[00:51] <dkn> i had bonding setup on my network interfaces eth0 and eth1, i unplugged eth0 and attached it to another port on my switch when i was booting off of recovery mode CD, now the bond is getting an ip but not working, i can't ping anything, and eth0 is getting it's own ip address.... what did the boot cd do to my network config?
[01:29] <jrwr> How stable is glibc crypt() (using $6$) from ever changing?
[01:44] <jMCg> Hey folks
[01:45] <jMCg> I got my NFS server up and running and I put according to https://help.ubuntu.com/community/NFSv4Howto server:/export/homes /home nfs4 _netdev,auto 0 0 in the clients' /etc/fstab -- but it doesn't mount it on boot.
[01:46] <jMCg> "Currently Ubuntu Server does not come with the scripts needed to auto-mount nfs4 entries in /etc/fstab after the network is up." oh joy.
[01:47] <jMCg> I was wondering about the missing /etc/network/if-up.d/$stuff...
[01:49] <jMCg> for fs in $( awk '/_netdev/{print $2}') < /etc/fstab ); do mount $fs ; done'
[01:49] <jMCg> That should do it.
[02:09] <jMCg> Yeah baby.
[02:10] <jMCg> Two machines, equal setup. One mounts nfs on boot, the other does not.
[02:27] <SpamapS> jMCg: can you boot with '--verbose' in grub on the one that doesn't, and share your /var/log/boot.log and grep init: /var/log/syslog ?
[02:54] <jMCg> SpamapS: I'd rather not touch grup - I only have SSH access.
[02:55] <jMCg> SpamapS: but adding sleep 5 to my script fixed the issue -- I tried it before with ifdown eth0; ifup eth0 -- and it said no route to host when trying to mount nfs.
[03:23] <scalability-junk> hey I'm trying to use automatic memory allocatin with ubunutu 11.04 and kvm, but can't find any docs, can anyone help please
[03:23] <scalability-junk> *allocation
[03:40] <ejv> greetings, I have apcupsd monitoring my APC UPS battery backup, upon a fresh restart of the server, 'apcaccess' responds promptly with data dump of all the relevant information; after several hours, it stops responding correctly and my dmesg log fills up with "generic-usb 0003:051D:0002.0001: control queue full" why is this happening, and how do I fix it?
[03:59] <Takyoji> RAID 1 is really the only possibility for software RAID (to boot from), right?
[04:00] <patdk-lap> no
[04:01] <Takyoji> then how the does the boot loader magically read the harddrives (if it exceed a sector) if it doesn't even have software RAID running? :P
[04:02] <jrwr> I think it does some magic sector mapping
[04:03] <patdk-lap> Takyoji, that would be the magical raid grub2 boot driver :)
[04:03] <Takyoji> ahh
[04:03] <Takyoji> So it's only present in GRUB2 then, yes? Whereas prior to GRUB2, it wasn't?
[04:03] <patdk-lap> yep
[04:04] <patdk-lap> but as everything, from 10.04+ has grub2
[04:04] <Takyoji> also, any thoughts on software RAID versus hardware RAID?
[04:04] <patdk-lap> that depends on your needs
[04:04] <Takyoji> How exactly do you configure the hardware RAID controller?
[04:04] <patdk-lap> how exactly do you configure your bios?
[04:04] <Takyoji> So it's modularly through the BIOS process?
[04:05] <patdk-lap> unless you get one with network/serial support
[04:06] <Takyoji> basically the need here is generally fault-tolerance
[04:06] <patdk-lap> then anything would work
[04:06] <patdk-lap> if the need is quicker burst ability, hardware raid + battery backup
[04:07] <patdk-lap> but there generally is not much difference from software and hardware raid unless your using a battery backup
[04:08] <patdk-lap> except that more of your pci bandwidth is used writing to the disks, that the hardware raid would be doing
[04:08] <Takyoji> because apparently there's a school that had a backup harddrive, and would rsync things nightly from the main harddrive (y u no RAID? D: ), and the backup harddrive failed, and they didn't even know, and apparently they also had a single copy of some data on there as well that's now pretty much lost. xP
[04:08] <Takyoji> ahh
[04:08] <Takyoji> there's really nothing else on the PCI bus anyway
[04:08] <patdk-lap> well, nothing will help with that
[04:08] <Takyoji> I know, but just for preventive future measures.
