[00:00] <cellofellow> thanks
[00:15] <OscarTG> thanks
[02:02] <kb244> I currently have PHP Version 5.2.6-2ubuntu4.1 is there any way i can get it to 5.2.9? seems last time I tried to manually compile it, it reverted back to 5.2.6 a day later
[02:03] <kb244> using ubuntu 8.10 server
[02:04] <twb> kb244: 5.2.9 is not available for Ubuntu yet.
[02:04] <kb244> ... despite I could compile it from source?
[02:04] <twb> kb244: this is probably for a good reason (i.e. installing it breaks something, or is insecure).
[02:04] <kb244> but 5.2.9 fixes security issues in 5.2.6
[02:04] <twb> You *could* compile it from source, but I strongly discourage that.
[02:05] <twb> If Ubuntu are doing their job, the Ubuntu version of 5.2.6 will include any relevant security fixes that upstream added to 5.2.9.
[02:05] <kb244> so 5.2.6-2unbuntu4.1 is a special build by the ubuntu team?
[02:05] <kb244> and not just a compile of 5.2.6 like you'd get from php.net
[02:11] <PhotoJim> kb244: yes
[02:11] <kb244> wont really worry bout it then :p
[02:12] <kb244> right now I got apache disabled, and just running nginx + php (fast-cgi) + mysql5
[02:16] <twb> kb244: all Ubuntu packages are gently caressed so they fit better into the overall distribution
[02:16] <kb244> nginx was installed with apt-get so I assume thats the case as well
[02:16] <twb> For simple packages like mg, that doesn't involve much.  For packages like PHP you can be sure that a LOT of fixing is needed to get things "right".
[02:17] <twb> It depends how silly upstream is, amongst other things.
[02:17] <kb244> i been using landscape , which is kind of nice. it would be interesting if it were extended into some sort of off-server cpanel :P
[02:18] <kb244> i been wanting however to make users only see their home folder and unable to see above it via SFTP/SSH, but I don't want to have to create a jailed enviroment and such, nor use vftpd etc
[02:22] <|rt|> anyone here familiar with iscsi...I'm having problems discovering a iscsi target being served by opensolaris
[02:22] <|rt|> windows xp's iscsi initiator works with it so I'm assuming that this is a configuration issue on the initiator side on linux
[02:23] <|rt|> http://pastebin.com/d3ebfbfac
[02:25] <twb> Wow, landscape is not just vapourware now?
[02:42] <kb244> twb: been using it for a couple weeks
[02:42] <kb244> just gota request a demo
[02:42] <kb244> its like a 60 day trial I think
[02:42] <kb244> then its 150$ a year per node after that
[02:42] <kb244> I only knew about landscape cuz I keep seeing this when I SSH into my box : Graph this data and manage this system at https://landscape.canonical.com/
[02:42] <twb> kb244: uh, what?
[02:43] <twb> Isn't landscape free software?
[02:43] <kb244> no
[02:43]  * twb loses interest
[02:44] <kb244> It'd be really nice if it was a free service, but no
[02:44] <kb244> I'm confused though
[02:44] <kb244> on their website if you dig a lil "Free 'Landscape' Web-based Systems Management"
[02:45] <kb244> but on the register page you're getting a 60 day free trial
[02:45] <twb> There has been a stub package for it in Ubuntu's package tree for ages
[02:45] <kb244> ahh I think I must have misread it
[02:45] <kb244> "
[02:45] <kb244> Landscape is free of charge to all of Canonical’s support customers. As part of your annual subscription you are allowed unlimited access to this powerful tool that will reduce the cost of ownership in your Linux environment significantly. Landscape is also available by subscription for those who do not require support contracts.  Subscription to the Landscape service is priced at $150 per node per year.
[02:45] <kb244> "
[02:45] <twb> "Description: Placeholder for the Landscape client"
[02:46] <kb244> naw I signed up for the landscape thing, then it told me what apt-get to install and how to register my machine
[02:46] <kb244> and once I did that, it was sending reports to my account and such
[02:46] <twb> Ah, so I guess this is not something I can actually provide to *my* customers; Canonical has full control over the back end.
[02:46] <kb244> however
[02:46] <kb244> "
[02:46] <kb244> To buy Landscape as a stand-alone service please go to www.canonical.com/contact/sales
[02:46] <kb244> "
[02:46] <kb244> you could , but they don't show prices
[02:46] <kb244> just request form
[03:38] <oh_noes> I have the following in a --exec script for vmbuilder, but as you can see it doesnt work.  Anyone have any suggestions?  http://pastebin.com/mf101449
[03:38] <oh_noes> All I'm trying to do is a automated install of java without the license prompt, I fail to see why it cant find apt-get though
[03:57] <a1fa> hello.. i have one of Appro servers. Funny thing is.. When you hookup USB keyboard or PS2 keyboard it does not recognize neither
[03:58] <a1fa> anyone have experience with this brand?
[04:11] <twb> a1fa: that's almost certainly a hardware issue that you need to take up with your hardware vendor.
[04:12] <twb> PS/2 and USB HID drivers are fairly commoditized, so I can't imagine how they could fail as a result of using Ubuntu
[04:17] <a1fa> twb: shouldnt the serer not care about ps2 2 usb adapter
[04:21] <twb> Ah.
[04:22] <twb> PS/2 to USB converters are just plug adapters.  The do not work with keyboards.  They rely on the mouse itself speaking BOTH protocols.
[04:23] <a1fa> so technically if legacy USB is not enabled it will be enabled when system bots?
[04:24] <a1fa> so thats only for mouse
[04:24] <a1fa> wierd
[04:24] <a1fa> it has keyboard pictured
[04:44] <PhotoJim> some older systems need a PS/2 keyboard to boot without error
[04:44] <PhotoJim> USB keyboards are relatively recent
[04:58] <SockPants> hello
[04:58] <SockPants> what's the easiest way to identify and temporarily mount a usb-ide drive (8.04)?
[05:02] <PhotoJim> easiest?  figure out what device name and partitions it has (dmesg | tail after you plug it in will help)... then mount /whateverfolderyoucreatedforit /dev/whateverdeviceyoufound
[05:02] <PhotoJim> so if it's partition 1 on drive sdg, mount /media/externalhd0 /dev/sdg1 for example
[05:04] <SockPants> yeah i can't seem to find which device name it is. i only have sda and thats the harddrive, not the usb drive
[05:05] <PhotoJim> browse through dmesg and see.  it will be there.
