[01:44] <sond> howdy all - i have one of these: Ubuntu 12.04.1 LTS (GNU/Linux 3.2.0-52-virtual x86_64) that appears to have its /boot partition %100 full due to automatic security updates ... do i just delete the earlier ones and edit the grub ? or whats the "best" way to sort this ?
[01:45] <sarnold> sond: best is to delete a kernel or two by hand, then use dpkg --purge to remove the corresponding packages for the kernels you've removed once dpkg will function again.
[01:45] <sarnold> sond: best of course is to keep at least two kernels around -- whatever you're currently running, and maybe the newest kernel or the next newest kernel.
[01:52] <sond> sarnold: thanks.. its an interesting situation which i can foresee reoccurring on other VMs - there's approx 18 of them in there..  the /boot partition was automatically assigned 200mb during the install which is now of course full.. been trying to figure out why things wernt working
[01:52] <sond> approx 18 kernels *
[01:53] <sarnold> sond: it's frustrating, I know I've cleaned this up at least once or twice, but most of the time apt seems to prune /boot on its own without any manual intervention.
[02:00] <sond> sarnold: in my /boot should i delete all of these : abi-3.2.0-XX-virtual,  config-3.2.0-XX-virtual,   initrd.img-3.2.0-XX-virtual,  System.map-3.2.0-XX-virtual,  vmlinuz-3.2.0-XX-virtual ? for each unwanted kernel ?
[02:01] <sarnold> sond: I only delete the 'big' things and let dpkg --purge clean up the mess
[02:01] <sond> sarnold: cool i was kinda hoping you would say that
[02:01] <sarnold> :)
[02:02] <sond> sarnold: i'll report back in a bit..
[02:07] <sond> sarnold: dpkg: error: --purge needs at least one package name argument ........ what do i give it ?
[02:08] <sarnold> sond: it'll be linux-image-... ... check dpkg -l | grep "^ii  linux"  for a quick list of choices...
[02:10] <Beatstreet> I have a HDD giving me problems. sdd1 and sdd2 are both part of different RAIDs. the RAID with sdd2 is running strong but sdd1 keeps getting kicked out as failed device. How can I format just that partiton?
[02:10] <sond> sarnold: thanks..
[02:10] <sarnold> sond: is dpkg happy again?
[02:11] <sond> brb..
[02:29] <sond> sarnold: dpkg not happy yet .. having probs with the syntax .. want to do a dpkg -P --force --nodepends
[02:29] <sarnold> sond: hrm :( what kinds of packages depend upon your linux packages?
[02:30] <sond> sarnold: ok.. i did this dpkg --purge linux-headers-3.2.0-45 and it didnt like it
[02:31] <sond> sarnold: possibly because i manually removed the ramdisk and vmlinz
[02:31] <sond> *corresponding
[02:33] <sarnold> sond: I don't think that'd be it; I just did dpkg --no-act --purge ... on some of my linux-headers- packages -- linux-headers-version-generic required the linux-headers-version package
[02:34] <sarnold> sond: in the end, also deleting linux-headers-generic and linux-generic did the job
[02:34] <sarnold> sond: you can put linux-headers-generic and linux-generic back..
[02:36] <sond> sarnold: weird.. its working now
[02:36] <sond> i think.. brb
[02:38] <sond> sarnold: yep , its seems to be behaving now..  i appreciate your assistance : )
[02:41] <sond> sarnold: it would appear that i had to remove the linux-headers-3.2.0-XX-virtual   before  the linux-headers-3.2.0-XX
[02:42] <sarnold> sond: aha! :) thanks
[02:42] <sarnold> there has been plenty of churn in the kernel packaging over the years, I forget some of the details :/
[02:43] <sond> sarnold: you mentioned earlier that you thought that apt may have / should have sorted out bloated /boot partition..
[02:43] <sarnold> sond: yeah
[02:44] <sond> sarnold: coz if it doesn't this could be a major if you dealing with lots of VMs
[02:44] <sarnold> sond: yeah :)
[02:45] <sond> Beatstreet: is it real raid or fake raid (mdadm) ?
[02:45] <Beatstreet> mdadm
[06:14] <leif_> greetings
[06:35] <leif_> greetings
[06:56] <leif_> :)
[07:08] <leif_> patdk-lap, by any chance danish
[07:14] <LeMike> helloo. i got a server, that crashes every morning during backups and sometimes even turn off. the system.log has no hint what went wrong. any other things I can lookup?
