[00:04] <resno> question for you guys
[00:04] <resno> if you were looking for a server admin, what would you look for.
[00:05] <resno> i want to be a server admin, and im not sure what an employeer would be looking for
[00:31] <DevoKun> resno: a decent Linux skillset, decent networking skills, Scripting (BASH and Python), version control (subversion, bazaar), great documentation skills in mediawiki.
[01:34] <resno> thanks DevoKun
[01:39] <resno> im having a problem with crons not running, at all. but others run without issue
[02:36] <joebob777as7> Having issues with two 10.04 systems: Error mounting: mount exited with exit code 1: helper failed with: mount: only root can mount
[02:36] <joebob777as7> can't mount usb ntfs drives by plugging them in
[02:36] <joebob777as7> added them to fstab and that works but if they aren't plugged in the system won't boot and that won't do.
[02:37] <joebob777as7> plz help been googling for hours lots of people have the issue no one has solved by not modifying fstab
[03:33] <kieppie> hi guys (& gal's). how's everyone doing on this fine day?
[03:38] <twb> concentrating
[03:38] <twb> don't distract me further
[03:51] <lifeless> twb: would we do that?
[03:53] <kieppie> HI guys. I have a 1TB HDD I've not used for a while (think I was pondering using it for backups), but when I plugged it in, fdisk reports there is no valid partition table. I think I've loaded data on there, so I want to check it before I blow it. Is there a way to best-guess & force a mount of the disk?
[04:03] <jmarsden> kieppie: testdisk may help you out.  http://www.cgsecurity.org/wiki/TestDisk
[04:04] <twb> kieppie: fdisk is obsolete
[04:04] <twb> kieppie: you should be using something libparted-based, like parted or gnu-fdisk
[04:06] <twb> ARGH ARGH ARGH.  Flipping upstart munchkins used absolute symlinks in /etc/init.d/foo -> /lib/init/upstart-job
[04:07] <twb> Absolute baaaaad, relative gooooood
[04:11] <kieppie> jmarsden & twb: nice, guys. thanks for the advice
[04:13] <kieppie> twb: I've used parted in the past. odd that none of those systems (AFAIK) deal with LVM
[04:14] <twb> LVM is not a partition table
[04:14] <twb> It's like saying "odd that most cars don't float on water"
[04:15] <Hypoglybetic> Can anyone point me in the right direction of Ubuntu-Server that supports i686 technology? Is that 10.04?
[04:17] <jmarsden> twb: Debian Policy says a link crossing a top level directory like that *should* be absolute: "In general, symbolic links within a top-level directory should be relative, and symbolic links pointing from one top-level directory into another should be absolute."
[04:18] <jmarsden> http://www.debian.org/doc/debian-policy/ch-files.html#s10.5
[04:18] <twb> jmarsden: hm, OK
[04:18] <twb> jmarsden: it plays merry hell when the root of the tree moves
[04:19] <twb> (e.g. chroots)
[04:19] <jmarsden> Yes :)  I think it is because people often symlink the top level dirs ... there was large mailing list discussion on this stuff in... 2008 or so.
[04:19] <twb> Ah, OK.
[04:20] <twb> Those people should probably rbind or something :-/
[04:22] <jmarsden> Hypoglybetic: Do you mean you want Ubuntu server that run 64bit amd64 instruction set code?  Any version of Ubuntu server, including 10.04, will have such images available.  Look for the ones with amd64 in the name instead of i386.
[04:24] <Hypoglybetic> jmarsden I don't think so.  I tried to install ubnutu on this Pentium M and it said something about a Kernel that supports i686 technology (it isn't x-86)
[04:25] <jmarsden> Get the exact error message and we'll see what we can do.  Is the image you currently have (that gives an error) an i386 image, or an amd64 image?
[04:26] <Hypoglybetic> Oh...Thats probably the friggen issue....Yeah, its the 64 bit, which is the amd64 edition.
[04:26] <jmarsden> Then get the 32bit i386 version and you will be fine :)
[04:26] <Hypoglybetic> I've been in 64 bit mode for years, I rarely do anything in 32 bit. Doh.
[04:27] <Hypoglybetic> First thing I am going to do is re-solder this DC jack on the laptop.
