[00:41] <SpamapS> oh nice.. maverick 32 bit vm doesn't even show plymouth
[00:41] <SpamapS> barely flashed on before the login prompt showed up
[01:14] <jkugler> Trying to do updates on a 10.04 server box.  have googled the problem, not gotten anywhere.  While extracting the archives, I am told "No space left on device."  "device" in this case being /usr.  /var has 18GB free, /usr has 464MB free, /tmp has 873MB free.  Anyone have any idea what's going on?
[01:20] <jkugler> must go for now...only have one network connection at this site.
[01:51] <The_Paco> yo. I'm finding Ubuntu 10.04 Server's Samba options/setup seems about twice as involved as setting up an OpenSuse box was. Is there any recommended reading/info for all the new options/setup that the smb.conf is throwing at me?
[01:52] <sebikul> @The_Paco check this page, https://help.ubuntu.com/community/SettingUpSamba
[01:55] <The_Paco> sebikul, thanks. I'm digging through it's links for more info on setting up a WINS server, but I'm not sure if I should bother doing that again
[01:57] <The_Paco> I'm trying to serve one shared folder of files across a VPN tunnel (that doesn't support subnet bridging, unfortunately) so that all our windows xp boxes can see it. WINS seems to be unsuited for that. Is there something I should look to instead? We're a small business with no AD service or anything like that yet
[01:59] <sebikul> mmm, i dont know how to do that, i only have experience with small home networks
[02:01] <The_Paco> *nods* ditto
[02:01] <weekly> Question: I have a machine thats spewing out kernel messages into my terminal, i can still use the terminal but these line keep appearing, how do i make them stop!?
[02:03] <The_Paco> I was halfway through using OpenSuse 11.2 and having it set up, but the X11 crap kept crashing and it was just too much for the hardware I had. I was finding there that I'd need to have a server on each side serving wins and sharing wins routing tables between them. Ubuntu's Samba seems to have bigger ideas, and a lot of options I'm not seeing in the old online oreilley books, either.
[02:08] <sebikul> you can try using gadmin-samba to set-up your network, you can find it in the synaptic, i am almost sure you will be able to do it using it
[02:08] <pmatulis> The_Paco: one share?  that's pretty easy
[02:09] <pmatulis> The_Paco: what's the problem?  there are a zillion guides out there for that
[02:12] <pmatulis> The_Paco: but i would recommend the ubuntu server guide
[02:12] <The_Paco> sebikul, I'll check it out, thanks; pmatulis, well, in this case I was trying to enable and allow all the windows machines to see each other using SFS, with a set of crap routers that won't bridge subnets. I had half a solution with one wins server that both routers would point to, but the other subnets systems would not transmit to or acknowledge it, regardless of their settings. Rather than have two systems using 2% of their power running wins and k
[02:12] <The_Paco> eeping a decentralized and insecure windows SMB environment, I'm now checking out my options for using something a little more centralized with individual logins/shares/security levels.
[02:13] <The_Paco> they could resolve share names on the other side of the router, but wouldn't share their own or 'see' the shares in network neighborhood, I mean
[04:15] <vbabiy_> Hey guys any idea why this would be happening: http://dpaste.com/250320/ look at lines 5, and 8
[05:09] <n3kl> Hello.
[05:09] <n3kl> Nice to see there is a server chanel as well.  To much man ls going on in the other.
[05:09] <n3kl> I am getting strange errors with ubuntu-vm-builder on lucid.  Can someone in here helo me figure out whats going on, please?  http://pastie.org/1188100
[05:10] <n3kl> Fresh install of the package
[05:10] <n3kl> Networking is setup and ready to go with a br0 interface and all
[05:13] <twb`> n3kl: I think it's pissed off because $LANG isn't set
[05:13] <n3kl> I wondered?  should that get passed somewhere, or do I need to setup my locales?
[05:13] <twb`> It should be set automagically by pam_environment or pam_<something else> in lucid
[05:14] <twb`> How did you log in?
[05:14] <twb`> More to the point: check if $LANG is set in that shell.
[05:15] <n3kl> ssh
[05:15] <n3kl> no
[05:15] <n3kl> $LANG  is not set
[05:16] <n3kl> so, what should it be set to, and where do I set it?
[05:16] <n3kl> seems strange that I should have to change anything with pam to build a vm, no?
[05:16] <twb`> en_AU.utf8, where "en" is your language code and "AU" is your country code
[05:17] <twb`> Check "locale -a" for a list of valid languages
[05:17] <n3kl> C and Posix.
[05:17] <n3kl> so I need to modify /etc/locales and run locale-gen, eh?
[05:18] <twb`> Just run this for now:
[05:18] <twb`> LC_ALL=C LANG=C ubuntu-vm-builder kvm lucid
[05:18] <n3kl> Looking much better
[05:18] <twb`> If that works, you can deal with locales later.  If it still fails, you need to report the bug against ubuntu-vm-builder because I don't know what's going on
[05:18] <twb`> OK, then just report a bug against it saying that it assumes $LANG is set
[05:19] <twb`> and include that backtrace that you pastebinned
[05:19] <n3kl> Cool, thank you
[05:28] <Exploiter> hello everyone
[06:12] <_Techie_> is there a max speed set on the repo's?
[06:15] <_Techie_> as im getting about one tenth the speed ive clocked downloads running at
[06:17] <twb`> Are you using your ISP or university's tertiary mirror?
[06:17] <_Techie_> im using whatever mirror that the installation chooses, its a server located in a datacenter
[06:19] <twb`> Well, obviously the first thing to do is confirm that it's pointing to a nearby mirror
[06:20] <Exploiter> yes
[06:20] <_Techie_> im using us.archive.ubuntu.com
[06:23] <_Techie_> how can i determine the closest mirror to the server?
[06:24] <twb`> In Debian there's a machine-readable list of mirrors that apt-spy(8) uses, and synaptic appears to have something similar built-in, but I don't know how you're supposed to do it on Ubuntu Server.
[06:24] <twb`> As for me, I know /a priori/ where my nearest mirror is.
[06:25] <SpamapS> twb`: where is the list for debian?
[06:25] <twb`> I can't remember
[06:25] <twb`> somewhere on www.d.o
[06:25] <SpamapS> I tried netselect-apt once
[06:26] <SpamapS> but it didn't find any mirrors
[06:26] <twb`> http://www.debian.org/mirror/index.en.html
[06:26] <twb`> ...except I thought it was text/plain, so maybe I misremember or it has changed
[06:44] <Kaelten> just verifying here, if I use mounting options to mount a directory to another directory ont he disk, then it's affectively that the one directory is in two palces at once
[06:44] <Kaelten> right?
[06:45] <joschi> Kaelten: if you mean `mount --bind [...]` then the answer is yes
[06:46] <Kaelten> joschi /var/www/uploads/attachments /var/www/uploads/170 bind defaults,bind 0 0
[06:46] <Kaelten> in fstab
[06:46] <Kaelten> assuming thats the same thing
[06:46] <joschi> it is
[06:46] <Kaelten> fun.....