[09:52] <rbasak> hallyn_: did you have any other fixes to SRU along with bug 1248394? I've prepared a debdiff for that one and attached to the bug. libvirt isn't in the precise server packageset so I can't upload it without poking Colin.
[09:53] <rbasak> hallyn_: looks like the previous SRU is still in flight?
[11:39] <xnox> Are there cloud images for trusty yet? https://cloud-images.ubuntu.com/query/trusty/server/released-dl.current.txt ?
[11:45] <rbasak> xnox: not released. Daily: https://cloud-images.ubuntu.com/query/trusty/server/daily-dl.current.txt
[11:45] <rbasak> xnox: the first "release" will be alpha-1 I think. Not sure.
[11:47] <xnox> rbasak: but we have daily trusty images ? Ah, i guess i need to change to daily channel instead of using released on.
[11:49] <xnox> yeap, daily stream works.
[12:27] <martinrame> hi, I need to create a new partition after /dev/sda5. My layout is this: http://pastebin.com/fGU2BA9u, can anyone help me?
[12:28] <martinrame> my problem is that if I create a new primary partition, it only has 1Gb, and I need to get at least 100Gb.
[12:28] <martinrame> sorry it only creates 1Mb.
[12:37] <geser> martinrame: your extended partition (sda2) isn't large enough to create a larger partition
[12:40] <martinrame> geser: what can I do, then?
[12:43] <geser> you would need to resize sda2 first and make a backup first in case something goes wrong
[14:34] <hallyn_> rbasak: i was going to sru 1245322 next week.  fix isn't yet upstream
[14:37] <zetheroo> I would like to setup a secondary DNS server at the workplace using Ubuntu  -- is this simple to do?
[14:56] <rbasak> hallyn_: that's not for precise though, right?
[14:56] <leif_> grettings
[14:59] <hallyn_> rbasak: oh, no. saucy.  forgot you were doing precise
[15:15] <leif_> any one wana chat casual on severs / network stuufff
[15:27] <leif_> :)
[15:42] <leif_> hmmm'
[15:53] <leif_> damm this place is dead
[16:01] <leif_> gret
[16:01] <leif_> greetings
[16:05] <hallyn_> rbasak: so if you get a chance to verify bug 1190387 that'll speed along the precise libvirt sru i guess
[16:06] <hallyn_> (i can next week, but not this week)
[16:06] <leif_> evning hallyn_
[16:10] <hallyn_> leif_: hi - about to change locales, will be afk for a bit
[16:10] <leif_> ok
[16:10] <leif_> just borred
[16:11] <hallyn_> leif_: play around with juju+lxc :)
[16:11] <rbasak> hallyn_: np. AIUI, it's not quite that urgent. At least nobody has been chasing me about it. I think we'll care more by the next juju upstream release.
[16:12] <leif_> :)
[16:12] <leif_> still to new to launch juju
[16:14] <leif_> is still trying to learn apache and lamp
[16:38] <leif_> looking for a newbie friendly moneterig tool
[16:50] <hallyn_> leif_: monitoring?
[16:50] <hallyn_> if so, what do you want to monitor?
[16:50] <leif_> well is a newb but is rearly diiging the sever / nework stuff
[16:50] <leif_> so just a basic thing
[16:52] <hallyn_> rsyslog for local activity, and i guess nagios is the cool kids' tool
[16:52] <leif_> ok thanks
[16:52] <leif_> is that open source or
[16:52] <hallyn_> tail -f /var/log/syslog | ccze -A is my old standby :)
[16:53] <hallyn_> yes
[16:53] <hallyn_> SpamapS_ used to be a big nagios proponent iirc
[16:55] <leif_> thanks mate
[17:00] <leif_> hmmm maybe i should start whit firing up a game sever
[17:39] <HiddenDjinn> how much room do i need in a vps for family mail, some personal cloud storage, and a couple of low traffic vhosted domains?
[17:44] <patdk-wk> HiddenDjinn, 20tb
[17:44] <HiddenDjinn> patdk-wk, i doubt that
[17:44] <patdk-wk> why?
[17:44] <patdk-wk> that is how much I'm using for the same stuff
[17:45] <patdk-wk> and since I am not you, I can't know how much YOU need, only how much I need
[17:45] <HiddenDjinn> i'm basically pulling a "take back my data" moment, and i'm doing all but the mail and cloud storage with a 20Gig/512MB ram vps fine
[17:45] <HiddenDjinn> i'm just wondering how much or if i need to upgrade
[17:46] <patdk-wk> not going be running spamass or clamd in there then
[17:46] <patdk-wk> not enough ram
[17:47] <HiddenDjinn> patdk-wk, oh, upgrading is definitely an option atm
[17:49] <HiddenDjinn> patdk-wk, how much ram do i need to run spamassassin and clamd?