[04:30] <Hypoglybetic> oh and thanks for your help jmarsden
[04:30] <jmarsden> Hypoglybetic: You're welcome.
[05:46] <a7ndrew> I am setting up workstations to authenticate to ldap. Domain users can log on but are receiving an "Your account has expired; please contact your system admnistrator" Does anyone know what causes that and how to fix it?
[05:46] <gobbe> is the ldap-server microsoft activedirectory?
[05:47] <a7ndrew> no, its openldap
[05:51] <gobbe> ok
[05:52] <gobbe> hmmh, does this happen with every account?
[05:53] <a7ndrew> one thing i noticed is that getent shadow for a local user (who doesn't receive that message) shows a value in the third field, "date of last password change" wheras there is no value for ldap users
[05:53] <a7ndrew> so every directory account, not for local accounts
[05:59] <gobbe> ldap-server should have information also for shadowExpire
[05:59] <gobbe> check out the howto for openldap
[06:00] <gobbe> https://help.ubuntu.com/community/OpenLDAPServer might be this one if i remember url correctly :)
[06:00] <a7ndrew> I'll check it out, I've been looking more at the client side of things.
[06:01] <gobbe> yep
[06:01] <gobbe> but ldap-server needs to have all information like shadow-expire
[06:14] <a7ndrew> I'm not sure we've got shadowexpire set in our current slapd.conf, its definitely something to look for, thanks
[06:28] <twb> gobbe: those attributes need no be set
[06:28] <twb> A valid RFC2307 user object need no have any shadowFoo attributes at all
[06:29] <twb> Personally I prefer to apply slapo-ppolicy, so that you don't need to give read access to password hashes to workstations' root user
[06:31] <gobbe> ok
[06:31] <gobbe> our ldap has them and working fine
[06:32] <gobbe> my bad
[06:32] <gobbe> anyway you cannot set ecpiredate without those
[07:02] <twb> gobbe: I don't like giving out passwords to anyone with a laptop and a network cable
[07:06] <gobbe> that's true
[07:09] <twb> Admittedly, slapo-ppolicy means (AFAICT) you can't hook samba into it
[07:10] <twb> But I don't need samba enough to roll out kerberos, which is the only way to do it (secure, centralized auth) properly.
[07:14] <Hypoglybetic> I just disassembled a dell laptop and put it back together in only 2 hours! I only have 3 screws left over.  I'm a genius.  G'nght #ubuntu-server
[07:30] <klochner> anyone willing to help me sort out system services?
[07:39] <twb> !ask
[07:39] <twb> !anyone
[07:58] <AlanBell> I am trying to explain why a KVM guest VM is outperforming the host, would like to compare a little benchmark result with others if possible.
[07:59] <AlanBell> I am computing a bunch of digits of pi just as a CPU load test using the command: time echo "scale=5000; 4*a(1)" | bc -l -q
[08:00] <AlanBell> on the host it takes 42 seconds, on the guest it does it in 28 seconds
[08:01] <AlanBell> there is no other running load on the box, just a couple of dormant VMs, there is plentiful memory and 4 cores hyperthreaded all idling normally
[08:04] <AlanBell> top shows that when the guest is running the benchmark it is maxing out just one of the 8 processors, the rest are idle
[08:05] <AlanBell> same when running on the host, maxes out just one CPU
[08:08] <twb> AlanBell: what are you using, dhrystone?
[08:09] <AlanBell> no, just: time echo "scale=5000; 4*a(1)" | bc -l -q
[08:10] <twb> Well perhaps it's because your test is bogus
[08:10] <twb> http://homepages.cwi.nl/~steven/dry.c <-- try that
[08:22] <AlanBell> 20920502 on the host and 19379844 on the guest
[08:22] <AlanBell> which is the right way round
[08:22] <AlanBell> so why was my braindead benchmark bogus?
[08:33] <twb> NFI
[08:34] <twb> What's a()?
[08:34] <twb> It can't be Ackermann's...