[06:46] <twb`> Usually a symlink is a better idea
[06:47] <twb`> If only because it's less confusing for other sysadmins
[06:47] <Kaelten> ya, I'm reversing someone elses work so I can move it to a new server
[06:47] <Kaelten> if I mount bind a parent dir binding identical sub dirs would have no effect I'm assuming
[06:47] <Kaelten> like if /x and /y are bound and then I did /x/a and /y/a
[06:47] <Kaelten> it'd really just be a wasted mount
[06:48] <joschi> yes
[06:48] <Kaelten> but what would happen if I went after that and then did /y/a /y/b
[06:48] <joschi> twb`: symlinks have some short comings though. not every program likes symbolic links plus a bind mount doesn't need another inode
[06:49] <joschi> Kaelten: you mean `mount --bind /y/a /y/b`?
[06:49] <Kaelten> yes
[06:49] <Kaelten> after having already bound /x and /y and then binding /x/a and /y/a
[06:49] <joschi> Kaelten: you'd have /x/a -> /x/b also since /x and /y are the same directory
[06:50] <joschi> Kaelten: just think of it as /x and /y being "views" on the same on-disk data
[06:50] <joschi> binding /x/a to /y/a is not possible, since when there's /x/a there's automatically /y/a
[06:50] <Kaelten> persumably ubuntu just ignores the impossiblity
[06:51] <Kaelten> or lets it do it anyway
[06:51] <joschi> that has nothing to do with ubuntu itself but with the linux kernel (or more precisely the posix layer of the respective file systems)
[06:52] <Kaelten> right
[06:52] <Kaelten> I really wish this was more straightforward
[06:52] <Kaelten> it was done ass backwards
[06:52] <twb`> joschi: OTOH NFS doesn't like bind mounts
[06:52] <twb`> And df, find -xdev and friends will not necessarily do what you expect.
[06:53] <twb`> I guess I'm just saying be careful
[06:53] <Kaelten> I just need rsync to work properly
[06:54] <twb`> rsyncd or rsync-over-ssh?
[06:54] <Kaelten> the latter
[06:55] <twb`> In what way is it failing?
[06:55] <Kaelten> it's not yet
[06:55] <Kaelten> but I just realized that instead of writing something that'd copy the files betwen the dirs or any number of other soltuions
[06:56] <Kaelten> they setup 8 interconnecting binds
[07:35] <\sh> why do we still enable apache2 httpd TRACE method by default? Any Reason? This Method is only useful in testing environments, but not for production ready systems, and will always be raised by External Security Audits...
[07:43]  * pennyless is away: Gone away for now
[08:08] <kees> \sh: I don't think trace has been allowed for a long time. do you have some examples of it working?
[08:09] <\sh> kees: looks like it was still enabled on jaunty, but lucid does has a default of TraceEnable Off...
[08:10] <\sh> I was looking on a jaunty webserver ... sorry
[08:10] <kees> \sh: ah, good. yeah, I was looking at /etc/apache2/conf.d/security and couldn't tell if I changed it manually or not
[08:11] <\sh> well, I got the results of our Security Audit and they were complaining about that setting...but I'm still running jaunty on our production servers, which will be changed during the next three months...
[08:12]  * kees goes to bed for real now.
[08:12] <\sh> kees: btw...do you have any clue about rdesktop on linux in combination with rdp SSL/TLS connections towards Windows Servers?
[08:13] <kees> \sh: I haven't used it myself, but let me ask a friend that uses rdesktop a lot
[08:13] <\sh> kees: that would be great...thx for that :)
[08:14] <kees> np, I'll email you when he answers me tomorrow :)
[08:14] <\sh> kees: thx :)
[08:16] <kees> \sh: oh, he's still up. says he just uses regular rdp, no ssl
[08:16] <kees> \sh: he's on a vpn so he hasn't bothered digging into it when it was introduced
[08:16] <\sh> kees: oh well...I need to enable ssl/tls on the windows server side, and I don't know if that works...well, I have to test it somehow then
[08:16] <kees> cool. good luck! :)
[08:16]  * kees really really going to bed now. g'night
[08:16] <\sh> kees: have a good night :)
[08:18] <kees> \sh: maybe check http://forum.nginx.org/read.php?30,101161   really^3 off now
[10:27] <koltroll> Hey guys. On my server I've set up a lamp install with ssh and ftp. Now I'm thinking, is there any reason for me to have a ftp-server when I can just use sftp?
[10:27] <koltroll> Are there any downsides with skipping the vsftp-server ?
[10:27] <twb> koltroll: no
[10:27] <twb> Indeed, it's a move I heartily support
[10:29] <koltroll> twb, another question. Is it somehow possible to set default permissions when using sftp just as you can configure the vsftp-server? In a way I think it sounds strange, but I'll ask anyway.
[10:29] <twb> You mean the umask?
[10:30] <koltroll> yea I guess (?) :) I want the uploaded files to be readable and writeable for the user & group.
[10:31] <twb> That should be handled via pam_umask.so
[10:32] <koltroll> oh ok. I'll look that up. thanks
[10:32] <twb> man pam_umask
[10:33] <twb> pam is not for the faint of heart
[10:33] <koltroll> haha then maybe I should ignore it :)
[10:33] <twb> You could use /etc/login.defs if you want it to apply to EVERYBODY
[10:34] <twb> But I suspect that's probably bad juju
[10:34] <koltroll> I think I rather think through how I want the permissions, and how I need it. It's just that I got everything working now, and I'm really not a server guy :)
[10:35] <koltroll> Right now I got my user, web. All web pages are in /home/web. web is in the group www-data. FTP-files uploaded with web can be accessed and changed by the www-data user as well.
[10:36] <koltroll> Also I got a git user /home/git that's also in the same group, and therefor can unpack repos into /home/web for projects we work on.
[10:36] <koltroll> But I belive the only thing that would be changed by this, if I just ignore the pam_umask thing, is that on folders that the user www-data needs write-access to I'll set them manually.
[10:37] <twb> That is, I confess, what I do for one site
[10:37] <koltroll> Ah ok.
[10:38] <twb> It's pretty fugly; doing it "properly" with access control would be preferable, but I don't know how
[10:38] <koltroll> :)
[10:38] <twb> chmod -vR go-rwx /home/*/ |
[10:38] <twb> # Suppress printing of (i.e. whitelist) some expected false positives.
[10:38] <twb> egrep -v 'retained as|nor referent has been changed'
[10:39] <twb> ...that's what I run to prevent users from sharing data clandestinely
[10:39] <koltroll> oh ok
[10:39] <koltroll> Yea I'll think this through. Thanks alot!
[10:39] <twb> Actually it should read go-rwx,a-st -- somehow the code has regressed :-/
[11:08] <doko> cjwatson, ttx: is libgeronimo-jacc-1.1-spec-java found on a server CD?
