[00:01] <mklappstuhl> sarnold, what do you think about this: http://hack2live.blogspot.de/2009/07/force-reboot-linux-server.html
[00:02] <sarnold> mklappstuhl: ah! that's easier than writing a little tiny shutdown program, which is where I was headed next.
[00:02] <sarnold> mklappstuhl: be sure to echo "s" first to _sync_, then "u" to umount
[00:03] <mklappstuhl> [2510048.769174] SysRq : HELP : loglevel(0-9) reBoot Crash terminate-all-tasks(E) memory-full-oom-kill(F) kill-all-tasks(I) thaw-filesystems(J) saK show-backtrace-all-active-cpus(L) show-memory-usage(M) nice-all-RT-tasks(N) powerOff show-registers(P) show-all-timers(Q) unRaw Sync show-task-states(T) Unmount show-blocked-tasks(W) dump-ftrace-buffer(Z)
[00:04] <mklappstuhl> sarnold, I just tried with "b" in the meantime — oops :)
[00:04] <sarnold> mklappstuhl: any success?
[00:04] <mklappstuhl> sarnold, dmesg output above. server still running
[00:06] <mklappstuhl> this is way to fucked up for 2am alread
[00:07] <sarnold> mklappstuhl (!) maybe you need to enable the sysrq sysctl first? Try sysctl -w kernel.sysrq=1
[00:08] <mklappstuhl> sarnold, and then echo "s" > sysrq-thing
[00:08] <sarnold> mklappstuhl: yes
[00:12] <mklappstuhl> looks like im fucked...any new ssh session isnt responding except for a login message
[00:13] <IdleOne> Please stop swearing
[00:13] <sarnold> ouch :/ do you have any serial access? or are you stuck driving?
[00:14] <mklappstuhl> sarnold, no serial access and I'm in my bed right now so I might just do it tomorrow in the morning
[00:15] <sarnold> mklappstuhl: ooof. Good luck.
[00:20] <mklappstuhl> sarnold, thanks for your help... lets see if I'm back in 7 hrs :D
[00:22] <sarnold> mklappstuhl: sorry we couldn't get it to work; hopefully you can get some sleep this way though :)
[01:34] <codepython7771> does it help to have a swap on a SSD drive? Is anyone using that here?
[01:34] <slyboots> codepython7771: You shouldnt really
[01:34] <codepython7771> slyboots: why is that? glad i asked
[01:35] <slyboots> SSD's you want to reduce writes where you can.   Plus your better just.. you know, buying more RAM
[01:35] <slyboots> RAms cheap :)
[01:35] <slyboots> And faster
[01:36] <qman__> SSDs are fast, and suited to certain purposes
[01:36] <qman__> but they're not reliable yet
[01:37] <qman__> more RAM is always better, and it is cheap
[01:38] <fdge> with a SSD there are NO signs of impending failure
[01:38] <fdge> it will just stop working
[01:42] <qman__> it's still bleeding edge to be sure
[01:42] <qman__> that said, it does have its uses
[01:42] <qman__> when speed is everything, SSD is a good choice
[01:42] <qman__> and they're really nice in laptops for the battery life
[01:43] <fdge> they are very nice, you just need some form of backup
[01:45] <qcjn> hi, anybody knows about quotas ? i've got error " can't guess filesystem" and "can't find a filesystem to check" but when i do a mount, the "usrquota" is there ?
[01:46] <qcjn> thats when i do the quotacheck command
[02:47] <codepython7771> for running a server, how essential is the RAM be ECC?
[02:54] <ikonia> codepython7771: nothing to do with a server, that's a hardware requirement
[02:55] <ikonia> and that's not really anything to do with ubuntu either
[02:58] <codepython7771> ikonia: is there an easy way to find out if ubuntu is running on ecc or non-ecc?
[02:58] <codepython7771> also do you use ecc or non-ecc memory for your server?
[02:58] <ikonia> it's nothing to do with the OS
[02:58] <ikonia> it's a hardware requirement
[02:58] <ikonia> codepython7771: research your hardware platform and it will tell you if it NEEDS eec ram or not
[06:29] <lordievader> Good morning
[06:38] <linocisco> hi all
[06:38] <linocisco> hi all
[06:39] <linocisco> According to this http://iso.qa.ubuntu.com/qatracker/milestones/219/builds, do I have to download .iso file daily?
[06:46] <Kentos> hey all
[06:48] <Kentos> getting very slow speeds on proFTP through Filezilla, other LAN traffick methods work well enough (roughly 2Mb/s) though i believe i should be getting a much stronger connection in either case on a 10/100 router connection
[06:49] <Kentos> what might be the source of a 32KB/s transfer speed on filezilla but getting 2MB/s on other transfers
[06:52] <theguywithanaxe> lol
[07:35] <linocisco> According to this http://iso.qa.ubuntu.com/qatracker/milestones/219/builds, do I have to download .iso file daily?
