[00:07] <tauren> I cannot perform any write operations on my drive anymore. This is a KVM host. Other KVM hosts on the server are operating properly.
[00:08] <tauren> I tried stopping the host (virsh shutdown hostname),and it stopped.
[00:09] <tauren> But now it won't start back up:  virsh start hostname says: "error: failed to create logfile /var/log/libvirt/qemu/hostname.log: Read-only file system
[00:09] <SpamapS> tauren: maybe you need to fsck it?
[00:10] <tauren> And trying "sudo fsck /dev/vol/partitionname" returns with "Superblock invalid"
[00:10] <tauren> fsck.ext2: Bad magic number in super-block while trying to open /dev/mapper/vol-hostname
[00:10] <tauren> The superblock could not be read or does not describe a correct ext2
[00:11] <tauren> SpamapS, that's what I'm thinking too. Any ideas why fsck would fail?
[00:11] <tauren> again, the rest of the server is working fine, so I think its related to just this one partition.
[00:12] <SpamapS> I've not done such things with kvm.. but I'd suspect there's a partition table on the lvm block device maybe?
[00:13] <tauren> hmm, maybe so. not sure how to check that.
[00:14] <tauren> SpamapS, yeah, from my notes I built it with a root, swap and /var partition
[00:15] <tauren> so do you know how I would run fsck on those partitions from the host system?
[00:18] <SpamapS> tauren: maybe kpartx?
[00:18] <tauren> ah, yes. it's been years since i used that. I'll have to figure it out again. thanks for the idea!
[00:19] <SpamapS> http://ubuntuforums.org/showthread.php?t=1262207
[00:19] <SpamapS> got the idea from that
[00:20] <tauren> thanks, looks like that will help
[00:52] <tauren> I have a remote ubuntu 9.10 server and I need to run fsck on its partitions. Do i have to physically go to the server with a recovery CD, or can I remotely change to single user mode, run fsck, then switch back to normal operation?
[01:02] <bogeyd6> tauren you can reboot into safe mode with ssh
[01:02] <bogeyd6> its on the server guide
[01:02] <bogeyd6> g2g
[01:03] <tauren> bogeyd6, I'd like to force fsck to run at boot, but my drive is read-only now, and I can't change /etc/default/rcS or touch /forcefsck
[01:07] <tauren> bogeyd6, sorry, misread your post. I'm not finding info in the server guide about doing that. Without write access on my drive, how would I remotely reboot into safe mode with ssh?
[02:14] <cybrocop> Hello all, I'm having libvirt problems. Can anybody help me?
[02:39] <jiboumans> cybrocop: best to just ask your question. The local expert is soren though who should be soon asleep right now  (he's in CEST)
[02:50] <cybrocop> jiboumans: Here is my issue http://open.eucalyptus.com/forum/libvirt-operation-failed-failed-retrieve-chardev-info-qemu-info-chardev#comment-11359
[02:51] <cybrocop> I'm also discussing this on #eucalyptus, but they are puzzled as well. This is I guess more of a libvirt issue than eucalyptus anyway
[03:13] <SamuelPeterson> I ran "sudo ifconfig eth2 hw ether 00:00:00:00:00:10" on a computer to spoof it's mac address, but it returned "siocsifhwaddr: too many open files in system." I had eth2 down before doing this. I ran ifconfig and the mac did change, but I can not connect to any networks.
[03:14] <SamuelPeterson> Would anyone happen to know what the issue is?
[03:42] <solarion> question: how do I change the upgrade policy to make sure to stya with LTS releases?
[03:42] <solarion> /etc/update-manager/release-upgrades Prompt?
[03:43] <ScottK> solarion: If you are on an LTS release you won't be automatically offered the chance to upgrade except to the next LTS.
[03:43] <solarion> aah, ok
[03:43]  * solarion is a fan of things Just Working. :)
[03:43] <solarion> in 2 years hopefully I'll be buying the 12-way Atom nettop. :)
[03:43] <solarion> with 128GB RAM
[03:43] <solarion> thanks
[03:43] <ScottK> solarion: If you want to check, that file should have "Prompt=lts"
[03:44] <solarion> yeah, that's what it has
[03:44] <ScottK> (if you're on an LTS)
[03:44] <solarion> yeah
[03:59] <Geekman> Hi everyone, need some help regarding RAID on Ubuntu server 10.04.  I'm working with Ubuntu's software RAID 1 right now, and it works fine, however I'm running into problems when needing to partition a fresh drive to add to the array for re-building -- having just removed one of the original drives.  I've used the command sfdisk -d /dev/sdb | sfdisk /dev/sda to partition the new drive and it works fine, but I'm doubting that's going to wo
[04:00] <Geekman> I'm trying to use fdisk to partition the new drive, but can't seem to create the extended partition as per the other drive -- the original drives were partitioned automatically by the installer.
[04:03] <Geekman> The following shows the partition table: http://pastebin.com/ajyDieuk -- I can create the partition sda1 just fine, but then when trying to create sda2 as an extended partition, cylinder 120675 is out of range I.  I am forced to enter 120676 but this means the second partition is slightly too small and hence rejected by MDADM.
[04:05] <Geekman> I've tried using sectors as the view in fdisk hoping that would be more specific, I was able to create the extended partition, but then had the same issue when trying to create the logical partition.  Is there some way I can force fdisk to do what I want?  Or is there a better utility for me to partition the drive like this?
[04:06] <Geekman> Or even, is there a hardware RAID controller that someone can suggest which is compatible with Ubuntu server?
[04:41] <bogeyd6> Geekman, almost any storage works card like a 400 or 420
[04:42] <Geekman> What Vendor are we talking about?  I'm fairly new to Hardware RAID.
[04:43] <Geekman> I was looking at MegaRAID which seems to be used by Dell servers, but it looks like the drivers only natively support SUSE/Redhat.
[04:54] <twb> I wouldn't trust anything with third-party drivers.
[04:54] <twb> (OTOH, lots of people seem to LIKE using ATI and Nvidia GPUs, so I'm pretty conservative.)
[04:58] <Geekman> I want to obviously try and ensure that at 2am, when we are going to have to do a re-build, that things just work.  Same goes for installing a new machine, because who knows if we'll have a time crush when doing so.  So I definitely don't want to go with anything that isn't isn't natively supported, if I can help it.
[04:59] <Geekman> ATI and NVidia have worked well for me in the past, but I also remember how much I had to tweak things back in 6.06, so I wouldn't use them in a production server environment.
[05:03] <Geekman> So if anyone knows of a RAID card they've used with Ubuntu that "just works", I'd appreciate it.
[05:04] <foxbuntu> Geekman, are you talking enterprise Cards or Desktop cards such as you mentioned?
[05:05] <Geekman> Well, depends on the cost.  But it's for our servers which are about to be put in some rack space in a data centre - so I'd assume enterprise.
[05:05] <foxbuntu> Geekman, i.e. Adaptec/LSI vs ATI/nVidia
[05:06] <Geekman> I had no idea that the MegaRAID stuff was for desktops.
[05:06] <foxbuntu> Geekman, I would assume fairly good results with many recent Adaptec/LSI Logic cards
[05:07] <Geekman> Thanks man, I'll look into that.
[05:07] <foxbuntu> np
[05:08] <foxbuntu> Geekman, I would suggest googling the model number of the card you look at prior to purchase just to double check
[05:08] <twb> Personally, I use md raid.
[05:08] <Geekman> twb, you mean Ubuntu's software MDADM RAID?
[05:09] <twb> It means I don't need to buy backup controllers, or worry about drivers.  The servers aren't doing anything for which I/O is critical -- just mail, file sharing, serving website, etc.
[05:09] <twb> Geekman: the kernel's md layer, yes.
[05:09] <Geekman> I've set that up at the moment, but as I stated above, I'm having issues with partitioning a clean drive to match the partition table correctly.
[05:10] <twb> Geekman: RAID1 or RAID5?
[05:10] <Geekman> Other than that, it does work fine.  But I/we just don't want to have to mess with it at 2 in the morning if it doesn't work just right.
[05:10] <Geekman> RAID1
[05:10] <twb> Are both disks from the same batch?
[05:11] <Geekman> The disks are identical - if I use sfdisk to clone the partition table onto the fresh drive it works, but trying to use fdisk to do it manually, I run into problems.
[05:11] <twb> Why are you using fdisk?  It's unmaintained.  Use something based on libparted -- parted, gparted, gnu fdisk
[05:12] <Geekman> Which, if I could trust that all drives in future were identical, using sfdisk would be fine - but I obviously can't gauruntee in future that a spare drive is going to be identical.
[05:12] <Geekman> fdisk is just what I've always used.
[05:12] <twb> Or: how, exactly, is fdisk failing
[05:12] <Geekman> http://pastebin.com/ajyDieuk
[05:12] <Geekman> That's the partition table on one of the original disks.
[05:13] <twb> What of it?
[05:13] <Geekman> As you can see, sdb2 starts on 120675, with sdb1 ending also on 120675
[05:13] <Geekman> fdisk won't let me do that.
[05:13] <Geekman> It says 120675 is out of range.
[05:14] <twb> Are you using the right units?
[05:14] <twb> Does /proc/partitions reports that both disks are the same size?
[05:14] <Geekman> I got around that by setting fdisk to use sectors instead, but ran into a similar issue when trying to setup the logical partition sdb5
[05:14] <twb> In the past I've found e.g. 250GB drives varying in size by a few MB.