[08:52] <AlanBell> arctan
[08:54] <twb> Pft
[08:55] <AlanBell> I know, not a real benchmark
[08:56] <AlanBell> thanks for the dry.c test btw
[08:57] <twb> It's not a real benchmark, either
[08:57] <twb> But ICBF buying specINT
[08:58] <twb> yay, my VM's a little faster: 25290844
[08:59] <AlanBell> cool
[08:59] <twb> It probably only tests one core
[09:00] <AlanBell> it does
[09:00] <twb> http://en.wikipedia.org/wiki/Dhrystone is the reference
[09:01] <AlanBell> this has revealed that a core i7 @2.8Ghz is about twice as fast as a Xeon @3.2Ghz
[09:01] <twb> Again, it's a SYNTHETIC benchmark
[09:02] <twb> http://en.wikipedia.org/wiki/Dhrystone#Shortcomings
[09:02] <AlanBell> indeed, this all comes from a customer machine with a big PHP app that we put on a little dev VM and it runs twice as fast as their production box
[09:04] <AlanBell> been tracing back caching settings, mysql tuning and all the rest but it looks like raw CPU power is the factor
[09:06] <twb> flops or mips?
[09:07] <AlanBell> hmm, probably mips
[09:08] <AlanBell> there is a big database of products and it calculates prices, stock levels, order requirements etc, just PHP and mysql crunching
[09:09] <twb> IME the best way to improve performance in such a system is to find someone who actually groks PHP and SQL to review the code
[09:10] <twb> Because there are so many stupid PHP monkeys, and they tend to write O(fⁿ) algorithms when there is an obvious O(f) algorithm
[09:10] <AlanBell> that was what we were about to dive into
[09:11] <twb> I had one guy last week, his perl script all of /var/log/daemon.log into memory before searching it... so when a dnsmasq bug made it grow to 1GB, his perl script was using about 12GB of RAM and swap
[09:11] <AlanBell> I was just trying to understand the baseline performance, took a copy onto our new dev machine and was a bit surprised at how much faster it was than the production one with the problem
[09:13] <AlanBell> I know there are algorithm problems in the code, might be repetitive stuff that could be dumped to memcached
[09:13] <twb> Hit them with a hammer labelled "`optimize LATER' doesn't mean `optimize NEVER'!"
[09:13] <twb> (The devs, I mean.  Not the bugs.)
[09:23] <AlanBell> heh, yeah
[09:28] <AlanBell> I think the original devs followed this algorithm http://xkcd.com/844/
[09:39] <awanti> I want to monitor my Apache web server. how? plz.
[09:40] <awanti> through cmd
[09:40] <twb> !English
[09:41] <awanti>  I want to monitor my Apache web server. how? please. (from command line)
[09:42] <twb> You can use netstat or nc to attempt to connect to it
[09:42] <twb> Or w3m, to actually browse, of course.
[09:46] <awanti> and how to recover /etc/fstab if its corrupt
[09:46] <twb> awanti: just write a new one
[09:47] <oCean> /etc/fstab is just an ascii file. You would have to restore it or correct the faulty line by hand
[09:47] <awanti> is any other way to recover or is their any tool?
[09:47] <oCean> no
[09:48] <twb> sysadmins aren't tools
[09:48] <oCean> some people do see 'em that way
[09:49] <awanti> 'em?
[09:49] <twb> awanti: it's called a "contraction"
[09:49] <oCean> them
[11:04] <yann2> hello! I am missing the lmhost file in my samba install - is this normal?
[11:15] <a7ndrew> whoops, sorry about that :P
[11:17] <a7ndrew> yann2: I just installed samba (3) yesterday and dpkg -L tells me there is no lmhosts
[11:27] <yann2> but in the man smb.conf:
[11:27] <yann2>            Default: name resolve order = lmhosts host wins bcast
[12:43] <randomdave> hi
[13:02] <yann2> arf.
[13:02] <yann2> was going to ask "Is this resolved: http://www.mail-archive.com/samba@lists.samba.org/msg103373.html"
[14:13] <mwd> Hi there, I've just installed Ubuntu server 10.04 for testing purposes. Unfortunately I've selected "german" als language. How can I switch back to "english"?
[14:14] <slestak> good morning guys.  I have an lts vmware vm that I am trying to reinstall postgres on.
[14:14] <slestak> the purge did not remove /etc/postgresql or /etc/postgresql-common, so I removed them manually
[14:15] <slestak> the following install, biut not put them back.  pretty frustrating.