[05:05] <SockPants> hmm, theres errors
[05:10] <SockPants> i unplugged it and plugged it in without the hub but nothing has changed in dmesg
[05:13] <SockPants> ah, works now.
[05:13] <SockPants> thanks
[05:13] <SockPants> i didn't know about dmesg :D
[05:17] <oh_noes> how do i change timezone to UTC without user prompt?  dpkg-reconfigure tzdata Etc/UTC still prompts 'chose a tz'
[05:22] <Alysum> hi should I be concerned that an ubuntu server with 1.7GB RAM only has 192MB free ?
[06:21] <oh_noes> Is there a package for open-vm-tools for hardy?
[06:23] <drspin> oh_noes: I had to compile it
[07:37] <owh> I successfully migrated a server from Gutsy to Hardy, now my fstab smbfs mounts fail with "CIFS VFS: cifs_mount failed w/return code = -22" -- any suggestions?
[07:38] <drspin> owh: what kernel are you running
[07:38] <owh> drspin: 2.6.24-23-server
[07:42]  * owh thinks that the smbfs format is different from the cifs format of the fstab line.
[07:43] <drspin> owh: did you try reinstalling smbfs ?
[07:43] <owh> drspin: No, I didn't know it had been uninstalled. All I did was upgrade the server. No documents I saw indicated that it would be removed.
[07:46] <drspin> owh: hmmm - feels like a kernel issue, my original 8.10 kernel crashed when I mounted CIFS. Kernel update fixed it
[07:48] <owh> drspin: Uhm according to apt-cache policy I'm running the most recent kernel.
[07:49] <drspin> owh: I'm on 2.6.11-server
[07:49] <drspin> own: imean, 2.6.27-11-server
[07:50] <owh> Hmm, with Hardy?
[07:50] <drspin> 8.10. intrepid
[07:50]  * owh has no intention of running anything other than LTS on a production server.
[07:50] <drspin> ok
[07:51] <owh> Any Idea on what the error code -22 is?
[07:51] <owh> Or where I can find out?
[07:52] <drspin> did you see this? http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=518113
[07:56] <owh> It seems to indicate that the problem is my ip=mumble stanza. The mumble is defined in /etc/hosts - I wonder if this is changed behaviour. My remote device does not have multiple addresses by the way - not that I know anyway.
[07:57] <owh> Crap, I hate it when working stuff just breaks for fun.
[07:59] <owh> This makes no sense to me. I'm reading reports where I should install smbfs, but isn't cifs the replacement for smbfs and ultimately the new name of the same thing?
[08:01] <owh> Not to mention that it's also already installed.
[08:12] <owh> In case anyone else comes across this. smbfs can use ip=mumble, when mumble is defined in /etc/hosts, but cifs cannot. In addition it appears that smbfs does a fallback on the guest account if the supplied credentials don't work, cifs doesn't.
[08:19] <kraut> moin
[08:27] <reid> I'm trying to set up an ssh server on a ubuntu-desktop install.  connecting from an external network I keep getting :No route to host
[08:27] <reid> can anybody help me?
[08:33] <cemc> reid: do you have ping to the ubuntu-desktop with the ssh on it /
[08:34] <cemc> is ssh server running? 'ps ax |grep ssh', 'netstat -nlp | grep :22'
[08:35] <cemc> do you have some firewall on which would block the ssh ? 'iptables -nxvL INPUT | grep "22\|DROP\|REJECT"'
[08:38] <owh> reid: Can you actually ping the other end?
[08:41] <ikonia> ssh localhost on the server - test it can actually accept connections
[09:07] <twb> If you have the ssh server in front of you, run "tail -n 0 -f /var/log/auth.log", then try to ssh in again
[09:07] <twb> It will either report WHY you can't log in, or print nothing.
[09:07] <ikonia> he's gone
[09:07] <twb> If it prints nothing, then ssh is not seeing your connection, so look at firewalls, routing, ssh -vvv, etc.
[09:07] <twb> ikonia: oh, sorry.
[09:07] <ikonia> no need to apologise
[11:00] <Doble> how can I see what the ACL properties are of a folder/file? when I do "ls -lha" I only get the drwxrwxrwx+
[11:01] <Kamping_Kaiser> Doble, file attributes? lsattr
[11:01] <Kamping_Kaiser> ls -d will list a directory, not its contents btw
[11:04] <Doble> Kamping_Kaiser: thanks, but lsattr isn't doing it for me ... i just get a bunch of dashes, and they are all blank despite the files/directories having ACLs applied
[11:04] <Kamping_Kaiser> Doble, what filesystem is this? how did you change the acls?
[11:05] <Kamping_Kaiser> AnRkey, your giving me quake3 flashbacks
[11:06] <Doble> Kaiser: it is ext3, i followed the "Securing Samba Guide" at https://help.ubuntu.com/8.10/serverguide/C/samba-fileprint-security.html - I used setfacl to change the permissions
[11:08] <Kamping_Kaiser> as i dont 'deal with samba or acls, i can only suggest man setfacl to check for a 'list' option, or look at 'related commands' at the bottom of the man page
[11:09] <Doble> ahhh, I didn't spot that when reading the man page the first time - "getfacl"
[11:09] <Doble> cheers
[11:10] <Doble> tried everything else! showfacl, infofacl, etc :)
[11:11] <Kamping_Kaiser> hehe
[11:11] <Kamping_Kaiser> Doble, no worries
[11:28] <Doble> I dont understand this ... I have a directory set to group 'admin', getfacl lists the owner as admin, and ls -lha lists the group permissions as "d---r-x---+" I am a member of admin, the owner of the file is 'nobody' - why can't I access the folder?
[11:36] <Doble> can anyone help ?
[11:39] <simplexio> Doble: does group have x right to directory
[11:41] <Doble> simplexio: I'm not sure ... do ACLS take precendence over unix permissions? according to unix the group has R and X permissions, but according to ACL's it doesn't ...
[11:41] <Doble> im guessing they must, because I just tried setting the ACL group to RWX and it works
[11:45] <simplexio> far as i know unix and acl permission are totally separated.., so if you set acl file "owner" you need give permission using setfacl
[11:46] <Doble> okay ... i think you're right, is there any way of viewing permissions under windows? I'm trying to make this samba file server as windows-environment-friendly as possible. With windows servers you can open properties and see the users that can access the file, is there any way of doing that with samba?
