[00:00] <kirkland> mathiaz: yeah
[00:00] <kirkland> mathiaz: i meant to test that today
[00:00] <kirkland> mathiaz: is it there?
[00:00] <mathiaz> kirkland: yes
[00:00] <mathiaz> kirkland: could you add a test case to cover it on https://wiki.ubuntu.com/Testing/Cases/ServerInstall ?
[00:00] <mathiaz> kirkland: I haven't tested though.
[00:01] <kirkland> mathiaz: k
[00:25]  * flyback bbl
[01:18] <mathiaz> kirkland: seems that the new virtualization task works correclty
[01:18] <kirkland> mathiaz: awesome
[01:18] <mathiaz> kirkland: it pulls in kvm, libvirt-bin and openssh-server
[03:51]  * flyback feels like he is about to have a stroke
[03:57] <ScottK> Any particular reason?
[04:00] <flyback> na
[04:04]  * flyback bbl
[05:18] <nomingzi> I plan to download and to install Ubuntu server, may I know if it is come with GUI, unlike Debian has no GUI.
[05:21] <[Solars]> debian has goo-ey just gotta install it
[05:21] <mathiaz> !servergui | nomingzi
[05:21] <[Solars]> ubuntu server doesn't come preinstall gui
[05:22] <[Solars]> you can easily take the 'desktop' version and use it as a server
[05:22] <[Solars]> just got to install the packages you want
[05:23] <nomingzi> Solars: does the Ubuntu Desktop version has 64-bit edition ?
[05:23] <nomingzi> Solars: sorry i m newbie
[05:23] <[Solars]> yes
[05:24] <[Solars]> get the one marked amd64
[05:24] <nomingzi> solars: many thanks
[06:03] <edgy> Hi, I am using Ubuntu 8.04.1 hardy and  logwatch --mailto me@mysite.com just prints to screen. I test postfix and it's working already
[06:04] <kgoetz> edgy: not sure i understand your question
[06:05] <edgy> kgoetz: logwatch is a log analyzer and reporter that when invoked like this: logwatch --mailto me@gmail.com should send the report to me@gmail.com
[06:06] <edgy> instead of sending the report by email, it prints it to stdout instead!
[06:06] <edgy> may be it's a bug I am not sure
[06:07] <edgy> any one had hardy on x64 to test?
[06:07] <kgoetz> edgy: its working fine for us - admitedly i'm running it via cron
[06:07] <edgy> kgoetz: 32 or 64?
[06:07] <kgoetz> 14:37:50 kgoetz@paver:~$ logwatch --mailto karl.goetz@rocksoft.com
[06:07] <kgoetz> 14:38:09 kgoetz@paver:~$
[06:07] <kgoetz> thats worked
[06:08] <kgoetz> edgy: 64bit
[06:08] <edgy> kgoetz: 8.04.1 or 8.04?
[06:08] <kgoetz> .1
[06:09] <edgy> my system is upgraded from a previous version and not a fresh installation, so what do you think the problem can be?
[06:09] <kgoetz> logwatch_7.3.6-1ubuntu1_all.deb is the package i used
[06:09] <kgoetz> some configuration issue
[06:11] <edgy> kgoetz: but i tried --mailto from the command instead of depending on logwatch.conf to troubleshoot it. Shouldn't it work whatever values in config files is, beside I reinstalled logwatch so it's the default configs
[06:12] <kgoetz> no idea then
[07:42] <[HU]gnanet> Hello
[07:43] <[HU]gnanet> i am looking for a good dedicated server provider in UK , of Ireland, preferrably with preinstalled ubuntu, any suggestions?
[07:44] <[HU]gnanet> erm .. of-> or
[09:02] <felimwhiteley_> [HU]gnanet: I've heard good thing about these guys http://bitfolk.com/
[09:02] <felimwhiteley_> but I've not used them personally
[09:03] <[HU]gnanet> Thank you
[10:40] <skep> whats the default syslogd in intrepid?