[14:15] <pmatulis> slestak: what was your purge command?
[14:15] <slestak> sudo aptitude purge postgres
[14:15] <slestak> going from memory, that was last night, I can look in bash history
[14:16] <Slyboots> shouldnt it be sudo apt-get remove --purge postgres
[14:16] <slestak> i will try that.  is aptitudes use frowned upon?
[14:16] <sander^work> What is the virtual kernel meant for?
[14:17] <pmatulis> slestak: there are other postgresql packages, for instance, /etc/postgresql-common is from postgresql-client-common probably
[14:17] <slestak> i removed all of them
[14:17] <slestak> the aptitude purge said it was removed the whole list
[14:17] <patdk-lap> you have to purge them all
[14:17] <slestak> ahh
[14:17] <patdk-lap> if it removes then, configs and stuff stay
[14:17] <patdk-lap> and you can't purge them if they are removed
[14:18] <patdk-lap> apt-get purge postgres.*
[14:19] <geekbri> if you want to have a user who can't login with a shell do you use /bin/nologin or /bin/false?
[14:19] <patdk-lap> anything that isn't in /etc/shells
[14:19] <RoyK> patdk-lap: shells don't neede to be listed there for login to work
[14:19] <RoyK> geekbri: both will work
[14:19] <patdk-lap> oh?
[14:20] <geekbri> RoyK: ok thanks !
[14:20] <RoyK> patdk-lap: just try creating some shell script under /usr/local/bin and point to it - works like a charm
[14:21] <patdk-lap> heh, evil
[14:21] <slestak> patdk-lap: ty, i think i found my example needed to give up aptitude for good :)  that looked effective
[14:22] <mwd> How can I change the language on the command line in ubuntu server 10.04 from "german" to "english"?
[14:22] <slestak> i think that would be a locale settign wouldnt it?
[14:22] <slestak> i am not sure
[14:22] <mwd> Can this be changed without reinstalling?
[14:22] <slestak> each user can have there console in a different language
[14:23] <mwd> How can I change the default setting from german to english?
[14:24] <slestak> i am not sure, I am one of those unfortunate English-only speakers (barely at that)
[14:25] <slestak> check out package language-selector
[14:32] <mwd> Ok, in /etc/default/locale I have the setting: LANG="de_DE.UTF-8"
[14:33] <mwd> Now I need the correct setting for "standard engish"
[14:34] <patdk-lap> en_US?
[14:34] <slestak> sec, let me look at mine
[14:34] <patdk-lap> en_UK?
[14:34] <slestak> LANG="en_US.UTF-8" is in mine
[14:35] <slestak> i am catching all kinds of grief with permissions with rsyslog-4.2
[14:35] <mwd> Ok, when changing the entry do I have to execute a command ?
[14:35] <patdk-lap> man there are a lot of them, en_AG, en_HK, en_NG, ....
[14:37] <sander^work> How can I see from witch apt repository a kernel i'm using is installed from?
[14:37] <mwd> Ok, LANG="en_US.UTF-8" seems to be the default setting. Do I have to execute a command that the changes take place?
[14:38] <slestak> i dont think so.  after chanign it just open a new shell and see if it is effective
[14:40] <mwd> It works, thanks :)
[15:39] <pmatulis> patdk-lap: i believe you can purge after removal
[15:40] <patdk-lap> I never can :(
[15:40] <patdk-lap> it just tells me package not found
[15:41] <jpds> patdk-lap: dpkg -P <package name> ?
[15:42] <patdk-lap> never tried it that way
[15:57] <tgwoollard> Good afternoon all. I am running Ubuntu 10.04 and have just made a change in fstab to automount a new partition. On reboot the system won't startup and just hangs. I therefore have no terminal in order to reverse the change. How can i get access to a command line on this installation so that i can reverse my fstab entry? Many thanks in advance
[16:04] <compdoc> tgwoollard, Ive had that happen. You can boot the live cd, mount the root, and try to correct fstab
[16:04] <compdoc> I would just comment out all the changes you did
[16:05] <tgwoollard> Thanks compdoc. That's exactly what i want to do. I think it's a problem i've seen before that actually relates to ISCSI connections not starting properly. Forgive me but i've never actually used a liveCD. Would the gparted livecd suffice in this instance?