[11:48] <simplexio> no idea.. i tried to define acls into my linux->linux samba server but it didnt work that good. for somereson it doesnt keep default permission when i copy stuff into dir, etc.
[11:49] <Doble> did you set the default mask?
[11:49] <simplexio> yeah.
[11:49] <Doble> :(
[11:50] <simplexio> its my home fileserver and it has about 1 user, so its not a big problem, tried to learn that stuff too
[11:57] <Doble> Is there any way to see permissions in this fashion when accessing a samba file share from a windows PC? - http://tinyurl.com/c2nq7v
[12:16] <Doble> arghh these permissions are driving me mad !!
[12:20] <Doble> I'm a member of the group 'admin' - the group admin has RWX permissions on the directory acl ... and yet I can open and read, but not write ???
[12:35] <kinnaz|w> when i try to install lokkit. somehow the apt wants to install xorg+gnome, can i force apt not to install deps somehow ?
[12:35] <kinnaz|w> thou lokkit is ncurses based application, so its strange in the first place it requires gnome.
[12:40] <ernstp> is there any way to only download packages for do-release-upgrade?
[12:48] <kinnaz|w> gnome-lokkit - basic interactive firewall configuration tool (GNOME interface)
[12:48] <kinnaz|w> lokkit - basic interactive firewall configuration tool (console interface)
[12:48] <kinnaz|w> why on earth does that "lokkit" depend on gnome
[12:48] <kinnaz|w> apt-get install lokkit, wants to bring me gnome
[12:49] <infinity> Recommends: gnome-lokkit
[12:49] <infinity> apt-get --no-install-recommends install lokkit
[12:49] <kinnaz|w> hmmmz tnx strange
[12:49] <kinnaz|w> i have installed lokkit several times
[12:49] <kinnaz|w> never had to tell him not to install recommends
[12:50] <infinity> Installing recommends by default is a reasonably recent change to apt.
[12:50] <kinnaz|w> okey, thou is it really the smartest thing to recommend that gnome version
[12:50] <infinity> (Reasonably recent to the server crowd, anyway, who are used to not upgrading to new releases very often)
[12:50] <infinity> The recommends should probably be a suggests.  Feel free to file a bug.  *shrug*
[12:51] <infinity> We cleaned up most of the overzealous recommends in main, but universe stuff like lokkit still needs some work.
[13:02] <Doble> I'm having problems with samba ... since I set up ACLs, when I create a user, I go to log in as that user (which usually 'primes' samba to allow that user to connect to file shares) and as I log in as bill, I recieve the message "Failed to add entry for user bill." - so bill can't access any file shares, any ideas ???
[13:12] <SockPants> hi, can anyone tell me if this is a severe problem? its on a server thats hardly in use:
[13:12] <SockPants> Tuesday, April 14, 2009 PM01:02:09 HKT  **WARNING** Address resolution queue is full [4096 slots]
[13:40] <Doble> why do I have to log into the terminal as a new user I created BEFORE they are able to access shares on my samba file server ?
[13:43] <foxbuntu> Doble, not totally sure on this, but I think that samba wont pick them up as a valid user until they have logged in once, unless you manually add them as samba users
[14:52] <jcastro> soren: want to do a virtualization session for openweek?
[14:55] <soren> jcastro: When is it? I'm still catching up on e-mail since... too long ago.
[14:56] <jcastro> week after release: https://wiki.ubuntu.com/UbuntuOpenWeek/Prep
[14:59] <soren> jcastro: I'd like to do several, actually. I need to consider whether I have time for anything at all, though.
[14:59] <jwendell> hi, folks, is there any plan to make ubuntu server homologed to run oracle db?
[15:01] <soren> "homologed"?
[15:01] <jwendell> certified
[15:01] <soren> Ah.
[15:01] <jwendell> as RHEL, Suse...
[15:02] <soren> SuSE is certified for Oracle? I didn't know that.
[15:02] <jcastro> soren: that would be great, can you think hard today? We'd like to get the schedule out and published.
[15:02] <jwendell> soren, yep
[15:02] <jcastro> soren: this is openweek, not developer week so the sessions don't need to be hardcore, intro sessions are fine.
[15:02] <jcastro> eg. you won't need to do much prep
[15:02] <soren> jcastro: "today" being your today? So I have another 7-8 hours?
[15:02] <jcastro> yeah something like that. :)
[15:03] <soren> jcastro: Oh.. Right, ok. The sessions I was thinking about were probably more developer week material.
[15:03] <jcastro> right, save those
[15:03] <jcastro> how about like, an intro to kvm or something?
[15:04] <soren> Let me think about it for a bit.
[15:26] <SockPants> hi all, what's a nice way to make regular backups of an entire system?
[15:26] <SockPants> onto a smb share
[15:27] <giovani> smb? uh
[15:28] <PhotoJim> if you were using an nfs share I'd suggest rsync
[15:28] <giovani> backuppc does smb, I think
[15:28] <PhotoJim> but smb doesn't retain the ownership and permissions of files that LInux filesystems record (not without some work at least)... so that implies using tar
[15:28] <giovani> but it's not really oriented at backing up servers
[15:29] <SockPants> i can live with using tar if its usable for this
[15:29] <PhotoJim> the bad thing about tar is that you'll have to fully backup your files each time
[15:29] <giovani> well tar just combines files ... so it's not really a backup facility
[15:29] <PhotoJim> rsync is smart enough to just backup the differences
[15:29] <friartuck> SockPants looked here? https://help.ubuntu.com/community/BackupYourSystem
[15:30] <PhotoJim> giovani: it's not optimal, but it retains file ownership and modes, which I think is important here.  assuming that SockPants for some reason can't use ext2/ext3/ext4 filesystems for the backup, which would be far better.
[15:30] <giovani> PhotoJim: I wasn't disagreeing with you
[15:30] <PhotoJim> giovani: oh, I know. :)
[15:30] <giovani> I was just informing SockPants that tar was not a "backup solution" the way he probably wants
[15:31] <PhotoJim> That's a good article, friartuck
[15:32] <giovani> indeed
[15:32] <giovani> glad he linked it
[15:33] <PhotoJim> I have my server on RAID1... I don't bother backing up the / and /usr drives (I just copy the hard-to-replicate config files onto an NFS share on another machine, or one could even use floppy).  and I have a third hard disk as my /public and /home partitions, and that drive is on RAID1.  plus, I backup all the data via rsync to an identically-sized drive in an external USB case once a week or so.