[05:16] <Geekman> Just a sec, SSHing in to check /proc/partitions.
[05:16] <Geekman> I do recall sfdisk reporting that both drives had the same cylinder count, though.
[05:19] <Geekman> So you said fdisk is out of date, is there any of the utilities you listed that you can suggest to replace it?
[05:19] <Geekman> I always assumed gparted was GNOME parted and hence required a GUI - don't have one of those.
[05:26] <twb> C/H/S is totally meaningless in a modern system
[05:27] <twb> gparted is a GUI for libparted; parted is a CLI for it.
[05:27] <twb> The parted UI isn't very nice, so maybe you would prefer to boot a live CD or move the disks into another host to run gparted -- I don't really care.
[05:28] <Geekman> I probably would like that, but the server's don't have CD drives, so that would be a hassle.  I'll try doing it with parted and see how it goes.
[05:28] <Geekman> Thanks for all your help.
[05:35] <twb> I just think switching to hardware RAID because you can't partition a disk is a bit of a leap
[05:38] <Geekman> I think you're right, but nobody wants me to sink any more time into trying to get it to work ok - so we figured hardware RAID would work better without a fuss.
[05:38] <Geekman> But I'll take another look when I get a spare moment.
[05:41] <twb> BTW, for RAID1 you don't *need* each node to be exactly the same size
[05:41] <twb> md will just use the smallest
[05:49] <Geekman> I know, but the way fdisk was working for me, the swap partition - sda5 always ended up being 1 cylinder too small or something.
[05:49] <Geekman> Oh and for the record, both the drives are reported as being the same size in /proc/partitions.
[05:54] <twb> There's usually no point RAIDing swap
[05:54] <twb> Especially RAID0 -- linux swap is automatically striped
[05:58] <Geekman> So, if one of the disks die at runtime, it's not going to corrupt the swap partition?  How can that be if it's not mirrored?
[05:59] <Geekman> .../part of the RAID array.
[06:00] <twb> I suppose so
[06:01] <twb> To be honest, since 2.6 came out, I have lost faith in swap *at all*
[06:01] <Geekman> Heh.
[06:01] <Geekman> In terms of what?  Application crashes when they fall back to swap?
[06:02] <Geekman> The only thing I want to avoid is having the OS crash because something's using swap and one of the drives has died, which has destroyed the swap.
[06:02] <twb> AFAICT when my hosts start swapping, they're too fucked to respond to fork a "telinit 6"
[06:02] <Geekman> I figure if the swap partition is part of the array, it'll prevent that.
[06:02] <Geekman> Well yeah that's true.
[06:02] <twb> So I might as well not have swap at all
[06:03] <Geekman> But just because you're swapping doesn't necessarily mean that you're swapping constantly, in my experience.
[06:25] <SpamapS> swap is *completely* worthless with RAM prices of today
[06:25] <SpamapS> there's no server app that does well when memory page fetches take longer than 1ms
[06:26] <SpamapS> even 1ms is pushing it quite a bit
[06:27] <SpamapS> Geekman: I prefer to think ahead.. build a system to run at 30% capacity 95% of the time.. and the other 5%, never over 50% capacity. If you breach either of those thresholds, the system cannot stand up when the business becomes a success.
[06:27] <twb> swap really means "I need more RAM than I can afford"
[06:28] <twb> And as you say, RAM is pretty damn cheap now
[06:31] <SpamapS> http://www.crucial.com/store/mpartspecs.aspx?mtbpoid=FF101E53A5CA7304
[06:31] <SpamapS> 16G for $1400 US ...
[06:31] <SpamapS> and thats 2x8G sticks
[06:33] <Geekman> How have you guys found Ubuntu to handle 32-bit apps in 64-bit?  The decision was made to stick with 32-bit for now as it's only going to be a web server, we don't think any of our core servers should really need more than 4GB.
[06:33] <twb> I don't run any 32-bit apps.
[06:33] <Geekman> But personally I'm a little uncomfortable with being limited to 4GB.
[06:34] <Geekman> You've never run into a situation where you've had to search the net because some app doesn't natively support 64-bit well?
[06:34] <twb> There are only three cases where biarch is useful: proprietary software, compilers that only support -m32, and development jails.
[06:34] <Geekman> I don't mind having to do that myself - it is Linux - but I guess business is business and we can't be having to do that if we need something running urgently.
[06:34] <SpamapS> Geekman: you're not limited to 4G
[06:35] <twb> scsh and scheme48 fall into the second category, and on some hosts I use the third.
[06:35] <twb> SpamapS: yeah, but PAE is such a kludge
[06:35] <SpamapS> kludge or not..
[06:35] <SpamapS> its better than replacing entire working systems
[06:35] <twb> Geekman: IMO your "rule of thumb" should be to use 64-bit
[06:35] <SpamapS> nobody should *design* to use PAE
[06:36] <SpamapS> but its certainly a decent plan B
[06:36] <SpamapS> twb: depends on how you want to scale
[06:36] <Geekman> PAE, I assume that's what I was reading about regarding the ability to use more than 4GB under 32-bit using some kind of emulation, albiet more slowly?
[06:36] <twb> SpamapS: granted
[06:37] <SpamapS> twb: 32-bit has some advantages when you need crazy high concurrency
[06:37] <twb> 64-bit also means you can simply assume things like SSE2
[06:37] <twb> That's less of a problem on Ubuntu, where i386 really means i686 nowadays :-/
[06:37] <SpamapS> Geekman: its a small memory access trade off that allows the OS as a whole to use more than 4G, but processes still only have a limited 32-bit address space.
[06:38] <Geekman> I see.
[06:38] <SpamapS> twb: yeah I think as of Maverick i586 is dead in ubuntu
[06:38] <twb> SpamapS: I saw the GCC changelog go through
[06:39] <SpamapS> I'm sure someone somewhere is cursing their Geode SOC design. ;)
[06:39] <foxbuntu> Geekman, PAE can break some apps too, although its usually poorly written apps misusing memory marshaling and threading
[06:39] <Geekman> Well in our case I think that if we're hitting a wall where we need more than 4GB, it's probably time to start looking at a higher spec server overall.  I don't have the specs with me, but they're not high end.
[06:39] <twb> Fortunately my Atom not-quite-SOC is x86-64.
[06:39] <SpamapS> foxbuntu: PAE *exposes* buggy apps... it doesn't break them. :)
[06:40] <foxbuntu> SpamapS, thats also a wa of looking at it
[06:40] <foxbuntu> way*
[06:40] <twb> foxbuntu: if apps were well written they wouldn't need >4 GB!
[06:40]  * twb kicks the entire java community
[06:40] <foxbuntu> twb, lmao
[06:40] <foxbuntu> twb, im not sure thats quite the case with things like Oracle
[06:40] <twb> Yeah, I know
[06:40] <SpamapS> twb: SOLR is the first java server app that I trust completely... somehow, they got it right, a nd it doesn't even eat *ALL* of your RAM. ;)
[06:41] <twb> I'm just sick of those kids with the mindset that making everyone buy more RAM is cheaper than their time to solve a problem properly.
[06:41] <foxbuntu> twb, agreed
[06:41] <SpamapS> twb: PHP devs have the same mindset. ;)
[06:42] <foxbuntu> twb, many apps do not/should not require that much RAM
[06:42] <twb> Especially when the first-year java lecturers explicitly TEACH that mindset >:-/
[06:42] <twb> foxbuntu: Emacs is only "Eight Megabytes and Constantly Swapping"
[06:43] <SpamapS> yeah they'll say things like 'RAM is 500 times faster than disk, so you should never use disk unless you absolutely must'
[06:43] <twb> SpamapS: ugh.  Block caching is the kernel's job, not the apps
[06:43] <SpamapS> its ok for an app to swap a little out.. I don't use the calendar in evolution, so it probably drops any bits I'm not using in swap. Its just not ok to PLAN for that.
[06:43] <foxbuntu> twb, unfortunately I work in the windows world (as many do :( ), and I have seen some really bad .Net apps (yeah yeah, bear with me here) that will crush a system under load to the point where x64 can even save the system
[06:44] <SpamapS> twb: tell that to InnoDB. :)
[06:44] <twb> Or like the Zimbra support company that tells us "you should deploy zimbra on its own host.  Zimbra assumes that the host its on is dedicated to JUST running zimbra"
[06:44] <foxbuntu> twb, more java!
[06:44] <foxbuntu> twb, feed the monster
[06:44] <twb> Well, Zimbra is mainly postfix and openldap.  The java is just their shitty "value added" glue
[06:45] <twb> Flipping calendaring
[06:45] <foxbuntu> twb, yeah, tis true, but the java magic is the main sludge in the system
[06:45] <twb> Granted.
[06:45] <foxbuntu> wtf?!
[06:45] <foxbuntu> speaking of witch
[06:46] <foxbuntu> which rather
[06:46] <foxbuntu> maybe both
[06:46] <foxbuntu> lol
[06:46] <foxbuntu> trying to convert a VM from Xen to VMWare and it keeps imploding
[06:46] <twb> foxbuntu: qemu-img convert ? ;-)
[06:46] <foxbuntu> twb, its windows
[06:46] <foxbuntu> :(
[06:46] <twb> You poor bastard
[06:47] <foxbuntu> twb, lol
[06:47] <foxbuntu> something like that
[06:48] <foxbuntu> doesnt help that the VM host that Xen is on is a turd machine
[06:48] <twb> Tell me about it
[06:48] <foxbuntu> the client expected miracles on an array of 5400 SATA disk
[06:49] <twb> We have two hosts that can run KVM, but I can't use either.  One is running a Windows VM to talk to the tax office (sigh), and the other was stolen to run a flipping pppoe session (FFS!)