[04:08] <patdk-lap> software and hardware riad both need to be monitored
[04:09] <patdk-lap> ya, if they had raid, they would unlikely know one drive failed
[04:09] <Takyoji> and I'm just wondering what alert system would be used, or how you'd even tell
[04:09] <patdk-lap> and would notice when the second one did
[04:09] <patdk-lap> normally, it emails
[04:09] <patdk-lap> but that only works if email server works to get out the email address is still in use, and all that :)
[04:09] <patdk-lap> and someone does monitor that email account, and it doesn't go to junk :)
[04:10] <Takyoji> but regardless, the reason of RAID would be: better read (don't really care much about write speed, xP), no need for doing the redundancy nightly instead of on-instant, etc
[04:10] <patdk-lap> only if people have a habit of deleting things
[04:10] <patdk-lap> that they need
[04:13] <Takyoji> I wonder if there's anything for along the lines of "soft deletion"
[04:13] <Takyoji> as for the files itself, it's an LDAP server with NFS/SAMBA
[04:13] <Takyoji> moreso focusing on the NFS/SAMBA part
[04:14] <Takyoji> so there can be precautions server-side that are transparent to the client.
[04:14] <thevinci> I've set up transmission-cli and transmission-daemon, and I've just accessed the web interface, i went to download a torrent, and it gave me an error, saying permissino denied.
[04:15] <thevinci> is this because transmission doesn't have write properties in the folder I created for my downloads?
[05:12] <scalability-junk> is there any way to enable automatic memory ballooning with ubuntu (kvm) couldn't find any docs about it
[05:30] <nronksr> What are people using for email servers these days?
[05:32] <Takyoji> Postfix and Dovecot?
[05:33] <patdk-lap> heh, that is software
[05:33] <patdk-lap> I normally use hp blades
[06:13] <srk9> I have been given a virtual machine running Ubuntu Server. On Gentoo, I can run "glsa-check -l" to view all security advisories that apply to installed software. IS there an equivalent command on Ubuntu?
[06:32] <Olotila> Is there a HDD benchmark similar to CrystalDiskMark for server?
[06:32] <Olotila> I tested in windows, would be interesting to see ubuntu server comparison
[06:33] <Olotila> disk utility tests mostly just cache, like gives over 1GB/s writes for 4x1TB HDD raid0 with spinning disks
[06:34] <Olotila> http://img1.uploadscreenshot.com/images/orig/10/28101590141-orig.jpg
[07:03] <Olotila> here is test in windows side, is there a program that is easy to compare to those results?
[07:03] <Olotila> http://img1.uploadscreenshot.com/images/orig/10/28103022752-orig.jpg
[07:05] <Takyoji> I'm not sure if you can use 'palimpsest' (which is actually called the 'Disk Utility' in the Admin menu in Ubuntu) to do such remotely.
[07:05] <Takyoji> I know you can use the tool for checking details of a filesystem remotely, but not sure if the read/write test is possible remotely.
[11:38] <dgrf> Hello,can somebody give me a guide for setting up postfix mail server along with dovecot and a webmail interface?
[12:09] <ejat> can someone help me with this driver on ubuntu server
[12:09] <ejat> 1c:00.0 VGA compatible controller: Matrox Graphics, Inc. MGA G200e [Pilot] ServerEngines (SEP1) (rev 02)
[12:10] <dkn> what do ya need?
[12:13] <ejat> xorg .. is it supported in oneiric .. the driver?
[12:13] <ejat>     13.617] (EE) Failed to load module "mgag200" (module does not exist, 0)
[12:13] <ejat> [    13.617] (EE) No drivers available.
[12:13] <ejat> i hv tried "mga" also not work
[12:16] <dkn> looks like that should work
[12:20] <dkn> if the xorg.conf is setup for mga in the chipset, and mgag200 in the device section, what version are you on?
[12:22] <RoyK> ejat: X isn't really a server issue, is it? ;)
[12:23] <ejat> RoyK :) .. yeah .. agreed .. just wondering "if" u guys can help me :)
[12:23] <ejat> ive tried using vesa … but its flickering ..
[12:24]  * RoyK had an rsync running all night for this apple time machine disk and only realized now that it uses hard links
[12:26] <dkn> buy gui's can good for quick reference :)
[12:27] <dkn> but*
[12:27]  * RoyK never uses X on servers
[12:28] <dkn> of course, i'm certain extra junk that came with my gui is causing the network problems i'm having...
[12:30] <dkn> http://pastebin.com/atPqJk4k if anyone is feeling brave :)
[12:32] <dkn> one of my bonded interfaces is still getting it's own IP... i think when i booted from the recovery CD it messed something up with the networking cause i'm setup for eth0 & eth1 --> bond0-->br0
[12:32] <RoyK> the gui isn't very likely to mess up networking
[12:33] <RoyK> probably PEBKAC ;)
[12:33] <dkn> when i restarted Network Manager (used to ignore the eth1 and eth0 interfaces), it started poping up with auto eth0 and eth1 getting their address....