[17:49] <patdk-wk> depends
[17:49] <patdk-wk> I normally end up with 150megs for clamv or other virus program, and 200megs for spamass
[17:54]  * hallyn_ is confused by thsi conversation (but that's ok)
[17:54]  * patdk-wk is confused too
[17:54] <hallyn_> I occasinoally have my web and mail server on an efika mx, which has 8G disk and 512m ram
[17:54] <hallyn_> patdk-wk: just the switch from 20tb disk to ram requirements :)
[17:55] <patdk-wk> guess since he figured there was no way we could guess diskspace, ram is easier
[17:55] <patdk-wk> you could get away with 512m ram, but you have to be careful
[17:55] <HiddenDjinn> patdk-wk, so a gig would keep everything probably performing the same as it has been?
[17:56] <hallyn_> patdk-wk: yeah it's tight and sometimes slow.
[17:56] <hallyn_> but at 5W i can live with ti :)
[17:56] <patdk-wk> a gig will give you enough room with alittle slack, sure
[17:56] <hallyn_> (right now it's in a container on a 2T/16G system)
[17:56] <HiddenDjinn> well, i have 20 gig of disk, and that's more than enough
[17:56] <HiddenDjinn> hallyn_, even
[17:57] <HiddenDjinn> patdk-wk, thanks
[18:00] <HiddenDjinn> and i'm guessing at that low level of ram, running a 64 bit os doesn't make any sense
[18:01] <bekks> HiddenDjinn: Having 64bit hardware and not running 64 bit OS is just a waste of resources with no savings at the power level, etc.
[18:01] <HiddenDjinn> bekks, well, the hardware is virtual
[18:01] <bekks> HiddenDjinn: The said still applies.
[18:01] <HiddenDjinn> bekks, if there's no real advantage in memory management or the like, i see no reason to do so
[18:02] <bekks> You see no reason for not wasting resouces? Odd.
[18:02] <HiddenDjinn> bekks, i don't see it as a waste of resources unless there is a performance advantage
[18:02] <HiddenDjinn> bekks, and, iirc, the performance differences were minor below 4g
[18:03] <bekks> Using only half if the registers of the CPU without any savings - is wastingvresources.
[18:03] <HiddenDjinn> or was that below 2 gig? i forget
[18:04] <bekks> HiddenDjinn: http://developer.amd.com/community/blog/2008/03/06/myths-and-facts-about-64-bit-linux/ :)
[18:05] <hallyn_> as i recall it was on powerpc that you really saw performance benefits to 32-bit
[18:10] <sarnold> heh, I thought x86 32 bit was so register-starved that we couldn't have nice things
[20:01] <CAPTAIN_CAPSLOCK> test
[20:08] <MrSmithGer> hi
[20:10] <xibalba> how do i disable ipv6?
[20:12] <bekks> !ipv6
[20:12] <MrSmithGer> i am wondering, why do kernel updates always get released on fridays?
[20:14] <bekks> MrSmithGer: Maintainers got some time to do so at friday night ;)
[20:14] <MrSmithGer> And leave the boxes unattended over the weekend?
[20:16] <bekks> MrSmithGer: Huh?
[20:17] <MrSmithGer> bekks: i don't work at the weekend. Upgrading kernels and leaving for two days...
[20:17] <MrSmithGer> not a good idea.
[20:19] <toabctl> how does canonical create the uec images for ubuntu? is there some documentation for that?
[20:20] <bekks> MrSmithGer: No one forces you to actually update your kernels over a weekend.
[20:21] <MrSmithGer> bekks: I know. But i hate leaving unfinished business :-)
[20:22] <bekks> MrSmithGer: I dont see unfinished business at that point at all. Updates have to be tested, planned, executed, controlled, rolled back if necessary, etc. - thats not done at a friday afternoon.
[20:23] <MrSmithGer> bekks: exactly. That is why i don't do that on fridays. Why not bundle updates up and release them on monday or tuesday?
[20:24] <MrSmithGer> beeks: The last kernel update was in QA for over two weeks. A few more days won't hurt.
[20:24] <bekks> MrSmithGer: Why not releasing updates over the weekend and let people download it instead? Planning phase can begin at Monday then.