[07:39] <lordievader> linocisco: That is the daily build, in other words everyday a new version gets uploaded without any testing. Would be strange if you need to download the iso daily, I assume that 12.10 updates daily. However I don't know, I guess you should ask in #ubuntu+1 (I believe that is the channel)
[08:01] <DoomGuy> this looks wired but true.. I am using ubuntu 12.04 in 3 machines, Ubuntu machines can't ping each other without restarting the network interfaces
[08:01] <DoomGuy> but Windows machine can ping ubuntu's one
[08:02] <DoomGuy> can someone tell me what's happening ?
[08:07] <koolhead17> hi all
[08:14] <ikonia> DoomGuy: can they use any other services on the machine
[08:14] <DoomGuy> ikonia, who ?
[08:15] <DoomGuy> it's 3 different machine : 1 server and 2 clients
[08:15] <ikonia> DoomGuy: whoever can't ping
[08:15] <DoomGuy> ikonia, of course !
[08:15] <DoomGuy> but they work normally
[08:16] <DoomGuy> ikonia, I am getting crazy really :(
[08:16] <ikonia> DoomGuy: ok, what services is the server running for example
[08:16] <ikonia> DoomGuy: (why do you care about ping )
[08:17] <DoomGuy> ikonia, Git, httpd, mysql
[08:17] <DoomGuy> it's web server
[08:17] <ikonia> ok, so they all work fine
[08:17] <ikonia> so you know there is base network connectivity
[08:17] <DoomGuy> ikonia,  nothing of them until the ping works
[08:18] <ikonia> so it has to be ICMP - and the only thing I can see that would behave like that is either a.) wake on lan b.) firewall - allowing established connections
[08:18] <DoomGuy> but from windows machine everthing woks fine
[08:18] <ikonia> DoomGuy: so one linux client, one windows client ?
[08:18] <DoomGuy> ikonia, 2 linux clients and 20~ windows clients
[08:19] <DoomGuy> ikonia,  the linux machine cannot access any service on the server. windows machine can
[08:19] <ikonia> DoomGuy: that really sounds like iptables
[08:19] <DoomGuy> I can do anything I want from a windows machine (ssh, commiting..)
[08:20] <DoomGuy> ikonia, does Ubuntu server come with iptables configured on it
[08:20] <ikonia> DoomGuy: yes
[08:20] <ikonia> DoomGuy: I asked you to check this the other day and you said it had no firewall
[08:20] <DoomGuy> ikonia,  hmm I will check that to see if any rules is setup
[08:21] <DoomGuy> ikonia, I didn't touch iptables
[08:22] <DoomGuy> ikonia, Policy Accept in every chain
[08:22] <DoomGuy> INPUT, FORWARD....
[08:22] <DoomGuy> by default it allow anything I guess
[08:23] <ikonia> can th eserver ping the client
[08:24] <DoomGuy> ikonia, yess !
[08:24] <DoomGuy> ikonia, if so then the client  start ping too
[08:25] <DoomGuy> ikonia, it looks like it's relative to standby functionality
[08:25] <ikonia> I wonder if network manager is putting the card to sleep
[08:25] <DoomGuy> if the server is in standby linux client lost connections
[08:25] <ikonia> ICMP is then waking it up
[08:25] <DoomGuy> ikonia, maybe it's that !
[08:26] <ikonia> it's an odd one if it is that
[08:26] <koolhead17> Daviey: around
[08:26] <thierry> hi everyone, i configured an adhoc network on my ubuntu-server image using this tuto ( https://help.ubuntu.com/community/WifiDocs/Adhoc) , the prob is that i see that in my system that wlan0 is activated connected and that iwconfig returns the correct parameters but i cant find the network on anyother device!
[08:26] <DoomGuy> ikonia, any idea about that subject ?
[08:27] <ikonia> DoomGuy: have a look at the card in network manager, also look at it in ethtool
[08:28] <DoomGuy> ikonia, have a look at the card in network manager => how to do so ?
[08:28] <DoomGuy> for ethtool I will check in the net
[08:29] <DoomGuy> ikonia, you're right dmesg show me that messages
[08:29] <DoomGuy> eth0 NIC is Down
[08:29] <ikonia> perfect
[08:29] <DoomGuy> later on eth0 NIC LINK is up
[08:29] <ikonia> so ethtool can disable sleep
[08:30] <ikonia> DoomGuy: may also be speed negotiation making it flap
[08:30] <DoomGuy> ikonia, OK this is the real issue
[08:30] <DoomGuy> thanks a lot I will check for that subject
[08:30] <ikonia> no problem
[09:19] <theguywithanaxe> minute**
[09:38] <freakynl> Hi, I'm in need of some assistence. We have a 7 disk RAID-5 (mdadm) that's exported over LIO (ISCSI / buffered) to a windows machine
[09:39] <freakynl> especially the write performance is horrible, read is pretty much ok. Currently I'm only copying a DVD (contents) to the storage. It's doing a - hold your horses - whooping ~3MB/s.