[12:35] <dkn> most likely, i had i just run grub before restarting i could have saved myself soooo much trouble...
[12:37] <dkn> i'm about ready to dump the whole bonding config and just go back to one interface and deal with it later :(
[12:38] <dkn> i just don't understand how eth0 and eth1 are getting an IP? i removed network manager, and /etc/network/interfaces has them as slaves for bond0.... does not compute!
[12:44] <RoyK> dunno - never used bonding on linux - sorry
[12:46] <dkn> ya? i just liked the idea that one port could go bad, or one cable could get un pluggd and everything would keep working
[12:47] <dkn> so the hard links filled things up did they?
[12:47] <RoyK> the lack of hard links
[12:47] <RoyK> 1,5TB used, while the source had 1,3TB
[12:47] <RoyK> stopped it and added -H to rsync
[12:50] <RoyK> installed 4 WD Black drives for holding this storage and found out a few hours later that adding a fan in front of them would be a jolly good idea - 65˚C isn't healty for a drive :P
[12:51] <dkn> did google post a report.... 45 or something was better then 35?
[12:52] <RoyK> I think I read something like that too
[12:53] <RoyK> but 65 is a bit over the top
[12:53] <RoyK> theyære currently at ~40˚C
[12:54] <dkn> the fans in our server could get used in a vacuum.... ear plugs NOT optional
[12:56] <RoyK> 1U servers have very small fans, spinning at a very high rate... :P
[12:56] <dkn> 16 disk 4U monster
[12:56]  * RoyK has a couple of 45 disk 4U monsters :P
[12:56] <dkn> they're hovering around 32˚C
[12:57] <dkn> ouch
[12:57] <dkn> 2.5"
[12:57] <dkn> ?
[12:57] <RoyK> well, it's two servers - 36 drives in the server chassises and 45 drives in their JBODs
[12:58] <RoyK> front and back
[12:58] <RoyK> supermicro stuff
[12:58] <dkn> i think i saw that thing
[12:59] <RoyK> 100TB net storage in each
[12:59] <dkn> is the crane optional? or included to help you mount it in the rack?
[12:59] <RoyK> hehe
[12:59] <RoyK> it's not that heavy if you install it without the drives
[12:59] <dkn> :P i know
[13:00] <dkn> and you just run single nic's?
[13:00] <RoyK> those ones uses 10Gbps
[13:00] <dkn> fiber? or copper?
[13:00] <RoyK> copper to the switch, single mode fiber from that datacentre to our main datacentre
[13:01] <RoyK> bacula backup storage
[13:01] <dkn> ahhh
[13:01] <dkn> alright alright.... i'll drop the bonding...
[13:02] <RoyK> bonding should work, though
[13:02] <RoyK> that is, if it's correctly configured on the switch
[13:02] <dkn> switch hasn't changed since i set it up last year, and it's been working great since yesterday :(
[13:03] <dkn> i mean.... until
[13:03] <RoyK> http://paste.ubuntu.com/704869/
[13:03] <RoyK> dkn: that's one of the servers' raid config ;)
[13:04] <dkn> wow.... 2 spares?
[13:04] <RoyK> should suffice - all VDEVs are RAIDz2, so in theory, we can lose 22 drives without losing data
[13:05] <dkn> ahhh
[13:05] <dkn> and.... raidz2...? is that like raid 60?
[13:05] <RoyK> raid6
[13:05] <RoyK> and no, it's not running linux :P
[13:06] <dkn> c'mon get gnome on there!
[13:07] <dkn> imagine the media library you could have :)
[13:07] <RoyK> http://paste.ubuntu.com/704872/
[13:07] <RoyK> this is the real beast
[13:07] <RoyK> only 24 drive slots, 24 gigs of memory, but with striped mirrors and SSDs for caching both reads and writes
[13:08] <RoyK> fastest file server I've seen so far
[13:08] <RoyK> cost us some NOK 43k
[13:08] <dkn> is that the solaris system for ssd caching?
[13:08] <RoyK> openindiana, so yes, solarisish
[13:08] <dkn> cool
[13:09] <RoyK> it's zfs' caching system
[13:09] <RoyK> works on freebsd too
[13:09] <RoyK> and there's a linux port, but I don't think that's very good (yet)
[13:11]  * dkn dreams about SSD caching in linux
[13:12] <RoyK> IIRC there's some in ext4
[13:12] <RoyK> you can place the journal on an SSD
[13:12] <RoyK> that won't affect reads, though
[13:13] <dkn> i think i read something to that effect....
