And feel++ I don't appear to have attempted, possibly due to a dependency build failure
[09:11] <x2x> A mail server is setup in your office and you want to make it public and can be accessed from the network out of your office
[09:12] <x2x> am I right?
[09:12] <x2x> 19xx port is the port for you mail server incoming use?
[09:12] <x2x> Whether you can apply that port from your ISP?
[09:12] <x2x> If not
[09:12] <x2x> beside the port forwarding
[09:12] <x2x> you may have a workaround method
[09:13] <x2x> setup a webmail
[09:13] <x2x> 443 or 8080 doesn't matter for a webmail.
[09:13] <linocisco> x2x . we have webmail
[09:13] <x2x> I am not sure whether you want to use mail client to retrieve the email.
[09:14] <x2x> webmail can use port 443 or 8080
[09:14] <linocisco> x2x . actually we wanna use mail client software that only works with 19xx port only. and that port is blocked by government
[09:14] <x2x> what mail server do you use?
[09:14] <linocisco> x2x. webmail is too slow. old emails can't be searched easily
[09:14] <x2x> what a big challenge...
[09:15] <linocisco> x2x, novell groupwise
[09:15] <jamespage> rbasak, do you want me to work on the justification for the FFe in the bug report?
[09:15] <x2x> oic
[09:16] <x2x> so only port 8080 and 443 are allowed?
[09:16] <rbasak> jamespage: Is this definitely the route we want to follow?
[09:16] <rbasak> jamespage: I didn't think we'd decided yet?
[09:17] <jamespage> rbasak, OK so what factors drive this decision
[09:17] <jamespage> ?
[09:19] <rbasak> 1) Whether the remaining FTBFS (that 1.5 introduces) can be fixed - I count 3? And 2) Whether the OpenMPI test suite passes (if I can get access to it) or if we should proceed without one (if I can't)
[09:20] <jamespage> rbasak: 1) I think we should only care about NEW FTBFS
[09:20] <jamespage> 2) agree that would be sensible
[09:20] <jamespage> I'm going to start sticking information in the bug report for the FFe - that way once/when we are happy its all ready to go
[09:21] <jamespage> on 2) we might be able to sniff a few of the rbd's manually to see if they continue to work.
[09:21] <jamespage> rbasak, I also took a look at the bugs for openmpi - there are a few
[09:22] <jamespage> I would suspect that these packages are not hugely used by the ubuntu user base
[09:22] <jamespage> they are generally quite specialist in application
[09:23] <uksysadmin> howdy ubuntu folks
[09:23] <rbasak> OK
[09:23] <uksysadmin> anybody seeing kernel panics in Precise A2? log.c:786: Assertion failed in log_clear_unflushed: log->remote_closed
[09:24] <rbasak> jamespage: so we plan to go ahead? What should I focus on?
[09:24] <jamespage> rbasak, 1) & 2)
[09:24] <jamespage> but in reverse order
[09:30] <koolhead17> hi all
[09:38] <jamespage> rbasak, how would you feel about me contacting the openmpi maintainers in Debian for an opinion on the saneness of this upgrade?
[09:39] <rbasak> jamespage: sure, go ahead. I've been in touch already - they know we're trying it
[09:40] <jamespage> rbasak, ack - I'll get something out now
[09:40] <jamespage> that would be nice to reference in the FFe if we go ahead
[09:41] <jamespage> rbasak, fair to say other than this work neither of us are close to openmpi as a project?
[09:42] <rbasak> Yep
[09:48] <jamespage> rbasak, so I'm asking for general opinion on sanity of moving to 1.5 (based on the work we have done so far)
[09:48] <rbasak> OK
[09:49] <jamespage> and any guidance they can provide in test cases we might execute - sound OK to you?
[09:49] <rbasak> Yep
[09:49] <rbasak> I've found some tutorials on ubuntu+openmpi that I can try. I'll give that a go today.
[09:49] <jamespage> rbasak, Are you subscribed to pkg-openmpi-maintainers? or should I CC you?