[15:34] <SockPants> instead of tar i could even use dd
[15:35] <SockPants> its only a 20gb drive of which a few gigs are used at best, the files are all on a different machine
[15:36] <PhotoJim> dd backs up every byte of the drive, so it's slower
[15:36] <PhotoJim> does work though
[15:36] <PhotoJim> every byte of the partition, technically
[15:36] <PhotoJim> dd would preserve all your filesystem information though
[15:37] <SockPants> i could restore a dd backup using a livecd right?
[15:38] <SockPants> mm, i suppose if i didn't compress the file, even empty space on the drive would take ages to copy over 100mbit/s
[15:39] <giovani> SockPants: you can gzip a dd image
[15:39] <giovani> which should eliminate all of the large blocks of zero data
[15:39] <giovani> in fact, you can stream a gziped dd image (while it's being taken) if you don't have enough space locally
[15:39] <giovani> I do it all the time
[15:39] <PhotoJim> SockPants: why don't you just get another hard disk and a USB case, and dedicate that drive to backups?  much easier.  you can use native Linux filesystems, and no other systems can affect your data.
[15:39] <giovani> with netcat
[15:40] <friartuck> SockPants if you were going that far, maybe look at partimage on http://partedmagic.com .
[15:40] <giovani> PhotoJim: dd does entire drives, or partitions ... user's choice
[15:41] <giovani> it's the best way to have an identical copy of everything
[15:41] <giovani> and it's faster than a filesystem-level backup, if you're going to backup everything
[15:41] <SockPants> can't i just dd the disk onto a smbfs mount and be done with that, even if it copies all the 0-blocks
[15:41] <giovani> SockPants: yes ... pipe dd through gzip
[15:41] <giovani> and the zero blocks are a non-issue
[15:42] <MTecknology> How do I setup my system so when a user changes their password the samba password for that account will update too?
[15:42] <PhotoJim> giovani: oh, yeah, that's true.  I always think of it as a partition-level command.
[15:42] <SockPants> that, or i could just leave it, because the filesystem is compressed anyway. it would just take time to copy all the 0-data to the other machine before it gets compressed
[15:42] <giovani> SockPants: the filesystem is compressed?
[15:43] <SockPants> yeah on the remote machine
[15:43] <giovani> uh, ok
[15:43] <giovani> well, you're still wasting time/bw moving the zero blocks over the network
[15:44] <SockPants> hmm
[15:44] <SockPants> but i'm saving the server processing time by not compressing it locally
[15:44] <giovani> and I don't know how your filesystem is compressing itself
[15:44] <SockPants> how much would that be?
[15:44] <giovani> hard to say
[15:44] <giovani> try it out
[15:45] <giovani> you can dd and gzip it in one command, right to the smb mount point
[15:45] <giovani> so it doesn't take up space locally
[15:45] <SockPants> copying the entire disk would copy swap space as well, right? would that cause any problems with dd?
[15:45] <giovani> no ...
[15:45] <SockPants> ok
[15:45] <giovani> dd only knows about hard drive blocks
[15:45] <giovani> it doesn't care what the data is
[15:45] <giovani> it doesn't know what the data is either
[15:45] <SockPants> so i'd just dd if=/dev/sda > /mnt/backups/backup1.bin or something
[15:46] <giovani> the advantage of dd is that if the server dies, you can literally restore the ENTIRE thing, in one step
[15:46] <SockPants> yeah
[15:46] <giovani> no reinstalling the OS and manually moving back configs, etc
[15:46] <giovani> no
[15:46] <giovani> dd if=/dev/sda of=/mnt/backups/backup1.img or whatever
[15:46] <giovani> however, I'd pipe it through gzip
[15:47] <SockPants> is there a fast setting for gzip that won't compress as much but still get rid of the 0-data
[15:47] <giovani> probably
[15:47] <giovani> read the manpage
[15:47] <SockPants> yeah, gzip --fast
[15:47] <giovani> or just try it out
[15:47] <SockPants> :D
[15:47] <giovani> you never know what the load will be
[15:48] <SockPants> i'll start now
[15:48] <SockPants> then lastly
[15:48] <SockPants> is there a way i could add something to the command to make it add a time/date to the filename?
[15:49] <giovani> use bash command substitution
[15:52] <PhotoJim> SockPants: yeah, copying swap space is of no benefit, but nor is it harmful
[15:53] <PhotoJim> SockPants: technically a gig of my terabyte data drive array is swap, but I don't care.  it's not significant enough space to treat it separately.
[15:53] <PhotoJim> gzip is pretty efficient, even on slower machines I find.
[15:54] <PhotoJim> 7zip has much better compression, but it will bring a slower machine to its knees.  works best on machines with multiple CPUs or multiple cores.
[16:04] <SockPants> ive got this now
[16:04] <SockPants> screen sudo dd if=/dev/sda | gzip --fast > /mnt/backup/ubuntu/dev-sda_on`date +%a%d%b%y-%s`.dd
[16:05] <SockPants> it creates the file, but outputs a lot of garbage to the screen and the file doesnt grow
[16:05] <giovani> why are you screening it?
[16:05] <SockPants> past 20 bytes
[16:05] <yann2> uh. interesting backup procedure :)
[16:05] <SockPants> because i want to do other stuff at the same time :P
[16:05] <giovani> uh ... it's called BACKGROUNDING
[16:05] <giovani> do not use screen for this
[16:06] <SockPants> how do i do this
[16:06] <SockPants> i always use screen =)
[16:06] <giovani> well that's a bad habit
[16:06] <giovani> break it now
[16:06] <giovani> and I said to pipe it to gzip
[16:07] <SockPants> i did.., what did i do wrong
[16:08] <giovani> uh, sorry, you are
[16:08] <giovani> so what's being outputted to the screen?