[16:06] <compdoc> It might - it allows you a command line. and it has most utils to work with drives
[16:06] <compdoc> like mount, etc
[16:06] <pmatulis> tgwoollard: for iscsi you first need to import the block device prior to mounting it's filesystem
[16:07] <tgwoollard> THanks pmatulis. I've setup the block device fine, partitioned it and mounted it manually. I guess i just screwed up fstab somehow. Will report back shortly
[16:08] <pmatulis> tgwoollard: ok, so next time you should test the fstab entry before rebooting (mount -a or mount /dev/whatever)
[16:08] <tgwoollard> Oh yeah that would've been sensible. It's been a long day!! Thanks for the tip :)
[16:25] <zul> hggdh: around
[16:25] <zul> ?
[16:27] <hggdh> zul: yes
[16:27] <hggdh> getting there now
[16:27] <zul> hggdh: or i can come down
[16:27] <hggdh> nah, I will go up, so that the rest of the team will still remember me ;-)
[16:27] <hggdh> zul: give me 5 min
[16:28] <tgwoollard> compdoc. Pmatulis. Many thanks for your suggestions. I'm back in business
[16:28] <zul> hggdh: gotcha too many context switches right now
[16:28] <compdoc> cool. gparted worked?
[16:28] <hggdh> zul: heh. What's one more, then?
[16:29] <tgwoollard> compdoc i actually just booted in to a shell prompt in recovery and made the root disk writeable
[16:33] <bpgoldsb> I'm taking a run at preseeding.  I'm using the netboot version of Lucid.  I'd like to host my preseed file on a webserver.  However, doing that means it doesn't get loaded until I manually configure the keyboard and such.  Is there a way to fix that?
[16:36] <pmatulis> bpgoldsb: yes, include the information as a kernel boot parameter for the installer
[16:42] <hggdh> zul: just found I have a meeting now, can we postpone it for (say) 40 min?
[16:42] <zul> hggdh: lol sure :)
[16:42] <hggdh> zul: merci beaucoup
[16:42] <zul> hggdh: bientot
[16:42] <gtaylor> Anyone know how to make a chmodding of /tmp persist across restarts?
[16:43] <gtaylor> There was some way, just can't remember.
[16:45] <patdk-lap> you don't
[16:45] <patdk-lap> you just stop the script from deleting things in it
[16:45] <gtaylor> I'm not concerned about its contents
[16:45] <gtaylor> I just need the mount point to be writable by all
[16:45] <patdk-lap> heh? that is default
[16:46] <gtaylor> Looks like it's getting changed: drwxr-xr-x
[16:46] <pmatulis> yeah, what's up with that?
[16:46] <gtaylor> I'm running this on EC2 with /tmp being mounted from ephemeral0
[16:46] <pmatulis> drwxrwxrwt 4 root root 4.0K 2011-01-12 11:46 /tmp
[16:46] <gtaylor> So I'm wondering if it's getting changed at mount time
[16:46] <patdk-lap> chmod 1777 /tmp if it's screwed
[16:47] <gtaylor> patdk-lap: That only persists until I restart the machine
[16:47] <gtaylor> Pretty sure it's getting chmodd'd while mounting that /tmp
[16:47] <patdk-lap> what do you mean by mounted from ephemeral0?
[16:47] <gtaylor> Amazon EC2's ephemeral devices, contents are not retained across restarts.
[16:48] <gtaylor> But it's a "free" good chunk of temporary storage
[16:48] <patdk-lap> oh, that is why
[16:48] <gtaylor> the contents don't have anything to do with this, though
[16:48] <gtaylor> at start time, it's an empty ext3 partition
[16:49] <gtaylor> yeah, weird. I unmount it, tmp is 777, I mount it, goes back to 744
[16:49] <gtaylor> but unmount it again and it's back to 777
[16:49] <patdk-lap> ya, that works that way
[16:49] <patdk-lap> one min, have to find where I'm doing it
[16:49] <patdk-lap> I had to do that for tmpfs /tmp
[16:50] <gtaylor> ah, ok
[16:50] <patdk-lap> same issue
[16:50] <patdk-lap> tmpfs		/tmp		tmpfs		mode=1777
[16:51] <gtaylor> hmm, that creates the tmpfs from memory though, right?