[06:49] <twb> "modprobe kvm" on the vmware/openvz host causes the entire system to hard hang
[06:49] <foxbuntu> pppoe, ugh
[06:50] <foxbuntu> awesome
[06:50] <twb> foxbuntu: exactly.  The pppoe could be done by a 200MHz MIPS system -- it doesn't need a 3GHz quad-core
[06:50] <foxbuntu> twb, lol
[06:50] <twb> But that's what was sitting on the test bench when $coworker needed a host
[06:50] <twb> And we can't take it out because it's mission critical
[06:50] <foxbuntu> twb, I am swaping this server over to a box with SAS 10k Spin and Dual 6-Core Opterons
[06:51] <twb> Nice
[06:51] <foxbuntu> yeah, gonna run much better
[06:51] <twb> Hopefully not running 2k8, tho
[06:51] <foxbuntu> not to mention the hypervisior wont suck either now
[06:51] <twb> Or whatever the hell is the current Windows Server
[06:51] <twb> Oh, right, you said vmware, so it's probably esxi
[06:52] <foxbuntu> well the guests at Win 2k8, but the host is VMWare ESXi
[06:52] <foxbuntu> yea
[06:52] <twb> I never tried that, only "vmware server".  Gods, that's an awful piece of shit
[06:52] <foxbuntu> they only have the one box, so no reason to buy ESX
[06:52] <foxbuntu> VMWare server == full blow job
[06:52] <foxbuntu> (and not the good kind)
[06:53] <foxbuntu> ESX/ESXi are really nice
[06:53] <twb> 2.x wants you to run tomcat and a client-side java app just to talk RFB
[06:53] <foxbuntu> lol
[06:54] <foxbuntu> if you are gonna run a bunch of *nix guests stick to KVM (or similar) but if there are any windows hosts get ESXi
[06:54] <foxbuntu> er guests that is
[06:55] <twb> I like qemu a lot more because I can run it entirely in userspace (including userspace networking), and it's easy to get serial or the 80x25 VGA console inside your xterm -- nice and fast, rasterizing the ascii
[06:55] <twb> *no rasterizing
[06:55] <twb> (kvm just being fast qemu)
[06:55] <foxbuntu> yeah
[06:56] <twb> All my $coworkers love vbox because it has a GUI and generally targets desktop-in-desktop virtualization :-/
[06:56] <foxbuntu> yea
[06:56] <twb> At least qemu 0.12 can read/write to vbox images natively
[06:56] <foxbuntu> Im thinking more along the lines of bare metal hypervisor
[06:56] <foxbuntu> like ESXi
[06:57] <foxbuntu> though kvm really isnt either
[06:57] <foxbuntu> I am just not a huge fan of the userspace VM hosts
[06:57] <foxbuntu> for production systems
[06:57] <twb> Yeah, I agree
[06:58] <twb> I have lots of ephemeral VMs for scratch work, so userspace is simple and convenient for me
[06:58] <foxbuntu> indeed
[06:58] <twb> For permanent systems my preference would be for jails, or failing that a lightweight hypervisor (like xen)
[06:58] <foxbuntu> yea
[06:59] <foxbuntu> is the ubuntu xen kernel still around these days?
[06:59] <foxbuntu> I havent looked in a long time
[06:59] <twb> No idea; I'm still on 8.04
[06:59] <foxbuntu> ah
[06:59] <foxbuntu> I run dev on my laptop
[07:00] <foxbuntu> always dive into the ver. around A2
[07:00] <foxbuntu> mostly because its a PITA to deal with VMs for my little bit of devel
[07:00] <twb> I run sid on my laptop because I'm not a GUI weenie :-P
[07:00] <foxbuntu> lmao
[07:01] <foxbuntu> while I *could* run sid, I like things that *just work* I spend too much time fixing c*&$ already
[07:02] <twb> AFAICT non-LTS releases are as flaky as debian/testing
[07:02] <foxbuntu> I learned on gentoo a long time ago, and that was enough of the whole build it when you need it thing for me
[07:02] <jmazaredo> i need to load balance 2 backend servers (web) what could be the best on ubuntu
[07:03] <twb> e.g. "oh hey, we moved openldap config into the database against upstream's advice.  HAND!"
[07:03] <twb> I was running testing with a handful of sid pins, but it was too reliable and  boring. :-)
[07:04] <foxbuntu> jmazaredo, define what you see as load balanced
[07:04] <foxbuntu> twb, lol, ok you got me on that one, things do like to change
[07:04] <foxbuntu> ...in Ubuntu
[07:04] <foxbuntu> ...allot
[07:05] <foxbuntu> not that I have anything against it
[07:05] <twb> More to the point, they change without much discussion or testing, AFAICT
[07:05] <foxbuntu> twb, no, not true
[07:05] <twb> Which is actually convenient for me, because it means that Ubuntu users are on the front line, testing stupid ideas before they hit Debian/testing
[07:05] <jmazaredo> balancing http request
[07:05] <foxbuntu> twb, people assume that but lots of those (most) decisions get made in UDS
[07:07] <foxbuntu> jmazaredo, go on, do you mean one request comes in and the next goes to another server
[07:07] <foxbuntu> jmazaredo, or do you mean the load gets calculated and weighted to pick the correct server
[07:07] <jmazaredo> request ----> goest ro balancer ---> spreads to two servers
[07:08] <jmazaredo> like balanceNG and nginx
[07:08] <jmazaredo> free balanceNG only has 1, Niginx need to configure every balance i do
[07:09] <jmazaredo> can iptables do the trick?
[07:09] <foxbuntu> jmazaredo, dont think for what you want it will
[07:10] <foxbuntu> jmazaredo, you might be able to do it with squid
[07:15] <twb> netfilter/iproute can perform some kinds of load balancing, yes.
[07:16] <twb> An application-level (e.g. nginx, pound, haproxy) load balancer is more conventional.
[07:21] <|corpse|> my PC must hate me, iv been trying to set up a file srver for 5 days now and i can not, for the life of me, get an istallation to work
[07:42] <killown> is there any option to encrypt entire / during ubuntu-server install?
[07:44] <twb> Block-level encryption is possible
[07:45] <twb> I don't know the details, especially I don't know if the release you're using supports it directly in the d-i UI
[07:45] <killown> twb, i need do something to not allow peoples of chroot my / partition from a live distro
[07:45] <twb> Do you *really* want to encrypt the OS filesystem of a server?
[07:46] <killown> twb, there is another way to block chroot instead encrypt?
[07:46] <twb> Why do you want that?
[07:47] <twb> What attack are you trying to prevent?
[07:47] <killown> twb, i need protect the information into my system
[07:47] <twb> What information?  The user data, or the OS?
[07:47] <killown> i am doing a proxy solution
[07:48] <killown> i thinks /etc /var
[07:48] <killown> think*
[07:48] <twb> Why?
[07:49] <killown> twb, to not allow peoples to copy conf content...
[07:49] <killown> /var for openldap
[07:50] <twb> What configuration could you possibly have that needs to be protected against physical attack?
[07:50] <killown> and etc for confs
[07:51] <killown> twb, i am just figure out anyway to protect my conf content
[07:58] <killown> twb, think about, you have a proxy solution for a vpn with 15 networks, so if you configured the master and the first network slave, why would anyone will need of you for finish the job??
[07:59] <twb> If your core routers aren't behind a locked door, you're probably screwed
[08:02] <killown> twb, encrypt etc and var is not a good idea?
[08:06] <killown> twb, hello
[08:07] <twb> Sorry, I don't have time to help you now
[08:08] <killown> change the phrase for: sorry, i don't know about that.
[08:11] <killown> twb, you are good for questions, but answers....
[09:05] <RoyK> any idea how I can find out which swap device this is about? http://pastebin.com/Hvmg4SUY
[09:08] <cemc> RoyK: you have more swap devices?
[09:08] <cemc> I mean more than one?
[09:33] <RoyK> cemc: ubuntu 9.04 standard, somehow - /dev/ramzswap0 and a partition
[09:34] <RoyK> swapping to compressed memory before using the disk
[09:36] <cemc> huh? :)
[09:36] <cemc> swapping to where?
[09:36] <twb> RoyK: /proc/swaps?
[09:37] <twb> compcache is ramzswap, so it's clearly not talking about /dev/sda2 or whatever
[09:38] <twb> The 251 and 0 looks like major and minor block device numbers, so you could also ls -l /dev and look for 251,0
[09:40] <RoyK> from /proc/devices 251 ramzswap0
[09:40] <RoyK> wierd
[09:42] <twb> Whoops, I can't modprobe ramzswap until I reboot.
[09:42] <twb> Stupid ricer kernel...