[16:08] <giovani> try running the command without prepending screen
[16:08] <SockPants> a constant stream of stuff
[16:08] <giovani> it should work fine
[16:08] <giovani> I just verified the command, it works perfectly
[16:09] <SockPants> ok, it works, but now it waits until it finishes
[16:09] <SockPants> how do i background it
[16:09] <giovani> learn bash :)
[16:09] <giovani> (add " &" onto the end
[16:09] <giovani> i.e sudo dd if=/dev/sda | gzip --fast > dev-sda_on`date +%a%d%b%y-%s`.dd &
[16:09] <SockPants> hehe yeah i will :) and thanks
[16:10] <giovani> it'll print the PID
[16:10] <SockPants> 2.0 gb copied in that short time
[16:10] <SockPants> it says
[16:10] <giovani> yep ...
[16:10] <SockPants> interesting
[16:10] <SockPants> that's faster than my network connection
[16:10] <giovani> heh
[16:10] <SockPants> how can i see how much harddisk space is actually in use
[16:10] <giovani> 2GB from DD
[16:10] <giovani> is not 2GB over the wire
[16:10] <giovani> because of gzip
[16:10] <giovani> df -h
[16:11] <giovani> to see disk space
[16:11] <SockPants> but if dd processes 2gb but gzip doesn't yet then where does it go
[16:11] <giovani> huh?
[16:11] <giovani> gzip and dd are stream-processing
[16:12] <giovani> meaning dd is outputting data to gzip all the time, not all at once
[16:12] <giovani> and gzip is live-compressing it
[16:12] <giovani> so DD read in, and read out 2GB off the HD, and gzip compressed it to ... whatever -- check the file size
[16:15] <SockPants> ok
[16:15] <SockPants> great!
[16:16] <PhotoJim> I use screen for that sort of stuff, so I can monitor it.  worse performance than backgrounding?
[16:16] <SockPants> now, "0 2 * * 1 <command>" in crontab will run this every monday at 2 a.m. right?
[16:17] <giovani> PhotoJim: screen won't let you monitor it any better than backgrounding will
[16:18] <SockPants> how do you access a backgrounded operation's output?
[16:18] <PhotoJim> giovani: how can you monitor a backgrounding task?  I & tasks sometimes, but I am really a bit of a n00b at it.
[16:19] <giovani> PhotoJim: depends on what you mean by "monitor"
[16:19] <giovani> in dd's case, you can send a SIGUSR1 signal to the process to have it dump it's info so far to the console you started it on
[16:20] <PhotoJim> giovani: ok, that's something I didn't know you could do.
[16:20] <giovani> that's dd-specific
[16:21] <PhotoJim> so what's the logic that using screen is inferior practice?
[16:21] <giovani> it has zero function ... that's why
[16:21] <giovani> it complicates something that doesn't need to be
[16:21] <giovani> screen is useful for other things
[16:21] <giovani> but not this
[16:21] <RichardP> Hi, I have a little problem installing the 8.04 LTS server version - its stuck on 'Installed MySQL-Server' at 85%, been there for the past 30 minutes
[16:23] <SockPants> giovani: 'it's info so far' being bytes transfered?
[16:23] <giovani> SockPants: yes
[16:23] <SockPants> how do i do this?
[16:23] <RichardP> SockPants: hey
[16:23] <SockPants> hi RichardP
[16:23] <giovani> SockPants: "ps aux | grep dd" -- find the dd process' PID
[16:23] <giovani> then "sudo kill -SIGUSR1 DDPU
[16:23] <SockPants> i have the pid
[16:23] <giovani> DDPIDHERE
[16:24] <giovani> that wasn't the PID that it printed when you backgrounded it, I was incorrect
[16:24] <SockPants> oh
[16:24] <giovani> you can bring apps out of background (althrough with DD that has no function) with %X where X is the job ID -- which you can find by running jobs
[16:28] <SockPants> after i do the kill command it keeps waiting, can i just control c without hurting dd?
[16:28] <giovani> is the process in the background?
[16:29] <genii> Doesn't sound like it
[16:29] <SockPants> it should be
[16:29] <SockPants> it was
[16:29] <PhotoJim> giovani: yeah, I guess I agree, once you know what you're doing and that the task works as expected.
[16:29] <giovani> SockPants: ps aux | grep dd
[16:29] <SockPants> then i did sudo kill -SIGUSR1 pid
[16:30] <giovani> PhotoJim: there's zero advantage to using screen with dd -- it has nothing to do with the app functioning as expected
[16:30] <giovani> SockPants: pastebin the output of the ps command I just gave you
[16:30] <ScottK-laptop> ivoks: Here so as not to futher disturb the meeting ...
[16:30] <ScottK-laptop> ivoks: What RFC?  I generally consider backscatter bad and bounces to any non-authenticated address to be avoided.
[16:30] <PhotoJim> giovani: yeah, no output I suppose except the benchmarks at the end.
[16:31] <giovani> PhotoJim: you'll get that with a backgrounded app too, if you want it
[16:31] <SockPants> http://pastebay.com/8266
[16:31] <SockPants> to get out of the 'kill' command i had to control-c
[16:31] <PhotoJim> giovani: I just have to learn more about backgrounding, evidently.
[16:32] <ivoks> ScottK-laptop: http://www.faqs.org/rfcs/rfc3834.html iirc
[16:32] <giovani> SockPants: did you try just pressing enter?
[16:33] <SockPants> no :P
[16:33] <giovani> it's not 'taking over the terminal' it just printed it, you can theoretically just type a command right there, without hitting enter
[16:33] <SockPants> that works too =)
[16:34] <SockPants> im impressed with the speed, and the file only takes up about 180 mB in the end even though it also copies swap
[16:34] <ivoks> ScottK-laptop: anyway, if message is not delivered, sender should know that
[16:34] <giovani> SockPants: how big is the drive?
[16:34] <SockPants> then again, only 212 bytes of swap is used
[16:34] <ivoks> ScottK-laptop: unless message is not delivered cause of non-existing user or bad communication from sender
[16:34] <SockPants> 120 GB
[16:35] <giovani> how much of it is used?
[16:35] <ScottK-laptop> ivoks: There's no guarantee of any relation between mail from and the actual sender.
[16:35] <SockPants> 1.0GB
[16:35] <ivoks> ScottK-laptop: i agree, but that's SMTP - it's broken
[16:35] <ScottK-laptop> In fact in virtually all cases where you bounce due to bad content it's forged.
[16:35] <giovani> 1GB of 120GB is used?
[16:35] <SockPants> yes
[16:35] <ivoks> ScottK-laptop: i allways change that, but this is something one shouldn't do :/
[16:35] <ScottK-laptop> ivoks: Which causes me to say don't backscatter no matter what.