[11:12] <cjwatson> ubuntu-server/daily/current/maverick-server-amd64.list:/pool/main/g/geronimo-jacc-1.1-spec/libgeronimo-jacc-1.1-spec-java_1.0.1-1fakesync1_all.deb
[11:12] <cjwatson> ubuntu-server/daily/current/maverick-server-i386.list:/pool/main/g/geronimo-jacc-1.1-spec/libgeronimo-jacc-1.1-spec-java_1.0.1-1fakesync1_all.deb
[11:12] <cjwatson> (yes)
[11:15] <doko> are these .list files in a convenient location to check them myself?
[11:15] <doko> still need http://packages.debian.org/changelogs/pool/main/g/geronimo-jacc-1.1-spec/current/changelog :-/
[11:16] <doko> ahh, ttx not here
[11:59] <cjwatson> doko: http://cdimage.ubuntu.com/ubuntu-server/daily/current/
[12:33] <BiG-M> hi, anybody knows if we can stop kthreadd or not  ??
[12:35] <jpds> kthreadd?
[12:35] <BiG-M> there was no process like before update, now it appears with lots of spawns
[12:36] <BiG-M> its 10.04 lst headless
[12:36] <BiG-M> * LTS
[14:06] <raubvogel> Anyone using vsftpd?
[14:06] <_ruben> !anyone
[14:08] <raubvogel> Is anon_root still available in vsftpd or there is a better way?
[14:25] <kinygos> hi...probably a daft question, but is it possible to change the partitions on an already installed ubuntu 10.04 server if i don't mind losing the data in the current mount points?
[14:29] <gentooxer> kinygos: yes its possible
[14:29] <gentooxer> kinygos: you can even preserve the data ...
[14:30] <gentooxer> kinygos: do you want to shrink, or enlarge?
[14:30] <kinygos> thanks :) from what i've read, that's what i thought...but i now have my bigger problem...
[14:31] <kinygos> i have a remote server with a 2nd physical hard disk that has no partitions defined or mounted...so what i was hoping to do was create partitions on that second drive, mount them, then configure RAID1 across the physical disks (for a given partition)
[14:33] <ttx> Daviey: everything looking good ?
[14:33] <Daviey> ttx: Yup.. we had a respin
[14:33] <Daviey> (not our request, late last night)
[14:33] <gentooxer> kinygos: you have to set up the 2nd device as a md device
[14:33] <ttx> Daviey: I saw that this morning
[14:33] <Daviey> ttx:  so the euca upgrade issue is resolved.
[14:34] <kinygos> according to fdisk -l, i have a partition whose system is extended...but i don't see it mounted anywhere in fstab...is extended system special?
[14:34] <ttx> I hit the print server issue, its resolved for post-RC
[14:34] <kinygos> gentooxer: ooh, i didn't realise that...thanks :)
[14:34] <Thelz> Hello. I've got a network here that consists of windows computers and one ubuntu server running the samba file server. Earlier today, something went wrong, and the file system must've gotten partially corrupted. Other computers were no longer able to access the ubuntu server, and the ubuntu server went in read-only mode.
[14:34] <kinygos> gentooxer: i thought i had to do that after creating the partitions
[14:35] <Thelz> Now I'm busy copying the files on the harddisk to an USB drive, but this seems to generate a lot of errors. I'm not sure if the files are really gone, or if it's a permission problem, and if there's some way to save the files.
[14:36] <gentooxer> kinygos: no you set up a RAID1 md device with only 1 device present; copy the data from the 1st device to your new RAID device
[14:36] <gentooxer> kinygos: then you have to "rebuild" the RAID, and you are ready to go
[14:36] <Thelz> To top it all off, the Ubuntu Server is running on two harddisks that are in mirror mode due to a FakeRaid setup using a Sweex controller.
[14:36] <kinygos> gentooxer: do i need to create corresponding partitions on the new device?
[14:37] <gentooxer> kinygos: but I guess a new installation is much easier
[14:37] <Daviey> ttx: Yeah... saw that.. that sounds good - those upgrading will get the fix and it's a transitional issue.. not an explosion :)
[14:37] <gentooxer> kinygos: they have to be big enought to store you data, thats all
[14:37] <kinygos> gentooxer: i know a new install is much easier, but i'm having a lot of problems arranging that, esp as the server is remote
[14:38] <kinygos> gentooxer: thank you very much for the steer...your time is very much appreciated :)
[14:50] <cjwatson> Daviey: for the record, the respin was a server team request
[14:51] <cjwatson> 21:25 <mathiaz> hi - I've just upload a fix for eucalyptus to maverick
[14:51] <cjwatson> 21:25 <mathiaz> and ttx fixed a problem in the seed which lead to cups not working in the installer
[14:51] <cjwatson> 21:26 <mathiaz> The later will require a -server iso respin
[14:51] <cjwatson> 21:26 <mathiaz> and I'd like to have the new eucalyptus included in it
[14:51] <cjwatson> 21:26 <mathiaz> so - could the eucalyptus package be accepted in maverick?
[14:52] <Daviey> cjwatson: Oh, sorry - i wasn't pointing fingers.  I spoke to mathiaz not long before that, and thought we agreed we weren't going to push for one.
[14:53] <Daviey> but i am pleased that it did :)
[15:01]  * zul pauses
[15:06] <ScottK> \sh: Are you coming to UDS again?
[15:10]  * zul returns
[15:12] <Thelz> Does anyone know? If I configured a Sweex controller to set 2 SATA disks in RAID1 mode, and then install ubuntu server, is the data only on the first drive, or on both drives? (Since Ubuntu Server still seems to detect both drives)
[15:12] <slestak> i am trying to forward root's mail on my 10.04 lts server install to my gmail account.  I havent setup an MX record or anything because I do not want to provide mail services, just forwarding
[15:13] <ruben23> hi guys i get to start my linux server when i startup i get grup loading 1.5, grub loading please wait.....error 15 ----> any idea please..?
[15:13] <slestak> i have an /etc/aliases entry for root : steve and an /etc/email.addresses entry for steve : myeml@gmail.com
[15:14] <cemc> slestak: you could just have root: myeml@gmail.com in /etc/aliases if I'm not mistaken
[15:14] <slestak> i thought this would be sufficient.  using mailx to send a mail to my work exchange server works fine
[15:14] <slestak> cemc: i tried that first but was getting rejections from gmails smtp
[15:14] <slestak> using mailx to send to local user root fails
[15:14] <cemc> slestak: rejections?
[15:14] <slestak> i have an example
[15:15] <slestak> sec
[15:16] <slestak> http://paste.pocoo.org/show/269001/
[15:16] <slestak> this looks like I am trying to offer up slestak.homelinux.org as a mail service, which is overkill for my needs
[15:17] <slestak> i could prob remove exim and just use bsmtp or sth liek that
[15:18] <cemc> slestak: I don't get it... do you want to send root emails to google? why is google trying to send email back to you?