[13:14]  * dkn drives all the way back to work get the networking..... working
[13:15]  * dkn grumbles.... stupid netnotworking
[13:17] <dkn> exit
[13:27] <scalability-junk> I'm looking for a way to have automatic memory ballooning, but couldn'T find any docs. is it even possible?
[13:30] <StevenR> scalability-junk: can you be more specific?
[13:31] <scalability-junk> I want the memory of my kvm ubuntu setup to be dynamicly allocated so if guest a needs 1 gb in peak and it is available it will get it.
[13:32] <StevenR> ahh. kvm. I'm not sure the hypervisor will do that
[13:33] <scalability-junk> any other hypervisor capable of this?
[13:33] <RoyK> scalability-junk: using virt-manager, you can set a minimum and maximum amount of memory available. is this what you mean?
[13:33] <RoyK> that's on kvm
[13:33] <scalability-junk> RoyK: sort of, I just want it to be automated, always check and allocated the needed memory, I don't think virt-manager is automating it
[13:34] <StevenR> scalability-junk: http://www.linux-kvm.org/page/FAQ#Is_dynamic_memory_management_for_guests_supported.3F
[13:34] <StevenR> might help
[13:35] <scalability-junk> StevenR: not really the link is moved and not available anymore, the one for the balloon monitor
[13:37] <scalability-junk> or is this something just supported and I wouldn't notice?
[13:40] <StevenR> scalability-junk: I think so. according to that url, it depends on your guest OS
[13:40] <scalability-junk> ubuntu 11.04 host and guest.
[13:41] <StevenR> then it probably does
[13:41] <scalability-junk> but I just don'T think it is automatic and I can't really test it
[13:46] <StevenR> scalability-junk: virt-top may help you
[13:47] <RoyK> scalability-junk: why would you use a non-LTS version for the server?
[13:47] <scalability-junk> RoyK: I always upgrade anyway
[14:11] <RoyK> scalability-junk: I usually find it wise to stick to LTS for servers
[14:11] <scalability-junk> RoyK: I'm more the always the latest kind of guy ;)
[15:19]  * dkn jumps in victory
[15:19] <dkn> network manager had cleared out the nameservers in /etc/resolv.conf
[15:36] <palt> If I have created some groups on a server, is it possible to also create a user with the same name as the group? For example  have group02, can I also make a user name group02?
[15:36] <dkn> yup
[15:36] <palt> When I use adduser group02 I get an "error" saying that the group group02 already exists
[15:37] <palt> That is "adduser group02"
[15:38] <palt> Can I use any other commands or is there a switch I have overlooked for the adduser command
[15:38] <episteme> exit
[15:38] <episteme> oops
[15:40] <dkn> adduser group02 group02 should create the user and the group at the same time and assign group02 as the primary group of the user group02
[15:40] <dkn> errr... sorry
[15:41] <palt> I still get the error message. The problem is that I have already created the group group02
[15:41] <palt> But I think it could be usefull to be able to login as the group also
[15:41] <palt> It's for an exercise many people will be doing on a server
[15:42] <palt> So they have individual accounts, but I have created groups for them. But I think it would be convenient to be able to login as their group also
[15:42] <dkn> ok
[15:44] <palt> But if it is too much a hassle they are probably allright with just a user accunt and memebership in the same groups
[15:44] <dkn> create the user group02 sudo adduser group02
[15:44] <dkn> then assign the user to the group02 sudo adduser group02 group02
[15:45] <dkn> to create a new group sudo adduser --group group03
[15:46] <dkn> it's confusing though :)
[15:47] <palt> But the command "adduser group02" complains that the group02 already exists. Yeah I know, I try to create the user and it complains about the group
[15:47] <palt> But it seems like the more low-level command useradd worked
[15:48] <palt> Oh... wait!
[15:48] <dkn> no, it's saying the user group02 is already created, so now create the group with sudo adduser --group group02
[15:48] <palt> adduser group02 --ingroup group02 seems to work :D
[15:49] <palt> When I run "adduser group04" I get this out adduser: The group `group04' already exists.
[15:50] <palt> But it seems to work if I use the useradd group04 --ingroup group04
[15:50] <palt> But yeah, quite confusing :P
[15:51] <palt> hurray :D
[15:52] <dkn> indeed
[15:53] <palt> It probably tries to be smart about users and groups and prevent me from mixing them up
[15:54] <Bigbucks> Hello, any clue why a lot of commands on my VPS are not working correctly
[16:00] <dkn> ok..... why would restarting cause the /etc/resolv.conf file to get regenerated by network manager..... when NetworkManager isn't installed anymore...