[10:31] <qman__> killown, encrypting your root directory won't help against online attacks of any kind, as the files have to be readable by the system in order for it to work
[10:31] <qman__> it would only offer protection against attacks from physical access
[10:32] <qman__> but, if you don't have physical security, you're basically screwed anyway
[10:32] <qman__> I suggest you get a locked room
[10:32] <qman__> it's a great option for laptops, but not so much for servers
[10:34] <xperia> does anybody know how can i reset the bios password on a hp proliant ML 530 G2
[10:35] <qman__> probably a better question for google, that sort of thing tends to get very hardware specific
[10:35] <qman__> if it's not a simple jumper setting, anyway
[10:37] <qman__> perhaps searching for a manual for that server would help
[10:39] <xperia> qman: thanls for the info. looking at google since hours. thinked maybe some person is here with experience background about this topic
[10:39] <qman__> yeah, sorry, never used one of those
[10:39] <qman__> I've got a DL380 but I've never had to reset the BIOS
[10:41] <xperia> qman: me stuipid have set extra a password for the bios and forgot it now. need to enable the netwrok card for installing ubuntu-server
[10:44] <cemc> RoyK: is there any howto on setting that thing up?
[10:56] <RoyK> cemc: setup what?
[11:01] <xperia> qman: i was able to reset the bios password with the smartstart cd. needed to erase everything.
[11:02] <qman__> ah, figured it'd take more than pulling the battery
[11:02] <twb> Usually there's a jumper on the motherboard
[11:02] <twb> Which is why physical security is critical
[11:03] <twb> Jumper the bios password away, then change the boot device, then you have root access to the hard disks (unless there's block-level encryption).
[11:07] <cemc> RoyK: nvr mind, found it
[11:31] <Pirate_Hunter> I keep getting 404 error every time I try to connect to localhost through lightpd on  port 8080. lighttpd restarts without a problem so I cannot understand why I can't connect through its port can anyone suggest something
[11:39] <cloakable> Pirate_Hunter: 404... not found. Check to see if your docroot exists.
[11:50] <xperia> qman_: very cool ! ubuntu server LTS Edition runs now on my professional server proliant ML530G2 :-)
[11:51] <xperia> qman_: is there any possibility to controll the speed of the main fans. they run allways full and poduce a lot of noice. system temperature is 8 Celcius
[11:51] <xperia> so the fans dont need to run 100% when the temperature is such low
[12:04] <Pirate_Hunter> maybe im doing something wrong but if I wanted to access lighttpd through my browser wouldn't it be something like httP://localhost:8080?
[12:04] <lenios> xperia, aptitude show cpufreqd
[12:05] <lenios> or maybe with acpi
[12:05] <cloakable> Pirate_Hunter: 404 means "File Not Found" lighttpd is not finding what you pointed it at.
[12:06] <lenios> aptitude show lm-sensors too
[12:06] <lenios> to change fan speed
[12:11] <twb> Bugger the cpufreqd package.
[12:12] <twb> Current kernels default to the kernel-space ondemand governor, which is a reasonable default.
[12:13] <twb> xperia: fan control depends on what kind of fan setup you have.  Sometimes it's uncontrolled, sometimes it's controlled by firmware, sometimes it's controlled by the kernel.
[12:14] <twb> xperia: if the fan connects to the motherboard with only two pins, you're probably screwed.
[12:14] <twb> xperia: otherwise you could find /sys/ -iname '*fan*'
[12:30] <jo-erlend> how do I find all files and folders that have been created or changed during the last 30 minutes?
[12:38] <twb> find -mtime -1 is "last 24 hours"
[12:38] <twb> Dunno if it takes other units
[12:38] <twb> Ah, -mmin 30
[12:39] <pmatulis> jo-erlend: man find
[12:39] <jo-erlend> -mmin did the trick. Thanks. :)
[12:39] <twb> Yeah, manpage is your friend
[12:40] <jo-erlend> of course I read it before I asked. I didn't understand it, that's all. :)
[13:01] <jo-erlend> I'm writing a script to automate creation of a virtual machine I've created. I need to reconfigure all the services before the first boot. What I'm doing now, is to simply install the services one by one and noticing which files change as I configure them, and how they change.
[13:01] <jo-erlend> any better ideas?
[13:02] <xperia> twb: lenios: thanks for the tips. will report back what i have for a fan
[13:36] <sommer> morning
[13:39] <bogeyd6> When I put username=blah,password=blah in my fstab file for mounting a samba share in my log it says it is connecting as user nobody. How can this be?
[14:27] <rgreening> kirkland: hey. In the upstart job for libvirt-bin, when it shut's down, are the currently running VM's shutdown or is there a way to make them come down gracefully?
[14:27] <rgreening> rather than manually going to each VM and shutting down?
[14:44] <dantti> I'm looking for a web tool to handle virtual machines (KVM or Xen), but one important thing would be that it has webbrowser plugins like VmServer (since I'm moving from it), does some one knows a tool that has something like that?
[14:44] <dantti> I'm trying openQRM but it's not populating my postgres db.. :/
[14:52] <fairview_> Thank you sommer for creating the ubuntu server guide.  very helpful
[14:53] <sommer> fairview_: welcome :-) glad it helps
[14:54] <fairview_> I have lots of learning to do so it will
[15:03] <Knifa> PHP is disabled by default for Apache 2 user directories. Despite following the instructions to re-enable it again, PHP files still don't work in user directories. I followed these instructions: https://wiki.ubuntu.com/UserDirectoryPHP
[15:03] <Knifa> Am I missing anything out?
[15:06] <Knifa> It tries to make me download the file instead of running it.
[15:06] <Knifa> But PHP works fine in the root web folder.
[15:07] <ivoks> did you restart apache?
[15:08] <Knifa> yes
[15:08] <Knifa> about eighty times.
[15:08] <Knifa> but it doesn't make a difference
[15:09] <ivoks> and it offers downloading .php file
[15:09] <ivoks> ?
[15:09] <Knifa> yes
[15:09] <Knifa> rather than processing it
[15:09] <Knifa> but it processes it on /var/www
[15:09] <ivoks> which ubuntu?
[15:10] <ivoks> version
[15:10] <Knifa> 10.04
[15:11] <ivoks> paste /etc/apache2/mods-enables/php5.conf to pastebin
[15:12] <ivoks> enabled
[15:12] <Knifa> http://pastebin.com/KWs66qyd
[15:13] <Knifa> I tried setting it it to On with it uncommented
[15:13] <Knifa> but that didn't help either.
[15:13] <ivoks> do you have .htaccess file in public_html/?
[15:14] <Knifa> nope
[15:14] <ivoks> ok, create a file names
[15:14] <ivoks> named
[15:14] <ivoks> rnd246.php in public_html
[15:14] <ivoks> put in it:
[15:15] <ivoks> <? phpinfo(); ?>
[15:15] <ivoks> and try accessing it
[15:15] <Knifa> it tries to download it
[15:15] <ivoks> hm
[15:16] <Knifa> really no idea what's going on
[15:16] <Knifa> it should work
[15:17] <ivoks> let me try it
[15:17] <ivoks> works
[15:17] <Knifa> kk
[15:17] <Knifa> man what :[
[15:18] <ivoks> sudo service apache2 stop
[15:18] <ivoks> sudo killall -9 apache2
[15:18] <ivoks> sudo service apache2 start
[15:18] <Knifa> i tried rebooting the server for the sake of it (it is not an important machine)
[15:18] <Knifa> so apache should have been proeprly restarted
[15:18] <ivoks> and access that file with w3m
[15:19] <LinuxAdmin> hi  guys
[15:19] <ivoks> Knifa: do you have additional configuration in conf.d/ or sites-enabled/?
[15:19] <Knifa> nope
[15:19] <Knifa> this is a fresh apache2 install
[15:19] <Knifa> i had lighttpd before.
[15:20] <ivoks> prefork?
[15:20] <Knifa> i have no idea
[15:20] <LinuxAdmin> I'm trying to use ocfs2 on top of drbd. everything went well on first node but when I tryed to mount the partition on the second node I get this error:
[15:20] <LinuxAdmin> mount.ocfs2: I/O error on channel while opening  device /dev/drbd0
[15:21] <ivoks> Knifa: dpkg -l | grep apache2-mpm
[15:21] <LinuxAdmin> can someone help me?
[15:21] <Knifa> yeah it's prefork
[15:21] <ivoks> LinuxAdmin: is drbd master/master?
[15:21] <ivoks> er... primary/primary
[15:21] <LinuxAdmin> I think so. Where can i see that?
[15:22] <ivoks> cat /proc/drbd
[15:22] <Knifa> ...
[15:22] <Knifa> I tried reinstalling the php5 module for apache2
[15:22] <Knifa> now it spontanously does not work at all.
[15:22] <Knifa> anywhere.
[15:22] <ivoks> Knifa: libapache2-mod-php5 is the name of the package
[15:22] <LinuxAdmin> ro:Secondary/Primary
[15:23] <Knifa> yeah
[15:23] <Knifa> that's the one.
[15:23] <LinuxAdmin> where can I define this?
[15:23] <ivoks> LinuxAdmin: there you go, second node isn't primary
[15:23] <ivoks> LinuxAdmin: and cannot write to device
[15:23] <ivoks> Knifa: check /var/log/apache2/error.log
[15:24] <LinuxAdmin> I wan to have the capability to write to the shared storage from both of nodes, that's why I created an ocfs2 clustering file system
[15:24] <ivoks> LinuxAdmin: drbdadm primary <resource>
[15:24] <ivoks> LinuxAdmin: but you have to set up drbd to work in primary/primary node first
[15:24] <ivoks> LinuxAdmin: ocfs is just a filesystem
[15:25] <LinuxAdmin> ok
[15:25] <ivoks> LinuxAdmin: drbd is the device
[15:25] <LinuxAdmin> yeah
[15:25] <ivoks> have you followed any documentation?