[15:18] <slestak> i want my root mails to be sent to my gmail address.
[15:18] <ruben23> how do i boot my server to get on a recovery mode whern i have commandline login as root..?
[15:18] <slestak> that is all i am trying to accomplish
[15:18] <slestak> so I am beginning to thing that exim or postfix are overkill
[15:20] <slestak> ruben23: there is a recovery mode avaiable in yor grub menu, iirc
[15:20] <slestak> i think you can do a sudo -i to get a root prompt in that shell
[15:21] <slestak> what do you guys do with root mail on non-mail servers?
[15:22] <\sh> ScottK: nope...this time I don't have the time, nor I get a sponsoring of my company
[15:22] <ruben23> slestak: is this upon reboot po to..?
[15:22] <cemc> slestak: I think you still need some basic postfix, even if it's not open to the internet
[15:22] <\sh> ScottK: hopefully next year again somewhere in europe
[15:22] <slestak> i think i have used bsmtp in teh past for that.  since it is a multiverse app i wanted to see what the common best practicve is
[15:23] <\sh> now lets have a look at the rcbug list
[15:24] <zul> ttx: why are we still testing esx?
[15:26] <ttx> zul: I think it's done in the certification labs
[15:26] <zul> ttx: okies
[15:30] <cemc> slestak: do you have static IP address with a good reverse on that host?
[15:30] <slestak> looks like esmtp is in apt, but is end-of-lifed per its own webpage.  msmtp looks liek a winner
[15:30] <slestak> cemc: no, i have dynamic dns.  its a home server
[15:31] <cemc> slestak: that could be a problem then, when sending mail to the internet, it might get rejected
[15:31] <ScottK> slestak: The odds of you being able to deliver mail to gmail from a dynamic IP are very low.
[15:31] <cemc> slestak: you should use a smarthost for sending mail, you shouldn't send directly
[15:32] <slestak> would you recommend postfix?
[15:32] <slestak> or exim?
[15:32] <cemc> slestak: me personally, definitely postfix
[15:33] <alok_rocks_> hello everybody
[15:33] <alok_rocks_>  have a problem with phpmyadmin, i have installed it but not working    http://localhost/phpmyadmin>Not  Found  The requested URL /phpmyadmin was not found on this server. Apache/2.2.14 (Ubuntu) Server at localhost Port 80
[15:49] <ttx> JamesPage: just assigned you to bug 651155
[15:50] <ttx> I won't have time to look into it, so it's better if you can have a look
[15:53] <ScottK> ttx: We'll need someone to be in charge of server boot experience.  It seems very popular on ubuntu-devel.  I could run a UDS session on what people would like to see, but it would be more useful to get someone who might actually be able to do the work involved from the start.  I added it to the wiki page.
[15:54] <slestak> it looks liek ssmtp is providing what i need
[16:00] <JamesPage> ttx: no problem - looking at it now - feel free to keep throwing this sort of stuff at me !!
[16:01] <blackxored> I know this is the simplest of questions, but still can't remember now, what's the rewrite rule I can use for when I go to http://myhost It actually goes to http://myhost/dir ???
[16:01] <blackxored> apache, TW
[16:06] <ttx> ScottK: did you get my answer ? I'm having connection issues
[16:06] <ScottK> ttx: I did not.
 ScottK: yes, that's what I had in mind under the "upstart/plymouth for servers" heading
 ScottK: I long advocated for a server "low level" position (aka server foundations) or for a Foundations "server" position to bridge the gap
 ScottK: not completely sure which track that would land to, though :)
[16:07] <ScottK> ttx: OK.  Thanks.  I'm not sure either, that's why I didn't reply to one of the track specific emails.
[16:07] <ScottK> (BTW, I think this is a bug in the new track definitions).
[16:07] <blackxored> neverming I spot it
[16:15] <ttx> JamesPage: do you have an opinion on JOnAS
[16:15] <ttx> ?
[16:18] <zul> ttx: i do its a good weezer song :)
[16:30] <JamesPage> ttx: just talking ISO testing with mathiaz
[16:31] <kinygos> hi...i wonder if someone could spare me a few moments...i have a ubuntu 10.04LTS server with 2 disks, only 1 is partitioned, and i'd like to implement RAID1 across both drives...following some earlier advice, i've come up with a series of commands to create the raid on the second disk, copy the first disk to it, then add the first disk logically to the raid.  the commands are listed here http://dpaste.com/250536
[16:32] <kinygos> my question is, when i reboot, how can i tell the server to use the raid on the second disk?
[16:32] <RoyK> in bios
[16:32] <RoyK> if the server is fairly reinstalled, reinstalling directly on software raid1 will probably be of less hassle
[16:33] <RoyK> kinygos: also, remember to install grub on that drive
[16:33] <kinygos> bugger...it's a remote server, and i don't have budget to get external media mounted on it :(
[16:34] <kinygos> RoyK: thanks for that...i knew i'd miss something
[16:34] <RoyK> you might be able to configure grub to boot from the new disk, but I would recommend doing that onsite
[16:35] <RoyK> the chances of something going bad are always high when fiddling with these things from remote
[16:35] <kinygos> RoyK: i do have a lights-out board in there, so i can reinstate the ISP base image for the server wuite easily
[16:35] <kinygos> RoyK: i don't have any data on it yet...just preparing the box for my application
[16:35] <RoyK> with the lights-out board, don't you get remote console with bios for it?
[16:36] <kinygos> RoyK: ooh..i didn't check that...lol (feeling foolish)
[16:36] <RoyK> I would guess it should be quite easy to make someone insert an ubuntu cd on it without too much cost?
[16:36] <RoyK> my ISP does that for free :P
[16:38] <kinygos> RoyK: unfortunately not...the ISP has 9.04 as base image, and i'm on a very tight budget...it will cost £60 per half hour engineer time to give me some disk space for an .iso image somewhere :(
[16:38] <kinygos> (and yep, i value my time more, but my sponsor doesn't see it that way)
[16:39] <ttx> smoser: how is image testing going ?
[16:39] <SpamapS> kinygos: where is the ISP base image mounted?
[16:39] <RoyK> kinygos: iirc 9.04 can be installed on raid1
[16:39] <SpamapS> yeah
[16:39] <SpamapS> kinygos: install 9.04 on RAID1, then upgrade to 9.10, then 10.04
[16:39] <ttx> Daviey, mathiaz: will shut down in 15 minutes, let me know if you have any question
[16:40] <SpamapS> kinygos: and tell your ISP to update to 10.04!
[16:40] <ttx> Daviey, mathiaz: looks like we are good to go for RC
[16:40] <mathiaz> ttx: great
[16:40] <mathiaz> ttx: Daviey: catchup on mumble?
