[00:45] <xibalba> how do you check mbuf's on ubuntu ?
[00:45] <xibalba> `netstat -m` is what would usually show it
[02:52] <KI7MT> .
[03:06] <jkitchen> KI7MT: sup greg.
[03:07] <jkitchen> didn't even need to guess :(
[03:07] <KI7MT> Hey what's up .. just writing some Ubuntu QA test cases.
[03:08] <KI7MT> In fact, doing Virtual Minimal and Virtual Minimal system installs via, server-iso
[03:09] <KI7MT> minimal system && virtual minimal system .. lol .. even confuzzed myself there fer a minute.
[10:02] <yolanda> morning jamespage, have a lot of MP ready for you :)
[13:57] <zul> hallyn:  ill get on libvirt and libvirt-python today
[13:57] <hallyn> cool
[13:57] <hallyn> thanks
[14:20] <rbasak> stgraber: re: bug 1204902, would a bind mount be similiarly objectionable?
[14:42] <zul> hallyn:  1.2.1 builds fine locally for me (that was too easy)
[14:44] <hallyn> zul: that's a pleasant change :)
[14:45] <hallyn> rbasak: what would we bind mount?
[14:46] <YamakasY> someone using partman schemes ?
[14:49] <hallyn> rbasak: Task t1 (lxc-start) clones task t2 (which becomes init) in a new mount ns.  any mounting done by t2 is not seen by t1.  To let t1 see t2's mounts we would have to do a scheme with mounts propagation, which would be ok if we know exctly how we are starting, but in the general upstream sense would be very fragile
[14:50] <hallyn> look at what we're having to do to account for hosts which either have root in ramfs, or have / mounted ms_shared
[14:50] <zul> smb/hallyn:  should i put these in a ppa for you to test
[14:51] <hallyn> zul: sure
[14:52] <hallyn> one day i'll need to build a new test image with acpi enabled for qrt so i can have 100% tests passing again..
[14:52] <smb> zul, would be nice. In theory you should have been able to drop two of my recent patches at least
[14:53] <smb> zul, And hopefully not have to fiddle with the ones we still need
[14:53] <zul> smb: i did and didnt need to fiddle
[14:54] <smb> zul, Makes me feel happy :)
[14:54] <zul> smb: ditto then i can work on other things :)
[14:54] <smb> hehe
[15:03] <zul> smb/hallyn: will be in here for trusty https://launchpad.net/~zulcss/+archive/libvirt-testing
[15:03] <ns5> what version of erlang does 12.04.03 have?
[15:04] <hallyn> zul: ok
[15:07] <smb> zul, Ok, I'll give it a whack in a bit
[15:08] <zul> thanks
[15:20] <stgraber> rbasak: AFAIK a bind mount is impossible
[15:21] <stgraber> rbasak: hmm, apparently it's not anymore, odd, I'm pretty sure this failed on ubuntu-touch when we tried...
[15:21] <ogra_> yep
[15:21]  * ogra_ remembers that too
[15:21] <stgraber> rbasak: so anyway, the problem you get with the bind-mount then is that it'll only show you the / of the container, not any of the sub mounts
[15:53] <phillw> Hi, I've just done do-release-upgrade -d but I do not have the 3.13 kernel (still 3.8) :(
[15:53] <phillw> https://help.ubuntu.com/12.04/serverguide/installing-upgrading.html
[15:55] <yctn> does anyone know if this issue still exsist in ubuntu 12.04 https://bugs.launchpad.net/ubuntu/+source/cyrus-sasl2/+bug/875440
[15:55] <Pici> phillw: so you're on 14.04 now?
[15:57] <phillw> Pici: that VM was told to upgrade... it ran for  couple of hours and is still on kernel 3.8 ... I'm guessing this may be a bug as my desktop upgrade has pulled in the new kernel series.
[15:59] <phillw> picking up bugs early is always good, especially as we have a lts --> ;ts release :)
[16:00] <Pici> phillw: Looks like linux-image-server is still at 3.13 on trusty.
[16:01] <phillw> Pici: so, how do I tell a server to get it?
[16:01] <Pici> phillw: You don't, there is not an updated package available.
[16:02] <phillw> (it is a VM and you are welcome to have a wander aroubd)
[16:03] <phillw> Pici: how do invoke drwxrwxr-x 23 ali ali     4096 Jan 12 10:04 linux-3.13-rc8
[16:03] <phillw>  to be used?
[16:03] <Pici> phillw: What do you mean by invoke?
[16:05] <phillw> use,... it's okay. I'll pop back to -kernel as there has been some issues with -rc7 that rc8 may have solved. that particular VM is for trialling 14.04 community bulds, so I do expect breakage as we learn how do them :)
[16:36] <smb> zul, Ok, completed a PXE/HVM install through libvirt-1.2.1/Xen/xl and it did not explode. Sounds good enough from that side ;)
[16:37] <zul> smb: awesome just need one more +1
[16:38] <phillw> zul: do you have the link to the test case?
