=== locust is now known as Guest22937 [02:56] does anyone use lynx browser? do you know what the blue bar near the bottom of the page is called? [02:57] I'm trying to setup a replacement for Gmail since they're no longer doing the whole Exchange ActiveSync thing, can anyone recommend anything? [02:57] I was thinking dovecot and z-push, anyone use those? [02:57] maxped: did you try http://linux.die.net/man/1/lynx to see if hwat you are looking at is mentioned? [02:58] dovecot works well. I didn't try it with z-push. [02:59] ScottK: You able to push to iOS? [03:00] holstein: i sure did. i know what is displayed in the bar can be changed via keyboard shortcut, but i cant find it anywhere. thought maybe if i knew exactly what lynx called that bar/area i could google a bit better [03:05] billy_ran_away: No. No need to for me. [03:06] ScottK: : ) === aarcane_ is now known as aarcane [08:04] \Can anyone help with getting a instance running on kvm ontop of Ubuntu to be able to see my tape drives (instance is a Linux Net Backup server) === smb` is now known as smb [08:17] Is there anyone alive in here or are you all sleeping?? [08:19] zzzZZZzzz [08:20] zvm-vlam: Haven't played with kvm too much so sorry i'm of no help. [08:27] k [08:31] morning [08:32] *yawn* [08:32] guys I'm figuring out how I shall partition my VM's for servers. Normally I had /boot / /usr/var /tmp and /home seperated but I wonder why I should do this when my Vm already is 30GB in this case without any storage in it [08:33] YamakasY2: How is the VM 30GB with no storage? What the hell do you ahve installed? [08:34] I personally wouldn't be worried about partitioning inside of a VM. The Virtual disks are just all stored in one big file on the host [08:34] how can I assign tape drives to a VM [08:36] histo: no just make sure I can't have a lockup [08:36] histo: my usr and var are mostly 10GB [08:36] I'm figuring out if I shall make them 5 GB [08:37] YamakasY2: It's your system do what ever you want. [08:37] zvm-vlam: I'd assume it would be the same way as sharing any other devices liek cdrom etc... Tape should be no different [08:37] histo: that is kinda simple to say... is that the ubuntu way these days ? [08:39] histo: the question is if it's wise [08:40] !best | YamakasY2 [08:40] YamakasY2: Usually, there is no single "best" application to perform a given task. It's up to you to choose, depending on your preferences, features you require, and other factors. Do NOT take polls in the channel. If you insist on getting people's opinions, ask BestBot in #ubuntu-bots. [08:41] YamakasY2: I told you my opinion already which you obviously didn't want to hear. So i'm not sure why you are asking. [08:41] histo: I'm not sure why you answering ? [08:41] *are [08:41] You're worried about non existent lockups [08:41] Don't run a VM then. [08:42] Run bare metal with backup or load balancing [08:42] histo: huh ? it doesn't have anything todo with a VM... but you want to keep Vm's as small as possible [08:46] YamakasY2: What lockup are you worried about? I asked why you are creating all these separate partitions and you said you are worried about lockups. [09:05] histo: filling up log, some application that decided to store in /usr, etc [09:07] YamakasY2: So how is putting it on a separate partition going to help solve that issue? [09:15] histo: it will not lock up the rest [09:15] That application will crash correct? [09:15] histo: you never had a failing mailserver because /var/log was full and your queue was not able to spool ? [09:15] Yes but I wouldn't use a 5GB partition for spool [09:16] You are the one doing that. [09:16] You're creating your own problems. IMO [09:17] histo: no I'm thinking about what happens when you do or don't [09:22] So what happens when your 5GB spool partition fills? [09:22] Same result [09:23] Use quota and other means to fix the issue. Don't put a hardware limitation on the problem. You are putting a bandaid on open heart surgery. [09:28] histo: yeah agree, but what would you do for server partition ? [09:31] I wonder, one / partition, a /tmp /boot and swap only ? [09:46] yolanda, I saw the ceilometer failure as well; fixed up the backports in the lab so is OK now [09:46] ok [09:47] yolanda, I also updated pecan on raring as the package we had fails to install [09:47] bug in the source code [09:47] you are always faster than me :) [09:47] yolanda, I updated the list to include what I did so we can cover it all for the CA [09:47] was it giving a bug? i was normally using that pecan package to install it manually [09:49] hi i have folder ruben23 @ /home and ruben23 also @ /var/www, now the folder on /var/www have new chnages and files added taht needs to be retain, how do i merge this folder into one that the new files added will not be lost any of this server, all will be retained and the same exsiting also [09:49] any idea guys..? [09:50] merging that intelligently will not omit the new files added and also retain teh duplicate === acidflash_ is now known as acidflash === fire is now known as megha [11:39] Hello guys, has anyone had any experience troubleshooting an AWStats installation? [12:24] rbasak, hows mongodb looking? [12:24] jamespage: I