[15:25] <LinuxAdmin> yes
[15:25] <ivoks> then you added needed stuff to drbd's config?
[15:26] <LinuxAdmin> ubuntu server official book, and ocfs2 how to
[15:26] <LinuxAdmin> I configured drbd and it synchronized well in both hosts
[15:27] <Knifa> ivoks access logs for php files in user directories aren't showing up
[15:27] <Knifa> but there are no errors.
[15:27] <ivoks> LinuxAdmin: but did you add allow-two-primaries to drbd's config?
[15:27] <ivoks> and become-primary-on both
[15:27] <LinuxAdmin> no
[15:28] <ivoks> LinuxAdmin: configs for drbd for ext3 and ocfs2 aren't the same :)
[15:28] <ivoks> LinuxAdmin: https://wiki.ubuntu.com/ClusterStack/LucidTesting#Pacemaker,%20drbd8%20and%20OCFS2%20or%20GFS2
[15:28] <LinuxAdmin> should I configure it on common section?
[15:28] <ivoks> check '3. Configure drbd'
[15:28] <Knifa> ivoks sigh. turns out firefox is caching the fucking download mimetype thing.
[15:28] <Knifa> ugh.
[15:29] <ivoks> Knifa: i said use w3m
[15:29] <ivoks> Knifa: for a reason :)
[15:29] <Knifa> oops :p
[15:29] <Knifa> thanks :3
[15:29] <jo-erlend> this was informative.. I didn't know drbd allowed both hosts to be primary at the same time.
[15:31] <ivoks> jo-erlend: it does; but then you need clustered file system
[15:31] <ivoks> jo-erlend: and stonith, and lrm and everything
[15:31] <jo-erlend> then it will simply work both ways?
[15:32] <jo-erlend> I'm not familiar with lrm. What's that?
[15:33] <LinuxAdmin> ivoks, do I have to configure pacemaker?
[15:33] <LinuxAdmin> can I avoid it?
[15:33] <ivoks> LinuxAdmin: you don't have to, but it might be a smart thing to do :)
[15:33] <LinuxAdmin> can you please tell me what pacemaker do?
[15:34] <ivoks> uh
[15:34] <ivoks> http://clusterlabs.org/
[15:34] <LinuxAdmin> it's another tool like heartbeat?
[15:34] <LinuxAdmin> ok
[15:34] <ivoks> it's not another
[15:34] <LinuxAdmin> I go read there
[15:34] <ivoks> it replaces heartbeat
[15:34] <LinuxAdmin> ok
[15:47] <cdubya> We have a Domain Controller setup with Exchange on the same machine. I'm wondering if there is a good method of setting up a fallback for a failure to that machine, but obviously Exchange is an issue. Are there any good methods to resolve something like this using ubuntu?
[16:15] <tschundeee> how can I add the usergroup www-data?
[16:15] <tschundeee> I havent installed nginx via apt-get
[16:16] <ivoks> addgroup --system www-data
[16:16] <tschundeee> instead i compiled it myself... so i need to set its default user and chown the www-data folder to this user and its group... so I need to know if the group exists
[16:17] <tschundeee> ivoks: thx man :)
[16:19] <tschundeee> btw... where is the equivalent file to /etc/passwd?
[16:19] <tschundeee> for groups?
[16:19] <ivoks> guess :)
[16:19] <tschundeee> hmm
[16:19] <ivoks> what would be a first thing you would think of?
[16:20] <ivoks> /etc/webserver?
[16:20] <tschundeee> etc/groups
[16:20] <ivoks> :D
[16:20] <ivoks> there you go
[16:20] <ivoks> without s
[16:20] <tschundeee> okay sry
[16:20] <tschundeee> i tried to edit with s and so I found nothing lol
[16:22] <ivoks> tab is your friend
[16:40] <ivoks> anyone knows where are defined options libvirt passes to kvm?
[16:40] <ivoks> how does it generates command line from xml
[17:56] <MTecknology> Do you guys know if it's possible to setup the user/group that's accessing a samba share on the filesystem?
[17:58] <MTecknology> or does samba default to existing?
[18:02] <webPragmatist> is there a sample partition schema available for using drbd
[18:02] <webPragmatist> Is it common to put /var /home /etc on these?
[18:02] <MTecknology> webPragmatist: I'd exclude var
[18:03] <MTecknology> webPragmatist: I'm going to play with that nifty little thing eventually.. but it mostly just depends on what needs to be sync'ed - probably /home and /etc only - maybe /var/www or something specific in /var - you don't want to sync /var/logs - that could get ugly
[18:04] <webPragmatist> right hrm
[18:04] <webPragmatist> MTecknology: okay my next question is how am i mounting each of these specific folders to this separate partition
[18:04] <MTecknology> webPragmatist: hm?
[18:05] <MTecknology> webPragmatist: just make one partition for each thing you want to sync
[18:05] <MTecknology> probably easiest
[18:05] <webPragmatist> so a bunch of logical volumes?
[18:06] <MTecknology> ya - probably easiest
[18:06] <webPragmatist> seems kinda hard to tell what the partition size should be for each
[18:06] <MTecknology> beyond that though - i can't tell you much because i haven't done it yet
[18:07] <ivoks> i responded on #ubuntu-ha :)
[18:07] <webPragmatist> oh okay
[18:13] <R0d> Hi all
[18:14] <R0d> does anyone know how can I configure Ubuntu to route trafic between 2 private nets, like 192.168.0.X and 192.168.1.X
[18:14] <ivoks> route add net 192.168.0.0/24 dev eth0
[18:14] <ivoks> route add net 192.168.1.0/24 dev eth1
[18:15] <ivoks> or -net
[18:15] <ivoks> -net :)
[18:15] <mcas> ivoks: echo "1" > /proc/sys/net/ipv4/ip_forward
[18:17] <MTecknology> Is it possible to have a Win7 user not have to type in credentials to log into a share? (Their system account) I have this in the config http://dpaste.com/199062/  I want adam to be able to seamlessly connect to the share.
[18:17] <R0d> ivoks, and with these 3 steps (route add & ip_forward), Clients from 0.X could access to clients from 1.X and vice versa?
[18:18] <zul> hey mathiaz
[18:18] <ivoks> mcas: /etc/sysctl.conf
[18:18] <mcas> ivoks: right
[18:19] <ivoks> R0d: if your server is their router, yes
[18:20] <R0d> ok, perfect :)
[18:21] <hackeron> hey, now that ubuntu has done away with hal - I'm trying to use gudev to get information about devices - I tried query_by_subsystem("sound") under gudev Client, but how do I filter just the capture devices from the list?
[18:22] <zul> hggdh: ping
[18:22] <hggdh> zul: pong
[18:22] <ivoks> hackeron: udevadm monitor
[18:22] <R0d> and... If after that, there is a Gateway to internet? For Example... [A Clients] - [Ubuntu Router] - [B Clients] - [Gateway] - INET... Being A= 1.X, B= 0.X and Gateway LAN IP= 0.1
[18:22] <zul> hggdh: can you nominate 551097 for lucid for me please?
[18:22] <hggdh> bug 551097
[18:23] <hackeron> ivoks: so how do I get alsa capture devices with udevadm?
[18:23] <ivoks> R0d: than ubuntu router will have default route to 0.1
[18:23] <hggdh> zul: you actually want me to nominate 551130, correct? 551097 is a dup of 551130...
[18:23] <zul> hggdh: yes
[18:23] <ivoks> hackeron: i'm afraid i don't understand you :)
[18:24] <R0d> ivoks, Wow... Awesome! Thank you very much!!
[18:24] <hackeron> ivoks: something similar to: hal-find-by-capability --capability alsa | grep capture
[18:24] <ivoks> alsa, hal... what are these words? :)
[18:24] <zul> hggdh: thanks in advance
[18:25] <hggdh> zul: done, and you are welcome
[18:25] <ivoks> hackeron: don't know
[18:26] <hackeron> anyone? - before lucid, I used to do hal-find-by-capability --capability alsa | grep capture -- what do I do now to get alsa capture devices?
[18:31] <xperia> hello to all. i was able sucessfull to install ubuntu-server LTS on a proliant ML530 G2 Server. it works everything like it should beside one small problem. everytime i poweroff the server and start it from new it hangs on the boot proccess asking me if i want to go into bios settings or if i want to continue. as long as keyboard and Monitor are attached to the Server this is no Problem but...
[18:31] <xperia> ...if this devices dont exist i cant boot really in Ubuntu. anybody know how to solve that ?
[18:33] <SpamapS> xperia: this actually sounds more like a BIOS issue
[18:33] <xperia> yeah it is na
[18:33] <xperia> a bios thing
[18:33] <SpamapS> xperia: you saying when you halt windows/redhat/etc it doens't do that?
[18:34] <darkscrypt> hey. I did not install lamp in the installer when it came up
[18:34] <darkscrypt> but now i want it
[18:34] <darkscrypt> how do i install the lamp package with apt-get
[18:34] <darkscrypt> is there like a group install?
[18:35] <ivoks> sudo tasksel install lamp-server
[18:35] <ivoks> is that right?
[18:35] <ivoks> it is :)
[18:36] <xperia> Spamaps: yes if i sutdown the server with poweroff and after a while power on the server i get three lines on the server that ask me if i want to continue or to go in to bios. normally after 30 seconds the server boot direct in to ubuntu but in my case it just sit there without doing anything. now i have changed a setting in the bios and need to look if this will help to resolve the Problem.