[16:39] <zul> phillw:  what test case?
[16:40] <phillw> (16:36:37) smb: zul, Ok, completed a PXE/HVM install through libvirt-1.2.1/Xen/xl and it did not explode. Sounds good enough from that side ;)
[16:40] <zul> phillw:  no this is testing libvirt
[16:41] <smb> phillw, This is me running it
[16:41] <zul> phillw:  we never had a formal test case for libvirt
[16:42] <phillw> I run KVM, in 14.04 ubuntu and centos 6.4..... what would you like testing (apart from https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1080674 and a couple of others that are no-fix) ? :)
[16:44] <zul> uh?
[16:45] <phillw> smb: I connect to the server using kvm (and libvirt) from 14.04 to a 6.4 CentOS system... no bugs :)
[16:46] <phillw> virt-manager is a local GUI that makes life a bit easier :) It was the libvirt devs who told me that I was doing it all wrong when connecting :)
[16:50] <smb> It certainly beats doing most things manually... it has some drawbacks (virt-manager) in the implementation. At least when one is using a lot of lvs
[16:55] <phillw> smb: I don't know what lots are.... I have 9 VM's running. But, I'm not a crappy company like godadday :P
[17:02] <smb> phillw, Its not about the number of running VMs. I have 37 LVs defined and virt-manager becomes quite slow displaying them for selection.
[17:04] <phillw> smb: what do you use? i do dabble with virsh for some admin jobs, but am always open to new ideas :)
[17:07] <smb> phillw, Partially virsh, or working around that limitation by defining the LVs by hand and type in the path instead of using the search box. For my small scale testing that is good enough
[17:13] <hallyn> zul: all tests passed, that's a go :)
[17:24] <zul> hallyn:  cool thanks
[17:33] <phillw> smb: feel free to ping me if there is ever a requirement for a 2nd kvm to confirm things. I still retain one that is not allocated to any team.
[17:39] <smb> phillw, The KVM side usually is covered more. But you are always welcome to check out the early daily images for the next release with your use cases. And report bugs (ask bjf said). They really help not to forget things. And you get feedback independently from when things are done
[17:42] <phillw> smb: I run lubuntu 14.04 on my laptop and centos 6.4 on my server... both machines have kvm running. all any one need do is give me a ping (I also live on #phillw).
[18:19] <StathisA> anyone have experience with MSMTP?...i did the symbolic link for sendmail - but apticron reports: send-mail: invalid option -- 'r'
[18:20] <StathisA> msmtp should befully compatible with sendmail switches and exit codes...:-S
[18:38] <sarnold> StathisA: http://www.sendmail.org/~ca/email/man/sendmail.html
[18:38] <sarnold> StathisA: "-rname      An alternate and obsolete form of the -f flag."
[18:40] <sarnold> wow, -rname was listed as obsolete even back in freebsd 1.0 days: http://www.freebsd.org/cgi/man.cgi?query=sendmail&apropos=0&sektion=0&manpath=FreeBSD+1.0-RELEASE&arch=default&format=html
[18:41] <sarnold> wow, 4.2 BSD NET/2 even. it's been obsolete for a very long time indeed :)
[18:47] <StathisA> hmmm...how to get around it then...:-S
[18:53] <sarnold> StathisA: ideally, file a bug on apticron, and then if you've got the ability and inclination, download the sources, find the -r, write a patch to fix it, and ask for help to push the fix through the SRU process: https://wiki.ubuntu.com/StableReleaseUpdates
[18:54] <sarnold> StathisA: but be sure to file the bug regardless :) hehe
[18:55] <StathisA> ok thank you!
[19:26] <blz> Hello, I'm trying to install GRUB to an installation which currently does not boot (doing this from the rescue option of the Ubuntu server installer).  `grub-install /dev/sda` complains that "embedding is not possible, but this is required for cross-disk install".  What should I be doing differently?
[19:27] <hitsujiTMO> blz: did you chroot to the install os?
[19:27] <holstein> i always just refer to https://help.ubuntu.com/community/Boot-Repair and use a live CD when i can
[19:27] <blz> hitsujiTMO, I believe so.  At some point it asked me what I wanted to use as my root and I selected /dev/md0
[19:27] <hitsujiTMO> blz: wait soory never mind misread you question
[19:28] <blz> hitsujiTMO, no worries =)
[19:28] <hitsujiTMO> blz: what partition layout is it?
[19:28] <hitsujiTMO> mbr or gpt?