[13:30] <sommer> morning all
[14:12] <ivoks> zul: i've sent you kernel commit for drbd
[14:25] <zul> ivoks: cool ill have a look
[15:09] <Abracadabra> Hi
[15:09] <Abracadabra> do you have a command to test write speed on a scsi disk?
[15:10] <Abracadabra> such as hdparm for ide
[15:53] <kirkland> mathiaz: section added to https://wiki.ubuntu.com/Testing/Cases/ServerInstall
[15:53] <mathiaz> kirkland: awesome - thanks
[15:54] <mathiaz> kirkland: do you have access to an intrepid host for doing some kvm testing ?
[15:54] <kirkland> mathiaz: yup
[15:54] <mathiaz> kirkland: I've run into a crash when trying to test raid installation
[15:54] <kirkland> really
[15:54] <mathiaz> kirkland: could you try to setup a guest with two qcow files and go through the installer to setup a raid mirror array ?
[15:55] <mathiaz> kirkland: the kvm process dies when the installer tries to activiate the array
[15:55] <kirkland> mathiaz: interesting...  i've always pre-created my disks with dd rather than qcow for my raid testing
[15:55] <mathiaz> kirkland: or at least around that time
[15:55] <kirkland> mathiaz: i would not have expected qcow to work properly
[15:55] <kirkland> mathiaz: but i'll give it a shot
[15:56] <mathiaz> kirkland: I use qemu-img
[15:56] <kirkland> mathiaz: i do too, except when i'm doing raid testing
[15:56] <mathiaz> kirkland: it works fairly well - I'm doing all of the iso testing with qcow2 files
[15:56] <mathiaz> kirkland: It never crashed.
[15:56] <kirkland> mathiaz: qcow2 doesn't crash, but qcow does, when doing raid testing?
[15:57] <mathiaz> kirkland: hm - why would there be a difference when doing a raid testing ?
[15:57] <mathiaz> kirkland: qcow2 crashes
[15:57] <kirkland> mathiaz: well, i want to simulate a "real" disk more closely
[15:58] <kirkland> mathiaz: so i do 'dd if=/dev/zero of=disk1.img bs=1M count=1000'
[15:58] <mathiaz> kirkland: interesting - I'll try with dd then.
[15:58] <kirkland> mathiaz: that's what my detailed testing notes on the BootDegradedRaid blueprint specifiy
[15:58] <kirkland> mathiaz: i suspect that the dynamic growth of qcow2 images don't work well with synchronous raid writes
[15:59] <kirkland> mathiaz: but i've never tried
[15:59] <kirkland> mathiaz: i will now though ...
[15:59] <kirkland> mathiaz: perhaps after the meeting
[16:00] <mathiaz> kirkland: sure - I'll give a try to your dd suggestion later today
[16:13] <ivoks> zul: does that commit look ok? (that's first serious commit i created :)
[16:13] <zul> ivoks: I havent gotten the email yet
[16:14] <ivoks> no?
[16:55] <kaushal> hi
[16:57] <kaushal> I have installed tomcat 5.5.25 on Ubuntu 8.04
[16:58] <kaushal> the issue is that it doesnot get logged
[16:58] <kaushal> in catalina.out file
[16:58] <kaushal> its really really annoying
[16:58] <soren> Koon: ^^
[16:58] <Koon> soren: i'm on it :)
[16:58] <soren> :)
[16:59] <Koon> kaushal: what are you using as a JVM ?
[16:59] <Koon> kaushal: the default one (gcj) ?
[16:59] <Koon> kaushal: or did you install something else?
[16:59] <Koon> (Sun's JDK, openJDK...)
[17:00] <kaushal> JAVA_HOME=/usr/lib/jvm/java-6-sun
[17:00] <laga> soren: i talked to the vbox guys. there is no nice way to create those VM description files. VBoxManage can only register empty xml files. we could cobble together a working file from the examples we have, but it's not exactly supported by sun/virtualbox. and there is no way of importing them
[17:01] <Koon> kaushal: I seem to remember the logging issue as gcj-specific
[17:01] <soren> Hey, what we're doing with vmware isn't supported either. It could fall apart at any given time.