[16:40] <RoyK> SpamapS: if they charge GBP 60 for half an hour fiddling, they will probably charge GBP 1k for burning out a new 10.04 CD
[16:40] <ttx> mathiaz: I can try -- though I don't have my headset
[16:41] <SpamapS> true
[16:41] <SpamapS> kinygos: how much are you paying per month? Maybe you should look into an IaaS provider. ;)
[16:41] <RoyK> kinygos: can you install 9.04 from scratch without paying them much?
[16:41] <kinygos> £79 per month...but i've paid for a year :(
[16:42] <smoser> ttx, i've run the tests, i have to go through and look at results. there was one set of failures that i think must have been test case failures.
[16:42] <SpamapS> kinygos: WOW
[16:42] <smoser> i'll start posting results soon.
[16:42] <kinygos> unfortunately it's an image blat...not an installation process that i can get involved in
[16:42] <SpamapS> kinygos: I hope thats a really powerful machine.
[16:42] <kinygos> i'm starting to feel nervous
[16:43] <RoyK> kinygos: man, that's a lot, is this a VM or a PM?
[16:43] <SpamapS> kinygos: Ok, you *can* build a broken RAID1, copy the system to it, and then boot from that, and add the other disk to it....
[16:43] <kinygos> dedicated server
[16:43] <RoyK> k
[16:43] <Daviey> mathiaz: sure, give me a few mins tho
[16:43] <Daviey> ttx: no questions here.
[16:43] <kinygos> i need a dedicated server as i'm gonna put postgresql and django on it
[16:43] <SpamapS> kinygos: a lot can go wrong, and the grub bit may get.. scary. ;)
[16:44] <RoyK> kinygos: ic, but with the lights-out board, you can probably do a lot from remote
[16:44] <SpamapS> kinygos: FYI, the cloud providers out there, like EC2 and Rackspace, will run a db service for you so you don't need such a powerful machine.
[16:44] <kinygos> SpamapS: i know :(  i've just started looking at how to install grub on the second disk and i'm flapping a bit
[16:45] <SpamapS> kinygos: its not that bad. we'll hold your hand.
[16:45] <SpamapS> kinygos: and as long as you have that lights out board, you can fix whatever breaks. :)
[16:45] <kinygos> :) thanks i really do appreciate your time...i'll go figure out the grub piece and report back
[16:46]  * RoyK wonder wth an ISP will be using 9.04 as the standard where LTS releases are so much better supported
[16:46] <SpamapS> yeah
[16:47] <SpamapS> they sound completely nutty
[16:47] <SpamapS> in fact, isn't 9.04 dead next month?
[16:47] <kinygos> it gets worse...they're planning on dropping ubuntu and continuing with CentOS!!!!
[16:47] <RoyK> SpamapS: not for servers - three years support for servers
[16:47] <RoyK> kinygos: heh - idiots
[16:47] <ScottK> RoyK: No.
[16:47] <ScottK> SpamapS: Yes.
[16:47] <SpamapS> RoyK: you sure? I don't think thats true.
[16:48] <RoyK> so far the only reason we have 9.04 on some boxes is that 10.04 doesn't ship with g77 :þ
[16:48] <ScottK> RoyK: Server support is 5 years on LTS, 18 months on everything else.
[16:48] <kinygos> upgrading from 9.04 to 9.10 then 10.04 was a breeze (thanks to the ubuntu community)
[16:48] <SpamapS> ScottK: FYI, mtaylor and I decided it may be better to not have drizzle in maverick, and just publicize the PPA.
[16:49] <ScottK> SpamapS: Then please file a bug asking for package removal.
[16:49] <SpamapS> ScottK: when the decision is final, will definitely do so.
[16:49] <JamesPage> ttx: don't have an opinion of JOnAS yet - I'll take a look and compare with JBoss
[16:49] <ttx> JamesPage: cool
[16:49] <RoyK> ScottK: you're right
[16:50] <RoyK> ScottK: I was mixing up desktop and server support....
[16:51] <RoyK> desktop support for LTS is 3yr
[16:51] <ScottK> Yep.
[16:51]  * ScottK notes Dapper on servers has 9 months to live.
[16:52] <RoyK> kinygos: remember to configure update manager not to upgrade to 10.10 when that arrives (vi /etc/update-manager/release-upgrades ...)
[16:52] <RoyK> well, unless you want inter-lts-releases
[16:52] <ScottK> That's actually default for LTS releases.
[16:53] <RoyK> ScottK: not if upgrading from non-LTS-releases
[16:53] <RoyK> upgrading from 9.10 to 10.04 won't change that
[16:54] <RoyK> I just upgraded a set of servers from 9.04 to 10.04 and had to change that thing myself
[16:58] <afeijo> I created a .bashrc at my home folder, but its not working when I log in.  What else need to be done?
[16:59] <RoyK> afeijo: normally there is a .bashrc in all user folders (copied from /etc/skel with useradd -m)
[16:59] <RoyK> and the work well
[16:59] <RoyK> afeijo: pastebin the .bashrc file, please
[17:00] <afeijo> RoyK: it has just 2 lines with alias cmd
[17:00] <afeijo> my folder had no .bashrc, pretty weird
[17:01] <afeijo> I will copy from skel and try again
[17:01] <RoyK> afeijo: cp -a /etc/skel/. $HOME
[17:01] <RoyK> or -R even
[17:01] <RoyK> doesn't matter - just make sure your user owns the file
[17:01] <RoyK> s/file/files/
[17:03] <afeijo> it didnt work :( something is missing in this server
[17:03] <afeijo> it is not reading my .bashrc
[17:05] <afeijo> and why when I run sh .bashrc it dont execute my aliases ?
[17:08] <hggdh> because when you run 'sh .bashrc' you are running on a new shell. You should instead source it: '. .bashrc'
[17:08] <afeijo> hggdh: thanks :)
[17:08] <afeijo> that help, but it would be better if it ran automatically at login :(
[17:09] <hggdh> .bashrc should executed on every new shell you open. Close your current (or logoff) and open it again, it should be there
[17:09] <afeijo> I did that
[17:10] <hggdh> afeijo: what is your login shell (look at your /etc/passwd entry)
[17:11] <afeijo> /home/lg:/bin/bash ?
[17:13] <hggdh> OK, yes
[17:14] <kinygos> RoyK: i was afk...just got your message about configuring update manager...thank you for that
[17:14] <hggdh> afeijo: on *login* ~/.profile will be executed. In there (default .profile) you should have a . .bashrc
[17:15] <hggdh> afeijo: so... is it there?
[17:16] <afeijo> bingo!
[17:16] <afeijo> hggdh: someone changed that file, it had a typo :)
[17:16] <afeijo> fixed, thanks!!!!!
[17:16] <afeijo> I fixed in the /etc/skel too
[17:17] <hggdh> afeijo: cool
[17:19] <afeijo> hggdh: last question if I may. Where would I add new glogal aliases (i.e. alias l="ls -lah --color=auto") so my users wont have to do it at their .bashrc ?
[17:22] <hggdh> afeijo: probably on /etc/profile.d -- create a new file there, based on the existing ones
[17:27] <ScottK> Good server upgrade test on i386.