[19:28] <blz> hitsujiTMO, gpt
[19:28] <blz> hitsujiTMO, they're 3Tb drives (x3)
[19:28] <hitsujiTMO> then its just grub-install
[19:29] <blz> hitsujiTMO, haha that simple, huh?
[19:29] <blz> =)
[19:29] <blz> hitsujiTMO, hmm actually it's printing the help when I just do `grub-install`
[19:30] <hitsujiTMO> blz: this is a uefi install right?
[19:30] <blz> uefi?
[19:30] <blz> I installed this with  ubuntu-server's text installer
[19:31] <blz> hitsujiTMO, The thing is, my /  is on a RAID5 partition and my /boot is on /dev/sda1
[19:31] <blz> So I suspect this is why I'm having issues
[19:32] <hitsujiTMO> what type of drive is /dev/sda1 ?
[19:32] <blz> 3 Tb HDD
[19:32] <blz> hitsujiTMO, let me give you the whole layout... might be clearer
[19:32] <hitsujiTMO> blz is the a /boot/efi ?
[19:34] <blz> hitsujiTMO, `ls /boot` returns only a directory called `grub`
[19:34] <blz> in said directory, there is efiemu.mod, efiemu32.o and efiemu.64.o
[19:34] <hitsujiTMO> blz ok. then you should have a reserved boot partition.
[19:35] <blz> hitsujiTMO, yes, that would be /dev/sda1
[19:36] <hitsujiTMO> blz: thats not /boot right?
[19:37] <blz> hitsujiTMO, aaaah you have me stumped.  how can i check?
[19:37] <hitsujiTMO> parted -l
[19:37] <blz> hitsujiTMO, all I know is that I defined a 256 mb partition on /dev/sda1 and set it's mountpoint to /boot
[19:37] <blz> but yeah, let me check that
[19:39] <blz> hitsujiTMO, I have a line that reads `1   1049kb    256mb    ext2                         boot`
[19:41] <blz> boot being under the 'flags' column... I don't see any indication of it being mounted to /boot, though
[19:41] <blz> hitsujiTMO, unless I'm missing something obvious
[19:42] <hitsujiTMO> blz thats just sda1. hmm.
[19:42] <blz> hitsujiTMO, I only copied one line, there.  There's much more
[19:43] <hitsujiTMO> blz is there any partition thats only like 1mb in size?
[19:43] <blz> hitsujiTMO, no
[19:44] <blz> hitsujiTMO, this is a fresh install, so maybe I should just start over.  I'm just scratching my head at where I went wrong =/
[19:44] <hitsujiTMO> blz: does the server support uefi at all?
[19:44] <blz> hitsujiTMO, how can I check?
[19:45] <hitsujiTMO> blz: how old is it?
[19:45] <blz> hitsujiTMO, a few years old -- circa 2007, maybe?  It's a socket 775, I believe
[19:45] <hitsujiTMO> nah then it wont
[19:46] <blz> well shoot =/
[19:47] <hitsujiTMO> hmm.. not sure exactly it would have done. on debian it installs to a reserved bios partition on gpt.
[19:48] <zlatan> Hi! I'm having some trouble opening ports on my VPS to the outside world, would I be able to ask for some guidance here?
[19:49] <hitsujiTMO> blz: does parted -l list any unusual partitions types?
[19:49] <sarnold> zlatan: yes, but it might be worthwhile to shoot a question to your vps provider to find out if they're blocking ports in their network
[19:50] <blz> hitsujiTMO, other than raid partitions? no
[19:50] <hitsujiTMO> blz: sorry out of ideas then.
[19:50] <blz> hitsujiTMO, no worries.  thanks for your help =0
[19:50] <blz> =)
[19:50] <zlatan> sarnold: that sounds reasonable. I didn't think it would be a probable explanation for the problem, but I'll check with them just to be sure :-)
[19:51] <hitsujiTMO> blz: anything partition that says bios_boot ?
[19:52] <sarnold> zlatan: it happens surprisingly often :) hehe
[19:52] <zlatan> sarnold: thanks, I wouldn't have guessed that haha
[19:52] <blz> hitsujiTMO, nope
[19:53] <zlatan> sarnold: btw, is it likely that they seem to block basically all ports except 80?
[19:53] <sarnold> zlatan: it would be a bit draconian, but I suspect _someone_ out there does just that :) hehe
[19:53] <hitsujiTMO> blz: maybe this might help: http://bootinfoscript.sourceforge.net/
[19:54] <blz> hitsujiTMO, great, I'll read through it.  Thanks again!