[18:36] <xperia> booting ubuntu is no problem
[18:37] <darkscrypt> .thanks
[18:37] <hackeron> anyone? - before lucid, I used to do hal-find-by-capability --capability alsa | grep capture -- what do I do now to get alsa capture devices?
[18:37] <darkscrypt> never used tasksel before is that specific to ubuntu?
[18:37] <xperia> the thing is that it dont do it automatic after 30 seconds or so
[18:38] <ivoks> darkscrypt: all debian derivates have it
[18:40] <darkscrypt> <-- rpm / red hat guy
[18:44] <xperia> SpamapS: fixed it in the Bios settings. there was one Line with "F1 prompt" that was activated have now deactivated it and it works. so now after poweron of my server ubuntu boot automatic without any user interaction on my Proliant ML530 G2 Server :-)
[18:47] <killown> ehy, i would like know how encrypt /var and /etc partitions.. does ubuntu server allow it during installation?
[18:49] <pmatulis> killown: in combination with lvm, yes
[18:49] <killown> i will try , thanks
[18:51] <SpamapS> xperia: excellent, please if you would give us warning in here before you take over the world. :)
[18:58] <eagles0513875> how can i change the sources list to a better mirror?
[18:58] <eagles0513875> im currently on lucid
[18:59] <Overand> Has anyone had any luck with using 10.04 exporting an NFS datastore for VMWare ESX/ESXi?
[18:59] <pmatulis> eagles0513875: with a text editor edit /etc/apt/sources.list
[18:59] <ikonia> eagles0513875: it's a text file
[19:00] <eagles0513875> ikonia: i know so i would have to look at the ubuntu mirrors list to determine the address of the mirror i want then?
[19:00] <ikonia> eagles0513875: it's that simple, yes
[19:01] <eagles0513875> ikonia: i have kinda gotten spoiled using kpackagekit
[19:03] <Overand> Oh for pete's sake, I had a type-o in my exports line.
[19:04] <pmatulis> typo, typographical error
[19:05] <Overand> pmatulis: thanks, that correction is both very helpful to my frustrated state, and very topical.
[19:06] <RoyK> anyone here using hardy on a Xen DomU?
[19:09] <MTecknology> Any ideas what's wrong with this config? A user can authenticate to the samba server, but the shares don't seem to exist. http://dpaste.com/199082/
[19:13] <RoyK> MTecknology: which user is authenticating?
[19:14] <MTecknology> RoyK: kalliki-docs
[19:14] <RoyK> dies it work with adam?
[19:15] <MTecknology> RoyK: hm.. yes it does
[19:16] <RoyK> perhaps trying with a username without a dash works better
[19:16] <RoyK> just a hunch
[19:16] <MTecknology> RoyK: I also removed the printable part - i guess i'm not sure what that did
[19:17] <RoyK> shouldn't mean much - see man smb.conf
[19:22] <Italian_Plumber> are there any disadvantages to using the 64 bit OS instead of 32?
[19:23] <smoser> flash will be your biggest sticking point
[19:23] <smoser> but for server... no
[19:26] <atomic__1> \
[19:26] <danutz> hello all ,I have a question..I have a vps and a domain...I created the name server at the domain registers ns1.mydomain.com and ns2.mydomain.com that goes to the vps ip and then set the domain to these nameservers the problem is that is not resolving..what do I have to do next?
[19:27] <Italian_Plumber> I just got a new motherboard/CPU and I thought it was 32 but now I've discovered it's 64, and I'm trying to decide if I want to install 32 bit or 64 bit Ubuntu (server)
[19:30] <hggdh> Italian_Plumber: probably you would want 64
[19:31] <Italian_Plumber> mmkay
[19:36] <sabator2> Does anyone uses Promise RAID card with Ubuntu ?
[19:39] <Overand> sabator2: there are many different Promise RAID cards - most of which are not 'really raid' - i.e. they're "FakeRAID" - only a few are hardware-assist or really RAID
[19:39] <Overand> sabator2:  take a peek at https://help.ubuntu.com/community/FakeRaidHowto
[19:40] <Italian_Plumber> BTW... you can kill the Fake Kraid with a simple ice beam or wave beam shot -- no need for missiles.
[19:41] <Overand> Italian_Plumber: generally speaking, I'd go for 64 bit on a server, and 'debate it' only on a workstation
[19:42] <Overand> Italian_Plumber: Unless you have very esoteric hardware, that's the route I'd go - I've actually been running ubuntu-server 64 bit on 8.04 for quite some time - and if I recall correctly, the machine was actually originally installed with a prior release
[19:42] <sabator2> Promise provide "FastTrak TX4310 Partial Linux source code", do I need to compile driver or Lucid will support it?
[19:43] <zul> you will have to compile the driver
[19:43] <Overand> Italian_Plumber: some of the files on that system were created 2006-04-19 - I may have in fact installed 6.06 LTS on it
[19:43] <Overand> sabator2: May I make a suggestion?
[19:43] <Italian_Plumber> Hardware is ASUS M4A785-M, AMD Phenom X4 9650, 2048MB PC6400 DDR2
[19:44] <Overand> sabator2: http://www.lmgtfy.com/?q=ubuntu+TX4310
[19:44] <Pici> Overand: Thats not helpful.
[19:44] <Italian_Plumber> I'll be running Hardy Server ...  DNS, 24/7 bitTorrent, apt-cacher-ng, svn, vmware ... no webserver.
[19:44] <Overand> Pici: It's one click away from some discussions on the particular harwdare
[19:44] <Pici> !lmgtfy
[19:44] <Pici> er,
[19:45] <Pici> !google
[19:45] <Overand> Pici: Alright.
[19:45] <Overand> Would providing a direct link to the google response that lmgtfy proviced be acceptable?
[19:45] <ivoks> sabator2: is that the driver for 2.4 kernel?
[19:48] <Overand> Italian_Plumber: given what you've described, that's so nearly identical to what I run that I'd say 'go for it'
[19:49] <Overand> Italian_Plumber: just do a stress test on it before you put it into production.  Also - since you're using apt-cacher-ng, you should be aware that the 32 and 64 bit packages will be different, so you're not going to get much benefit for the 64 bit box if everything else is 32, etc.
[19:49] <sabator2> Do we need to setup Kernel Compile Environment with Lucid like we do in Suse with "make mrproper" command ?
[19:50] <mathiaz> zul: I've nominated/accepted bug 577165
[19:50] <zul> mathiaz: thanks
[19:50] <Overand> sabator2: Just a question - before you start digging into this- have you tried the card yet?
[19:52] <Italian_Plumber> what would the stress test tell me?
[19:53] <Overand> Italian_Plumber: it's just a good way to poke around for unexpected incompatabilities - it's probably unneeded 'specifically' for a 64 bit system, but it's never a bad idea
[19:53] <Italian_Plumber> gotcha
[19:55] <sabator2> Overand: No, I thought I had to load the driver prior to boot with the Lucid.
[19:57] <Overand> sabator2: have you verified that the array you build *doesn't* show up in the lucid installer?
[19:57] <Overand> I'm just trying to make sure you're not trying to fix something that's already working, that's all
[20:02] <BigThetan> Hello all
[20:03] <BigThetan> can anyone tell how to print from command line
[20:03] <BigThetan> I do not have gui installed on my server
[20:04] <ivoks> lpr
[20:05] <BigThetan> is ther a man for lpr?
[20:06] <pmatulis> BigThetan: i beg your pardon?
[20:06] <savid> Hi all.  I'm trying to get glusterfs set up.   When I run sudo /etc/init.d/glusterfs start,  nothing happens,  and when I run /etc/init.d/glusterfs status,   I get "GlusterFS server is not running"
[20:07] <BigThetan> i was trying to see if there was a manual for lpr but I founs it thanks
[20:07] <savid> using ubuntu 10.04
[20:07] <pmatulis> savid: strace perhaps
[20:09] <sabator2> Overand: I'm not installing Lucid on RAID5, I'm just accessing data on it.
[20:13] <smoser> name calling is just not nice
[20:13] <pmatulis> yeah
[20:15] <savid> pmatulis,  http://dpaste.com/199112/
[20:17] <webPragmatist> is a grow partition always logical?
[20:17] <imthenachoman> anyone used webmin?
[20:19] <pmatulis> savid: i would research the "Inappropriate ioctl for device" part
[20:23] <webPragmatist> if you guys were trying to cluster two servers would you try to share /etc between the two?
[20:31] <SpamapS> webPragmatist: /etc contains a lot of local configuration settings, so thats usually not advisable
[20:32] <SpamapS> webPragmatist: typically when clustering you need to identify shared, and local configs
[20:32] <webPragmatist> SpamapS: do you have an resources on doing this?
[20:32] <SpamapS> savid: do you have config files created in /etc/glusterfs ?
[20:33] <imthenachoman> is ufw better than Shorewall?
[20:33] <SpamapS> webPragmatist: clustering is a very broad topic... one which I don't have a definitive source for information on.
[20:33] <webPragmatist> SpamapS: well what I am using is drbd as that seems a common solution...
