[00:20] Hi everyone, I'm trying to install 12.04.2 LTS, however after the install finishes configuring DHCP, the install gui closes... Are the servers down and it doesn't tell me or is it something else? Thanks for the ehlp. [00:28] gui? [00:33] well the install screen [00:36] Rayfloyd: are there error messages that you can find anywhere? try looking through all the virtual consoles.. [00:37] I had left the pc open since it disappeared, just checked back and it's back [00:37] odd === zkxs_ is now known as zkxs === morph3k is now known as morph [02:29] what do you guys recommend to add an opensource application that helps me share documents with my staff as well as assign them task (online interface) [02:38] Ok so lets say i'm staring to build out prototype devices [02:39] and I want to automate configuration and setup ... includes sshing in, downloading files via github, specificiying branches [02:40] would this be a better use of puppet or just a shell script [02:40] I guess depends on the scale ? [03:51] 13.04 64x server had powerloss & memory damage, hangs on pci_bus resource mem x28-xfcffffffffff; any ideas? [03:52] i removed the damaged memory as well === Ursinha is now known as Ursinha-afk === Ursinha-afk is now known as Ursinha [05:19] ok...i have a pci_root PNP0A03:00: fail to add MMConfig information and then it hangs on pci_bus resource mem === jtv1 is now known as jtv [05:49] hi === highvolt1ge is now known as highvoltage === smb` is now known as smb === Ursinha is now known as Ursinha-afk === maxb_ is now known as maxb [08:02] can someon eplease help me with my body [08:02] ive been taking vicodin for 1 56 year [08:02] todayi started laughing outso ba [08:02] but they were good [08:03] i havent been asleep ainsce 10. slept for 1 hrs. got up at 12:30 [08:07] morph: [08:08] yo [08:08] oh woops [08:08] chatted that in the wrong channel [08:08] sry [08:11] lol === Ursinha-afk is now known as Ursinha [09:09] greetings all. in upstart (specifally the version on 12.04 LTS), is it possible for the job to have a "reload" action that sends a specific signal to the process? [09:10] reload sends HUP [09:10] if you need something else, I don't see a way to change it === DasIch_ is now known as DasIch [10:10] I'm looking at haproxy, which has a "reload" command [10:10] @sgran thanks === nmollerup is now known as mollerup [12:33] hardly surprising mysql updates this morning [13:11] Hey guys to update phpmyadmin, what else should I do besides moving the new folder? [13:21] what should the server 12.04 kernel be? [13:23] pehden-: That depends :) If you did a fresh install from a point you might have a backported kernel installed [13:23] pehden-: Otherwise the regular 12.04 kernel would now be something like 3.2.0-49* [13:24] 3.2.0-48-generic [13:24] hmm [13:25] it wont go past this one [13:25] every update says its keeping this one back [13:27] pehden-, do you use dist-upgrade (as you should for kernel upgrades) ? [13:27] just "ugrade" wont work ... [13:27] *up [13:28] eee [13:28] see i want to only use LTS [13:28] (see the description of upgrade ad dist-upgrade in the apt manpage) [13:28] nobody will stop you :) [13:28] somebody should give dist-upgrade another name :) [13:28] ++ [13:28] :) [13:28] !dist-upgrade [13:28] maybe an alias [13:28] A dist-upgrade will install new dependencies for packages already installed and may remove packages if they are no longer needed. This will not bring you to a new release of Ubuntu, see !upgrade if that is your intention. [13:29] ahh ok [13:29] apt-get dist-upgrade? [13:29] yes [13:29] ok [13:29] wait is there a list of things removed before I do this [13:30] yes [13:30] 3.2.0-48-generic #74-Ubuntu SMP Thu Jun 6 19:43:26 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux [13:30] is what I have now... [13:30] anyone know the link right off? [13:31] link for what? [13:31] the list [13:31] apt-get will show you [13:31] ok [13:31] just like it shows what it's about to install, upgrade or hold back [13:32] i only have ssh access to this server so I dont want to screw it up [13:32] most likely it will remove nothing [13:32] pehden-: Feels like you could use a dev environment, where you can do a bit of safe testing :) [13:32] pehden-: Something like a local virtual machine or something. [13:32] maybe, but this server is live [13:33] looks like its only upgradeing the kernel [13:33] you already upgraded the other packages with 'upgrade' [13:34] exactly [13:34] which is a good idea BTW, to always do upgrade before dist-upgrade [13:34] i made a script that i use to update everything at once but this command is not on there [13:36] best to check the results of every upgrade command anyway [13:37] oh yea, figured that out the hard way a year ago [13:37] ran