[19:55] <hitsujiTMO> zlatan: whats the output of: sudo iptables --list
[19:56] <zlatan> hitsujiTMO: it's currently a very convoluted list, but I used 'ufw allow 8000' to open port 8000
[19:56] <zlatan> which didn't seem to work
[19:57] <hitsujiTMO> zlatan: whats output of: sudo ufw status
[19:57] <hitsujiTMO> zlatan: there may be a rule that's superseding your rule
[19:58] <zlatan> hitsujiTMO: it's a very barebones VPS that I just setup, so I deem it highly unlikely, but I'll paste the output here just to be sure
[20:00] <hitsujiTMO> zlatan: sudo apt-get install pastebinit && sudo ufw status | pastebinit
[20:00] <sarnold> pastebinit++  :)
[20:00] <zlatan> hitsujiTMO: as a test, could I use 'ufw disable' and try to access the port (via normal http) to see whether it'd work without any firewall? Or would that still leave some barriers intact?
[20:01] <hitsujiTMO> zlatan: no that should work if you disable ufw
[20:02] <zlatan> hitsujiTMO: let me recheck that first, I recall that I tried to do so and it still didn't work
[20:02] <hitsujiTMO> zlatan: if you disable ufw. whats the output of: sudo iptables --list
[20:06] <zlatan> hitsujiTMO: it says Status: inactive
[20:07] <zlatan> hitsujiTMO: sorry, wrong command
[20:07] <zlatan> hitsujiTMO: http://paste.ubuntu.com/6763942/
[20:08] <zlatan> hitsujiTMO: I also tried using netcat and tcpdump to listen for incoming packages, but nothing seemed to happen
[20:08] <hitsujiTMO> zlatan: and its still blocked now?
[20:08] <hitsujiTMO> zlatan: looks like they've some external firewall up if thats the case
[20:09] <zlatan> hitsujiTMO: I just ran 'tcpdump port 8000' and tried to access my server ip:8000, but the connection times out and I don't receive any packages...
[20:09] <zlatan> hitsujiTMO: that certainly seems likely now, yes
[20:12] <zlatan> hitsujiTMO: just a strange idea: could I try logging on to the server on two separate sessions and trying to send a request to the localhost:8000 from one session and listen to it from the other? To see whether it does accept local connections?
[20:12] <hXm> hello, I want to create a cloud storage in my server, I know there is openstack but that is a bootable OS no? exists an alternative?
[20:12] <hXm> something like dropbox could be just perfect
[20:13] <sarnold> hXm: look into owncloud
[20:13] <hXm> oh, i'll google that, thanks
[20:13] <holstein> or, share via ssh or whatever, and connect to the share..
[20:13] <sarnold> I'm happy enough with sftp but can see why someone would want more :) hehe
[20:14] <hXm> for me I just use sftp too, but for regular users they just imagine something like \\domain.com\directory and copy,paste,execute from it as a shared folder
[20:14] <hitsujiTMO> zlatan: no need to do a second sftp session. background the listening app with ctrl + z, send your request to localhost:8000, then foreground the listener with fg
[20:15] <hXm> owncloud is via web only
[20:15] <zlatan> ah right
[20:15] <zlatan> forgot about that
[20:15] <hitsujiTMO> zlatan: you should also look into screen of tmu
[20:15] <hitsujiTMO> zlatan: you should also look into screen of tmux
[20:15] <hXm> ah, ah, desktop clients too, nice
[20:15] <zlatan> hitsujiTMO: a regular wget should do the trick i guess?
[20:16] <hitsujiTMO> zlatan: yup
[20:24] <zlatan> hitsujiTMO: this is interesting. Listening on port 8000 doesn't do anyting whatsoever. However, when I let nginx run on port 80, then execute 'tcpdump port 80' and consequently do 'wget localhost:80', it does retrieve index.html and tcpdump shows that packages were received. However, when nginx is not running, running the same wget command merely returns a 'connection refused'
[20:25] <hitsujiTMO> zlatan: try: sudo tcpdump -i lo port 8000
[20:25] <zlatan> hitsujiTMO: sure
[20:30] <zlatan> hitsujiTMO: when I run 'tcpdump -i lo port 8000' and consequently run 'netstat -ntlp | grep "LISTEN"', I guess I should see an entry for tcpdump listening to port 8000? Since it doesn't show up...
[20:30] <sarnold> tcpdump won't be listening
[20:31] <sarnold> it throws the NIC into promiscious mode and reads packets raw through a different mechanism
[20:32] <zlatan> sarnold: sorry, didn't know that. Anyways, now that I've setup nginx to listen on port 8000, it does in fact work when I approach it locally
[20:33] <zlatan> sarnold hitsujiTMO: however, it still doesn't listen to my connection from the outside
[20:33] <sarnold> zlatan: try changing the tcpdump -i lo port 8000 to just tcpdump port 8000  ?