[20:34] <webPragmatist> and i am just using typical web services
[20:34] <webPragmatist> apache2, mysql, pgsql, svn
[20:34] <SpamapS> webPragmatist: ahh drbd is a high availability failover solution
[20:35] <webPragmatist> right
[20:36] <RoAkSoAx> webPragmatist: if you want to keep config file in sync within various nodes, you can take a look to csync2
[20:38] <webPragmatist> RoAkSoAx: whats your take on using drbh
[20:38] <webPragmatist> drbd*
[20:39] <RoAkSoAx> webPragmatist: if've used it before but to replicate data as in Web Server data and MySQL
[20:39] <RoAkSoAx> webPragmatist: but never for /etc. However, I strongly recommend DRBD. I really like it.
[20:39] <RoAkSoAx> webPragmatist: for config files, I'd recommend you to take a look to csync2
[20:40] <RoAkSoAx> webPragmatist: here's a howto I did couple years ago: http://www.roaksoax.com/2008/06/cluster-sinchronization-tool-csync2
[20:41] <webPragmatist> RoAkSoAx: so you just replicate the individual service data?
[20:41] <webPragmatist> like /var/www
[20:41] <RoAkSoAx> webPragmatist: something like that yes
[20:42] <webPragmatist> did you just tell the service to store it's data in a different location?
[20:42] <webPragmatist> also how did you go about upgrading the service… it seems like you could easily fubar your data that way if you don't upgrade both services to the same version?
[20:43] <webPragmatist> (at the same time)
[20:43] <RoAkSoAx> webPragmatist: For multiple webservers, since they needed the same data, I use two nodes with DRBD, exporting the data with NFS, the same for MySQL. Two nodes. Then the webservers only mounted the exported NFS data and connected to the MySQl
[20:43] <RoAkSoAx> webPragmatist: in my case, I used csync2 to sync up apache config file's between servers
[20:44] <RoAkSoAx> webPragmatist: each apache web server, mounted the NFS share and set it as the root for apache
[20:44] <webPragmatist> does it sync the entire /sites-available?
[20:44] <RoAkSoAx> webPragmatist: check out the post. You can either sync single files, or complete folders
[20:44] <webPragmatist> oh okay
[20:46] <webPragmatist> so really theres three "nodes" because of your NFS?
[20:46] <webPragmatist> i mean whats the point of the nfs
[20:48] <awb> Hello
[20:48] <pmatulis> awb: 'wassup?
[20:48] <webPragmatist> RoAkSoAx: ?
[20:49] <awb> Does anyone know how to setup a Primary Domain Server on Ubuntu Server 10.04 from Scratch (I want to be able to join windows clients to it)
[20:51] <RoAkSoAx> webPragmatist: Two nodes in HA, replicating data with DRBD. used as NAS. Then various loadbalanced webservers accesing the shared data through the NFS export
[20:51] <pmatulis> awb: search "ubuntu server guide"
[20:51] <RoAkSoAx> webPragmatist: keep in mind that webservers nodes are in a different level of storage servers
[20:51] <guntbert> !serverguide
[20:52] <webPragmatist> RoAkSoAx: ohh i see… so you just used the drdb as a centralized file storage
[20:52] <webPragmatist> for some dummy web servers
[20:53] <RoAkSoAx> webPragmatist: The architecture I used is explained here: http://www.roaksoax.com/2008/07/ubuntu-in-my-thesis-part-2
[20:54] <webPragmatist> RoAkSoAx: what would you do if you just had 2 servers….. or rather $500 bucks a monthish heh
[20:54] <webPragmatist> best bang for buck
[20:55] <RoAkSoAx> webPragmatist: to servers for what? Web serverS?
[20:55] <webPragmatist> web/db
[20:55] <akincer> is there a way to create custom commands with eBox like you could with Webmin?
[20:55] <webPragmatist> postgresql, mysql, apache2, streaming video files (mp4s), other junk
[20:55] <webPragmatist> all related to the webserver
[20:55] <webPragmatist> svn
[20:56] <webPragmatist> thats it
[20:56] <RoAkSoAx> webPragmatist: webPragmatist you want loadbalacing?
[20:56] <akincer> similarly, is eBox the preferred Ubuntu web based config tool these days?
[20:56] <webPragmatist> RoAkSoAx: not really necessary
[20:56] <webPragmatist> availability is key
[20:56] <webPragmatist> it's an ecommerce website
[20:56] <RoAkSoAx> webPragmatist: if you only want two webservers in HA, with active/passive, I'd just go for DRBD to replicate the shared data (mysql, apache data and etc)
[20:58] <webPragmatist> could i do more with 500 bucks a month (dedicated hosting) … ? We send out a hefty bit of data…. about 8TB a month
[20:58] <webPragmatist> that's my original plan
[20:58] <webPragmatist> was to just get drdb working
[20:59] <RoAkSoAx> webPragmatist: well other thing that you could do, if you have really good hardware, is setup virtualization and set everything in VM's
[20:59] <webPragmatist> not sure if i'd do that… seems a bit pointless
[21:00] <RoAkSoAx> webPragmatist: not really
[21:00] <webPragmatist> unless i was just running something that was terrible at threading
[21:00] <RoAkSoAx> webPragmatist: setting multiple VM's gives you the possibility to have different VM's for each service
[21:01] <RoAkSoAx> webPragmatist: for example, you could even have a loadbalanced webserver that provides scabalability by adding VM's running apache
[21:01] <RoAkSoAx> and stuff like that
[21:03] <webPragmatist> I guess it could help allocate the resources better
[21:04] <webPragmatist> but in the long run you might end up just limiting one services resources to gurantee anothers
[21:04] <webPragmatist> atleast this is how i see it
[21:04] <webPragmatist> no matter what you have 8gb of ram to work with….. partitioning it off using a vm doesn't improve anything
[21:05] <awb> What do I actually need to install to make a Ubuntu Server PDC for Windows Machines, I have Samba 4 already is this all I need to install?
[21:06] <akincer> awb: I believe there is info on the Samba Wiki if you feel like digging into the documentation there. Once upon a time, you had to install Kerberos and a few other packages. I want to say that is all built in to Samba 4, but you will want to check the docs to be sure
[21:06] <webPragmatist> RoAkSoAx: did you end up with a crapload of logical partitions
[21:06] <webPragmatist> using lvm
[21:07] <webPragmatist> for your drbd partition
[21:07] <webPragmatist> oh wait just /srv or something
[21:07] <webPragmatist> hrm
[21:08] <webPragmatist>  hrmmmmmmmmmmmmmmmmmmmmmmm
[21:08] <webPragmatist> so much
[21:08] <webPragmatist> lets see if this idiot guy can setup a web server
[21:08] <webPragmatist> with drbd
[21:13] <RoAkSoAx> webPragmatist: i didn't use lvm
[21:13] <RoAkSoAx> webPragmatist: it is not hard :) it is fun
[21:14] <webPragmatist> RoAkSoAx: just a big ol' drbd part?
[21:15] <RoAkSoAx> webPragmatist: yep. I just used ext3 over drbd
[21:15] <RoAkSoAx> s/over/on top
[21:18] <webPragmatist> RoAkSoAx: with your scaling idea how did you propose to keep all the service updated?
[21:18] <webPragmatist> with each node
[21:23] <webPragmatist> RoAkSoAx: would you bother to raid mirror 1 your cluster
[21:23] <webPragmatist> nodes
[21:24] <webPragmatist> i'm thinking it's a waste of sapce
[21:24] <webPragmatist> space*
[21:25] <RoAkSoAx> webPragmatist: take a look to something like glusterfs
[21:25] <RoAkSoAx> webPragmatist: or a cluster filesystem on top of DRBD
[21:25] <webPragmatist> well this is a hardware raid
[21:25] <webPragmatist> so that would be only somewhat equivalent
[21:26] <webPragmatist> gfs though hrm
[21:26] <RoAkSoAx> webPragmatist: oh well then. I think you should evaluate what is best form you then given the hardware you have and what you want to implement
[21:26] <webPragmatist> i just don't want crap to break
[21:27] <webPragmatist> but in the 3 years i've been running this server it's only hardware mirror and it's never broke :(
[21:27] <webPragmatist> but we have another server that we are only using to backup to which i've never had to use either
[21:28] <MTecknology> my linode seems to be really slow on the disk :S
[21:28] <akincer> OK, a quick check of eBox shows me that #1 it is feature anemic and #2 it is NOT a suitable replacement for Webmin in any sense of the word. At least not for what I need it for
[21:29] <MTecknology> akincer: learn to use the command line?