[09:40] <freakynl> If you look here: http://pastie.org/4746029 you can see writes issues to md4, no read are issued to it however. Yet, the disks the RAID consists of do read 6x what they write and I think the whole issue comes from there. Especially since this is largely sequential write it should be writing entire stripes at once and then it doesn't need to read a single bit single the whole stripe is replaced
[09:41] <freakynl> sde-sdl is md4 btw, but sdl is hotspare so you don't see I/O on it
[09:42] <freakynl> beh last single should be since
[09:42] <xnox> freakynl: work out the bottleneck: try local writes, test network, try another client (e.g. a linux client). Then you will know who is the bottleneck / culprit.
[09:43] <xnox> freakynl: with mdadm linux-raid mailinglist is good at helping troubleshooting performance, if indeed even the local writes are slow.
[09:43] <freakynl> xnox: obviously something in the LIO/mdadm stack. There are no reads issued whatsoever currently. I don't have the knowledge to work out bottlenecks that deep in the kernel
[09:44] <freakynl> I've been running on 3.5.3 for a while - that was much faster. However, it was also unstable approx 1 - 1.5 days until crash... Might try 3.5.4
[09:44] <xnox> freakynl: no, it's not obvious. No need to have kernel knoweledge. Simply dd /dev/zero into a file on the mdadm e.g. 2 GB and see the throughput.
[09:45] <freakynl> dd local is fine
[09:45] <freakynl> the issue exists in the combination lio/mdadm. For some reason an immense load of reads are send to the devices, this is done by mdadm as otherwise they would show up on the md device in iostat
[10:21] <freakynl> 3.5 kernel also has a black screen btw, passing nomodeset to the kernel doesn't help. can ssh into it just fine however
[10:22] <ogra_> did you try switching to a tty and back ?
[10:22] <DoomGuy> ikonia, sorry for disturbing but I am still unable to wake up my e1000e ethernet card :(
[10:22] <ogra_> oh, wait i'm in #ubuntu-server, ignore that
[10:22] <ikonia> DoomGuy: no need to apologise
[10:22] <ikonia> DoomGuy: do you still see if going up / down in the syslog
[10:23] <freakynl> ogra_: :)
[10:23] <ikonia> DoomGuy: as in the interface going up/down
[10:23] <DoomGuy> ikonia, yes many messages like that
[10:23] <ikonia> ok so it's still flapping
[10:23] <ikonia> DoomGuy: I guess we need to work out why
[10:23] <ikonia> can you post the output of sudo ethtool $interface
[10:24] <ikonia> (pastebin)
[10:24] <DoomGuy> ikonia,  OK friend
[10:26] <DoomGuy> ikonia, http://pastebin.com/020BJR7b
[10:26] <ikonia> DoomGuy:
[10:26] <ikonia> Advertised auto-negotiation: Yes Speed: 100Mb/s Duplex: Full
[10:26] <ikonia> DoomGuy: is it possible it's flapping as it's set to auto neg ?
[10:27] <ikonia> DoomGuy: also check what the "wake-on" option "g" is
[10:28] <DoomGuy> ikonia,  I tried to make autoneg off in both side (server and client linux machines) but it has no effect and windows machine can't do any automatic wake up to the ethernet  card
[10:29] <ikonia> DoomGuy: do you see it flapping on the switch
[10:30] <DoomGuy> ikonia, how to do so ? I am not network guru
[10:30] <DoomGuy> :(
[10:30] <ikonia> DoomGuy: is this a "home" network, or a work network ?
[10:30] <freakynl> xnox: switching to 3.5.4 kernel (from mainline repository) increases throughput to ~25-30MB/s with no other changes whatsoever. Still seeing a lot of reads however, but much much faster now (still not anywhere near what I was hoping for... sequential should easily fill the GBe wire there's no other I/O on it currently)
[10:31] <DoomGuy> ikonia, work network with HP procurve  big switch
[10:31] <ikonia> ok, you need someone who can manage the procurve, I can't I don't know procurve well at all
[10:31] <DoomGuy> ikonia, what do you need exactly ?
[10:31] <DoomGuy> do you think it's a switch issue ?
[10:32] <ikonia> look at the switch port the machine is pluged into, see if it is disconnecting at the switch and the server end, or just the serer or switch end
[10:33] <DoomGuy> ikonia, hmm OK I will check this
[10:33] <ikonia> it's worth looking it can often give you a clue
[10:33] <eagles0513875> any email server experts in here willing to help me with an issue which has me pulling my hair out
[10:33] <ikonia> eagles0513875: you've been told what to do in #postfix and #dovecot, don't start in here
[10:34] <eagles0513875> ikonia: feel free to put me on ignore
[10:34] <ikonia> eagles0513875: no, feel free to stop wasting peoples time spamming channels with the same questions and ignoring the answer
[10:35] <ikonia> DoomGuy: when you put the machine on no-auto-neg did you set it to forced speed on the switch too ?