[20:34] <hitsujiTMO> zlatan: or: tcpdump -i eth0 port 8000                      if the external interface is eth0, but without it should work
[20:34] <zlatan> sarnold: now tcpdump doesn't see any packages arriving, even though nginx does send back index.html
[20:37] <miseria> "no estoy de acuerdo con la pena de muerte, al final las leyes sobrenaturales nos tienen condenados a morir" bienvenidos: http://castroruben.com *temo_a_un_ser_sin_rival*
[20:39] <zlatan> hitsujiTMO sarnold: only when i use '-i lo', it actually shows the packages coming in. Both of the others stay blank when i do 'wget localhost:8000'
[20:40] <sarnold> zlatan: perhaps when you just run 'tcpdump' it binds to the incorrect network interfae
[20:41] <sarnold> zlatan: check 'ip' output to find out the interfces on the system; you might need to select a different one
[20:42] <zlatan> sarnold: just 'ip'? It seems to need parameters?
[20:42] <zlatan> sarnold: 'ifconfig' maybe?
[20:43] <sarnold> zlatan: ifconfig also works
[20:44] <zlatan> sarnold: http://paste.ubuntu.com/6764105/
[20:44] <sarnold> zlatan: ha, that's easy enough. drat :)
[20:46] <zlatan> sarnold: just wondering, the things you're seeing so far... does it seem out of the ordinary (meaning that something like an external firewall might actually be in place) or is there still a good possibility that it's just my lack of skills in setting up server systems?
[20:48] <sarnold> zlatan: I'm not sure; packet captures can be funny things. it sounds like you've got your nginz running on 8000 now and it seems to be working, right?
[20:49] <zlatan> sarnold: only locally... when I try to approach it via a regular browser, the connection merely times out
[20:49] <sarnold> zlatan: times out, eh? that does sound like a firewall, one set to DROP packets rather than REJECT packets
[20:50] <zlatan> sarnold: connections time out on every port except 80, even when the local firewall is set to deny specific ports
[20:50] <sarnold> zlatan: that sounds very much like the VPS provider to me
[20:50] <zlatan> sarnold: that's good to know :-)
[20:51] <sarnold> zlatan: is it by chance amazon? if so you may need to configure new security groups
[20:51] <zlatan> sarnold: my friend says that it is in fact possible to run directadmin on the vps (via a custom installation process) which listens on port 2222... can this invalidate our conclusion, or might that be a side-effect of the custom installation process they offer?
[20:52] <zlatan> sarnold: no, a minor Dutch provider ;)
[20:52] <sarnold> zlatan: sorry, no idea on the direcadmin thing; if it is a web-based administration tool, please be cautious, those are notortiously insecure
[20:53] <rostam> Hi I am trying to start  dhcp server: service isc-dhcp-server. However, every time, I get the following message: init: isc-dhcp-server main process(...) killed by TERM signal.       why?
[20:53] <sarnold> rostam: is there anything in the logs that looks related?
[20:54] <rostam> sarnold, which log exacly I need to look for?
[20:54] <blz> I'm trying to install ubuntu-server to a software RAID volume, but I keep getting an error when installing grub.  The error reads "executing `grub-install /dev/sda/` failed".  What can I do to fix this?
[20:54] <blz> This is RAID5 on 3 gpt-partitioned HDDs
[20:54] <sarnold> rostam: /var/log/syslog -- and if the dhcp server has its own log file, that too..
[20:56] <zlatan> sarnold: it is indeed a web-based admin tool, but we're not installing that :-)
[20:56] <rostam> sarnold,  there is nothing about isc-server ...
[20:56] <sarnold> zlatan: oh, good. :)
[20:57] <sarnold> rostam: it'd probably have dhcp in the name
[20:58] <rostam> sarnold,  it seems it worked at some time, because I see DHCPREQUEST and DHCPACK but nothing else...
[21:01] <tc0nn> rostam: That's pretty normal. The "renewal" process is just that simple.
[21:02] <sarnold> rostam: hrm, so no messages that look out of the ordinary just before it blows up?
[21:02] <rostam> sarnold,  yes not other messages.
[21:03] <rostam> tc0nn, renewal process: how can I stop that..
[21:03] <tc0nn> That's normal. If it doesn't renew, it loses the DHCP lease.
[21:04] <tc0nn> Basically that's the DHCP client telling the server "hey dude, I'm still here, don't give away my IP address as I'm still using it"
[21:04] <rostam> tc0nn,  sarnold  thank you I think I understand now.
[21:05] <sarnold> rostam: if it happens again, I recommend running ubuntu-bug ich-dhcp-server
[21:06] <rostam> sarnold, thx
[21:09] <zlatan> sarnold hitsujiTMO: thanks for helping out, we'll see what our VPS provider can do for us :-) take care!