[17:27] <zul> JamesPage: ping
[17:27] <JamesPage> zul: pong
[17:27] <zul> JamesPage: why did the automated testing miss the samba amd64 one last night?
[17:28] <JamesPage> It didn't - the test has a bug which causes it to fail as nmbd does not start quickly enough.
[17:28] <JamesPage> I think the only test which is an issue is the print-server one which ttx is aware or.
[17:28] <zul> JamesPage: ah ok
[17:29] <JamesPage> Postgresql has a similar issue with the actual test script itself.
[17:37] <kinygos> i'm planning on installing RAID1 on a second physical disk, copying my first disk contents to the second disk, rebooting then adding the first disk to the RAID...i'm wondering if the following is valid for installing grub on the RAID that i've created  # grub-install /dev/md1 (i'm sure it can't be this simple)
[17:38] <kinygos> or should i be installing grub on the underlying disk (grub-install /dev/sdb) ?
[17:40]  * RoyK doesn't remember
[17:40] <RoyK> google for it
[17:42] <kinygos> ok, i will try it...i've read somewhat conflicting instructions...which makes it somewhat baffling
[17:57] <kinygos> lol...dd takes for ever
[17:58] <kinygos> that wasn't a criticism btw...just an observation that i should've kept to myself in hindsight
[18:07] <RoyK> kinygos: dd?
[18:07] <RoyK> kinygos: you can easily copy a filesystem with rsync
[18:07] <RoyK> without copying the empty bits
[18:09] <kinygos> d'oh!
[18:10] <kinygos> <-- will never make a sysadmin
[18:11] <kinygos> i guess you learn from experience
[18:12] <RoyK> hehe
[18:18] <SpamapS> kinygos: The only way any of us has learned, is by admitting that we know nothing. ;)
[18:19] <SpamapS> which is why we stop learning and turn into kurmudgeons ;)
[18:19] <kinygos> :)
[18:22] <SpamapS> kinygos: to answer your earlier question, yes install it on the underlying disk.
[18:22] <SpamapS> kinygos: are you still dd'ing ?
[18:24] <kinygos> i am indeed
[18:26] <SpamapS> kinygos: its probably *safer* to use rsync/cp than dd
[18:27] <SpamapS> kinygos: otherwise the volume sizes might be slightly different
[18:29] <kinygos> SpamapS: should i cancel and start again with rsync?
[18:30] <SpamapS> kinygos: I would
[18:31] <SpamapS> kinygos: you'll need to mess with a few things like UUID, but that shouldn't be so terrible. ;)
[18:31] <kinygos> SpamapS: i was thinking of destroying the RAID i'd created...go back to an unmounted, unpartitioned disk
[18:32] <SpamapS> kinygos: no, the RAID is fine
[18:32] <SpamapS> kinygos: I mean, it should be
[18:32] <SpamapS> kinygos: as long as you marked the partitions as type FD (raid autodetect)
[18:33] <SpamapS> hrm this is getting beyond my recent experience.. int he past I'd have chrooted into the new root filesystem and ran 'mkinitrd' ...
[18:35] <kinygos> all i did was create the RAID on the disk, update the raid config on the original disk, and then ran update-initramds -u
[18:36] <kinygos> i never even knew mkinitrd existed
[18:43] <SpamapS> on the disk?
[18:43] <SpamapS> so no partitions?
[18:43] <SpamapS> kinygos: update-initramds is better. ;)
[18:45] <SpamapS> kinygos: when you say you that you "created the raid on the disk".. did you create partitions, or just do the whole disk, like , /dev/sdb ?
[18:49] <kinygos> SpamapS: apologies for the delay, was afk...i used the following: mdadm --create /dev/md1 --level=1 --raid-devices=2 missing /dev/sdb
[18:50] <SpamapS> kinygos: bad idea
[18:50] <SpamapS> kinygos: create a partition
[18:50] <SpamapS> kinygos: otherwise your raid array may not be properly autodetected
[18:51] <elb0w> Does ubuntu have a built in firewall that would block me from connecting to a remote db on 3306?
[18:52] <kinygos> SpamapS: my first disk has 5 partitions (iirc)...should i match them, or just create any parition?
[18:54] <SpamapS> kinygos: the simplest thing is yes, to duplicate the partition table.
[18:54] <SpamapS> elb0w: it shouldn't be blocking anything outgoing by default
[18:55] <JamesPage> mathiaz or ttx: I just pushed a branch for bug 651155; please could you review (server-mrs)
[19:07] <JasonMSP> Ii've got squirrelmail working on my server.  Now I am trying to setup the virtualhosts to point each mail.mywebsite.com to the squirrelmail folder so that rather then the user typing www.mywebsite.com/squirrelmail they can just type mail.mywebsite.com  virtualhost file.  Going to the mail subdomain gets me a blank page.
[19:20] <DualProxy> Anyone know how to list members of groups that are admins (`gpasswd -A`)???
[19:22] <JasonMSP> you can cat /etc/group and read the names next to the group name.
[19:24] <DualProxy> yeah, I tried that but it does not specify if they are admin or just members.
[19:26] <JasonMSP> grep ^admin /etc/group  will also work
[19:29] <DualProxy> I'm not talking about the group "admin", I mean the group "status" of a member. If you use `gpasswd -A usr grp` that user becomes an admin of the group, and can add/remove members from only that group.
[19:34] <RoyK> ffs - the internet connection regularly drops around 8-9pm
[19:39] <bobslaede> Hi, which package is 'killall' in? apparently its not installed by default in my VM guest
[19:42] <RoyK> roy@lamia:~$ dpkg -S `which killall`
[19:42] <RoyK> psmisc: /usr/bin/killall
[19:43] <bobslaede> RoyK: yeah, its not installed on my guest os
[19:43] <bobslaede> ubuntu
[19:43] <ScottK> bobslaede: http://packages.ubuntu.com/search?searchon=contents&keywords=killall&mode=exactfilename&suite=lucid&arch=any
[19:43] <ScottK> (you'll want the second one on the list)
[19:44] <bobslaede> ah, i get it. thanks ScottK and RoyK :)
[19:44] <ScottK> You're welcome.
[19:52] <illytacos> Hola amigos, I have a bit of a situation and I know I've been in here asking a lot of questions and I appriciate the help!!!!!! I have some questions around configuring a SAMBA file server (using Ubuntu Server Edition 10.04) Right now all I need to do is be able to see the file server on a windows machine. I'm following this article: https://help.ubuntu.com/community/MountWindowsSharesPermanently am I more or less on the righ
[20:06] <illytacos> Hey folks, sorry to be a pest, anyone around to assist with this question? -> I have some questions around configuring a SAMBA file server (using Ubuntu Server Edition 10.04) Right now all I need to do is be able to see the file server on a windows machine. I'm following this article: https://help.ubuntu.com/community/MountWindowsSharesPermanently am I more or less on the right track or am I spiraling out in the wrong directi
[20:06] <st_iron> as i know you have to use the cifs.mount command
[20:07] <st_iron> and give the recommended parameters to it
[20:07] <illytacos> ahhh ok so not smbfs it's cifs
[20:07] <st_iron> shortly: you have to mount it like a samba share
[20:07] <st_iron> sure let's try
[20:08] <illytacos> hm ok, so I think I'm stuck on SAMBA share because this is my first time ever setting this up and configuring.