[10:35] <ikonia> DoomGuy: if the switch is set to auto neg, and the machines aren't that can cause you a problem
[10:35] <Kentos> figuring out LAN speed problems?
[10:35] <ikonia> Kentos: possible lan speed problem, card dropping off the lan
[10:35] <ikonia> Kentos: maybe a speed issue or a WOL issue
[10:35] <ikonia> just working it through for DoomGuy
[10:36] <Kentos> i have some fairly low LAN speeds myself, only gets about 2-3MB/s have a 10/100 ethernet router and the NIC's for the two machines im using should be the same if not the full GB speed
[10:37] <freakynl> anyone aware of tools that can move a partition table? I currently have a md4 device, with 2 partitions, the partitions itself are exported over iSCSI, so the partition contains a partition :). I want to make the partition that's currently on /dev/md4p1 the partition that is on /dev/md4. So basically I need /dev/md4p1p1 to become /dev/md4p1 so I can just export /dev/md4 instead of /dev/md4p1 if that makes sense :)
[10:37] <ikonia> Kentos: why are you expecting 1GB out of a 10/100 router
[10:37] <Kentos> sorry for the caps, gigabit, you know the highest internet speed for NICs
[10:38] <ikonia> freakynl: you can do it with dd workout the stop/start
[10:38] <ikonia> Kentos: yes, but the router they are plugged into is 10/100 - so they will never do 1GB
[10:38] <Kentos> and i didnt say i expected anything, just more the a 16megabit connection
[10:38] <freakynl> yea I could - just really worried I'm gonna f* that up :P
[10:38] <ikonia> freakynl: easy on the language please
[10:38] <ikonia> freakynl: yes, it is a concern / risk
[10:38] <freakynl> I could just calculate where md4p1p1 starts and where md4p1 starts and the offsets and start new partition there with old size
[10:38] <ikonia> Kentos: you asked why they where not using 1GB connection, because the router is 10/100MB
[10:39] <freakynl> and/add
[10:39] <ikonia> freakynl: I don't think it will work with software raid disks - but it's worth looking how the clonezilla process would work
[10:39] <DoomGuy> ikonia, "when you put the machine on no-auto-neg did you set it to forced speed on the switch too ? " I don't know how to configure the switch to do so
[10:39] <ikonia> freakynl: or look at how it clones
[10:39] <Kentos> no, i told you the specs of my NICs, which should be gigabit speed capable, not what i expected
[10:40] <ikonia> DoomGuy: may also be a problem too
[10:40] <freakynl> ikonia: don't get that, clones what?
[10:40] <DoomGuy> ikonia, for the LAN speed issue it's possible but tell me how you can explain this
[10:40] <ikonia> Kentos: ok,
[10:40] <Kentos> i also said im getting 2-3 MB, so roughly 16-24 megabit speed, i should be able to get 100 or at least somewhere close
[10:40] <ikonia> freakynl: there is a tool call clonezilla - it is veyr good at cloning disks
[10:40] <DoomGuy> ikonia, one linux machine is connected to the server via SSH and the other machine linux cannot ping the server
[10:40] <ikonia> Kentos: is that router connnected to the internet ?
[10:40] <freakynl> concerning networks, you either force both sides to the same speed / duplex or use autoneg on both sides. Don't ever force one side and autoneg the other - it will mess up
[10:41] <DoomGuy> ikonia, I have no problem with Windows machine connected to the server
[10:41] <Kentos> yeah it is
[10:41] <freakynl> ikonia: yea, I don't have the space to clone this, nor the time. ~11TiB
[10:41] <ikonia> DoomGuy: basically you need to try setting the speed and duplex on the server's interface card AND the HP swithc port
[10:41] <ikonia> freakynl: no, but looking at how it takes the partitions table maybe worth while
[10:41] <ikonia> Kentos: are you measuring performance via the internet ?