[23:01] <blz> Hello, I'm trying to install ubuntu-server to an LVM-on-RAID5 volume.  The installer runs its course and then fails when trying to install the bootloader.  I've tried installing to /dev/sda, /dev/mapper/md-0 and /dev/mapper/md-1, to no avail.  What gives?
[23:01] <blz> Also please note that these are 3 gpt-partitioned drives (>2Tb)
[23:04] <ikonia> blz: I assume you have a /boot partition that is a physical partition ?
[23:04] <blz> ikonia, no, I was under the impression that GRUB2 didn't require one
[23:05] <blz> ikonia, what's more, I tried it with a seperate /boot earlier and it still failed with the same message (although it was trying to install to /dev/sda, then)
[23:05] <blz> ikonia, google seems to suggest that using non MSDOS partition tables opens a lot of issues
[23:05] <maxb> blz: EFI or BIOS booting? Also, are you able to run 'parted --list' or similar and pastebin it to show us the full details of your partition layout?
[23:06] <blz> maxb, I'm not sure about EFI vs BIOS -- i suspect BIOS as the computer is a few years old.   As for the pastebinning, I'm still in the graphical installer, so cutting&pasting (let alone internetting) is a bit of a challenge
[23:06] <blz> maxb, but if you tell me what to look for, I can relay the information from parted -l
[23:07] <blz> Oh also, I should mention that I'm still in the installer.  I can drop to a shell via ctrl+shift+F2
[23:09] <maxb> So, you have sda, sdb and sdc, each with a GPT? What partitions do they contain? What makes up md-0 and md-1 (cat /proc/mdstat will help for this)
[23:10] <blz> maxb, yes, sd[abc], each GPT, each containing one partition marked as physical volume for raid
[23:10] <blz> I built a RAID5 array using the above partitions
[23:10] <blz> and an LVM on top of that
[23:10] <blz> maxb, the installer is presently hanging on the final bit where it installs the bootloader
[23:10] <maxb> Something seems a bit wrong if you have *two* md devices in such a setup
[23:11] <blz> maxb, yeah i thought that was rather strange, too
[23:11] <blz> let me copy down the output from ls -l /dev/mapper
[23:11] <blz> it'll be a bit quick & dirty...
[23:11] <maxb> Jump to a shell and have a look at /proc/mdstat - it should tell you which underlying devices make up each md device
[23:13] <blz> http://paste.ubuntu.com/6764728/
[23:13] <blz> maxb, did I say `md` ?  I meant `dm-0` & `dm-1`
[23:13] <blz> I think these are the logical volumes on my LVM
[23:14] <maxb> Oh, right, that makes all the difference :-)
[23:14] <blz> maxb, my bad =)
[23:15] <blz> /proc/mdstat appears empty, btw
[23:15] <blz> but that might be because i'm in the installer's shell ?
[23:15] <maxb> It sounds to me like /dev/sda would be the correct installation device
[23:15] <maxb> mdstat will be empty because it turns out you're not using md :-)
[23:15] <blz> maxb, oh right. yes that makes sense
[23:15] <blz> maxb, I'll try /dev/sda again, but that failed in the past
[23:16] <blz> let me try again, though
[23:16] <maxb> I'm not all that familiar with where the graphical installer logs to. Do you get any messages about *how* it is failing?
[23:16] <blz> yeah it's throwing an error.
[23:16] <maxb> which is? :-)
[23:16] <blz> maxb, well no -- that's the frustrating thing
[23:16] <blz> it says "Executing grub-install /dev/sda failed.  this is a fatal error"
[23:17] <maxb> How 'helpful'
[23:17] <blz> maxb, no kidding
[23:17] <maxb> Could you run parted -l and tell me what the "Start" value is for the partitions?
[23:17] <blz> sure, one sec
[23:18] <maxb> The partitions on sd[abc], that is, ignore any stuff about the dm devices
[23:18] <blz> hmm it's saying parted not found
[23:18] <blz> let me reboot into the rescue option
[23:18] <blz> and drop to a shell there
[23:18] <maxb> Or try fdisk instead
[23:19] <blz> just plain `fdisk` ?
[23:19] <blz> or fdisk -l ?
[23:19] <blz> eeh, fdisk not found either
[23:19] <blz> i'll bring up the rescue shell
[23:25] <blz> maxb, hmm I have some new information
[23:25] <maxb> ok
[23:25] <blz> maxb, I ran the 'repair a broken system' option in the ubuntu-server installer and opened a shell in /dev/root/system
[23:26] <blz> which is my LVM logical volume mapped to /
[23:26] <blz> and i tried running grub-install /dev/sda from there
[23:26] <blz> maxb,  I got a more informative error:  "/usr/sbin/grub-setup:  warn:  this GPT partition label has no BIOS Boot Partition; embedding won't be possible!"