[17:01] <kaushal> ok
[17:01] <Koon> kaushal: please check that it's really using sun's jvm
[17:01] <laga> soren: right. they're working on a better interface for that kind of stuff.
[17:01] <Koon> kaushal: that said, an update for hardy is in the pipe
[17:01] <kaushal> Koon, so whats the workaround
[17:02] <Koon> so that it works with openjdk
[17:02] <laga> soren: we *could* call VBoxManage to create the VM, but we don't necessarily want the VM to be registered for the user who built them
[17:02] <soren> laga: What I'm saying is: I don't give a hoot if creating the xml isn't supported. Let's just do it anyway if that's what works.
[17:03] <laga> soren: yes, but that still doesn't solve the problem that you can't import the XML.
[17:03] <Koon> kaushal: at that point the workaround would be to install sun-java6-jdk and run without specific JAVA_HOME (the init script should pick up sun's jdk)
[17:03] <soren> laga: We've not promised anyone that it'll work to any degree. It's a best effort thing.
[17:03] <Koon> in a few days the workaround woulfd be to install openjdk-6-jre
[17:03] <laga> soren: ah, so it's not a "people will bi... complain to canonical if it doesn't work"?
[17:03] <ivoks> ScottK: did you get my clamav patch?
[17:03] <Koon> since the init script would support running that
[17:03] <laga> thing.
[17:04] <soren> laga: If we think it's really and issue, and vmbuilder goes to main, I can just split out the vbox plugin into a separate package. No biggie.
[17:04] <laga> right.
[17:04] <Koon> kaushal: see https://bugs.launchpad.net/ubuntu/+source/tomcat5.5/+bug/251004
[17:05] <Koon> kaushal: running with TOMCAT5_SECURITY=no is also a workaround (if your environement is not very exposed)
[17:05] <laga> soren: still, how it's useful if you can't import that piece of xml?
[17:05]  * Koon leaves for a couple of hours
[17:06] <soren> laga: Just because they don't offer a supported way of doing it, doesn't mean there isn't a way we can do it.
[17:07] <soren> laga: Last I worked with virtualbox they had all-xml configuration files.
[17:08] <soren> laga: ...which isn't terribly difficult to change programmatically.
[17:08] <soren> ..by design, even.
[17:09] <laga> yeah. it's easy to create an xml file describing the VM itself. but it's not that easy to add it to virtualbox, because you need to mangle the main virtualbox configuration file.
[17:09] <soren> I'm quite sure VMWare doesn't support whipping up vmdk with qemu-img and handcrafting .vmx files. We do that anyway. *shrug*
[17:09] <laga> does u-v-b actually import other VMs?
[17:09] <soren> laga: I don't understand the question.
[17:10] <laga> soren: as in, "creates .vmx and feeds it to vmware"
[17:10] <laga> soren: Yes. but in virtualbox, there is no nice button to import that piece of XML generated by u-v-b. nothing.
[17:10] <soren> laga: I realise.
[17:11] <soren> laga: Correct me if I'm wrong, but here's how I remember it:
[17:11] <soren> It has a main config, which refers by uuid to a set of vm's.
[17:11] <soren> Each vm has a separate XML file, which in turn refers to some disk images, identified by uuid.
[17:11] <laga> yes.
[17:11] <soren> These disk images have a separate XML file pointing at the disk images.
[17:12] <soren> What's to stop us from whipping up the disk xml, vm xml, and add the vm to the main xml?
[17:12] <soren> From where I'm sitting, it seems the answer is: Not a thing.
[17:13] <soren> This is all Linux and even free software. It can't hide stuff from us, even if it wanted to. We can just go and see what it would do itself and replicate that.
[17:13] <laga> indeed, not a problem. but is that useful? do we *want* to add VMs directly to virtualbox?