[10:42] <eagles0513875> ikonia: whats the difference between the server guide and help.ubuntu.com
[10:42] <freakynl> ikonia: do you happen to know where gpt is stored? with mbr I could just copy the first 1MiB of the device and restore it with dd in case of emergency and I could test with ro I suppose
[10:42] <eagles0513875> cuz i have noticed they have 2 different things when it comes to setting up certain things like dovecot and postfix
[10:42] <EMKO> What's your routers speed.? I had to get a gigabit router to use my 100mbit connection
[10:42] <ikonia> eagles0513875: look for yourself
[10:42] <ikonia> eagles0513875: I'm not reading them for you to tell you the answer for you to then ignore
[10:42] <Kentos> im measuring performance based on transfers in filezilla and uploads via a web interface on the local network
[10:42] <eagles0513875> ikonia: all im asking is what is the difference im not expecting you to read them
[10:42] <ikonia> eagles0513875: how do I know the difference without reading them
[10:43] <eagles0513875> ok
[10:43] <ikonia> Kentos: Hmmm I wonder if that routes through the public interface
[10:43] <ikonia> Kentos: mabe explaining your slower performance
[10:43] <ikonia> "maybe"
[10:44] <Kentos> ikonia: could be right, but i dont see why, i was using a different router yesterday, still got around the same speeds though, up 5MB at times
[10:44] <Kentos> ikonia: not sure where the setting differences are with that
[10:45] <Kentos> far as router configuration on the new one goes
[10:45] <ikonia> I guess it would be router specific
[10:46] <EMKO> Are you transferring large files or a bunch of small files
[10:46] <freakynl> anyone aware of a tool that will dump partition table for gpt in a way it can be imported again? Like you can with sfdisk -l (but that doesn't support gpt)
[10:47] <ikonia> freakynl: did you say you where going to check out clonezilla and how it dumped the partition table ?
[10:47] <ikonia> freakynl: (apologies too many conversations lost track)
[10:48] <xnox> freakynl: you switch to mainline 3.5.4 from what kernel/which release? is there a performance regression in the ubuntu delta? or where you using an older kernel?
[10:49] <freakynl> xnox: compared to the latest 3.2 that 12.04 has
[10:50] <xnox> freakynl: ok. there are quantal kernel backported to 12.04 somewhere, which will be supported after quantal is released.
[10:50] <freakynl> ikonia: I'll have to see if I can boot that - however if there's something like sfdisk dumping it'll be way easier, I can just mod the current partition table, see if I can mount it ro and see my data, if not just sfdisk -l equivalent restore it
[10:50] <ikonia> freakynl: I was wondering if you could see what it called and then just use that tool
[10:50] <xnox> freakynl: such that throughout 12.04 support lifespan you can use newer and newer kernels.
[10:51] <freakynl> xnox: not sure what you mean by that but it's running quantal: http://kernel.ubuntu.com/~kernel-ppa/mainline/v3.5.4-quantal/
[10:54] <xnox> freakynl: that one is without ubuntu-specific patches I think. Just vanilla kernels.
[11:03] <jamespage> zul, do we still need to maintain the delta against debian for less.js?
[11:03] <jamespage> 'Drop BD on uglify whatever'
[11:05] <DoomGuy> ikonia, I think it was resolved now.. The problem was that the switch used for the server was a little D-LINK switch connected to the biggest one HP procurve
[11:05] <DoomGuy> now I have plugged the cable directly to the biggest switch and use the port of Gigabite connection and it works just fine
[11:06] <DoomGuy> ikonia, thank you a lot for your advices
[11:07] <DoomGuy> it looks like it was conflict between 100Mb port and 1G port
[11:12] <zul> jamespage: nope
[11:21] <xnox> freakynl: https://launchpad.net/~ubuntu-x-swat/+archive/q-lts-backport are the kernels that have ubuntu-patches applied and will be supported.
[11:21] <xnox> supported in precise that is.
[11:51] <freakynl> xnox: thx
[11:52] <freakynl> xnox: the weird thing is, I never had such performance issues when I was still using ietd
[11:54] <freakynl> what's also weird is that it seems to be between lio's buffer and mdadm. That is, the first copy after restarting or flushing buffers goes really fast until the buffers are full. At that point it becomes horribly slow. Moving from buffer to disk is thus where the issue seems to exist, but as stated, I didn't have that with ietd and that was using write back (thus buffers too)
[12:06] <freakynl> xnox: went back to ietd, buffers are full now, instead of copying from local to iscsi I'm copying from iscsi to iscsi now (so it both reads and writes instead of only writing) and it's still over 10 times faster... there's about 110-120% reads (kB/s compared to writes) instead of 600-700% reads vs writes. This is whilst reading from the disk too, whereas with lio I wasn't even reading from the disk, only writing, which would come down to 10-2
[12:11] <jotterbot> Is someone able to help me configure Apsis Pound on ubuntu?
[12:12] <ikonia> I've never even heard of Apsis Pound
[12:12] <ogra_> thats likely not a british pound :)
[12:13] <jotterbot> http://www.apsis.ch/pound
[12:13] <jotterbot> it is a reverse-proxy similar to nginx
[12:14] <jotterbot> anyone?
[12:15] <jotterbot> I am trying to make Apache on OSX log the CLIENT IP of requests
[12:15] <jotterbot> instead i am getting the local PRIVATE IP address of the ubuntu reverse proxy server
[12:15] <jotterbot> any help?
[12:17] <freakynl> in #osx perhaps?