had an idea, but it took hours to build and then it didn't work :-/ [12:24] What kind of timescale are you hoping to get this fixed in? I would like to keep digging but I don't have an ETA now. [12:25] rbasak, please keep digging; raring would be nice :-) [12:25] It seems to me that the compiler should dtrt if it can be told that this particular operation could need manual alignment [12:26] And it also looks like it is already being told how to do this. It was in a way that I thought wasn't technically correct ("packed" isn't exactly the same thing as "aligned"), but telling it "aligned(1)" didn't seem to help [12:26] I want to reduce this to a test case and then experiment with options and look at the assembly generated [12:26] rbasak, sounds like a good approach [12:26] working with mongodb on arm is slow... [12:26] Indeed! [12:27] Also "nocheck parallel=4" didn't' work. I've patched debian/rules and submitted a bug [12:27] (to Debian) [12:35] yolanda, hmm - the mongodb sync in the ceilometer packages is not working well for me in the charms [12:37] yolanda, its a packaging issue rather than a charm issue; ceilometer-common does the db sync [12:37] but does not have a depends on MongoDB [12:37] maybe we need to re-think that a bit [12:37] I also think that MongoDB should be a suggests [12:38] so that I can deploy mongodb elsewhere [12:38] just like we do with nova [12:38] but then, we cannot do the sync in the package? [12:39] yolanda, hmm [12:42] yolanda, OK - so I think we have to take the sync out of the packaging - not ideal [12:42] yolanda, also the way the configuration file is being manipulated in the postinst script is not aligned to policy === acidflash__ is now known as acidflash [12:43] as it does not take into account fully whether the ceilometer configuration file has actually been changed [12:43] doesn't ceilometer have the ability to run on postgresql too ? [12:43] * jamespage looks [12:44] http://doughellmann.com/2012/10/grizzly-sessions-on-metering-openstack-with-ceilomter.html [12:44] lifeless, indeed it does - and mysql as well [12:44] and sqlite [12:44] yolanda, in which case the MongoDB stuff has to come out [12:45] its completely not safe [12:45] they have support for several databases [12:45] so i remove the mongo recommends, and the db sync? [12:45] user should have to do it manually?= [12:45] yolanda, all of the other openstack packages default to using sqlite [12:46] yes, but due to the nature of data it's better to have mongo [12:46] yolanda, yes - but that is a charm thing, not a packaging thing [12:46] yolanda, sqlite is a 'install and it works option', not a 'install and get huge scale' option [12:47] yolanda, sqlite is not the 'best' option for any of the openstack packages, but its a sane default. [12:47] so we have to depend on sqlite and do the sync using sqlite? [12:48] yolanda, broadley yes [12:48] yolanda, the sync should not happen in ceilometer-common either [12:49] no? where should it be? [12:49] yolanda, it should happen in the agent package that writes to it [12:49] ok [12:49] and then, in the charm, we should do the sync for mongo? [12:49] yolanda, actually ignore that comment [12:50] which one? [12:50] the bit about not doing it in ceilometer-common [12:50] yolanda, please use cinder-common.postinst as a reference [12:51] yolanda, it broadley does the right things [12:51] ok [12:51] yolanda, the charm should reconfigure ceilometer when mongodb is related to it and then do a db-sync [12:51] ok, i can work on it [12:52] yolanda, great - ping me when you have a MP - I don't always notice email [12:52] lifeless, thanks for expanding my knowledge of ceilometer [12:52] * jamespage is running to catchup [12:55] i need a while to finish something with the panels [13:13] I have been spending 2 whole days now trying to get 12.04 to run on a server here ... one of three identical servers ... I performed an install on one of the three a couple weeks ago with no issues ... but now with the second one I am installing for the 5th time now [13:15] basically no matter how I partition the RAID1 device, after the Ubuntu install completes successfully, the system reboots and then leaves me at "error: no such device" and grub rescue prompt [13:15] this is a basic install that worked perfect of one server but is not happening on this one!! [13:18] hi there [13:18] hey, i have a problem with latest ubuntu server edition, if i leave it on its own the screen goes black and it stops responding [13:18] i have eht0 defined as dhcp and eth0:0 as static. whenever the dhcp client (isc-dhcp-client) renews its configuration it removes/ brings down eth0:0. How can I fix this? [13:18] Abu: energy saving enabled? :) [13:19] well having hard time figuring that out :D [13:20] on default does the ubuntu come with ready configuration? [13:21] Abu: sry I don't know... :( [13:23] its annoying problem [13:24] it locks up in hour or something [13:24] but now i have stuff on keyboard so that it presses a button all the time and it has been succesfully on for 6 hours [13:28] smoser: What is the current best practice for building customised Ubuntu OpenStack images? [13:30] soren, my usual suggestion is http://cloud-images.ubuntu.com [13:31] ie, i wouldn't recommend it any more than i'd recommend you build your own kernel or glibc. let your distro do it. [13:31] smoser: I fire up a bunch of instances. Their average lifetime is 8 minutes. [13:31] smoser: 5 of those are spent on installing the exact same stuff. Every time. [13:31] smoser: I'm pretty tired of that extra wait. [13:32] well, one thing i've recently been doing which makes that particular pain less painful. [13:33] is 'fast_apt' at http://paste.ubuntu.com/5607702/ [13:33] smoser: Hm. [13:34] but that wont help your network pipe [13:34] smoser: Yeah, I guess that might have an impact. [13:34] smoser: I don't actually think I'm spending that much time on actual downloads. [13:35] It's quite possible that eatmydata would make a difference. [13:35] its a huge impact. [13:35] This is a recent addition to cloud-init, yeah? [13:35] but if you are interested in building your own images, i'd at least suggest that you start with ours. [13:35] Right, that was certainly the intent. [13:35] well, thats just a bootcommand that dpkg-dirverts [13:35] Sure. [13:35] so its not really cloud-init itself. i just feed that as user-ata [13:36] Oh, I see what you're saying. [13:37] then, i'd recommend either mounting and chrooting (mount-callback-umount http://bazaar.launchpad.net/~smoser/+junk/backdoor-image/files makes that part easy) [13:37] or, since that is sometimes finicky with grub (which i'd love for you to just figure out the issue and get it fixed "right") [13:38] the other way to do it is just to boot a cloud image in kvm 'cloud-localds' and given user-data that preps it the way you want. [13:38] What I wonder is what sort of cleanup I ought to do after doing that? [13:39] well, you can do 'rm -Rf /var/lib/cloud /var/log/cloud-init*' [13:39] I could probably work it out, but if you have a list already, I'd love to have it. [13:39] and 'dd if=/dev/zero of=/tmp/zero.img ; rm -f /tmp/zero.img' [13:39] but realistically, cloud-init *should* re-run everything that is needed "per-instance" [13:40] and when you boot it for real, it will have a new instance-id. [13:40] I'll try the eatmydata thing first. That's a great suggestion. [13:40] i think you will be very surprised with how much time storing data takes. [13:43] soren, if you have a better suggestion for the zero image thing, i'd love to know. i know that there should be more high-tech route for that. [13:44] smoser: There is one for ext2 images. [13:45] smoser: I'm trying to recall its name. [13:45] yeah. i think that actually with ext4 and kvm it should be able to poke the 'trim' requests down through virtio into the qcow image [13:45] i'm not sure though. [13:45] but at some point. [13:45] i think that "should work" [13:45] smoser: zerofree [13:46] smoser: That's not the one I was thinking of, but it looks to be the same thing. [13:46] I didn't know about zerofree. Thanks soren! I wonder if there's some kind of convergence with fstrim possible here? [13:47] http://intgat.tigress.co.uk/rmy/uml/index.html says "The utility also works on ext3 or ext4 filesystems. " [13:48] (fstrim works online) [13:49] rbasak, but does it correctly poke through to qcow2 disk format? [13:49] smoser: I doubt it. So I was thinking of some hack. But what would be really nice if it did pass through, and qemu could be configured to dtrt for thinly provisioned stuff and optionally zero out for directly mapped stuff. [13:50] rbasak, there has been work done in that area [13:50] Though for cloud image preparation perhaps offline operation would actually be better. More consistent for delivery. [13:50] i know i've read of that before. [13:51] rbasak, http://comments.gmane.org/gmane.comp.emulators.kvm.devel/66551 [13:51] at least a reference. [13:53] http://www.outflux.net/blog/archives/2012/02/15/discard-hole-punching-and-trim/ [13:55] As an aside, while we're talking about this, there's an argument here that says that cronned fstrim is preferable over a discard mount option for SSDs: http://en.opensuse.org/SDB:SSD_discard_%28trim%29_support [13:57] I finally got Ubuntu Server 12.04 booting up on a server here ... but now when I reboot the server it just hangs with a black screen ... I have to power it off with the power button and start it up again ... [13:57] how do I troubleshoot this? [13:58] at what point does it hang with a black screen? [13:58] after grub? before post? [13:58] during os boot? [13:58] when it's hung, are you sure it's not the console screen saver? [13:58] not sure actually because I reboot it remotely ... [13:59] when it's hung, is it pingable? [13:59] I will reboot it directly and see where it gets to ... [13:59] after you reboot it, what do the logs say? [13:59] that's how you troubleshoot it... you identify the point of failure before you attempt to identify the causative factors [14:00] but in syslog this is the last two lines before I manually turn the server off: [14:00] Mar 12 14:51:06 saturn kernel: Kernel logging (proc) stopped. [14:00] Mar 12 14:51:06 saturn rsyslogd: [origin software="rsyslogd" swVersion="5.8.6" x-pid="1411" x-info="http://www.rsyslog.com"] exiting on signal 15. [14:00] which logs should I be looking at? [14:01] syslog