[09:50] <rbasak> CC please
[09:51] <jamespage> rbasak, sent
[09:52] <rbasak> thanks!
[09:52] <jamespage> rbasak, hmm - it does not clean un-install at the moment - I'll try latest from experimental
[09:59] <linocisco> ok.
[10:01] <jamespage> rbasak, what FTBFS did you see?
[10:02] <rbasak> In my testing?
[10:02] <linocisco> 10.1.1.1 - 172.16.1.0<--->172.16.1.0 - 192.168.1.0  . I want to know how to ping from 192.x.x.x network to 10.x.x.x
[10:06] <Spanky> anybody know their mount commands and fstab editing?
[10:09] <_ruben> Spanky: you might wanna ask more specific questions
[10:09] <Spanky> I'm having a heck of a time with mounting NSLU2 on Ubuntu 10.04 Server permanently....
[10:10] <Spanky> running Zentyal....
[10:10] <Spanky> I can mount in Nautilus
[10:14] <Spanky> Here's my fstab entry....   http://pastebin.com/ctjAfJx8
[10:16] <jamespage> rbasak, yes
[10:18] <rbasak> jamespage: a whole bunch more than what you have. Mostly build-deps issues, probably due to the way I was build testing them. For the failures in your PPA, I have the same failures in my build logs. Except for feel++, wihch I never attempted presumably due to a dependency build failure.
[10:18] <jamespage> rbasak, so build-dep is  just order - I found some I had missed
[10:19] <rbasak> jamespage: I stuck your dependency tree into a Makefile and had that drive the builds. I think my failures are due to my apt pinning or something
[10:19] <rbasak> jamespage: ah. I just read what you wrote again. That would make sense.
[10:23] <jamespage> rbasak, there are some upgrade issue we will need to resolve as well - looking at those ATM
[10:54] <RoyK> my kvm suddenly fails to start any VMs, virt-manager reporting 'libvirtError: monitor socket did not show up.: Connection refused' - anyone that knows where to start debugging this?
[10:55] <sbarakos>  hey there guys, whats the command to limit connections 3 per ip /1 sec for 1 port in iptables?
[10:56] <RoyK> sbarakos: dunno if there's a specific limit for that, but I guess using conntrack and LIMIT, I guess it should be possible
[10:56] <RoyK> sbarakos: it's easier from within the server code, though
[10:58] <greppy> sbarakos: first hit on google was http://www.debian-administration.org/articles/187
[11:07] <sbarakos> command : -A INPUT -p tcp --dport 80 -i eth0 -m state --state NEW -m recent --update --seconds 1 --hitcount 3 -j DROP
[11:07] <sbarakos> something like that seems good?
[11:09] <greppy> sbarakos: sure, give it a shot :)
[11:18] <Parham> Hi everyone. I had JDK 1.6.0_U23 and Tomcat 6 (I'm not sure which one) installed on my Ubuntu 11.10 machine. However, last Wednesday I got notified that there are security updates. And after updating, Tomcat6 was really slow and took up 101% of memory (1 CPU core out of 4). I then installed Tomcat 7.0.21, but the problem still persists. This only happens with one application; other applications wo
[11:18] <Parham> rk fine.
[11:19] <oCean> Parham: ah, so it's not in the tomcat server
[11:20] <Parham> oCean: I have no idea. The application worked fine before the update.
[11:21] <Parham> oCean: The application as of this moment is not different than it was a week ago (when it was working fine).
[11:36] <jamespage> Parham, the last security updates applied in 11.10 for tomcat6 where quite intrusive
[11:36] <jamespage> https://launchpad.net/ubuntu/oneiric/+source/tomcat6/+changelog
[11:37] <Parham> jamespage: The interesting thing is that applications written with Grails 1.3.7 have this problem; the demo applications that come with Tomcat run fine.