[16:36] <ivoks> ScottK-laptop: it's up to you; it would be one less configruation hassle for me, but it might not be correct thing to do :/
[16:36] <giovani> SockPants: and it finished already?
[16:37] <SockPants> it has finished once, i'm restarting it just to test
[16:37] <RichardP> when you install 8.04 LTS and get the disk partitioning option, what exactly does 'Guided - entire disk with LVM' mean?  does it mean I get one partition for / taking the entire disk, under LVM?
[16:37] <SockPants> i could time it i think i'll do that
[16:37] <SockPants> do i just add 'time' to the command?
[16:37] <giovani> uh ... sure, except then you don't want to background it
[16:37] <giovani> to ensure you see the time when it's done
[16:37] <giovani> (just for the test)
[16:38] <ScottK-laptop> ivoks: From your RFC "A responder MAY refuse to send a response to a subject message which contains any header or content which makes it appear to the responder that a response would not be appropriate."  I think not resoponding to bad content is RFC legit.
[16:39] <SockPants> ah right
[16:39] <SockPants> so i'll screen it ;)
[16:39] <SockPants> oh, but that did'nt work out well
[16:39] <SockPants> nvm i wont screen it i'll just open anothe ssh to do other stuff
[16:40] <ivoks> ScottK-laptop: ok, then change it :)
[16:40] <ScottK-laptop> ivoks: Assuming I want to change it, do you agree that's the correct change?
[16:41] <SockPants> giovani: its being pretty sluggish when its running, gzip is using 20% cpu so it must be the harddrive then?
[16:42] <SockPants> meanwhile, how do i run a commmand at startup? i can background all these processes now but they'll all die if the power goes out some time
[16:42] <ivoks> ScottK-laptop: yes
[16:42] <ScottK-laptop> ivoks: Thanks.
[16:42] <ivoks> ScottK-laptop: but we will probably get some flame mails about this :)
[16:42] <ivoks> ScottK-laptop: but i'm willing to put up the fight
[16:43] <giovani> SockPants: sure ... it's maxing out the HD ... run "vmstat 5 5" while it's running (in another terminal), or if it's backgrounded, in the same terminal
[16:43] <giovani> or run iostat
[16:45] <ScottK-laptop> ivoks: OK.  I think no backscatter is a clear best practice these days.  People who don't like it are welcome to change their config locally.
[16:45] <ivoks> correct
[16:46] <SockPants> procs -----------memory---------- ---swap-- -----io---- -system-- ----cpu---- r  b   swpd   free   buff  cache   si   so    bi    bo   in   cs us sy id wa 1  1    212   5628 193652  22296    0    0   155    15  102  366  2  4 88  5
[16:46] <SockPants> blah, formatting
[16:52] <SockPants> can i do nested command substitution?
[16:57] <SockPants> giovani: nevermind, it can't have finished. dd is at 18 GB now, after about 15 minutes. so another 75 minutes or so before it finishes, but i'll leave it running
[17:04] <SockPants> giovani: thanks for all the help, i'm out for tonight :) i just checked it out, the file is 499MB in size (522895360 bytes) but it only takes up 416MB on the server (zfs compresses it further). nice way to get the best of both worlds :)
[17:05] <SockPants> its not done yet though
[17:13] <mathiaz> soren: wrt to bug 360825 - I think this is a regression
[17:14] <mathiaz> soren: it used to work with a previous revision of kvm 84
[17:18] <soren> I respectfully disagree.
[17:18] <soren> :)
[17:26] <giovani> SockPants: yeah, I wouldn't believe that gzip --fast would compress 1GB into 100-something MB
[17:46] <mathiaz> soren: disagree? why - it was working before, now it's not anymore. Isn't that a regression?
[17:46] <soren> i don't believe that it worked in a previous revision of kvm 84.
[17:47] <mathiaz> soren: well - it worked. I was able to sucessfully install a guest on a raid5 array.
[17:47] <mathiaz> soren: that involves defining 3 block devices in libvirt.
[17:47] <soren> Then you've switched to virtio in the mean time.
[17:48] <soren> This is a virtio problem only.
[17:48] <soren> virtio devices get added to an array based on command line options.
[17:48] <soren> When they get added to the virtual hardware, it's done by iterating through this array until an undefined index is found.
[17:49] <mathiaz> soren: ah ok. Let me check if I was using virtio for my raid5 test install.
[17:54] <mathiaz> soren: hm - I can't tell for sure. However it seems that I conducted the jaunty beta raid5 test case with virtio devices.
[17:54] <soren> mathiaz: And all that changed in between is the kvm-84 revision?
[17:58] <mathiaz> soren: well - jaunty beta raid5 was tested with kvm84 ppa3
[17:58] <mathiaz> soren: while jaunty rc raid5 was tested with kvm84 ppa5
[17:58] <soren> No libvirt changes in between?
[18:00] <mathiaz> soren: libvirt was also upgraded - 0.4.0-2ubuntu8.1 to 0.4.0-2ubuntu9~ppa1
[18:02] <mathiaz> soren: I think I should conduct more testing to make sure it was working before. I'm not 100% sure that I was already using virtio devices when running jaunty beta raid5 test installs.
[18:02] <soren> Do you still have logs or something?
[18:03] <soren> Either /var/log/libvirt/qemu/* logs or syslog from the guest or something?
[18:05] <mathiaz> soren: well the libvirt/qemu/ logs are overwritten
[18:05] <soren> Ah.
[18:05] <mathiaz> soren: whenever you start a guest with the same name
[18:05] <soren> Right.
[18:05] <mathiaz> soren: however I may be able to find the relevant log in my backups
[18:06] <beniwtv> hi al... I've installed bind9 and followed the server guide to make a caching name server, but I get query denied in the logs when I try to query the dns from a machine other than the DNS. Any ideas?
[18:12] <PhotoJim> beniwtv: if you're still stuck in an hour I can try to help you out.
[18:13] <beniwtv> PhotoJim: thanks, but I have just been called to a meeting :/ Will continue tomorrow
[18:20] <kirkland> mathiaz: i keep getting the following, trying to install to a raid 1 http://people.ubuntu.com/~kirkland/Screenshot-QEMU.png
[18:21] <mathiaz> kirkland: hm - never seen that before.