[21:30] <guntbert> akincer: yes (I see it similarly) but webmin *is* bound to damage your system some time
[21:30] <akincer> MTechnology: That is a complete non sequitur WRT my needs. I can use the CLI just fine
[21:31] <akincer> guntbert: Not using it for system configuration but rather to restart some services using custom commands. Was hoping to move to eBox since its interface is much nicer looking than Webmin. But eBox sticks you in a box and holds you there
[21:32] <guntbert> akincer: the services management will not work for long any more - the move from sys-v to upstart is fully under way and webmin has no idea of that
[21:33] <akincer> guntbert: The fact that Webmin will execute any shell command I want it to says otherwise
[21:34] <akincer> I really was just hoping eBox had the same flexibility
[21:34] <akincer> perhaps one day
[21:35] <guntbert> akincer: as long you not only know what you are doing but also what webmin is doing ... :-)
[21:36] <akincer> haha true enough. I just configure custom commands to be executed from the CLI and create a single click button to get it to do that. For example: /etc/init.d/myservice restart
[21:36] <qman__> if all you need is a web-based terminal, there's always ajaxterm
[21:36] <akincer> so long as it does nothing more than do what I tell it to on the CLI, we're all good
[21:37] <akincer> no, this is for people who know zero about the CLI to be able to do handy things like restart some custom software services I have
[21:37] <qman__> but frankly, I'd put my own interface together before I used any of the above
[21:37] <akincer> also to let them edit some config files for those services with the web based text editor
[21:37] <qman__> a simple PHP script with some exec()s wrapped around the service command
[21:37] <akincer> yeah, as soon as I get some time, I'm going to do just that
[21:37] <qman__> handle auth with apache
[21:38] <akincer> Nah, I'm going to make a Joomla plugin to do it
[21:39] <akincer> I just find it very puzzling that you can't restart system services with eBox unless there are predefined plugins specifically for those services
[21:40] <akincer> Seems like absurd overkill to essentially just have something behind the scenes do /etc/init.d/blahblah restart
[21:42] <qman__> well, that's not even the correct way anymore
[21:42] <qman__> because of upstart
[21:42] <qman__> so the service command, now included, is the 'right' way
[21:42] <akincer> I guess I don't know much about upstart
[21:42] <qman__> because it can handle both sysv scripts and upstart scripts
[21:43] <qman__> or at least it's supposed to
[21:43] <qman__> I haven't done enough investigating to find out whether it actually works
[21:44] <akincer> So if I have a script that has stop/start/restart defined, once upon a time I would use update-rc.d myscript defaults to register the service and use /etc/init.d to control it
[21:44] <akincer> You are saying that is now the wrong way. What's the right way?
[21:44] <qman__> well, that'll still work
[21:44] <qman__> but only with sysv scripts
[21:44] <qman__> upstart scripts are handled differently
[21:45] <qman__> I don't know how to register one, but an upstart script is handled via "start <script" and "stop script"
[21:45] <qman__> however, the service command is supposed to handle both upstart and sysv
[21:45] <qman__> service <servicename> [start|stop|restart]
[21:46] <akincer> Well I know next to nothing about upstart so if that's now the "preferred" way or whatever, guess I'll have to read up on it
[21:46] <qman__> sysv style still works for now
[21:47] <qman__> but it's getting phased out
[21:47] <qman__> and actually, the current setup is just a compatibility layer on top of upstart
[21:47] <qman__> actual sysvinit is gone
[21:47] <akincer> right, it's the "for now" part that moves me to think learning upstart would be wise
[21:48] <ScottK> Just don't get too attached to Upstart as it is now, since it's promised to change a lot by 12.04.
[21:48] <akincer> Well then I'll just wait a bit to worry about it
[21:48] <qman__> I think the safest bet is to write your stuff around the service command, since that should be updated to handle whatever comes
[21:49] <qman__> but, just my opinion
[21:49] <akincer> I won't have to worry about it for now since 10.04 seems to honor the old ways
[21:50] <akincer> I'll just test out the non-LTS versions as they come along, play with it and then when the next LTS nears, I'll see what I need to know
[21:50] <qman__> yeah
[21:51] <akincer> thanks for the info though
[22:03] <savid> Ok,  so on ubuntu it installs   /etc/glusterfs/glusterfsd.vol --   but, it looks like those are example files.  I have my own files in that folder that I generated using glusterfs-volgen.    The problem is that the init script seems to be using the glusterfsd.vol file when it starts.   How do i make glusterfsd just use all the config files?
[22:10] <Tweeda> savid, I move the default files in /etc/glusterfs/ out of the way and sym-link /etc/glusterfs/glusterfsd.vol to the file generated by glusterfs-volgen and refer to the glusterfs-client files by name in /etc/fstab
[22:14] <savid> Tweeda,  When I did glusterfs-volgen,  it generated three files.    I did "glusterfs-volgen --name=slurpee_media --raid 1 slurpee-web1:/export/slurpee_media slurpee-web2:/export/slurpee_media".     It created  slurpee_media-tcp.vol,  slurpee-web1-slurpee_media-export.vol,   and slurpee-web2-slurpee_media-export.vol.
[22:14] <savid> Not sure which of those is the client file,  or which of those is the server file :-P
[22:14] <savid> very confusing
[22:15] <Tweeda> savid, one sec
[22:17] <Tweeda> savid, the /etc/glusterfs/glusterfsd.vol should contain something like "option directory /export/data0/data0" in the volume block.  that eventually gets shared out as a 'brick'
[22:17] <jo-erlend> what is the dyngroup schema for in an LDAP directory?
[22:18] <Tweeda> savid the glusterfs.vol will list the servers and describe how to assemble the bricks into a filesystem
[22:18] <savid> Tweeda, are you talking about the default glusterfsd.vol that's included when installed?
[22:18] <Tweeda> I'm talking about what volgen generates
[22:19] <Tweeda> savid, honestly, this should probably be taken to #gluster :)
[22:19] <savid> Tweeda,  volgen doesn't generate a glusterfsd.vol file.      I did "glusterfs-volgen --name=slurpee_media --raid 1 slurpee-web1:/export/slurpee_media slurpee-web2:/export/slurpee_media".     It created  slurpee_media-tcp.vol,  slurpee-web1-slurpee_media-export.vol,   and slurpee-web2-slurpee_media-export.vol.
[22:21] <cybrocop> Hi all. Can someone tell me how I can troubleshoot a "suspected" kernel bug within Apparmor?
[22:21] <cybrocop> Symptoms:
[22:22] <cybrocop> a) Access to a file is denied.
[22:22] <cybrocop> b) Nothing is kern.log
[22:22] <cybrocop> c) I purged Apparmor modules
[22:23] <Tweeda> savid, slurpee_media-tcp.vol is the client description that gets listed in /etc/fstab or used by the mount command.  The slurpee-web1-slurpee_media-export.vol is the glusterfsd.vol file on slurpee-web1
[22:24] <cybrocop> to be specific on a) access is denied when I'm running KVM through virsh. And to be even more specific this happens when I'm providing a <serial> directive in order to define a serial port for my virtual machine.
[22:25] <cybrocop> kirkland: Sorry, I pinged you yesterday on #eucalyptus but I dropped off so I don't know if you answered my questions there. A kernel bug seems to be the only logical explanation for the problems that I am having.
[22:26] <savid> Tweeda,  Ahh,  I see.  So I need to move the files it generated to the relevant servers.  Gotcha.
[22:28] <cybrocop> Some more background on the problem I'm having: http://forum.eucalyptus.com/forum/libvirt-operation-failed-failed-retrieve-chardev-info-qemu-info-chardev#comment-11334
[22:38] <BigThetan> how do i test to see if my rsyslog server is receiving??????????
[22:39] <webPragmatist> RoAkSoAx: you still around?
[22:39] <webPragmatist> have any of you configured drbd
[22:40] <webPragmatist> i'm just wondering if i have to create a ext3 partition first
[22:40] <webPragmatist> or if drbd needs unallocated space
[22:43] <BigThetan> how do i test to see if my rsyslog server is receiving??????????
[22:45] <KillMeNow> bigthetan:  IIRC there should be a log
[22:46] <KillMeNow> also check to make sure the port active and listening
[22:46] <|corpse|> im having a problem trying to install ubuntu server from a usb drive on a machine that has no cd-rom. is a cd-rom required?
[22:46] <BigThetan> there are a few logs which one should I check?
[22:46] <|corpse|> actualy it has a cd-rom and its listen in the bios and post but the server install dosnt not recognize it and i cant get passed it
[22:47] <BigThetan> i can ping my server
[22:47] <BigThetan> from another machine
[22:53] <webPragmatist> i want to create a /srv directory to hold svn repositories, www, etc...
[22:53] <webPragmatist> with drbd should i create a partition for each?
[22:53] <webPragmatist> or can i just map the entire /srv dir
[22:54] <webPragmatist> to the drbd
[23:01] <RoAkSoAx> webPragmatist: you can map the entire /srv dir
[23:01] <webPragmatist> got it
[23:06] <|corpse|> im stuck at my install right after the keyboard setup,i get a progress bar that says Detcting hardware to find CD-ROM drives, and it just sits at 0%
[23:08] <webPragmatist> RoAkSoAx: hey i screwed up and put the wrong ip for my cluster1
[23:09] <webPragmatist> i fixed it and tried to restart drbd and it says the volume is in use
[23:09] <webPragmatist> oh detach drive0
[23:10] <webPragmatist> 0: State change failed: (-2) Refusing to be Primary without at least one UpToDate disk
[23:26] <kirkland> cybrocop: i'm sorry, i don't have the bandwidth to answer your questions this week
[23:27] <kirkland> cybrocop: please check #ubuntu-kernel, if you think it to be a kernel issue
[23:27] <RoAkSoAx> webPragmatist: try something like drbdadm -- --overwrite-data-of-peer primary data-lower
[23:27] <RoAkSoAx> webPragmatist: try something like drbdadm -- --overwrite-data-of-peer primary <drbd-resource>
[23:31] <webPragmatist> i need to read all these commands
[23:31] <webPragmatist> i'm just copy pasting random crap
[23:34] <riz0n> Hey guys, I have an ubuntu server 8.04LTS and am curious what steps I should take to upgrade to the new 10.04 LTS without having to do a fresh installation?
[23:37] <ScottK> riz0n: In general it's recommended to wait for 10.04.1 in two months to upgrade.  If you want to now, sudo do-release-upgrade -d should do it.
[23:37] <riz0n> Thanks ScottK!!
[23:38] <centaur5> How do you give sudo access on a ltsp fat client?