to start [14:01] hi, how can we install Sun JVM >=1.5 on a Ubuntu server running 12.10? [14:01] well this is what syslog has in it after the last powerup: http://paste.ubuntu.com/5607757/ [14:02] zetheroo: looks fine, then [14:02] or was that the good boot? [14:03] sw: google shows tons of results for oracle java 6/7 on ubuntu 12.10 [14:05] I have no knowledge of Java, so Sun JVM 1.5 = Oracle Java 6/7? [14:05] java 1.5 == java 5 [14:05] java 1.6 == java 6 [14:05] silly, but get used to silly with java [14:05] so, > 1.5 is java 6 or 7 [14:06] oracle bought sun in 2010, so oracle ~= sun [14:06] tedski: that was the output from the time that I manually powered the server on [14:07] zetheroo: now get it to hang and see if it's pingable while it's hung [14:07] zetheroo: then see if you can access any of the services that should startup (i.e. ssh) [14:07] tedski: so I did a reboot again and the last line I see on the screen is "Will now restart" ... then it goes to a black screen and hangs there ... [14:08] the system is then unresponsive to eveything [14:08] and it never resets ... [14:09] zetheroo: oh, then you have an acpi problem [14:09] so I have to do a hard-reset [14:09] acpi ... like in the BIOS or in Ubuntu? [14:09] well... that depends :) [14:10] is there a reboot= parameter in your grub kernel line? [14:10] cat /proc/cmdline if the system is up [14:10] BOOT_IMAGE=/boot/vmlinuz-3.2.0-23-generic root=UUID=a47c5e6b-583b-4bc3-b97e-37c8fee11df5 ro [14:11] what type of system is this? [14:11] just a rack server [14:11] Intel ... [14:12] jamespage: https://code.launchpad.net/~yolanda.robla/ceilometer/grizzly/+merge/152916 [14:12] the brand is some SuperMicro thing ... [14:12] try adding reboot=a,b,c to your kernel line [14:12] how do I do that?\ [14:12] default is reboot=kbd which means it will first try to reboot via keyboard, then bios, then acpi [14:12] so, a,b,c says try acpi, then bios, then cold reboot [14:13] zetheroo: when it's booting, interrupt the timer in the grub menu and follow the onscreen instructions for editing [14:13] ah like that .. ok will give it a shot .. [14:13] but this is not normal no!? [14:14] it's normal for some systems [14:14] I mean I have never had to do this before and we have 3 of these identical systems ... and one of them I setup with Ubuntu 12.04 a few weeks ago and it had no issues [14:14] oh [14:14] well, then no [14:14] it's not normal [14:15] and something is different in the reboot method between the working and non working system [14:15] but then nothing about this particular server has been very "normal" [14:15] go to your known good system and make sure all acpi settings are identical to this system [14:15] trouble is that other server is busy running several production VM's :) [14:16] so I cannot stop it to check the BIOS settings [14:16] you can still cat /proc/cmdline [14:16] ok [14:16] and check sysctl [14:16] BOOT_IMAGE=/boot/vmlinuz-3.2.0-37-generic root=UUID=34975010-34e2-477e-8b96-fca973948c74 ro [14:16] looks the same [14:17] check: comand not found [14:17] heh [14:17] no, don't run "check sysctl" :) [14:17] oh sorry .. :P [14:17] sysctl can show you all of the acpi settings [14:17] man sysctl [14:18] I don't see how this could have anything to do with Ubuntu actually ... but rather something from the BIOS [14:19] sure [14:19] i'm not saying that it is one or the other [14:19] i'm telling you how to troubleshoot :) [14:19] that was, afterall your question [14:19] there isn't a way to get BIOS info without rebooting the server is there!? [14:19] yes thanks :D [14:19] maybe I asked the wrong question ... LOL [14:23] ahh there is a difference in the BIOS versions of the two server! [14:23] yolanda, comments in MP [14:24] ok === zurich is now known as level15 === ogra_` is now known as ogra_ [14:28] ok, got it, better to do a patch to modify the default conf [14:38] jamespage, should i set synchronous mode for sqlite? or i leave that by default? [14:52] how do I change the number of seconds that grub counts down to before booting Ubuntu ? [14:53] zetheroo, /etc/default/grub [14:53] yolanda, leave everything else as defaults [15:11] hi everybody. having problems with slow performance of kvm on ubuntu 12.10. i suspect that hardware acceleration is not working. kvm-ok says that KVM can be used. after launching the virtual machine, the log file in /var/log/libvirt/VIRTNAME.log shows LC_ALL=C PATH=/usr/local/sbin:/usr/local/bin:/usr/bin:/usr/sbin:/sbin:/bin QEMU_AUDIO_DRV=none /usr/bin/qemu-system-x86_64 -name wheezy -S -M pc-1.2 -no-kvm -m 1024 -smp 1,s [15:11] when I do apt-get update I get this: http://paste.ubuntu.com/5607963/ [15:11] any idea why? [15:12] zetheroo: You have a DNS problem. [15:12] the machine is connected to the Internet [15:12] jamespage, recheck [15:12] ic [15:14] yes, DNS was the issue - thanks jpds [15:14] I still have weird issues with rebooting ... [15:15] yolanda, OK - lgtm [15:16] system now reboots and gets up to the GRUB menu ... waits 10 seconds ... and then seems like its going to boot the first entry in the menu ... screen goes black (but still on) and that's about as far as it gets [15:16] great [15:22] tedski: can you help what JDK or JRE