upgrade and it removed programs i used and replaced them with alternatives I was furious [13:38] until i learned how to use the alternatives [13:39] what programs were that? [13:39] i dont remember that much [13:42] it just removed nvidia driver i think [13:42] and virtualbox [13:42] that wasn't on a server, I suppose :) [13:42] it was [13:42] both were [13:43] from the original install [13:43] its a desktop converted to a server [14:03] anyone can help please. I have installed wordpress on nginx, user group www-data, and have created new ftp user and added user to group, but when I upload files over FTP client my wordpress cannot read files? [15:00] Daviey: o/! if you have a second could you please process the crmsh binary from the NEW queue? [15:02] roaksoax: done [15:02] Daviey: awesome, thanks!! [15:25] zul, hallyn: In case one of you feeling adventurous... chinstrap:~smb/4review/qemu+xen.debdiff (for Saucy) ;) [15:31] smb: looking [15:32] hallyn, It is a cherry-pick of three supposedly stable worthy changes on top of qemu to avoid it segfaulting when used for a PV Xen guest and libxl. [15:33] smb: the cp of commit 62fc403f11523169eb4264de31279745f48e3ecc drops a lot of the original commit [15:33] how come? [15:33] oh no [15:33] lol. weird formatting on my screen confused me [15:34] smb: +1 [15:34] Ah ok... I was not sure I had done something like this [15:34] :) [16:02] So, I upgraded my test lxc box to the newer 3.8 kernel from raring [16:02] and now it's saying the memory controller cgroup is missing [16:03] is there a work around? or should i go back to the quantal kernel? [16:13] Patrickdk/sarnold: I declare the issue fixed! [16:13] open files limit was the problem [16:30] Hi, the link in this message is in regards to an Open Source Computing to further the developments of hardware for the mobile industry. It's time for Linux to be the #1 Consumer Operating System, Ubuntu Edge movement! http://pastebin.com/j57Dc29E We can all make a difference for as little as One Dollar! Thank You for your time. [16:47] Hm. [16:48] jsonperl: great success! [16:48] totally… the relief I am feeling this friday is magical [16:48] thanks so much for the help [16:48] :) [16:48] man, I'm both glad and embarrassed it was something so much simpler than I feared.. [16:49] same [16:49] ya, normally that issue is easily solved, cause of error logging [16:49] I'm actually digging through logs some more to make sure I didn't miss anything [16:49] atleast the memory *leak* was fixed though too [16:50] and a few other scalability issues that haven't hit yet :) [16:50] Yea some great stuff happened here… [16:50] We're just getting rolling too… This is just the beginning of scaling issues ;) [16:50] I switched all machines to 64 bit after jemalloc magically fixed "the leak" [16:50] Also… btw… jemalloc NO LIKE 32 bit [16:51] I have never used jemalloc on 32bit [16:51] no wonder I never hit that issue [16:51] And you should not :D [16:51] I got random segfaults after running for a while [16:52] Yea really… NOTHING in the logs [16:53] not suprised [16:54] normally issues like this is what drive poeple to worry about spending time to add them [16:55] Found some good articles about tweaks for scaling [16:55] One from urban airship about 500,000 concurrents! http://urbanairship.com/blog/2010/09/29/linux-kernel-tuning-for-c500k/ [17:09] hah, nice, when c10k is just too damned cute :) [17:10] now THAT is some serious scale [17:12] wow, their guide has -way- less in it than used to be necessary for scaling out to even 1/100th that load. [17:12] and their advice for limits is just plain wrong ;) hehe [17:13] which ones do you feel are wrong [17:14] jsonperl: you found out the other day that limits.conf is only used if the user's processes are started via the PAM stack somehow, which isn't terribly likely for production servers.. [17:15] granted, I think they knew that they didn't know what they were doing with the caveat about "look at the manpage" but not say -which- of the myriad manpages would be enlightening. :) [17:15] correct… so obviously that one was way off! [17:15] jsonperl: did you set your maximum socket buffers to 16MB as well? [17:15] yep :/ [17:16] jsonperl: give this a skim and then consider dropping it :) http://en.wikipedia.org/wiki/Buffer_bloat [17:16] net.ipv4.tcp_rmem = 4096 4096 16777216; net.ipv4.tcp_wmem = 4096 4096 16777216 [17:16] k [17:16] what do you think are more sain levels? [17:16] sane [17:16] 16M might be fair enough for locally connected gigabit or huge latency fat pipes (transatlantic? transpacific?) [17:17] jsonperl: probably 512K or so. [17:17] We certainly get some pretty latent connections… but not THAT much data [17:17] maybe 2k/s [17:18] so more like 4096 4096 524288 [17:20] Looks like the default is 4096 87380 6291456 [17:21] that seems pretty sane. how many connections will you have? multiply that by the middle number and hopefully it'll still be reasonable.. [17:23] We're really not looking for more tha 400 or so concurrent per machine [17:23] At that point we'll start to see cpus chuggin [17:46] ya, I only use 256 to 1meg on mine [17:46] patdk-wk: maximum or default? [17:47] max [17:48] cool, thanks :) [17:48] now, I did adjust my tc rules for outgoing though :) [17:49] if more than 10 packets are waiting to go out, 11+ get dropped [17:49] so doubt I ever hit that limit anymore :) [17:49] haha [17:49] ya, used to really annoy me [17:49] tc has been on my todo list for a decade now.. [17:49] scp transfer done, and then wait 5min for the buffer to flish [17:49] flush [17:50] it's limited 10 per class [17:51] also figured out how to proplerly match my *prenat* ip's so they match correctly in outgoing [17:51] that really fixed my ruleset I had made === matsubara is now known as matsubara-lunch === Ursinha is now known as Ursinha-afk === Ursinha-afk is now known as Ursinha [20:46] Does anyone happen to know if restarting iscsid will cause existing connections to be dropped? [20:47] defently, why wouldn't it? [20:48] users love it when storage goes away [20:48] well, that shouldn't cause a user an issue [20:48] patdk-wk: Because the iscsi daemon is just provides the control plane. [20:48] as it should auto-reconnect [20:49] patdk-wk: The kernel itself provides the data plane. [20:49] hello what is errors=remount-ro in /etc/fstab? [20:49] patdk-wk: So it should necessarily cause the connections to be dropped. [20:49] soren, never been my experience [20:49] always dropped when I restarted [20:49] patdk-wk: Ok, thanks. [20:49] had to use the command thing to adjust it, if I didn't want to restart to drop it [20:49] but this might have changed [20:50] patdk-wk: I'm not saying they won't, I'm just responding to your "why wouldn't it". [20:50] considering the this is from 8.04 :) [20:51] streulma: It tells the kernel what to do in case of errors. [20:51] streulma: remount-ro means "remount the filesystem as read-only". [20:51] soren: on boot or on running? [20:51] as in, disk went missing, unable to read from disk, writes failing to disk, ... [20:51] anytime [20:52] if you yank the disk out, it will take 30seconds, then go ro mode [20:52] streulma: Other options: "continue" (pretend like nothing happened) and "panic" (make the entire system fall over) [20:52] streulma: Running. [20:52] what is /var/lib/dpkg/cmethopt used for? [20:52] s/^/does anyone know/ [20:53] I'm going through debootstrap and was just curious [20:53] pdevine: IIRC, it's a config file for dselect. [20:56] I love peeling the onion back on this stuff. it's like an archeological expedition [20:59] I discovered the other day that dpkg can't handle pre-depends correctly.. and today discovered a comment in debootstrap complaining about how dpkg doesn't handle pre-depends [21:00] I wish I'd known that last week. :-D === Ursinha is now known as Ursinha-afk [23:00] my server gets stuck at here time to time any ideas? http://pastebin.ubuntu.com/5916517/ [23:04] nate15329: did you do the usual suspect removal? [23:04] nate15329: aka noacpi acpi=off [23:04] ? [23:04] also which kernel is there? [23:07] 3.8.0-26-generic; i tried removing my pci-x sata card recently; but no change...weird part is that it hangs there even in recovery mode, but in normal mode this sometimes happens [23:09] ill try the noacpi & acpi=off on next boot; i got it up temporary for a quick backup [23:10] another option would be to go for a different kernel [23:10] pick something from the lower shelf. [23:10] tried that as well...the one below it [23:10] is it some ancient hardware? [23:10] which i think is 3.8.0-18 or so [23:10] commodity or server? [23:10] nate15329: by lower shelf I was thinking like reaaaaally lower ;) [23:11] server 13.04 hp dl385 g1...old hardware indeed [23:11] nate15329: 3.x & 2.6.x [23:11] G1? [23:11] a bit old, indeed. [23:11] lol yep g1 [23:12] acpi and HPET if available [23:12] disable and check [23:12] and with such hardware - go lower with kernels [23:12] and as I wrote - seriously lower ;) [23:12] does older kernels support sata II drives? [23:13] no problem [23:14] ok how would i install 2.6.x kernels on 13.04 xD [23:15] I haven't tried 13.04 [23:15] I did downgrade 12.04.02 once from 3.5 to 3.2 [23:15] nate15329: try booting a lucid livecd on it first or something similar, make sure it works at all.. [23:16] yes, that's a sane suggstion [23:16] (sorry, I am up at work for 14 hours now ;) [23:18] oh that's trouble :) [23:18] ;q [23:22] ah yeah i will...mostly i perfer running the latest though xD