[18:24] <mathiaz> kirkland: http://people.ubuntu.com/~mathiaz/raid1.preseed
[18:24] <mathiaz> kirkland: this is my raid1.preseed file
[18:24] <mathiaz> kirkland: d-i partman-auto/expert_recipe string       multiraid ::       400 500 1000000000 raid $primary{ } method{ raid } .       64 256 200% raid $primary{ } method{ raid } .
[18:28] <kirkland> mathiaz: in your raid testing, what iso are you using?
[18:28] <kirkland> mathiaz: amd64 or i386?
[18:28] <kirkland> mathiaz: today's server iso?
[18:28] <mathiaz> kirkland: both arch, yesterday's isos.
[18:29] <kirkland> mathiaz: have you ever seen the segfault that soren is reporting?
[18:29] <mathiaz> kirkland: nope
[18:29] <mathiaz> kirkland: I haven't seen any segfault on my system.
[18:29] <mathiaz> kirkland: all my test installs were successful.
[18:57] <mario_> Hi, how can I copy a logical volume between two servers connected with ethernet?
[18:58] <jmedina> mario_: what you mean?
[18:58] <jmedina> move data?
[18:58] <jmedina> or move whole LV to another VG?
[18:58] <mario_> a whole LV to another VG
[18:58] <mario_> but
[18:58] <mario_> the VG is on another server
[18:59] <jmedina> I not sure, but, unless you export the bock device where VG is runnning you can't migrate with LVM tools
[19:00] <jmedina> but you can use lvm snapshots and dd/rsync
[19:00] <mario_> ok thank you
[19:02] <jmedina> probably someone else have another solution
[19:06] <mario_> jmedina, excuse me for asking you another question but, can you suggest a manual about lvm that explains advanced features like snapshot and autobackup?
[19:06] <jmedina> mario_: I live official howto
[19:07] <jmedina> mario_: this redhat document also have very good documentation
[19:07] <jmedina> http://www.redhat.com/docs/en-US/Red_Hat_Enterprise_Linux/5/html/Cluster_Logical_Volume_Manager/index.html
[19:07] <jmedina> no
[19:07] <jmedina> sorry
[19:08] <jmedina> the topic says cluster, but most of info is for normal lvm
[19:08] <mario_> jmedia, thank you again
[19:12] <jmedina> you are celcom
[19:12] <jmedina> welcome
[19:36] <cellofellow> Dunno if I'm in over my head or not, cause I've never configured these apps before, but I'm trying to set up a community/networking server with Wordpress, phpBB3, and MoinMoin (most likely), and would like to have single-signon of some kind using LDAP. Also would like the LDAP to operate as a PGP Keyserver. Can I do this?
[19:38] <cellofellow> I was originally gonna use SKS for PGP but when I found out that LDAP is a way to run keyservers I figured I'd scrap that.
[19:50] <cellofellow> (why is all the info I can find on LDAP meant for old (pre RHEL/Fedora split) versions of Red Hat?
[19:51] <jmedina> cellofellow: because you haven't read official documentation
[19:51] <cellofellow> ok, looking there
[19:51] <jmedina> for example openlda administration guide
[19:51] <jmedina> *openldap
[19:51] <jmedina> it is current documentation
[19:51] <jmedina> Fedora Directory also has official documentation from redhat
[19:52] <cellofellow> well, thanks
[19:53] <jmedina> cellofellow: what you want to know about LDAP?
[19:53] <jmedina> cellofellow: I bought a good book from Packet Publishing
[19:54] <jmedina> http://www.packtpub.com/OpenLDAP-Developers-Server-Open-Source-Linux/book
[19:54] <jmedina> all you want to about ldap and openldap
[19:54] <jmedina> Mastering OpenLDAP: Configuring, Securing and Integrating Directory Services
[19:54] <mathiaz> soren: ok - I've managed to boot a guest with 3 virtio devices with kvm 62 and libvirt 0.4.0-2ubuntu8.1
[19:54] <jmedina> that is the topic
[19:55] <mathiaz> soren: the guest sees three virtio devics (vda, vdb and vdc)
[19:56] <soren> mathiaz: That's kind of funny, really.
[19:58] <soren> mathiaz: The change in the ~ppa1 version is changing from the patch I wrote that upstream ended up rejecting and reimplementing to what they put in.. So the patch I wrote apparantly works, while the one they decided to put in doesn't :)
[20:03] <mathiaz> soren: I've updated bug 360825
[20:04]  * soren looks
[20:05] <mathiaz> soren: it seems that the old version of libvirt doesn't use the index option on the kvm command line
[20:05] <soren> mathiaz: Ah, right.
[20:05] <soren> That's right.
[20:05] <mathiaz> soren: that would explain why it was working.
[20:05] <soren> What I did instead was to sort the list of the disks so that they were simply passed in the correct order isntead.
[20:05] <soren> That explains.
[20:06] <cellofellow> jmedina➜ thanks, I'll check that out. First I just want to know the basics, then I'll mess with getting those apps to work with it.
[20:06] <mathiaz> soren: kirkland: so what should we do with bug 360825?
[20:07] <mathiaz> is is a regression?
[20:07] <mathiaz> is it a regression?
[20:07] <soren> mathiaz: Since hardy, yes.
[20:07] <soren> Since Intrepid, no.
[20:07] <mathiaz> soren: OTOH virtio block devices weren't officially support in hardy
[20:08] <mathiaz> soren: intrepid was already broken?
[20:08] <soren> mathiaz: I would expect so, yes.
[20:08] <soren> The patch landed in 0.4.1.
[20:09] <mathiaz> soren: ok - so this is not high/critical for jaunty.
[20:09] <kirkland> mathiaz: i would say not critical, since there is a workaround
[20:09] <kirkland> mathiaz: should be fixed, though, since soren triaged the file and line where the bug is
[20:10] <mathiaz> kirkland: agreed.
[20:10] <mathiaz> kirkland: I don't see the workaround though.
[20:10] <mathiaz> kirkland: IMO it should be fixed if we plan to release a backport of kvm to hardy.
[20:11] <soren> mathiaz: It doesn't have anything to do with kvm.
[20:11] <soren> Well...
[20:11] <soren> It doesn't have anything to do with the version of kvm. It's libvirt that's causing it.
[20:11] <mathiaz> soren: well - it's an interaction between kvm and libvirt.
[20:11] <soren> mathiaz: ...which didn't change between kvm 62 and 84.