is? in the INSTALL I'm looking at it says the requirement is JVM but I don't see this anywhere [15:23] sw: JDK = java development kit; JRE = java runtime environment [15:23] sw: if you're not going to develop java programs, then install the jre [15:24] the jre has the jvm plus class libraries and some other supporting stuff [15:24] jre is for running java programs, jdk is for developing java programs [15:32] tedski: ah ok :-) [15:33] hey guys is there a way i can tweak amavis a bit to provide some better memory usage? [15:46] amavis doesn't use much memory [15:47] something like 5-10megs [15:48] in my case, each amavis thread is using 60megs of ram, plus 80megs shared [15:49] when creating a RAID1 setup should the swap be in the RAID or not? [15:49] and all of that is due to spamassassin rules [15:49] zetheroo, up to you [15:49] if a disk dies, do you want your system to crash? [15:49] patdk-wk: could it be problematic ? [15:49] zetheroo: would you like your machine to keep going, or die because something was in swap and the swap disk failed? [15:50] Hi, I'm new to ldap and trying to setup dovecot to use ldap, this is kind of a dumb question, but I've never really read what it means to bind to an LDAP server... [15:50] I have some idea of what I think it means… but I don't definitively know what it means. [15:50] I think it means... [15:50] so having swap in the RAID is not a bad thing ... [15:51] zetheroo: there is no point in RAID1 at all if swap is not part of it. With RAID1 you want your machine to continue to function if half of your disks fail, hence swap must be treated same as your rootfs. [15:51] For an authentication process to attach itself to a specific part of the LDAP schema to be able to search from the point down... [15:52] hello [15:53] xnox: true, but there is a second use case for RAID1. If you don't mind the machine failing, but do want to keep your data. === dumfries_ is now known as dumfries [15:53] xnox: ok good point :P [15:54] Though even in that cause I usually RAID1 my swap. There's no real cost to doing it. [15:54] that case [15:55] so I have been fighting with a server only to find out that the isp comcast is blocking port 25. So is there a way that I can enable the server to maybe get around the block? [15:55] rbasak: hence I said "same as rootfs" if the rootfs is not on RAID1 (as it can be trvially rebootstrapped), and only data partitions are on RAID1, having swap on raid1 improves things but not by much ;-) [15:55] lwizardl, upgrade to comcast business class [15:55] I have been struggling to get Ubuntu Server working correctly on a server which is using this UEFI stuff ... [15:56] should the EFI partition be in the RAID1 area or not? [15:56] patdk-lap, yeah I plan to do that but i'm trying to get the server working for now [15:57] yes, and that is the *only* solution [15:57] crap [15:57] unless your solutions involves not using comcast [15:57] then you can setup a vpn from somewhere that allows port 25 [15:58] I would love that option. but for my sucky town it is either comcast or AT&T uverse [15:58] well, there is a reason why, residential, says, no servers [15:58] yeah [16:03] is there anything wrong with rebooting using init 6 ? [16:06] soren, did you try the eatmydata? === matsubara is now known as matsubara-lunch [16:47] plars: so this is perplexing. Your i386 lxc utah run says mounted cgroups weren't found when running lxc-execute. [16:48] hallyn: looks like the amd64 one finished too: https://jenkins.qa.ubuntu.com/view/Raring/view/Smoke%20Testing/job/raring-server-amd64-smoke-lxc/119/ [16:49] hallyn: also showing cgroups not mounted [16:51] where do i see the actual test log then? [16:52] anyway that might be a utah error then. cgroup-lite is being installed, cgroups definately ought to be mounted. [16:53] plars: I run locally using: [16:53] sudo -u utah -i run_utah_tests.py -p lp:ubuntu-test-cases/server/preseeds/lxc.preseed -i /home/serge/raring-server-amd64.iso lp:ubuntu-test-cases/server/runlists/lxc.run [16:53] is that all right? suppose i should update my raring-server-amd64.iso... maybe there's an error int he new one [16:53] * hallyn updates [16:58] hallyn: we seem to be running run_utah_tests.py -p lp:ubuntu-test-cases/server/preseeds/lxc.preseed -i /var/cache/utah/iso/raring-server-amd64.iso lp:ubuntu-test-cases/server/runlists/lxc.run -x /etc/utah/bridged-network-vm.xml [16:58] but I think the networking config is not necessary for you [16:59] right i can't use that :) [16:59] plars: almost have the new iso, i'll start a utah run and log in and see if everything's kosher [16:59] ok [17:20] plars: by gollie, thsi seems to be a cgroup-lite bug [18:06] hello to every one I'm installing firehol and when I execute the command "frirehol try" I get a lot of errors like this: "iptables: Memory allocation problem." I where lokking at top command while I where doing and is a lot of memory free in the system [18:06] any ideas? [18:07] you excute "firehol try" i assume [18:07] lisandrop05: you can also pastebin you exact errors [18:07] any particular reason you don't use ufw? [18:07] !paste [18:07] For posting multi-line texts into the channel, please use http://paste.ubuntu.com | To post !screenshots use http://imagebin.org/?page=add | !pastebinit to paste directly from command line | Make sure you give us the URL for your paste - see also the channel topic. === maxb_ is now known as maxb [18:11] yes holstein [18:11] i execute "firehol try" [18:13] lisandrop05: the kernel cannot allocate memory for itself in the same way that applications can allocate memory. on 32 bit systems, the kernel actually has very tight memory requirements. over the run time of a system, memory gets more and more fragmented, making it harder to allocate large blocks of kernel memory. [18:13] lisandrop05: check dmesg; if you see something like "order 1 allocation failures" then your best bet is reboot. [18:13] plars: and now i'm seeing it on new raring cloud instances too! [18:13] lisandrop05: trying again - why don't you use ufw? [18:14] hallyn: well at least it's reproducible now I guess :) [18:14] seems firehol hasn't been much updated the last five years [18:15] Afternoon I was refered to this channel [18:15] RoyK: what's the difference [18:15] lisandrop05: dunno firehol [18:15] !ufw | lisandrop05 [18:15] lisandrop05: Ubuntu, like any other Linux distribution, has firewall capabilities built-in. The firewall is managed using the 'ufw' command - see https://help.ubuntu.com/community/UFW | An alternative to ufw is the 'iptables' command - See https://help.ubuntu.com/community/IptablesHowTo | GUI frontends such as Gufw (GNOME) and Guarddog (KDE from Lucid onwards) also exist. [18:15] I currently have Ubuntu 8.04LTS Server running a older kernel 2.6.24-21-server. I am trying to install the Kernel sources but it appears they have been deleted? I am getting Package linux-headers-2.6.24-21-server is not available [18:16] okok, I'm installing ufw [18:16] Testtube302: you better upgrade soon - 8.04 only have a month left of support [18:16] The only reason i need the Kernel source is because i am attempting to install a backup agent that needs the source to compile a module [18:17] It almost looks like the linux-header files for the kernel i am using has been wiped off the planet :/ [18:18] Testtube302: looks like you haven't updated that machine too much - the current version for 8.04 is 2.6.24-32 [18:18] Royk Thats correct this machine is in production use and pretty isolated [18:19] there are archives around for that sort of use, but the *supported* method is to upgrade to the last stable version [18:19] Any idea of where i can get the linux header files for my kernel [18:19] i have searched everywhere. [18:20] and as mentioned, 8.04 won't be supported for much longer [18:20] what sort of backup system is this? [18:20] The backup agent is r1soft [18:20] also, is there a good reason for not upgrading to the latest 8.04? === matsubara-lunch is now known as matsubara [18:21] RoyK as mentioned this is a production system and there is no room for downtime right now. We have some time slotted later this year to upgrade to 10.04 like the rest of our cluster [18:21] ok [18:22] it's already 2013, would it make more sense to upgade to 12.04 LTS instead? [18:22] hey now i got error message [18:23] mmc0 got data interrupt even though no data operation was in progress [18:23] Testtube302: http://packages.ubuntu.com/source/hardy-updates/linux-ubuntu-modules-2.6.24 [18:23] how do i fix that [18:23] Testtube302: just wondering a bit why a backup agent needs access to the kernel source [18:23] ufw only 1 minutes and I get my first error: ERROR: problem running ufw-init [18:23] lisandrop05: pastebin command and output [18:25] ERROR: Could not load logging rules [18:25] it's short [18:25] command is shoet too [18:25] and the command was? [18:25] short* [18:25] I dont believe that URL you provided actually has the Linux headers for my kernel [18:25] are you root? [18:25] ufw enable [18:25] and yes I add port 22 first [18:26] if I put it again it skip existing rule [18:26] lisandrop05: please *pastebin* the whole range of commands and output. and btw, which distro version is this? [18:27] RoyK: lisandrop05 is from cuba, probably the little iron curtain forbids getting to a pastebin site. I'm stunned lisandrop05 can get here... [18:27] oh [18:28] is Ubuntu precise [18:28] lisandrop05: sorry if you can't get to those services. first of all, are you root? can you do "ufw add 22/tcp" without issues? if so, the "ufw enable" command should work well [18:28] http://pastebin.com/Jtr9tm3J [18:28] ha :) color me wrong :) [18:28] Royk http://packages.ubuntu.com/hardy-updates/linux-headers-lum-2.6.24-21-server Says Package not available under depends [18:28] RoyK: yes, I'm root [18:29] I install the ufw [18:29] in the same shell [18:29] looks like there may be an issue on trying to add ipv6 rules on an ipv4-only system [18:29] https://bbs.archlinux.org/viewtopic.php?id=141008 [18:29] not sure, though [18:30] lisandrop05: try setting IPV6=no in /etc/ufw/ufw.conf [18:31] RoyK: hrm, I thought that one had been fixed. [18:31] * RoyK has no idea [18:32] lisandrop05: also, if sarnold is right - it might have been fixed - have you upgraded lately? [18:32] no I don't [18:33] try to disable ipv6, then [18:33] Royk what exactly is Source Package: linux-ubuntu-modules-2.6.24 [18:33] not sure [18:33] It work ok, I add IPV6=no to ufw.conf now, new question: where can I add an IP blacklist from a file [18:33] your not sure? [18:34] Testtube302: I think what you need is 'apt-get install linux-source' [18:34] no need to update, And i check and my system is updated [18:34] ok [18:35] lisandrop05: I normally use 'apt-get update && apt-get -y dist-upgrade && apt-get -y autoremove' - there's usually some new packages around [18:42] anyone have any experience working with Kamailio and kamailio-ims-modules? [18:44] hi, I'm using ubuntu server 12.04 for cluster computing machines (no services, just cli). Is it safe to remove dbus-daemons and all the other desktop-related stuff? especially does upstart still works fine without dbus? This appears to be the case but I'd like to get some more input before breaking things [18:45] Testtube302: still curious about why a backup solution needs to make kernel drivers - could you enighten me on this? [18:50] hi guys! Ive got a problem. When I run a script with crontab does not work but when I run it, it works. Why? [18:52] Diegonat: quite often that is due to directories in your PATH that are set for an interactive login but not set when executed via cron [18:52] sarnold, good shot [18:52] RoyK: how do I allow all outgoing traffic in ufw?? [18:52] basically when it fails [18:53] it when it tries to excute one python script in a for loop [18:53] in firehol is simple: client accept any, but in ufw I don't know [18:55] lisandrop05: $ sudo ufw status verbose [18:55] Default: deny (incoming), allow (outgoing) [18:55] that's the default [18:55] lisandrop05: it allows all outgoing [18:55] plars: fix for that breakage is uploaded to raring. [18:55] hallyn: you are awesome! [18:55] psivaa: ^^ should be fixed on tomorrows run [18:56] lisandrop05: as for ipsets, that isn't supported in ufw yet, but it is planned [18:59] plars: nm. there's an error in how i did it. grrr. [19:00] RoyK: if I disable the firewall ufw all my old rules disappear? [19:01] stgraber: grrr. my fix was bad. changing locale and then fixing it for real. (you threw me by calling the mount 'cgroup', i need to detect that moutn differently :) [19:01] biab [19:01] lisandrop05: ufw replaces iptables [19:01] I disable and enable ufw [19:01] lisandrop05: you need to configure ufw to use existing rules if you want to use it [19:01] and it block me out [19:02] my ssh http ftp all is not working [19:02] hallyn: mountpoint -q /sys/fs/cgroup? [19:02] lisandrop05: disabling it should flush all iptables rules - enabling it should enable ufw rules [19:02] I'm locked out of my server [19:02] I'll need to call support assistence [19:03] I'm on VPS [19:03] and ssh is all access I have [19:04] lisandrop05: VPS sometimes don't have everything that is needed in their kernels. if ufw is not working for you, try running: sudo /usr/share/ufw/check-requirements [19:05] jdstrand: that's awesome :D [19:05] jdstrand: I can't I'm blocked out of my server [19:06] lisandrop05: do you have kvm access? [19:08] RoyK: I'm from cuba a lot of ports are blocked here [19:08] including all cpannel ports [19:09] yeah, someone told me (above) that you were from cuba [19:10] lisandrop05: see pm [19:11] what is pm? [19:11] stgraber: well phooi. never heard of that [19:11] lisandrop05: private message [19:12] lisandrop05: try "/msg RoyK hi" [19:17] smoser: I fired it off, but didn't have time to stick around to see the results. I'm checking now. [19:18] stgraber: but that doesn't answer my problem actually - i need to know if anything under /sys/fs/cgroup is mounted [19:18] smoser: Well, it cut off ~75 seconds. [19:18] sarnold: heh, thanks :) [19:18] smoser: So significant for sure, but not enough to make me not want to build custome images. [19:18] out of ~ 300 ? [19:18] custome. [19:18] custom. [19:18] Shite. [19:18] right. [19:18] Err.. Yeah, something like that. [19:19] i guess just 'mount -t cgroup' actually works [19:20] hallyn: ah, I'd just have changed the current for loop to do: [19:21] [ ! -d /sys/fs/cgroup/$d ] && mkdir /sys/fs/cgroup/$d (or use -p) [19:21] mountpoint -q /sys/fs/cgroup/$d || mount -n -t cgroup ... [19:21] stgraber: no, not that bit [19:22] stgraber: if any actual cgroups have already been mounted, then i dno't want to run [19:22] but i'm just doing 'if [ -n "$(mount -t cgroup)" ]; then exit 1; fi" [19:22] hallyn: ah, well, good luck detecting that... because logind is now mounting an empty cgroup [19:22] so on systems with logind, you'll get: [19:22] systemd on /sys/fs/cgroup/systemd type cgroup (rw,noexec,nosuid,nodev,none,name=systemd) [19:23] which shouldn't prevent cgroup-lite from doing its job [19:23] why not? [19:23] systemd wants to do cgroups its won way... let it [19:23] /swon/own/ [19:23] gah [19:23] no [19:23] logind will ONLY mount the systemd cgroup [19:24] on a system with logind, you still want cgroup-lite to mount all the others [19:24] sigh [19:24] (talking about logind here, not systemd. It's just a bit confusing because the logind cgroup is called 'systemd') [19:24] there are people out there with funky pre-existing setups who will want cgroup-lite to not bethe one mounting cgroups [19:25] i.e., they may simp0ly not want some cgroups mounted at all (for speed reasons, whatever) [19:25] i suppose you can say so long as they are writing their own upstart job, [19:25] they can set cgroup-lite tomanual, [19:25] long as you'll back me up on that :) [19:26] well, we don't hard-depend on cgroup-lite IIRC, just recommend it [19:27] oh, you do Depend on it from libvirt... maybe you should change that to Recommend then? [19:27] anyway, I think it's fair to say that if you install cgroup-lite (directly or indirectly), you want it to do its job. If not, then remove it or turn it off (with an upstart override file) [19:28] all right i'll do it that way - the old safeguards were really for people who still had cgconfig.conf installed [19:28] so I guess we want cgroup-lite to always try to mount any cgroup that's not already mounted somewhere (so on mount failure, silence the error and remove the directory) [19:30] eh ? like 'mount -t cgroup -o $d /sys/fs/cgroup/$d || { rmdir /sys/fs/cgroup/$d } ? [19:30] i'm afraid that'll make local configuration errors look like package errors in lxc/libvirt... [19:31] do you think it's worth it? [19:32] i'd say at taht point, we're better off having cgroup-lite record that it failed, [19:32] and user can figure it out frmo there [19:34] let's see how this one goes [19:44] stgraber: do http://paste.ubuntu.com/5608710 and http://paste.ubuntu.com/5608711 look ok? [19:44] (tested ok on an instance, but that doesn't mean im' not doing something stupid) [19:50] whelp, this one seems to work, will wait a few mins for a NO YOU FOOL then push [19:51] hallyn: mountpoint -q /sys/fs/cgroup/$d || (mount -n -t cgroup -o $d cgroup /sys/fs/cgroup/$d || rmdir /sys/fs/cgroup/$d) [19:52] hallyn: so that if it's already mounted somewhere else we don't end up with an empty dir [19:52] hallyn: mountpoint -q /sys/fs/cgroup/$d || (mount -n -t cgroup -o $d cgroup /sys/fs/cgroup/$d || rmdir /sys/fs/cgroup/$d || true) [19:52] hallyn: actually, just to be safe ;) [19:53] you were supposed to say NO YOU FOOL [19:54] will do, thx :) [20:05] stgraber: ok so that leaves two potential issues (i.e. not sure if we care) [20:06] oh, no,j ust one [20:06] if someone decided to mount an empty tmpfs (for instance) under /sys/fs/cgroup/devices, cgroup-lite wont' mount it at start, but it will umjount it at stop [20:06] i think that's ok... [20:07] ah yeah, but someone would be pretty stupid to do that, so that's their own fault ;) [20:08] all right - pushing [21:10] smoser: I fired up a VM, logged in interactively, apt-get install'ed the relevant packages, ran "nova image-create" and am now using that snapshot. Seems to work like a charm. [21:11] smoser: I also added your eatmydata thing to apt-get as well as added it to the sbuild inside the VM. I don't have total numbers yet, but the difference for each step looks really good. Thanks for remininding me of eatmydata. [21:13] hi, im attempting to vnc to 12.04.1 server from a mac and keep getting a black screen. any ideas? [21:13] vnc to server? [21:13] i would just ssh over to it [21:14] yes, but i have a need for vnc....long story :) [21:45] I'm looking for help with Apache rewrite. System doesn't recognize .htaccess. I've read countless posts and tried all the proposed solutions. One clue is that my apache2.conf file won't accept "AllowOverride All", just errors with "AllowOverride not allowed here. Action 'configtest' failed. [21:47] News2Me: that means you put the 'AllowOverride' in the wrong location in the config file [21:48] News2Me: see http://httpd.apache.org/docs/2.2/mod/core.html#allowoverride [21:48] It has to be embedded in a tag? [21:48] "Only available in sections" [21:48] yes [21:48] So even if you don't have virutal sites... just use it witout and conditions? [21:53] Looking at the page link... THANKS Sarnold!!!... been banging my head for days on this. [21:53] News2Me: that sounds about right, you've got the real apache experience now :) === fire is now known as baba [22:54] helllo [22:54] if I am going to have some KVM managed VMs running in my desktop, which ubuntu server version should I use? [22:54] is there a jeos or similar or shall I just use -server? [22:55] this is to test varnish and additional tools [22:57] hubuntu: 12.04LTS would be a good start [22:57] hubuntu: -server is probably fine, but do note: http://cloud-images.ubuntu.com/ [22:57] !minimal [22:57] The Minimal CD image is very small in size, and it downloads most packages from the Internet during installation, allowing you to select only those you want (the installer is like the one on the !Alternate CD). See https://help.ubuntu.com/community/Installation/MinimalCD [22:58] there we go, minimal 12.04 is certainly what I was looking for :) [22:58] oh good :) three different answers :) haha [22:58] thanks holstein [22:58] RoyK and sarnold [22:58] :) [22:58] :) === baggar11_ is now known as baggar11 [23:02] :) [23:05] Sarnold: IT WORKS! ok, so there's a file for virtual servers that lives at: /etc/apache2/sites-available/default. and that one was set to AllowOverride None. It's hard to follow different flavors of Linux.. but thanks Sarnold! [23:06] News2Me: ah, yes, I know what you mean. :)