[17:17] <Daviey> smoser: can you check out bug 871297 please?
[18:11] <scalability-junk> how do I run the command (qemu) info balloon in ubuntu?
[18:41] <qman__> palt, when you adduser without specifying a group, it automatically creates a group by the same name and puts the user in it
[18:57] <palt> qman__: Aaaahh.. so therefore it didn't like that the group was specified from before :)
[18:57] <qman__> yep
[18:59] <qman__> adding the user to an existing group without confirmation could have big implications, using a different group name automatically would be inconsistent, and prompting the user would ruin scriptability
[18:59] <qman__> so erring out is the best option
[19:02] <palt> I see. Yeah, my use case is probably a bit strange. I though of adding a group user after I had created the users so
[19:03] <palt> *afet I had created the groups
[19:16] <jeeves_moss> I have a "Ethernet controller: Intel Corporation 82543GC Gigabit Ethernet Controller (Fiber) (rev ff)" and I can't get the OS to see it. ideas on how I can find it?
[19:21] <Kiall> Jeeves_: Odd - I would have thought that would be supported, since its one of the adapters VirtualBox can simulate (well .. its the same chipset anyway)
[19:22] <Kiall> when you say the OS doesnt see it, what do you mean?
[19:22] <jeeves_moss> Kiall, when I run an lspci, I get "00:0a.0 Ethernet controller: Intel Corporation 82543GC Gigabit Ethernet Controller (Fiber) (rev ff)"
[19:23] <Kiall> Okay, and you have tried adding the interface to /etc/network/interfaces?
[19:23] <jeeves_moss> but when I grep dmesg, I get "e1000: This is not a software error! Something bad happened to" "e1000 0000:00:0a.0: eth1: (PCI-X:33MHz:64-bit) 00:00:00:00:00:00" "e1000 0000:00:0a.0: eth1: Intel(R) PRO/1000 Network Connection"
[19:24] <jeeves_moss> and I can use "ifconfig eth1 up" to bring it up.  I've tried adding it to the interfaces list, and it won't come up
[19:24] <jeeves_moss> the reason I want to use it is for my NAS
[19:24] <Kiall> Oh .. Are you sure the card is working? ("This is not a software error")
[19:24] <jeeves_moss> Kiall, the card was pulled from a knowen working envoriment
[19:25] <jeeves_moss> the issue is that I have the 64bit card plugged into a 32bit slot
[19:25] <jeeves_moss> and when I use the "ifconfig eth1 up", I get "SIOCSIFFLAGS: Cannot assign requested address"
[19:26] <Kiall> a 64 bit card, assuming it fits (older PCI-X slots wont accept it), should *generally* would ..
[19:27] <jeeves_moss> yep, it fits.  and the card was functioning in a 32 bit slot in the device it was pulled out of
[19:28] <Kiall> Its possible - ah .. scratch that idea then
[19:28] <jeeves_moss> ??
[19:28] <Kiall> I was going to say maybe that card is incompatible with 32 bit slots ..
[19:28] <Kiall> but you beat me to it ;)
[19:29] <jeeves_moss> lol.  it's just weird that the OS "sees" it, but won't "use" it
[19:29] <Kiall> Well - the OS sees it, and believes there is a hardware error so is refusing to use it
[19:30] <Kiall> was it taken from another linux box?
[19:30] <jeeves_moss> hmmm.  I don't want to rip other hardware apart to get a gig card.  :-(
[19:32] <jeeves_moss> ok, let's try a "new" card.  <rolls eyes>
[19:41] <jeeves_moss> Kiall, now, let's see if it like a Sun systems MMF card any better
[19:43] <jeeves_moss> weird, it likes the Sun card.  LOL
[19:48] <Kiall> typical ;)
[20:06] <jeeves_moss> argh, it causes ssh issues
[20:06] <jeeves_moss> grrrr.  time to rip out the card in my other server that's a 32 bit
[21:15] <dkn> is it better to put the uuid, or the /dev/mapper/udisks-luks-uuis-yadayada in fstab if i want to automount an encrypted lvm?
[21:16] <dkn> automount on boot that is
[21:34] <SpamapS> dkn: I'd use UUID=xxx ... but it probably isn't that critical of a decision
[22:23] <dkn> i guess both will be visible if it's set to remember the pass on the server