[11:37] <jamespage> Parham: that said tomcat7 has not had the same security update
[11:37] <jamespage> so I would not suspect that as the cause
[11:47] <rbasak> jamespage: an MPI hello world fails with the same error as in the mpi4py FTBFS. It seems that openmpi 1.5 doesn't actually work currently. Though at least when I fix it I should get that FTBFS as well :-)
[11:47] <jamespage> rbasak, nice
[11:55] <kai> hi folks
[11:56] <kai> I'm currently playing with linux containers a bit, does anyone happen to know a way of moving guest containers around between hosts? I've got a shared storage I'm keeping them on, so data continuity shouldn't be a problem
[12:09] <jamespage> rbasak, anything I can do to help with that?
[12:10] <rbasak> jamespage: I'm doing OK with it - tracked it down with gdb to figure out what it's actually trying to do. The segfault is because it traverses an empty list but seems to assume that it's non-empty. I've just figured out where it populates the list (global variable fun!) and hopefully should figure out why it's empty soon.
[12:16] <uvirtbot`> New bug: #936916 in keystone (universe) "package keystone 2012.1~e4~20120203.1574-0ubuntu3 failed to install/upgrade: el subproceso instalado el script post-installation devolvió el código de salida de error 1" [Undecided,New] https://launchpad.net/bugs/936916
[13:30] <soren> smoser: Around yet?
[13:31] <jamespage> soren, unlikely - presidents day today - but I guess you never know
[13:31] <soren> jamespage: Ah, thanks for the heads up!
[13:31] <soren> Didn't realise.
[14:22] <NCommander> rbasak: ping?
[14:22] <rbasak> hey NCommander
[14:23] <NCommander> rbasak: so, re: openmpi, are we putting 1.5 side by side with 1.4? (I've heard a few condractory points on this)
[14:24] <rbasak> It's not definitely decided yet, but we're hoping to update to 1.5 if we can demonstrate that it works. Currently, it's broken, but I'm working on it.
[14:24] <rbasak> jamespage: ^^ - accurate?
[14:24] <NCommander> rbasak: right, thats a bit of a problem.
[14:24] <NCommander> We're past feature freeze at this point
[14:24] <jamespage> rbasak: yes - that about it
[14:25] <rbasak> https://bugs.launchpad.net/ubuntu/+source/openmpi/+bug/889644 - [FFe] Please update OpenMPI to the 1.5 upstream version
[14:25] <uvirtbot`> Launchpad bug 889644 in openmpi "[FFe] Please update OpenMPI to the 1.5 upstream version" [Wishlist,Confirmed]
[14:25] <NCommander> rbasak: ok, great
[14:25] <NCommander> rbasak: second question, whats the main blocker that has all of servercloud-p-arm-system-management blocked (aside from !IMPI hardware)
[14:26] <rbasak> Essentially, just IMPI hardware blocked. I had said that I can do pieces of work items without hardware, but I'd have to go over all of it with hardware again.
[14:26] <NCommander> rbasak: well, the binary image deployment with PXE is also blocked
[14:26] <NCommander> OMAP3/4 can PXE boot
[14:27] <rbasak> I can't see that - where is it?
[14:27] <NCommander> postponed	servercloud-p-arm-service-orchestration	Essential	Binary image deployment with PXE on ARM ‡₁
[14:28] <NCommander> Er
[14:28] <NCommander> its on servercloud-p-arm-service-orchestration
[14:28] <rbasak> Ah
[14:29] <rbasak> I put that down as no longer relevant. I can't recall the discussion, but I'm not aware of a use case for it that we're pursuing.
[14:29] <rbasak> Do you have one?
[14:29] <NCommander> rbasak: required for mass deployment to Calxeda hardware (1000+ machines in a single rack)
[14:30] <NCommander> There were concerns about orchesta even scaling to that bit
[14:30] <rbasak> I wasn't aware of this.
[14:30] <rbasak> As I understood it, the orchestra enhancements that will be delivered in 12.04 would deliver this and be able to cope with the scale.