[20:11] <mathiaz> soren: agreed. anyway it's not important for jaunty.
[20:25] <soren> It's a simple patch to work around it, though.
[20:26] <soren> mathiaz, kirkland: I can toss you a simple patch that should fix it, but if I were to fix it properly, I'd use a linked list instead of a static array or something.
[20:27] <mathiaz> soren: right - we should revisit that once jaunty is released;
[20:28] <kirkland> mathiaz: soren: if it's a simple patch with minimal chances of regression, i think it could make jaunty
[20:28] <soren> kirkland: Hang on, I'll pastebin.
[20:29] <kirkland> soren: i'll check it out when i'm back
[20:31] <soren> http://pastebin.ubuntu.com/151006/
[20:31] <soren> It's completely and utterly untested. I haven't even checked if it compiles.
[20:32] <soren> And the indentation is clearly screwed.
[20:33] <soren> But the general idea is there: Don't stop at the first hole in the virtio indexing, but keep going. It might increase vm startup time a few milliseconds.
[20:33] <soren> So meh.
[20:33] <soren> kirkland: Do with it as you please.
[21:38] <kochan> hi guys I have a strange problem
[21:38] <kochan> i've just installed ubuntu server 9.04 x86
[21:38] <Actium_> go ahead
[21:39] <kochan> and rails stack (apache, passenger, ruby, gems etc)
[21:39] <kochan> and sudo gem .. doesn't work
[21:39] <Actium_> sorry, but i've never used ruby on rails b4, so my help might be very limited.
[21:39] <kochan> but gem ... works
[21:40] <Actium_> any error messages?
[21:40] <kochan> command not found
[21:40] <kochan> i think there is a problem with path for sudo
[21:41] <Actium_> try "sudo -- echo $PATH" and "echo $PATH"
[21:41] <Actium_> and compare it with the result of "whereis gem"
[21:42] <kochan> strange
[21:42] <kochan> both outputs are exactly the same
[21:43] <kochan> and "ruby -v" works and "sudo ruby -v" doesn't :|
[21:44] <Actium_> what's the result of whereis? any unusual install dir?
[21:44] <Actium_> got any details? works and works not make debugging very difficult, you know.
[21:45] <kochan> sorry, for that
[21:45] <Actium_> no need to apologize
[21:45] <kochan> "ruby -v" gives me version of ruby, and "sudo ruby -v" gives me command not found
[21:46] <kochan> both "whereis ruby" and "sudo whereis ruby" gives me "ruby: /usr/lib/ruby /opt/ruby/bin/ruby /opt/ruby-enterprise-1.8.6-20090201/bin/ruby"
[21:48] <Actium_> try specifying a full path, when invoking ruby.
[21:48] <Vog-work> word of advice to anyone out there using dspam or rabl.nuclearelephant.com in you spam filters.... you might want to remove is since it just blacklisted a large portion of the internet.
[21:49] <Actium_> just like google did the other day? ;)
[21:50] <kochan> " sudo /opt/ruby/bin/ruby -v"  works, i mean it gives me a ruby version
[21:50] <kees> ivoks: heh, I like how you didn't link to the article.  :)
[21:51] <ivoks> :)
[21:51] <ivoks> it was a mistake
[21:51] <ivoks> but then i figured that it's better to show author's references, twice :)
[21:51] <kochan> is there a way to set path to work without  full path?
[21:52] <Actium_> i'd still like to know what makes the difference between with sudo and without sudo invocation when in both cases the $PATH is the same
[21:54] <Actium_> you could just add "/opt/ruby/bin" to your $PATH, that should do the job. however i do not like workarounds that much.
[21:56] <kochan> check this out: "ruby -v" - works, "sudo ruby -v" - command not found, "sudo su" and then "ruby -v" works also :|
[21:58] <Actium_> im outta clues right now
[21:58] <oruwork> ivoks, !
[22:00] <ivoks> sudo -i - takes root's environment
[22:00] <ivoks> sudo - takes user's environment
[22:00] <ivoks> echo $PATH
[22:00] <ivoks> sudo echo $PATH
[22:02] <andol> ivoks: Regarding su and /bin/false accounts. Actually, that's no problem if you feed su with the -s option. (in reply to ubuntu-server@)
[22:03] <ivoks> andol: but, in the end, what do we gain if we set /bin/false as shell?
[22:03] <kochan> http://pastie.textmate.org/private/yl7diifcavgkab9cxb1yw
[22:04] <andol> ivoks: Not much. That's why I didn't write the reply on the mailing list. Just thought I'd provide the info as a tip for some future situation.
[22:05] <ivoks> kochan: why do you have three /opt/ruby/bin?
[22:05] <ivoks> andol: right...
[22:06] <kochan> good question...
[22:21] <kochan> i've got rid of dupicates http://pastie.textmate.org/private/6xcj5og74vers07hkzcsow
[22:24] <kochan> any ideas guys?
[23:19] <mathiaz> soren: If I install libvirt-bin on jaunty, is there anything else I should do to get the default network up and running?
[23:20] <soren> mathiaz: It ought to just be there.
[23:20] <mathiaz> soren: hm - it's not.
[23:20] <mathiaz> soren: I only see eth0 and lo.
[23:20] <mathiaz> soren: where can I see the log for libvirtd?
[23:21] <soren> It doesn't have one, I think.
[23:21] <soren> virsh net-list?
[23:22] <mathiaz> soren: http://paste.ubuntu.com/151083/
[23:23] <mathiaz> soren: http://paste.ubuntu.com/151084/
[23:23] <soren> o_O
[23:23] <soren> Err.. Oddness.
[23:24] <soren> I need to get some sleep. I'll look at it tomorrow.
[23:25] <mathiaz> soren: bug 350780
[23:26] <Sergii> Is it possible to set up transparent Active Directory authentication on Apache server on Ubuntu? The idea is that I want internal users to open a web page, and have them authenticated with user name they logged in with without having them enter their login and password again.
[23:31] <foxbuntu> Sergii, I think this is what you are looking for: http://www.onlamp.com/pub/a/onlamp/2003/09/11/kerberos.html?page=2
[23:32] <Sergii> Foxbuntu, I'll take a look. Thanks
[23:32] <jmedina> Sergii: look at port25 microsoft site
[23:32] <jmedina> they have a good document about SSO using active directory and kerberos auth
[23:35] <Sergii> Will do! Thanks.