[23:27] <blz> maxb, followed by "Embedding is not possible, but this is required when the root device is on a RAID array or LVM volume
[23:27] <blz> certainly more "helpful" than "Error. it didn't work" :D
[23:27] <maxb> Ah, so it is that problem.
[23:27] <blz> I'm all ears =)
[23:28] <maxb> GRUB needs space to put the code which actually understands how to boot off LVM - but there is no space available in which to put it
[23:28] <blz> maxb, ah, so how would I go about fixing this?
[23:28] <blz> maxb, I'm willing to reinstall if that's simpler/cleaner
[23:30] <maxb> In a MBR partition table setup, the space is made by leaving unpartitioned space before the first partition. In a GPT setup, it's the same idea, but instead of unpartitioned space, you have to allow a partition of a special type, which GRUB knows belongs to it for this purpose
[23:30] <maxb> So unfortunately, you will need to repartition
[23:32] <blz> maxb, ok that's not a problem
[23:32] <blz> so basically I'll need to make a small partition at the begining of every disk and format it as a specific type of partition?
[23:33] <blz> how large and what's the type called?
[23:34] <maxb> If you are creating it using parted, the incantation to mark it as the right type is 'set (number) bios_grub on'
[23:34] <blz> maxb, i was hoping to create it during the partitioning menu of the installer
[23:34] <maxb> As for size, I'm not sure, but GRUB seems happy with only 1MiB on my LVM-without-RAID setup
[23:35] <blz> maxb, ok -- is this something I can do from the installer?
[23:35] <blz> or should I run parted first?
[23:35] <maxb> I'm unsure whether the graphical installer is going to give you the flexibility you need.
[23:35] <ikonia> blz: apologies I had to make a phone call
[23:35] <ikonia> blz: what's the current situation ?
[23:35] <blz> ikonia, no worries! Thanks for checking in =)
[23:36] <maxb> ikonia: Needing to repartition so that GRUB has room for embedding
[23:36] <ikonia> embedding ?
[23:36] <blz> what maxb said ^^
[23:36] <ikonia> apologies, I've missed part of the conversation, I'm certainly interested
[23:38] <blz> ikonia, I'm not sure how well I can explain it seeing as I just found out that this thing existed
[23:39] <blz> ikonia, but in a nutshell it would seem that GRUB2 expects a small partition at the beginning of each of my drives from which it can load the stuff it needs to boot from GPT
[23:39] <blz> and I'm now wondering how to set up such a partition from the installer
[23:39] <maxb> embedding is when GRUB needs to write some of its support code somewhere on the disk (in this case, the code that it uses to know how to read stuff from LVM)
[23:40] <ikonia> maxb: are you talking about the stage files ?
[23:40] <maxb> stages is what Grub 0.9x called it
[23:40] <ikonia> yes, but in essence the same thing
[23:40] <ikonia> that's why it needs a /boot partition
[23:40] <ikonia> (or it's better to have one)
[23:41] <maxb> It doesn't *need* a /boot partition
[23:41] <blz> ikonia, oh so this is /boot we're talking about?
[23:41] <ikonia> maxb: it needs somewhere, so /boot is the logical place.
[23:41] <ikonia> blz: yes,
[23:41] <maxb> A /boot partition would be solving the same general problem in a different way
[23:41] <blz> I tried installing a /boot at one point and I got the same error
[23:41] <ikonia> blz: a seperate boot partition or a logical volume /boot
[23:42] <blz> ikonia, seperate /boot partition
[23:42] <blz> initially I had a 256 mb /boot partiton on sda
[23:42] <ikonia> blz: that should be fine
[23:42] <blz> and 256 mb swap partitions on sd[bc]
[23:42] <ikonia> 256mb swap partition ??
[23:43] <blz> and then each drive had a (3tb - 256mb) physical volume for raid
[23:43] <blz> ikonia, two of them
[23:43] <ikonia> hardly seems worth it
[23:43] <blz> ikonia, well it was mostly to do something with the extra space
[23:43] <ikonia> blz: how big is your boot disk ?
[23:43] <blz> ikonia, I'm trying to boot from a 3-disk RAID5
[23:43] <ikonia> just put swap on an lv
[23:43] <ikonia> how big is the physical disk ?
[23:43] <blz> 3x 3TB
[23:43] <ikonia> so they are gpt partitions
[23:44] <sarnold> .. and yet, 256M swaps are 16 times larger than the first swap partitions I had..
[23:44] <blz> ikonia, yes
[23:44] <ikonia> is the motherboard using uefi ?
[23:44] <blz> sarnold, crazy isn't it?
[23:44] <hushnowquietnow> Quick question for you fine ubuntans
[23:44] <sarnold> blz: yeah :)
[23:44] <blz> ikonia, no, BIOS
[23:44] <ikonia> blz: how is your raid setup, mdadm ? fake raid ? hardware raid ?