[16:51] <patdk-lap> ya
[16:51] <patdk-lap> you want the mode=1777
[16:52] <gtaylor> hmm, looks like it's freaking out, but let me keep tinkering
[16:53] <gtaylor> EXT3-fs (sdb): error: unrecognized mount option "mode=1777" or missing value
[16:53] <gtaylor> shoot, does that only apply to tmpfs?
[16:53] <patdk-lap> dunnp
[16:53] <patdk-lap> dunno
[16:57] <RoyK> man mount
[16:58] <resno> ive got a cron that isnt working. and i havent been able to get to the bottom of why not.
[16:58] <gtaylor> RoyK: Where in there does it have this?
[16:58] <RoyK> gtaylor: just look for tmpfs
[16:58] <patdk-lap> he isn't using tmpfs
[16:58] <gtaylor> RoyK: I can't use tmpfs, that's memory-resident.
[16:59] <RoyK> you want tmp on a separate filesystem?
[17:00] <RoyK> I usually create a small /var/tmp and then bind mount that to /tmp
[17:00] <gtaylor> And /var/tmp stays 777?
[17:00] <patdk-lap> royk, he is using a tmpfs that looks like a normal drive
[17:00] <patdk-lap> so on reboot, the drive is wiped clean
[17:00] <RoyK> well, that's tmpfs for you
[17:00] <gtaylor> I don't care that it's wiped clean, it just mucks with the umask on /tmp
[17:01] <patdk-lap> ya, I think your going have to manually fix it
[17:01] <gtaylor> I tried setting up an upstart entry to chmod it, but I think the task happened after mount time
[17:01] <RoyK> gtaylor: if you can't use tmpfs, use a normal filesystem
[17:01] <RoyK> even root will do if there isn't much usage on /tmp
[17:01] <gtaylor> RoyK: It is, it's an ext3 fs
[17:01] <RoyK> then just chmod 1777 /tmp
[17:01] <RoyK> and you're done
[17:02] <patdk-lap> royk, http://victortrac.com/EC2_Ephemeral_Disks_vs_EBS_Volumes
[17:02] <gtaylor> RoyK: It only sticks until I restart, though
[17:02] <patdk-lap> your being overly dense :)
[17:02] <gtaylor> Which is bad, because these are EC2 instances that are instantiated all the time.
[17:02] <patdk-lap> gtaylor, how do you do a mkfs?
[17:02] <klochner> can anyone clear up an init.d question for me?  I want to confirm that the OS doesn't monitor daemon processes launched from init.d
[17:03] <gtaylor> patdk-lap: That's the weird thing, it just comes up as a valid ext3 partition. I guess that's what Amazon decided would be ephemeral storage for UNIX/Linux instances.
[17:03] <klochner> i.e., doesn't restart failed processes
[17:03] <RoyK> gtaylor: pastebin `mount`
[17:03] <gtaylor> patdk-lap: Perfectly valid, empty partition
[17:03] <gtaylor> RoyK: /dev/sdb on /tmp type ext3 (rw,_netdev)
[17:03] <patdk-lap> klochner, nope, that would be upstart
[17:03] <RoyK> gtaylor: iscsi?
[17:03] <klochner> thanks
[17:05] <gtaylor> If I could make my upstart event the very last thing ran in the boot process, this would be a non-issue, as I can just chmod it then.
[17:06] <patdk-lap> gtaylor, rc.local?
[17:06] <RoyK> gtaylor: what is /dev/sdb ?
[17:06] <klochner> do most people use upstart rather than init.d ?
[17:06] <patdk-lap> royk, read up, that was explained like 5 times
[17:07] <RoyK> oh, ec2
[17:07] <RoyK> but that shouldn't matter
[17:07] <patdk-lap> royk, it's an ec2 disk that is wiped at each reboot, much like tmpfs
[17:07] <RoyK> then just add chmod 1777 /tmp in rc.local
[17:08] <gtaylor> patdk-lap: Hmm. Will have to try that. But I think there's an equivalent to upstart, too. Annoying thing is, the _netdev mount option that it uses/requires means it can get mounted after everything else is already started/booted.