[14:30] <NCommander> rbasak: I thought you were standing next to me and davidm when we were discussing the specifics the calxeda design
[14:31] <NCommander> rbasak: orchestra needs to work on ARM, or the scale improvements are pointless for us
[14:31] <rbasak> In what way doesn't it work today?
[14:32] <NCommander> rbasak: I have no idea if it works today or not.
[14:33] <rbasak> cobbler works on my panda
[14:33] <NCommander> great, is it documented?
[14:33] <NCommander> (and is that LXC or baremetal deployment?)
[14:34] <rbasak> Bare metal. The only documentation I know of is scattered around
[14:34] <NCommander> right
[14:34] <NCommander> that's useful :-(
[14:34] <rbasak> I've tested LXC too, and fixed various issues with that, so it works on my panda as far as juju's local environment purposes need it anyway.
[14:36] <rbasak> I've been working on panda and automatic netinst etc. as a set of scripts that together Just Work, but it's not ready yet. I've been diverted by openmpi which is evidently far more work than I thought it would be.
[14:37] <rbasak> But cobber on panda is a panda-specific story really. It'll be different for real ARM server, and until hardware is delivered I don't see that there's any more work to do.
[14:40] <NCommander> so the server team is regularly using LXC then on ARM?
[14:40] <NCommander> rbasak: that's a fair viewpoint. We should (I hope) have a PXE enabled bootloader for armadaxp soonish though
[14:40] <NCommander> rbasak: though I'm curious on what's specific to clobber on armel+omap4
[14:40] <rbasak> Just me when I tested/fixed it, AFAIK. I'd like to set up automatic regression testing but I've  been distracted by openmpi
[14:42] <NCommander> rbasak: understandable, but if it works, I'm satisified; I got an alarming report it was hosed/completely broken/etc.
[14:43] <rbasak> Very little specific. The initrd/kernel image URLs, kernel parameters on the installer. And the preseed file is slightly different because s/archive/ports. Although pxelinux.cfg/default is broken because it uses menus which u-boot doesn't support, and also there's no arch detection.
[14:44] <rbasak> NCommander: let's just make sure I'm not giving you a false impression. What did you think was completely hosed/broken? cobbler automated netinst on panda, or something else?
[14:44] <NCommander> Well, pxelinux.cfg isn't strictly speaking part of the intel PXE spec when I looked; that was a third-party loader
[14:44] <NCommander> rbasak: LXC on ARM. and I was just it was broken, nothing more which I thought was bunk, and was planning on hammering today
[14:44] <rbasak> Yes - and u-boot implements exactly 0% of the intel PXE spec, and some part of the third-party loader.
[14:44] <NCommander> ....
[14:45] <NCommander> and they call it PXE?
[14:45] <NCommander> */no comment*
[14:45] <rbasak> I don't. I call it pxelinux emulation.
[14:45] <NCommander> Didn't Linaro implement "PXE" in uboot?
[14:45] <rbasak> u-boot implements a subset of what pxelinux.0 does after it is loaded.
[14:46] <NCommander> so it doesn't properly DHCPDISCOVER for bootservers and such?
[14:47] <rbasak> It does DHCP first, yes. To get the next-server to fetch /pxelinux.cfg/* from. It does not honour the DHCP options that allow that location to be changed.
[14:47] <NCommander> ....
[14:47] <NCommander> legally speaking, that can't be called PXE
[14:47] <NCommander> *sighs*
[14:48] <rbasak> That's what I keep saying! I thought you guys knew all this already :)
[14:48] <rbasak> Legally speaking, nothing on ARM can implement PXE.
[14:48] <rbasak> (by spec)
[14:48] <NCommander> First I heard of it
[14:49] <NCommander> obviously we've had a breakdown in communications
[14:49] <NCommander> Its been awhile since I broke the PXE spec out, but I thought 2.1 specification had proper arch tags and crap for ARM
[14:50] <rbasak> AIUI, PXE has no ARM architecture definition at all. But I may be wrong. I am sure though that u-boot does not implement PXE in that form. There is no arm binary image that it downloads. It starts at pxelinux.cfg stage, and thus I'm sure that it only emulates (a limited) pxelinux and nothing else.