[12:17] <jotterbot> It is Pound that i think needs configuring though. Which is running on Ubuntu Desktop 12.04.1
[12:18] <jotterbot> I am surprised no one has ever used Pound
[12:18] <freakynl> jotterbot: I take it you're not using %a but %A http://httpd.apache.org/docs/2.2/mod/mod_log_config.html but this is not an apple/osx help channel
[12:18] <jotterbot> for load balancing
[12:19] <jotterbot> I understand it is not an apple channel
[12:19] <jotterbot> I will follow that up though. thankyou
[12:19] <freakynl> xnox / ikonia : thx for the help - I gotta run unfortunately. Have it running on ietd now - way faster. I'll take it up with the mailing lists of mdadm and lio
[12:20] <freakynl> jotterbot: usually not that hard, just need the right LogFormat directive. %a should do it
[12:20] <freakynl> lower case thus :) capital A is local IP
[12:20] <jotterbot> Aha! I will hit up the OSX channel to help configure
[12:21] <jotterbot> I then have to sort out Rumpus ftp
[12:21] <jotterbot> :(
[12:21] <jotterbot> but thanks guys!
[13:03] <koolhead17> zul: zul zul zul zul
[13:04] <zul> koolhead17: seriously we are worknig on it calm down ill let you know
[13:04] <koolhead17> zul: okey :P
[13:04] <koolhead17> zul: marked the swift bug as invalid as it was our config issue
[13:05] <zul> koolhead17:  k
[15:46] <codepython777> I've 4  x 1TB drives and would like to run RAID on them. Whats the best way to load and run ubuntu server on it?
[15:58] <TheLordOfTime> anyone here able to set a bug as applying to a specific release of a package for me?
[16:11] <xnox> codepython777: simply install ubuntu-server on them.... there is raid options in the installer. See the server guide, advanced installation.
[16:12] <codepython777> xnox: thanks
[16:13] <jamespage> TheLordOfTime, bug #
[16:13] <TheLordOfTime> jamespage, standby
[16:13] <TheLordOfTime> jamespage, bleh system futz'd up :/
[16:13] <TheLordOfTime> one moment
[16:14] <TheLordOfTime> jamespage, https://bugs.launchpad.net/ubuntu/+source/nginx/+bug/1046150  can you set that to precise?  its possible quantal may already have that fix, i'm checking now
[16:14] <TheLordOfTime> since i plan on getting that SRU'd.
[16:14] <TheLordOfTime> i'm tired of not seeing that package be fixed :P
[16:18] <codepython777> xnox: what kind of raid do you recommend for running on a 4disk system? Can I build the raid before i install the ubuntu on that raid? I would prefer some kind in which if 1 disk fails, i can just replace it
[16:20] <gsteinert> codepython777: RAID10 would suit your needs... but its all a compromise between recoverability and capacity. check the wikipedia article... that's got a pretty good overview of the differences
[16:21] <sarnold> codepython777: hopefully useful: http://www.miracleas.com/BAARF/
[16:22] <xnox> codepython777: read wikipedia page about RAID, there are many raid levels. Each raid level has it's own: read/write performace increase/descrease, space efficiency and fault tolerance.
[16:22] <xnox> codepython777: http://en.wikipedia.org/wiki/RAID#Standard_levels
[16:23] <jamespage> TheLordOfTime, Priority: Low? really
[16:23] <TheLordOfTime> jamespage, upstream requested lowest priority
[16:23] <TheLordOfTime> i could set it higher
[16:23] <TheLordOfTime> at the time, i was headed out the door
[16:23] <xnox> codepython777: RAID 1, 10, 5, 6 all sound suitable, e.g. "replace one drive", but check the table and related links for more info as to which one you want.
[16:23] <TheLordOfTime> jamespage, its impossibly hard to replicate, afaict
[16:23] <TheLordOfTime> i even tried replicating
[16:24] <patdk-wk> actually, these days it's more complex
[16:24] <patdk-wk> the larger the disk you use, the longer it takes to recover from a failed disk
[16:24] <TheLordOfTime> it took upstream to confirm the bug
[16:24] <patdk-wk> and the more likely you will have a second failure, or more
[16:24] <TheLordOfTime> jamespage, i've been debating setting it higher, but i use 1.1.19 in a production environment with try_files, no segfaults in the 8 months I've had it running
[16:24] <TheLordOfTime> (pulled 1.1.19 from Debian before Precise was even out)
[16:26] <jamespage> TheLordOfTime, fixed in 1.2.1 right?
[16:26] <TheLordOfTime> jamespage, that's what its *reported* as in the bug, trying to confirm with upstream
[16:26] <TheLordOfTime> since they're evil with identifying what commit ties to a given version
[16:27] <TheLordOfTime> upstream tracker reports it was fixed but doesnt say *when* it was fixed
[16:27] <TheLordOfTime> i'm *assuming* it was fixed in 1.1.20, but...