[20:25] <bekks> MrSmithGer: You should do your own QA, since you are administering your servers. Canonical/Ubuntu is not responsible for your non-testing ;)
[20:25] <MrSmithGer> bekks: Or that. The thing is, getting in important updates on fridays and then leaving for the weekend just feels bad.
[20:26] <bekks> I dont feelbad about that at all. It will take weeks to apply those updates even if they would have been released on a Monday.
[20:26] <MrSmithGer> Depends on the kind of updates
[20:27] <MrSmithGer> Wouldn't you speed up the process for remote execution fixes?
[20:27] <bekks> No. I would speed up ITIL processes controlling the application of fixes.
[20:27] <jrwren> amd64 over i686 every time, because 4 times the register file  :)
[20:28] <MrSmithGer> granted.
[20:30] <MrSmithGer> I hope someone is going develop something similar to splice so that we can apply kernel updates without rebooting.
[20:31] <MrSmithGer> bekks: What is your ITIL process for applyinf fixes?
[20:33] <bekks> MrSmithGer: Get notified about fixes, setup a test environment, test the fixes, test dependencies to other required and optional applications, approve the fix, communicate a planned downtime, apply teh fix in the downtime, test it briefly, start up applications again, communicate the end of the downtime.
[20:33] <AleksEst> what is the difference between debian7 server and ubuntu server? I just don`t get it
[20:35] <bekks> MrSmithGer: You see, applying a fix is not a job for a friday afternoon, and it basically doesnt matter if it is released on a friday :)
[20:36] <MrSmithGer> bekks: hmm i disagree. We have a similar process for fixes. And if you spent all week getting system up to date it really pisses me off when i know that monday begins with the same stuff.
[20:36] <MrSmithGer> it's a psychological thing.
[20:38] <MrSmithGer> takes away the feeling of having accomplished anything.
[20:38] <bekks> MrSmithGer: ITIL doesnt care about psychologics ;) And since we have to plan downtimes, etc., we cant just spend a week for updating every bit and byte, since we are a bit... conservative, regarding updates.
[20:41] <MrSmithGer> bekks: Being conservative and careful is ok. But you how this works with the brass. Say you receive an update notification on friday, fix the test env and the leave for the weekend and someone hacks into your live boxes on sunday the suits will blame who? the admin of course who didn't fix live.
[20:41] <MrSmithGer> bekks: ITIL processes or not.
[20:42] <bekks> MrSmithGer: We run several hundred database servers, a lot of them mainatained and monitored 24/7.
[20:42] <bekks> MrSmithGer: And noone is directly reachable :)
[20:42] <MrSmithGer> bekks: my operation is bit smaller.
[20:46] <MrSmithGer> bekks: If noone in their right mind applies fixes on friday, why release them then? Why not wait one or two days?
[20:49] <byprdct> how can I find out if the server is serving files via gzip?
[20:49] <byprdct> is there a web service I can use or an ubuntu command?
[20:49] <byprdct> terminal command
[20:50] <bekks> MrSmithGer: Why waiting when the fixes are ready, and why no allowing people to download fixed already? :)
[20:50] <bekks> MrSmithGer: Thats what weekends are for - unattended, not-monitor-worthy, large-data transfers.
[20:51] <MrSmithGer> bekks: simple. less liability for admins.
[20:52] <bekks> MrSmithGer: Yu still have to pay the admin to apply the monday-afternoon-released fix on monday night.
[20:52] <bekks> Doesnt matter for the business, actually.
[20:53] <MrSmithGer> bekks: Actually with proper load balancing, replication and such you can do that tuesday morning
[20:53] <bekks> And your admin still gets liable when someone hacks your server at monday night. Same game, different day.
[20:53] <MrSmithGer> you're right.
[20:54] <MrSmithGer> but it's only one night. less exposure :-P
[20:55] <bekks> If you are afraid of that, hire some nerday students to watch out for updates and apply them hourly :P
[20:55] <MrSmithGer> LOL
[21:01] <MrSmithGer> well. i'll be leaving.
[21:18] <ph8> ubuntu images in Microsoft Azure somehow manage to mount /dev/sdb1 to /mnt (an annoying mount indeed!). Now it's not in /etc/fstab and i've grepped /etc/init.d for it (trying to remove the auto mount) - do you have any idea where else i might look?
[21:24] <ph8> ah found it
[21:24] <ph8> there's a super secret mount config