[17:13] <soren> What's the alternative?
[17:14] <soren> If you say it doesn't even have a button to import anything.
[17:14] <laga> if we want to just add a VM to virtualbox, we can use VBoxManage. that's a lot easier.
[17:14] <laga> soren: it's a completely different thing from what u-v-m does not
[17:14] <soren> I don't follow.
[17:14] <laga> soren: it's a completely different thing from what u-v-m does currently
[17:14] <laga> sorry, i'm tired.
[17:14] <soren> Does VBoxManage provide hooks to do this or not?
[17:15] <laga> if you tell it to make a vmware VM, it gives you a .vmdk file and a .vmx file. it does not import the VM into vmware.
[17:15] <soren> Not at the moment, no.
[17:15] <soren> ..but there's not reason it couldn't.
[17:15] <laga> soren: VBoxManage lets you import a VM into virtualbox. yes.
[17:15] <laga> alright, if that's actually desired behavior, then let's go for it
[17:16] <laga> it'd be neat to have an option not to do that, though.
[17:16] <soren> laga: I'm missing something. Clearly.
[17:16] <soren> 16:00:45 < ~laga> soren: i talked to the vbox guys. there is no nice way to create those VM description files. VBoxManage can only register empty xml files. we could cobble  together a working file from the examples we have, but it's not exactly supported by sun/virtualbox. and there is no way of importing them
[17:16] <laga> heh.
[17:16] <soren> Can you import a VM or can't you?
[17:16] <maswan> ugh. having a custom nfs server with a custom portmap kind of sucks.
[17:18] <laga> soren: you can talk directly to VBoxManage and tell it to create a VM. it *might* be possible to use "VBoxManage registervm <filename>" to import a VM, but i was told by a vbox developer that it only accepts empty files.
[17:18] <laga> for whatever reason.
[17:18] <laga> we're talking about two different things, though.
[17:18] <soren> Clearly.
[17:18] <soren> Have you ever actually used ubuntu-vm-builder?
[17:18] <soren> do you understand what it does?
[17:19]  * soren doesn't see the point of importing an empty file, though
[17:20] <laga> right now, u-v-b does one thing (for me): it creates a VM which i can import into my VM application, eg vmware. that's just not going to work for virtualbox *right* now (provided the 'empty file' thing is true)
[17:20] <laga> what you are talking about is immediately registering the VM on the current system, which is a different thing.
[17:20] <laga> and yes, i have used ubuntu-vm-builder :)
[17:20] <soren> Ok.
[17:21] <ScottK> ivoks: I did.  I haven't had time to look at it yet thought.
[17:21] <kirkland> mathiaz: hey
[17:21] <kirkland> mathiaz: okay, i tested using qcow
[17:21] <laga> .. and in the former case, i can just put the VM on my usb drive and use it on another computer.
[17:21] <kirkland> mathiaz: i actually didn't have any problem
[17:21] <laga> importing is not really useful in that use case.
[17:21] <kirkland> mathiaz: but i did think of something else....
[17:21] <soren> laga: The point of actually deploying is this:
[17:21] <mathiaz> kirkland: qcow or qcow2 ?
[17:22] <soren> laga: Well, there's nothing else we can do really, is there?
[17:22] <soren> laga: Anything short of that doesn't help you one bit.
[17:22] <soren> laga: Sure, we can add a --no-deploy option to the vbox plugin if that makes you happy, but I don't see the point.
[17:23] <mathiaz> kirkland: http://paste.ubuntu.com/52512/ <- this is what is printed in dmesg
[17:23] <soren> laga: If you don't actually deploy it (add it to the hypervisor) you're left with something that you can't import anyway, which seems rather pointless to me.
[17:23] <mathiaz> kirkland: when kvm segfaults (or at least the guest dies)
[17:23] <laga> soren: yes. my point is that it's different from the current behavior (as I see it in hardy). so, it'd be sensible to use an extra option, eg --register-with-vbox.