[17:08] <gtaylor> Pain in the butt.
[17:09] <RoyK> not everything else, just the network and associated services
[17:09]  * gtaylor facepalms.
[17:09] <gtaylor> RoyK: _netdev
[17:10] <gtaylor> and nobootwait
[17:10] <geekbri> does anybody know if there is an ubuntu repo that keeps the most up to date versions of php avialable?
[17:11] <gtaylor> Trying rc.local now. I imagine there are times when this will work and there are times when this will fail. If wherever they get these ephemeral devices from is laggy enough, it could conceivably be mounted after rc.local is ran. Meh.
[17:12] <gtaylor> patdk-lap: It looks like rc.local might be the trick. Gotta see if it consistently works, but that's promising. Worked on the first attempt. Thanks!
[17:13] <RoyK> geekbri: perhaps ubuntu backports?
[17:13] <geekbri> RoyK: thanks, ill take a peak
[17:13] <geekbri> peek
[17:18]  * Slyboots sighs.. blasted permissions are busting my balls
[17:18] <webroasters> hi guys. quick postfix question. I have a server with a few sites on it, and I'm trying to setup re-configure postfix. What Internet protocol should I use? All, IPv6, or IPv4?? I'm not real sure what any of that means, but what does everyone suggest??
[17:19] <RoyK> webroasters: if you don't know what that means, choose ipv4
[17:19] <RoyK> that's the current one
[17:19] <webroasters> ok thx
[17:31] <ogra> SpamapS, http://paste.ubuntu.com/553265/ thats the serial.conf file we copy into /etc/init in ubuntu arm
[17:31] <SpamapS> ogra: ack
[17:38] <EvilSushi> anyone here running encrypted backups via duplicity to s3?
[17:39] <SpamapS> ogra: honestly, that looks great. What are the objections to it?
[17:39] <ogra> SpamapS, parsing of /proc/cmdline
[17:39] <ogra> SpamapS, Keybuk always blocked because of that
[17:40] <ogra> i think we can do it in a distro patch to the upstart package, but it wont be accepted as upstream fix
[17:54] <\sh> hmmm...when I want to upgrade an ec2 lucid instance to maverick, and I did a dist-upgrade from lucid to maverick...what aki id do I need to set to boot the correct kernel (which should be linux-image-2.6.35-24-virtual)
[18:04] <kim0> \sh: might wanna ask in #ubuntu-cloud
[18:14] <TimR> hey guys can anybody help me out with perl module for Net/SMTP/TLS/ButMaintained.pm  for server 10.04LTS
[18:21] <DevoKun> TimR: what's the problem?
[18:37] <jongbergs> hi, i have ubuntu-desktop 9.10 running fo a while..how do i convert this to a server? that is..just like a server edition without applications that's part of standard desktop..
[18:42] <sako> hey guys I am trying to select-editor however I get macvim
[18:42] <sako> er
[18:42] <sako> /usr/bin/select-editor: 47: gettext: not found
[18:44] <sako> was there another way to select the system wide editor? I remember using another method
[18:46] <gobbe> sudo update-alternatives –config editor
[18:46] <gobbe> --config editor
[18:49] <bpgoldsb> Anyone here preseeding their Ubuntu mirror?
[18:54] <sako> gobbe: thats what it was thanks!
[18:56] <RoAkSoAx> kirkland: could you please sponsor these (in this order) :): bug #700161, bug #701233, bug #702023, bug #702013
[18:57] <sako> hey guys, what do you think about running ubuntu 10.04 on a VPS with 512mb ram.. Is the ram gained by 32bit worth the trouble?
[18:57] <sako> I'm assuming you give up the optimization of 64bit on some software like mysql for a little less ram used in 32bit?
[18:58] <sako> Can't figure out if this is worth it :?
[18:59] <gobbe> i didn't get your point
[18:59] <gobbe> you want to run 10.04 with 512MB ram?
[18:59] <sako> on a VPS
[18:59] <sako> gobbe: they offer both 32bit and 64bit..