[14:51] <jhobbs> you can setup your dhcp server to serve up different pxelinux.cfg files for arm than for intel
[14:51] <rbasak> jhobbs: I tried that; it doesn't work
[14:51] <jhobbs> how did you try it?
[14:51] <rbasak> RFC5071
[14:51] <rbasak> DHCP options to point it somewhere else.
[14:52] <rbasak> Is there any other way? Other than the DHCP server dynamically picking up the vendor id and making a note to serve it a different pxelinux.cfg when it TFTPs, or using a different TFTP server entirely?
[14:53] <NCommander> rbasak: ew, your right. ARM isn't defined in the proper PXE spec
[14:53] <jhobbs> look at CONFIG_VCI_STRING in U-Boot
[14:54] <jhobbs> you can setup your dhcpd to serve up different bootfiles for different VCI's
[14:54] <NCommander> rbasak: it sounds like it just reads the DHCPOPTIONS it gets directly; but strictly speaking, the host doesn't have to provide them if the client doesn't identify it as a PXE server ...
[14:54] <NCommander> jhobbs: that requires recompiling uboot
[14:54] <NCommander> Not exactly a desirable option :-P
[14:54] <jhobbs> well you have to recompile it to build pxe support in anyhow
[14:54] <rbasak> jhobbs: so dhcpd can differentiate on architecture. But the TFTP server cannot.
[14:54] <jhobbs> tftp doesn't have to.. it goes where dhcp points it to
[14:55] <NCommander> jhobbs: stock builds should have PXE now
[14:55] <rbasak> jhobbs: by IP only. It seems to ignore DHCP option flags. Unless there's an option flag I'm missing, which would be great. Is there?
[14:55] <NCommander> or PXE-like
[14:55]  * NCommander has a headache
[14:55] <NCommander> ugh
[14:55] <NCommander> thanks
[14:55] <jhobbs> yes there is, it's the VCI string option
[14:55] <jhobbs> look at rfc 4578
[14:56] <jhobbs> it's part of the PXE rfc, not pxelinux RFC
[14:57] <rbasak> jhobbs: I don't follow
[14:57] <jhobbs> http://codepad.org/Td7pOfAA
[14:57] <rbasak> jhobbs: how exactly should the dhcp server tell the machine to use an alternate pxelinux.cfg?>
[14:58] <rbasak> jhobbs: I understand that the dhcp server can differentiate between arm and intel this way. But that's useless unless it can also tell it to do something different once it has detected that difference.
[14:58] <rbasak> jhobbs: the only straightforward mechanisms I can think of are to serve a different ip range, or to point it at a different tftp server IP.
[14:58] <rbasak> jhobbs: is there any other mechanism?
[14:59] <rbasak> this is why Daviey raised https://bugs.launchpad.net/ubuntu/+source/u-boot-linaro/+bug/927781, and justinlw is looking at it AFAIK. If there's a better way, please tell us!
[14:59] <uvirtbot`> Launchpad bug 927781 in u-boot-linaro "PXELINUX implementation doesn't respect dhcp ConfigFile or PathPrefix values" [Undecided,In progress]
[15:01] <jhobbs> http://codepad.org/YINNBafr
[15:01] <jhobbs> i have to run
[15:01] <jhobbs> i'll look at that bug later, thanks for linking me
[15:02] <rbasak> thanks jhobbs - if that last paste works, then our problem is solved :-)
[15:03] <rbasak> So the pxelinux.cfg is overloading the filename parameter to mean the path to the pxelinux.cfg directory instead of the path to the bootloader binary?
[15:03] <rbasak> Daviey: ^^
[15:04] <rbasak> NCommander: are you absolutely clear on things now, or would it be worth a call for us to sync?