[17:23] <kirkland> mathiaz: when you created the partitions, did you mark them "Bootable" ?
[17:23] <mathiaz> kirkland: yes
[17:23] <laga> soren: you can import it by adding the .vmdk as a disk and creating a new VM, but it's not as neatly integrated as with vmware and .vmx
[17:23] <kirkland> mathiaz: that's the problem
[17:23] <soren> laga: Yes, the current behaviour in hardy for vbox is: break and fall apart in spectacular ways.
[17:24] <soren> laga: ...also not very useful :)
[17:24] <mathiaz> kirkland: because I want to be able to boot from RAID
[17:24] <kirkland> mathiaz: yeah, well, that causes seg faults
[17:24] <laga> soren: and for intrepid, it will be "futz with my Virtualbox config and possibly break it"? :(
[17:24] <kirkland> mathiaz: i might have a bug on that one open already, i'll look
[17:24] <soren> laga: For intrepid it says: I don't know what vbox is. Go away.
[17:24] <laga> soren: you get the point.
[17:25] <mathiaz> kirkland: ok - so how does this impact the test case for booting from degraded raid ?
[17:25] <soren> laga: But sure, we can add a --no-deploy option if you really think that's useful.
[17:25] <ivoks> ScottK: it's a very small contribution - an on-liner :)
[17:25]  * ivoks had way too much coffe today :(
[17:25] <soren> laga: I believe the sensible thing to do, though, is to add it to vbox.
[17:25] <laga> soren: users will pull their hair out in spectacular ways if the vbox plugin behaves completely different from the vmware plugin. if you deploy automatically, then the vmware images should also be added automagically to the vmware hypervisor.
[17:25] <soren> laga: Sure.
[17:26] <soren> laga: It's just not been implemented yet.
[17:26] <soren> laga: In part because I never bothered to figure out how.
[17:27] <kirkland> mathiaz: just don't mark the individual /dev/sda1 and /dev/sdb1 bootable ...  rather, /dev/md0 will be made bootable
[17:27] <mathiaz> kirkland: allright - I'll try that
[17:27] <laga> alright. i guess now i knew what that "deploy" thing in the wiki meant. actually adding a .vmdk file to vbox is not hard. i'll compile some information for you then
[17:27] <kirkland> mathiaz: i'll talk to cjwatson about this one
[17:27] <mathiaz> kirkland: great - thanks for the help.
[17:27] <laga> soren: and sorry for the long discussion, i should have made it clearer what i was talking about
[17:28] <soren> laga: I probably should, too :)
[17:29] <kirkland> mathiaz: no problem, thanks for the report
[17:30] <\sh> guys, A big Thank You from Netviewer AG/Webzooms AG to you for providing rock solid ubuntu server edition :)
[17:30] <mathiaz> kirkland: is there a bug report for this ?
[17:31] <laga> soren: btw, since you're the VM guy: latest KVM seems to support pci passthrough. do you know if special hardware support is required for that?
[17:32] <kirkland> mathiaz: not one i can put my finger on at this moment
[17:32] <kirkland> mathiaz: this is the issue, though
[17:33] <kirkland> mathiaz: i'm discussing with cjwatson in #ubuntu-installer
[17:33] <mathiaz> kirkland: ok - should I file one ?
[17:33] <kirkland> mathiaz: yes, please
[17:33] <kirkland> mathiaz: file against both kvm and partman-md
[17:33] <kirkland> mathiaz: the segfault part is clearly a kvm bug
[17:33] <kirkland> mathiaz: but partman-md might need something too
[17:34] <mathiaz> kirkland: ok.
[17:34] <soren> laga: I'm actually not sure, but I expect it requires an iommu.
[18:22] <mathiaz> kirkland: meh - kvm still segfaults even if I don't set the bootable flag when creating the partition.
[18:22] <kirkland> mathiaz: ???