[18:59] <gobbe> aah
[18:59] <gobbe> well, i don't see any point to use 64bit
[19:00] <sako> gobbe: so I am trying to decide which is better suited.. apparently 32bit will eat more ram as pointers are bigger etc
[19:00] <gobbe> if you don't have software that needs it
[19:00] <sako> gobbe: well, I guess the question is the benefits that software optimized for 64 bit like mysql.. is that a significant difference?
[19:00] <sako> I have been looking for benchmarks.. can't find anything solid
[19:01] <gobbe> i would say that you cannot see any difference on there
[19:01] <gobbe> maybe when you have lots of memory and you are running with high load
[19:01] <sako> i see
[19:02] <gobbe> but i'm running my servers with 64bit
[19:04] <kirkland> RoAkSoAx: will do just after lunch!
[19:48] <geekbri> does fastcgi have its own package? or is spawn-fcgi the package for it?
[19:57] <LowValueTarget> anyone know of a timeline for an update to fix the latest PHP bug (last update caused openbase_dir issues)
[19:57] <LowValueTarget> i knwo the fix has been committed to the trunk
[19:58] <ignarps> geekbri, maybe check libfcgi-* ?
[19:58] <pmatulis> LowValueTarget: i know there has been a recent security fix released for PHP
[19:58] <geekbri> ignarps: thanks ill give a look
[19:59] <LowValueTarget> pmatulis: how recent, the update the boke things was released on the 8 iirc
[19:59] <LowValueTarget> *that broke
[20:23] <geekbri> does anybody know a good way to obtain the ./configure used in the php install for ubuntu?
[20:27] <RoyK> geekbri: apt-get source
[20:27] <geekbri> RoyK: most excellent, thank you
[20:27] <RoyK> the configure line is in debian/rules iirc
[21:42] <geekbri> RoyK: i found it, thanks a ton!
[22:52] <BlaDe^> Hi guys ---- With a regular http request it takes around 322ms to connect to the server -- that's obscene (It doesn't always take that long.. it varies)
[22:53] <BlaDe^> any ideas what it could be?
[22:57] <pmatulis> BlaDe^: use tcpdump to capture the traffic on the wire
[22:59] <BlaDe^> pmatulis:  there's an awful lot of output---mostly from my ssh connection
[23:01] <BlaDe^> can I filter by port 80 ? can't see it in man tcpdump
[23:01] <Nafallo> BlaDe^: filter it, and potentially consider some better measurement than ping?
[23:01] <BlaDe^> im not using ping --- i'm looking at the chrome output of the http request
[23:01] <BlaDe^> and that's good measurement because it's a webserver and I want it to server web files, fast
[23:02] <BlaDe^> serve*
[23:02] <Nafallo> BlaDe^: what httpd are you using?
[23:02] <BlaDe^> nginx
[23:02] <Nafallo> I found thttpd to be quite efficient myself :-)
[23:03] <BlaDe^> benchmark wise nginx performs the best for what I need
[23:04] <Nafallo> fair enough
[23:04] <BlaDe^> but still something is going wrong
[23:04] <BlaDe^> the connection and wait time is too high
[23:04] <BlaDe^> it can be anywhere between 60 and 350ms which is extremely inconsistent
[23:05] <BlaDe^> the load avg is 0.17 0.08 0.02 1/266 7954 and I have Intel(R) Xeon(R) CPU           X3360  @ 2.83GHz * 4 (4 cores)
[23:07] <BlaDe^> is there any kernel tuning I can do to speed it up ?
[23:07] <pmatulis> BlaDe^: 'sudo tcpdump -qtn -e eth0 port 80' is a sample
[23:08] <Nafallo> -e ? not -i ?
[23:08] <BlaDe^> okay that filters web requests it seems - but doesn't show anything useful
[23:08] <BlaDe^> yeah I used -i
[23:18] <BlaDe^> Nafallo:  does this help much? http://pastie.org/private/y4gmvn9y58g75pyd2yab8g
[23:18] <BlaDe^> i filtered my request only
[23:18] <BlaDe^> one request but the keep-alive sends some ack's i guess
[23:20] <Nafallo> BlaDe^: I believe pmatulis is better off helping you. :-)
[23:21] <BlaDe^> google doesn't show anything much for that
[23:40] <smoser> kirkland, http://paste.ubuntu.com/553405/
[23:55] <pmatulis> BlaDe^: yes, '-i' sorry