[20:08] <kinygos> i suspect my server crashed during the dd, but on reboot, ignoring the serious errors found on each partition, i have a running os with RAID1 "configured" on the second disk...
[20:09] <st_iron> it is not difficult, just read the docs when you have free time :)
[20:09] <kinygos> unfortunately, when i try to install grub on that disk, i get the error "the file /boot/grub/stage1 not read correctly"
[20:09] <illytacos> st_iron: I've been reading https://help.ubuntu.com/community/MountWindowsSharesPermanently and going through it but is this the correct article? I'm scared I might have picked up the incorrect documentation
[20:11] <illytacos> learning on the fly. always fun lol.
[20:12] <st_iron> illytacos: I have used cifs at work for mount the windows and samba shares
[20:12] <st_iron> use the 'man cifs' command and you can get the correct configuration
[20:12] <st_iron> you learn more when you make it yourself like a puzzle
[20:12] <kinygos> i'm baffled...i've pasted my fdisk -l output to http://dpaste.com/250649/  can anyone spare me a few moments to suggest some checks i can make ?
[20:12] <illytacos> st_iron: thanks!
[20:13] <st_iron> illytacos: you're welcome
[20:16] <kinygos> is it possible to manually install grub on a hard disk?  i can find instructions on how to install on a floppy disk...
[20:17] <st_iron> kinygos: grub-install command ;)
[20:18] <kinygos> i did that, but it fails...it finds /boot/grub, but then says it cannot read /boot/grub/stage1 correctly
[20:20] <kinygos> i suspect that it's trying to read the file that has been dd'd onto that disk, but is unreadable as it is under raid
[20:25] <kinygos> wow...i have a real feeling i'm close...if SpamapS is still on, i could sure use some advice....
[20:25] <kinygos> i booted onto the second hard disk, and i now have a grub interactive prompt
[20:26] <kinygos> is it possible to install grub on that disk from that prompt?
[20:31] <JasonMSP> i figured it out
[20:32] <JasonMSP> wrong window.. sorry!
[20:33] <kinygos> can anyone confirm, if my second disk was listed as (hd1) in devicemap, root (hd1) will install grub on it?
[20:34] <kinygos> sorry, root (hd1) followed by setup (hd1) ?
[20:50] <mibocote> top and free indicate that the server is using 6.5/8G, but adding up the memory usage of all the processes gives about 1.6G. The 6.5G is without buffers/cache. Any ideas?
[21:14] <b0gatyr> hi everyone, what's a good content management system or portal for use in a corporate environment where they need to post files/docs and allow only certain people to view them?? I was thinking Moodle? any other suggestions??
[21:18] <kinygos> is there a way to tell if dd is working or hung?
[21:20] <shauno> kinygos: it'll tell you its progress if you send it SIGHUP
[21:24] <scrllock> running strace on the PID works too
[21:24] <kinygos> shauno: thanks...just trying to figure out how to do it :)
[21:24] <kinygos> scrllock: i'll look at that too...thanks :)
[21:26] <kinygos> hmm...i'm "running" dd in a remote kvm session (lights out board)...logging in through putty and i only see 2 processes running...bash and ps...am i really tired, or does that mean dd is not running?
[21:27] <scrllock> what does 'ps aux | grep dd'
[21:27] <scrllock> say?
[21:27]  * pennyless is back.
[21:27] <kinygos> scrllock: lots of baffling stuff...
[21:28] <scrllock> mind putting it in a pastebin?
[21:28] <kinygos> sure
[21:29] <kinygos> scrllock: http://dpaste.com/250696/
[21:29] <kinygos> scrllock: thank you for helping
[21:29] <scrllock> ok, the 1641 PID is the one you want
[21:29] <scrllock> strace -p 1641
[21:30] <scrllock> or sending it it SIGHUP
[21:30] <kinygos> strace -p 1641 has generated shedloads of stuff...and it's still going
[21:30] <kinygos> can i ^C that?
[21:30] <scrllock> sure
[21:31] <scrllock> so you know it's still going then
[21:31] <kinygos> lol...awesome :)  thanks scrllock
[21:31] <scrllock> np sir
[22:01] <SpamapS> kinygos: how is the raid setup going?
[22:01] <SpamapS> kinygos: I see you continued the dd... I'm pretty sure thats going to fail. :-/
[22:13] <kinygos> SpamapS: slowly...i thought dd had hung earlier when it hadn't...i've started it again since...waiting for it to complete (i know you said it's a bad idea...but i kinda had it in my head to try this way and if it fails, start again tomorrow morning)
[22:15] <kinygos> SpamapS: i pasted my fdisk -l dump http://dpaste.com/250649/...the disks have the same characteristics, so it may succeed
[22:15] <kinygos> SpamapS: if you have 2 secs, maybe you could have a quick look at them and tell me if anything looks obviously wrong?
[22:17] <SpamapS> kinygos: It really doesn't matter if the disks are the same characteristics
[22:18] <SpamapS> kinygos: Oh hmm.. MD seems smarter than I had thought
[22:18] <SpamapS> kinygos: I suppose it will work then. :-P
[22:18] <kinygos> SpamapS: maybe :)
[22:19] <kinygos> SpamapS: i hope i can install grub on that second disk...
[22:19] <SpamapS> kinygos: The usable blocks is the same, so yeah, maybe
[22:30] <SpamapS> kinygos: 250G is going to take a long time. ;)
[22:31]  * RoyK taps his 50TB setup
[22:31] <kinygos> SpamapS: am considering going to bed and checking back in the morning :)
[22:32] <SpamapS> RoyK: 50's pretty good.. I remember being excited with my first 1.5TB box.. now I have 1TB in my backpack. ;)
[22:32] <kinygos> SpamapS: i will be so chuffed if this works
[22:32] <RoyK> SpamapS: we're setting up a couple of 115TB systems soon for bacula
[22:33] <SpamapS> kinygos: if you've got a blog, post your experiences.. would be good as encouragement for the next person who needs to do that. ;)
[22:33] <mathiaz> SpamapS: o/
[22:33] <SpamapS> RoyK: removable disks too?
[22:33] <SpamapS> mathiaz: \o
[22:33] <RoyK> SpamapS: zfs
[22:33] <mathiaz> SpamapS: would you have time to do maverick-server-i386 RAID1 test case?