[18:23] <mathiaz> kirkland: I'll try with dd files instead of qcow2
[18:24] <kirkland> mathiaz: i'm baffled
[18:24] <kirkland> mathiaz: i'd like to somehow see exactly what you're doing
[18:24] <mathiaz> kirkland: I'm gonna record a screencast then
[18:24] <kirkland> mathiaz: cool
[18:24] <kirkland> mathiaz: are you using the testing instructions from https://wiki.ubuntu.com/BootDegradedRaid ?
[18:25] <mathiaz> kirkland: not exactly - I'm just trying to setup a RAID array from the installer.
[18:26] <kirkland> mathiaz: that's what my instructions describe....
[18:27] <kirkland> mathiaz: anyway, i'll watch your screen cast
[19:01] <mathiaz> kirkland: hm - seems that qcow2 files are the problem
[19:01] <kirkland> mathiaz: interesting
[19:01] <mathiaz> kirkland: using files created with dd, I was able to setup a RAID array.
[19:02] <mathiaz> kirkland: and the system was able to go the base package install.
[19:02] <kirkland> mathiaz: k
[19:16] <nealmcb> mathiaz: Do you know what's going on with the problems with includes in old team reports and moin 1.6?  see https://bugs.edge.launchpad.net/ubuntu-website/+bug/276414
[20:51] <Dave-X> need help configuring my sshd to accept only certain hashing/ encryption algorithms
[20:52] <DubAndy> Hi! I've been trying to configure vnc4server on my ubuntubox. When I run either tightvnc for a windows computer or chicken of the vnc for a macintosh computer. All I get is an X11 connection with a hatched gray background and an x-cursor. It seemes like gdm is unable to startup. I've checked /.vnc/xstartup but I can't find anything.
[21:11] <Dave-X> need help configuring my sshd to accept only certain hashing algorithms
[21:29] <kirkland> Dave-X: http://manpages.ubuntu.com/manpages/intrepid/man5/sshd_config.html
[21:31] <Dave-X> i set Cyphers
[21:31] <Dave-X> and the man page doesnt say anything about hashing
[21:31] <Dave-X> couldyou help me hout?
[21:33] <Dave-X> i would like it to only use hmac-ripemd160
[21:40] <kirkland> Dave-X: on that page, search for "hmac-ripemd160"
[21:40] <kirkland> Dave-X: there's a hit about halfway down
[21:41] <kirkland>  MACs    Specifies the available MAC (message authentication code) algo‐
[21:41] <kirkland>               rithms.  The MAC algorithm is used in protocol version 2 for data
[21:41] <kirkland>               integrity protection.  Multiple algorithms must be comma-sepa‐
[21:41] <kirkland>               rated.  The default is:
[21:41] <kirkland>  
[21:41] <kirkland>                     hmac-md5,hmac-sha1,umac-64@openssh.com,
[21:41] <kirkland>                     hmac-ripemd160,hmac-sha1-96,hmac-md5-96
[21:41] <kirkland> Dave-X: MAC is another word for "Hash"
[21:43] <Dave-X> ayty
[21:45] <Dave-X> so
[21:45] <Dave-X> adding these 2 lines
[21:46] <Dave-X> Ciphers aes256-cbc
[21:46] <Dave-X> MACs hmac-ripemd160
[21:46] <Dave-X> will enable only thoes hashes, algorithms?
[21:46] <kirkland> to /etc/ssh/sshd_config
[21:46] <kirkland> right
[21:47] <kirkland> and then `/etc/init.d/ssh restart`
[21:47] <Dave-X> of corse
[21:56] <Dave-X> ty kirkland
[21:57] <kirkland> Dave-X: welcome
[23:56] <Ergo^> hello
[23:56] <Ergo^> i want to run something like : @reboot for i in /var/www/fastcgi/startup/*.sh; do $i; done
[23:56] <Ergo^> in cron
[23:56] <Ergo^> but it appears that @reboot command doesnt work
[23:56] <Ergo^> how do i deal with that ?