[23:44] <blz> sarnold, my first gen iPod video has more RAM than my mother's PC
[23:44] <hushnowquietnow> I've got an Apache process that's eating up loads of cpu and memory.  How can I figure out which of my vhosts is causing the problem?
[23:44] <blz> ikonia, software
[23:44] <ikonia> blz: mdadm ?
[23:44] <sarnold> blz: haha, nice
[23:45] <blz> ikonia, i assume so?
[23:45] <blz> ikonia, I just went through the partition manager in the ubuntu server installer
[23:45] <ikonia> blz: the ubuntu installer is creating the raid ?
[23:45] <ikonia> ok, so mdadm
[23:45] <blz> and selected "physical volume for raid" and whatnot
[23:45] <blz> ikonia, yeah
[23:45] <ikonia> blz: so here is my suggestion
[23:46] <blz> ikonia, I'm all ears =)
[23:46] <sarnold> hushnowquietnow: my first thoght is that the vhost that is causing the problems is likely the most active vhost -- can you check logs to get a feeling for which one is most active?
[23:46] <ikonia> blz: make a 300mb partition on each disk, sda1, sdb1, sdc1 - put it in raid 1 + 1 spare create 1 partition on the other disks filling the whole disk sda2, sdb2, sc2 - raid them in raid 5, put that only (not the sda1/b1/c1 mirror) under lvm control
[23:46] <ikonia> blz: make the mirror /boot - write the grub to the mbr of your boot disk (probably sda)
[23:47] <ikonia> I suspect you'll be fine
[23:47] <blz> ikonia, OK,  let me say it all back to you to make sure I get it...
[23:47] <blz> ikonia, step 1)  create small partition on all drives, RAID1 it (with one spare drive)
[23:48] <ikonia> tick
[23:48] <blz> ikonia, do I make a /boot on top of that raid?
[23:48] <hushnowquietnow> sarnold: That's a pretty straightforward way to go
[23:48] <blz> or no?
[23:48] <ikonia> blz: so the mirror + space = /boot
[23:48] <ikonia> blz: yes
[23:48] <ikonia> + spare
[23:48] <ikonia> not space
[23:48] <blz> ok cool
[23:48] <blz> right right
[23:48] <hushnowquietnow> I was hoping there'd be some sort of command I can feed the pid into and get more info on it though
[23:48] <maxb> ikonia's method should work, but I still prefer the bios_grub method - it seems less complicated to not have to set up a separate /boot with a different raid level
[23:48] <blz> step 2:  make a partition using the remaining space on all three drives.  RAID5 that.  LVM on top of that.  install as usual
[23:48] <blz> step 3:  GRUB2 to /dev/sda
[23:49] <ikonia> maxb: I don't disagree
[23:49] <sarnold> hushnowquietnow: well now, you say that.. maybe there is a way to read the apache scoreboard
[23:49] <ikonia> blz: you got it
[23:49] <blz> ikonia, cool
[23:49] <blz> perfect
[23:49] <blz> maxb, ikonia so i shoudl go with that?  Or should i do the bios_grub approach?
[23:50] <blz> ikonia, more exactly, why do you recommend this as opposed to maxb's suggestion?
[23:50] <sarnold> hushnowquietnow: ooh, investigate how this works: http://www.apache.org/server-status
[23:50] <sarnold> hushnowquietnow: that might do it :)
[23:50] <ikonia> blz: it's hard to be %100 clear on this, I prefer my approach because I use it, maxb's is correct there is nothing wrong with his approach at all
[23:51] <blz> ikonia, haha ok, except that i'm not sure I understand how to do his
[23:51] <hushnowquietnow> sarnold: just might!  Thanks! :)
[23:51] <blz> oh well.  I'll just try yours, ikonia.  At least I'm clear on what needs to be done =)
[23:52] <maxb> So I've started the ubuntu server .iso in a VM. Does it even *have* a graphical installer?
[23:52] <ikonia> maxb: all ncurses based I thought
[23:53] <maxb> That's what I thought too, but blz was talking about a graphical partitioner
[23:53] <hushnowquietnow> sarnold: `apachectl fullstatus` looks to be what I want!
[23:54] <parallel21> maxb: Why do you want a graphical installer?
[23:55] <maxb> I don't. But blz was asking if it was possible to create a bios_grub partition in it
[23:57] <sarnold> hushnowquietnow: nice, thanks! :)
[23:57] <ikonia> maxb: that's half the reason I was sceptical, I'm not sure how the installer would deal with it
[23:58] <parallel21> maxb: should do it automagically I think. Just curious, is there a difference between a grub partition and a bios_grub partition?
[23:59] <parallel21> maxb: sorry... I just caught up to the conversation. Disregard