[22:33] <RoyK> SpamapS: 8 drives in each raidz2 vdev
[22:33] <mathiaz> SpamapS: http://iso.qa.ubuntu.com/qatracker/test/4609
[22:34] <mathiaz> SpamapS: it's on the last test case to go thourgh
[22:34] <SpamapS> mathiaz: was just about to start it. :)
[22:34] <mathiaz> SpamapS: I'll do the preseeded
[22:34] <kinygos> SpamapS: definitely...gotten so much help online, it will be so good to make a contribution, however small
[22:34] <mathiaz> SpamapS: o^5
[22:34] <RoyK> SpamapS: it should be quite safe
[22:34] <SpamapS> mathiaz: I wanted to discuss the test instructions w/ somebody before editing them..
[22:34] <SpamapS> mathiaz: I believe there are some bad practices suggested in the test.
[22:34] <mathiaz> SpamapS: for example?
[22:35] <mathiaz> SpamapS: are you referring to http://testcases.qa.ubuntu.com/Install/ServerRAID1?
[22:35] <SpamapS> mathiaz: disconnect 1 drive, boot, check, halt, reconnect drive, disconnect other, boot, check, halt, reconnect drive... produces a VERY unstable system, which I would actually expect.
[22:35] <SpamapS> mathiaz: yeah
[22:35] <SpamapS> mathiaz: you create diverged RAID1's
[22:36] <SpamapS> where its not clear which one is "correct"
[22:36] <mathiaz> SpamapS: right - I notice some issue with this one as well
[22:36] <mathiaz> SpamapS: last time I ran the test the RAID1 array would actually autosync
[22:36] <SpamapS> so we just need to plug instructions in there to reconnect and boot again.
[22:36] <SpamapS> mathiaz: its a race condition
[22:36] <SpamapS> mathiaz: SOMETIMES it can tell
[22:36]  * mathiaz nods
[22:37] <SpamapS> but we shouldn't test for something we would never support
[22:37] <mathiaz> SpamapS: the main point of the test case should be to cover a real-world use case
[22:37] <mathiaz> SpamapS: it seems that the sequence you've outlined doesn't necessarly make sense
[22:37] <SpamapS> if somebody did this in the real world, they just broke their RAID ;)
[22:37] <mathiaz> SpamapS: well - it may happen unintenionaly
[22:37] <mathiaz> SpamapS: bad luck
[22:38] <SpamapS> so it should be disconnect disk0, boot, check, halt, reconnect disk0, boot, check, halt, repeat w/ disk1
[22:38] <mathiaz> SpamapS: ie: reboot a system the controller goes flaky
[22:38] <mathiaz> SpamapS: right - that makes more sense
[22:38] <mathiaz> SpamapS: it may be worth thinking about corner cases though
[22:38] <SpamapS> mathiaz: right but the MD driver doesn't really have a way to guarantee it can recover from this split raid situation.
[22:39] <mathiaz> SpamapS: right - in that case it should just drop and no try
[22:39] <mathiaz> SpamapS: it seems that it still tries correctly
[22:39] <SpamapS> If nothing else..
[22:39] <SpamapS> it should drop to a "CRAP we can't tell which disk is right!" state and ask you to pick.
[22:39] <SpamapS> but thats really.. way way deeper than I think we are expecting from MD
[22:42] <mathiaz> SpamapS: yeah - we had a long conversation about what should be the safe behavior
[22:42] <mathiaz> SpamapS: it's a complicated issue
[22:42] <mathiaz> SpamapS: and corner cases matters a lot
[22:42] <SpamapS> It really, really, must not be complicated.
[22:42] <SpamapS> If it is, then we fail. :(
[22:42] <mathiaz> SpamapS: as we really don't wanna scratch data
[22:42] <mathiaz> SpamapS: it's complicated to detect in which state you are IIRC
[22:42] <SpamapS> Right, it should be a "fail early, by default" type of system.
[22:43] <kinygos> that test case as described without booting with 2 disks in between is a highly unlikely corner case that would probably be caused by some other part of the system (imho ofc)
[22:43] <mathiaz> SpamapS: kees and kirkland discussed that for hours in the intrepid cycle IIRC
[22:43] <SpamapS> mathiaz: indeed, all you have in the superblock is the number of mounts, IIRC, which may or may not be helpful.
[22:44] <SpamapS> the point is any time you have a system where you have two-way information flow, you need some kind of guide to ensure they are in sync/out of sync.
[22:44] <SpamapS> so I understand that we will, in fact, torch the disks right now if you do the scenario described
[22:45] <SpamapS> because our sync method is vulnerable to misinformation
[22:45] <mathiaz> SpamapS: right - it may be worth filing a bug about this issue
[22:45] <SpamapS> thats my second point..
[22:45] <mathiaz> SpamapS: and talk with kees and kirkland about it at the next UDS
[22:45] <SpamapS> mdadm has a ridiculous number of bugs open. :(
[22:45]  * kirkland has no interest in discussing RAID/failover implementations in Ubuntu any more
[22:45] <kirkland> there is no way of pleasing everyone
[22:46] <mathiaz> SpamapS: as for the test cases what would you recommend to do?
[22:46] <kirkland> and everyone has their own opinion
[22:46] <kirkland> most of which are horribly uninformed
[22:46] <SpamapS> 78 open, 40 new
[22:46] <kinygos> apologies
[22:47] <SpamapS> kirkland: I can understand. It stems from the conflicting goals Ubuntu users have.
[22:48] <kees> documentation is the answer.
[22:48] <kees> "here is what we did for RAID"
[22:49] <kees> mathiaz: does the Server Team have a FAQ?
[22:49] <mathiaz> kees: hm - not a *User* FAQ
[22:50] <kirkland> kees: +1
[22:50] <kirkland> SpamapS: sorry, not taking offense;  i've just been beat up on this one too much
[22:50] <kirkland> SpamapS: we're miles and miles and miles beyond where we used to be, pre-hardy, with respect to RAID
[22:51] <kirkland> SpamapS: if you or anyone else wants to take it from here, be my guest :-)
[22:51] <kees> though please don't change the existing defaults, as they were well discussed.
[22:51] <kees> the expected behaviors should remain the same; that's what took so much time to nail down.
[22:52] <SpamapS> No I don't want to change any defaults.
[22:52] <SpamapS> I want to change the test cases.
[22:53] <SpamapS> I'm just lamenting that we don't have an easy way to know the RAID1 is out of sync if you do what the test case suggests (write to each disk individually without ever syncing)
[22:54] <SpamapS> interesting
[22:54] <SpamapS> when I install from starbucks it guesses that I'm in New York, when I install from home, it guesses Los Angeles
[22:56] <mathiaz> SpamapS: well it's simple: you're connecting to a starbucks in New York via wifi
[22:56] <SpamapS> :)
[22:57] <SpamapS> actually ip2location.com guesses Newport, Rhode Island
[23:31] <SpamapS> heh.. I just typed this and it looked funny to me..
[23:31] <SpamapS> clint@ubuntu:~$ watch !!
[23:32] <SpamapS> like.. hey.. you... WATCH!!