[16:27] <TheLordOfTime> http://trac.nginx.org/nginx/changeset/4601/nginx  <--
[16:27] <TheLordOfTime> jamespage, i'm also treating it differently than, say, Apache, with priority setting, because its Universe
[16:28] <jamespage> TheLordOfTime, don't - if it breaks stuff its still important even if its in universe
[16:28] <TheLordOfTime> jamespage, therefore, what would you use?>
[16:28] <jamespage> TheLordOfTime, a good test case to reproduce will be important for the SRU
[16:29] <TheLordOfTime> jamespage, tell me about it
[16:29] <jamespage> TheLordOfTime, sounds like you have not see the issue right?
[16:29] <TheLordOfTime> i havent found *one* test case that works
[16:29] <jamespage> hmmm
[16:29] <TheLordOfTime> jamespage, only found one test case outlined on trac
[16:29] <TheLordOfTime> but they're using a nonstandard setup
[16:29] <TheLordOfTime> s/nonstandard/different/
[16:30] <TheLordOfTime> if you look at the trac bug, they've got some session tracker or something
[16:30] <TheLordOfTime> which may or may not contribute to the triggering of this bug
[16:30] <TheLordOfTime> jamespage, mind a /query?
[16:30] <jamespage> TheLordOfTime, feel free :-)
[16:33] <zul> jdstrand: quantum should be good now
[16:41] <jdstrand> zul: ok, I'll take a look
[17:25] <gholms> smoser: It seems that create_user in cloudinit/distros/__init__.py has a bunch of options with underscores, but the docs all use hyphens.  Is there some character mapping going on here that I just haven't found yet?
[17:26] <utlemming> gholms: yes...there is
[17:26] <smoser> gholms, utlemming did that. yeah, it does do mapping.
[17:26]  * gholms tries to find the code for that
[17:26] <utlemming> gholms: give me a minute and I'll find that for you
[17:28] <utlemming> gholms: cloutinit/config/cc_users_groups.py line 71
[17:53] <lordievader> Good evening
[17:54] <gholms> utlemming: Thanks
[18:41] <Joel_re> hey, does anyone know how to get rid of the prompt - as whats mentioned in this question - http://askubuntu.com/questions/187337/unattended-grub-configuration-after-kernel-upgrade
[19:38] <hallyn> zul: did i ask you last week to try an openstack testrun (whatever that means) with ppa:ubuntu-virt/virt-daily-upstream installed
[19:58] <zul> hallyn: im not sure if you did
[20:03] <hallyn> zul: could you?  i'm just looking for some stress testing.
[20:04] <zul> hallyn:  check with adam_g
[20:04] <hallyn> if jdstrand finds it works for him, i'd like to go to release team and ask for it to be considered a new upload
[20:04] <hallyn> zul: thanks
[20:04] <hallyn> adam_g: ^
[20:12] <jdstrand> zul, adam_g: hey, in order to check over the quantum updates, I just upgraded my quantal openstack vm and things all seem to be working except for horizon. I get this: http://paste.ubuntu.com/1213617/
[20:12] <jdstrand> Endpoint not found
[20:13] <jdstrand> is there something new I need to do to get horizon to work? (I'm using https://wiki.ubuntu.com/SecurityTeam/TestingOpenStack, like always)
[20:13] <jdstrand> cli stuff seems to work fine
[20:13] <jdstrand> well, I haven't tested everything
[20:14] <adam_g> hallyn: what needs to be tested specifically? i can probably just manually upgrade an existing installation and poke at it however you wanted
[20:15] <hallyn> adam_g: just spinning up some images, making sure they come up...
[20:16] <adam_g> jdstrand: meaning you can spin up instances, list images, etc using CLI but you cant use horizon? are you getting an error in the browser or just a 50x?
[20:17] <adam_g> hallyn: yah, i can do a basic test for you in a few
[20:17] <jdstrand> adam_g: I can list images using cli. I get a 401. no error in the browser
[20:18] <adam_g> jdstrand: you get a friendly authentication erorr within the dashboard? or something else?
[20:19] <jdstrand> adam_g: not friendly. the username and password just clear themselves
[20:19] <adam_g> jdstrand: but using the same username/password  with the CLI tools (asOS_USERNAME, OS_PASSWORD in env) works?