[15:05] <Daviey> rbasak: erm
[15:05] <Daviey> rbasak: there are two values
[15:06] <Daviey> rbasak: configfile and pathprefix
[15:06] <Daviey> rbasak: pxelinux binary is irrelevant, as this is the config file that pxelinux (or uboot's implementation), consumes
[15:07]  * Daviey goes afk
[15:07] <rbasak> Daviey: those are the option extensions that pxelinux uses. There's also a non-extension that has been around a lot longer, used to specify the bootloader binary path. If I understand jhobbs correctly, u-boot has overloaded this field to mean the pxelinux.cfg path, as in the case of u-boot pxelinux emulation a bootloader binary download is not required
[15:54] <brendan0powers> jamespage: Hey, thanks for reviewing the package
[15:55] <jamespage> brendan0powers, np - thanks for your work on it
[15:55] <brendan0powers> jamespage: What's the process for making updates to the package
[15:55] <brendan0powers> If I needed to fix a bug?
[15:55] <jamespage> brendan0powers, OK so once the package is accepted into Ubuntu
[15:55] <jamespage> you should follow the ubuntu development process for packaging updates.
[15:56] <jamespage> brendan0powers, http://developer.ubuntu.com/packaging/html/
[16:17] <magma> ubuntu server shows wich users are using the machine and mem usage when you logon though ssh. Is it possible to do the same thing with the normal ubuntu?
[16:29] <uvirtbot`> New bug: #937040 in dovecot (main) "Duplicate ufw rules installed by dovecot packages" [Undecided,New] https://launchpad.net/bugs/937040
[16:31] <jamespage> rbasak, hows openmpi coming on?
[17:30] <koolhead17> hi all
[17:31] <Lazerath> Hello.
[17:31] <Lazerath> Would someone help me fix grub2 so i can boot my server
[17:44] <Lazerath> no one
[18:02] <rbasak> jamespage: wooo, my hello world fix fixed the FTBFS on mpi4py too
[18:03] <jamespage> rbasak, sweet!
[18:03] <jamespage> what was the issue?
[18:03] <hallyn> !ask
[18:03] <rbasak> Missing shared objects, and/or shared objects in the wrong packages. I think requirements for where they are have changed from 1.4 to 1.5. So I've rejigged it a bit.
[18:06] <rbasak> Finding out _which_ shared objects was non-trivial though. The library reads the directory and then searches that cached list, so strace is no help. Nor does it actually tell you what it can't find.
[19:09] <hallyn> stgraber: are you around by chance?
[19:12] <stgraber> hallyn: yep
[19:12] <stgraber> hallyn: you are the one with the public holiday, not me ;)
[19:25] <hallyn> stgraber: heh - never mind, sorry.  i answered my own question
[19:25] <hallyn> (was goign to ask if, when a boudn user's shell doesn't exist, we should fail, or install the needed package)
[19:25] <hallyn> (decided installing makes more sense)
[20:01] <hallyn> stgraber: well, except it's not so simple
[20:01] <hallyn> stgraber: what would you say is the easiest way to determine which package provides /bin/zsh?
[20:02] <hallyn> (which is a symlink to a /etc/alternatives/ file...)
[20:02] <stgraber> tricky
[20:03] <stgraber> hallyn: dpkg -S $(readlink -m $(which zsh)) | cut -d ':' -f1
[20:03] <stgraber> kind of ugly though ;)
[20:05] <hallyn> i don't understand readlink -m
[20:05] <hallyn> looking at the manpage.  not groking
[20:05] <hallyn> oh.  ok
[20:06] <hallyn> it'll work.  thanks
[20:25] <hallyn> stgraber: i wonder if we should be automatically adding the bound users to the sudo group
[20:25] <hallyn> or if ppl will be happy to log in as ubuntu to do that
[20:26] <stgraber> hallyn: not sure, I don't think I've ever used the bound stuff (I usually just use arkose for that)
[20:30] <hallyn> i don't want to make things more fragile (complicated) if people are ok without it...  hm...
[21:53] <Owner> hey whats the proper way to change network settings in unbuntu server
[21:54] <guntbert> Owner: edit /etc/networking/interfaces