[20:20] <jdstrand> oh weird, 'nova list' says "Please enter password for encrypted keyring:"
[20:21] <jdstrand> adam_g: well, euca-describe-images works fine. nova list is prompting me for the keyring password. I have not seen that before
[20:23] <jdstrand> that seems to be bug #1020238
[20:23] <jdstrand> adam_g: ok, so now I have:
[20:23] <jdstrand> $ nova --no-cache list
[20:23] <jdstrand> ERROR: n/a (HTTP 401)
[20:24] <jdstrand> so apparently the ec2ools are working, but not nova
[20:24] <adam_g> jdstrand: i'd check the nova-api.log and keystone.logs to see whats going on. could be anything :|
[20:27] <jdstrand> ok thanks
[20:40] <adam_g> hallyn: sorry, i actually wont be able to get that libvirt version tested. im currently trying to get unblocked from https://bugs.launchpad.net/bugs/1035172
[20:41] <adam_g> hallyn: i can try that version to at least make sure instances startup, but beyond that not much
[20:43] <jdstrand> adam_g: what logfile should I be looking in for: http://paste.ubuntu.com/1213679/
[20:44] <jdstrand> nova-api.log gives me: http://paste.ubuntu.com/1213681/
[20:45] <hallyn> adam_g: ok, thanks anyway
[20:45] <adam_g> jdstrand: see if nova-compute.log reports anything. if theres no trace of any activity there, see nova-scheduler.log
[20:45] <adam_g> jdstrand: also, waht is the output of 'euca-describe-availability-zones verbose' assuming you're an admin user?
[20:46] <adam_g> jdstrand: the 'endpoint not found' error + that compute error leads me to believe nova-compute is not okay
[20:46] <jdstrand> adam_g: I fixed the endpoint not found
[20:46] <adam_g> jdstrand: what was it?
[20:46] <jdstrand> adam_g: the [filter:authtoken] in /etc/nova/api-paste.ini, /etc/glance/glance-api-paste.ini, and /etc/glance/glance-registry-paste.ini had these lines removed:
[20:47] <jdstrand> auth_host = 127.0.0.1
[20:47] <jdstrand> auth_port = 35357
[20:47] <jdstrand> auth_protocol = http
[20:47] <jdstrand> (on upgrade)
[20:47] <jdstrand> I put those in and nova list gives me output, but with that status error
[20:47] <jdstrand> nova-compute.log and nova-scheduler.log were silent when running nova --no-cache list
[20:48] <adam_g> jdstrand: id try to boot another, see that nova scheduler casts it to a compute node, and compute attempts to boot it
[20:50] <jdstrand> adam_g: ah, that seems to be working. the instance started
[20:50] <jdstrand> ok, so I had some crufty image I guess
[20:50] <jdstrand> instance
[20:51] <jdstrand> hrmm, now horizon gives me a 501
[20:58] <adam_g> jdstrand: apache error log should give you some info, maybe have to enable debug in /etc/openstack_dashboard/local_settings.py
[21:06] <jdstrand> yeah, that is what I'm trying
[21:10] <jeiworth> hi all, playing around with maas and juju, it appears that i can only deploy max as many instances as i have hw nodes, is this correct?
[21:11] <jdstrand> error_log just isn't giving me anything
[21:11] <adam_g> jdstrand: is debug enabled?
[21:11] <jdstrand> DEBUG = True in /etc/openstack-dashboard/local_settings.py
[21:13] <jdstrand> I didn't do anything else
[21:16] <jdstrand> adam_g: ok, it seems I did not solve the endpoint problem for horizon
[21:16] <jdstrand> http://paste.ubuntu.com/1213727/
[21:16] <jdstrand> if I restart my browser and try to login, I get that in error_log
[21:17] <jdstrand> keystone endpoint-list shows me everything
[21:18] <adam_g> jdstrand: what about 'keystone catalog' (with SERVICE_TOKEN, SERVICE_ENDPOINT unset)
[21:19] <jdstrand> adam_g: http://paste.ubuntu.com/1213732/
[21:19] <jdstrand> it seems ok...
[21:21] <jdstrand> /etc/openstack-dashboard/local_settings.py has OPENSTACK_KEYSTONE_DEFAULT_ROLE = "Member", should that be "admin"?
[21:21] <jdstrand> doesn't help
[21:22] <jdstrand> OPENSTACK_KEYSTONE_BACKEND. what is that...
[21:23] <adam_g> jdstrand: try clearing your browser cache
[21:24] <adam_g> jdstrand: horizon keeps a cache of the service catalog in the users browser, it might be stale
[21:25] <jdstrand> adam_g: didn't work... :\
[21:30] <jdstrand> ah, I needed to restart apache after setting Debug = True
[21:30]  * jdstrand should have known that
[21:31] <jdstrand> Exception Type: 	ServiceCatalogException
[21:31] <jdstrand> Exception Value: 	
[21:31] <jdstrand> Invalid service catalog service: volume
[21:31] <jdstrand> adam_g: so, is swift required these days?
[21:32] <jdstrand> seems so https://answers.launchpad.net/horizon/+question/189551
[21:33]  * jdstrand doesn't even have nova-volume installed
[21:33] <jdstrand> seems to be bug #946874
[21:34] <adam_g> jdstrand: hmm didnt know volume was a hard requirement now.
[21:36] <adam_g> jdstrand: i think you can get by with just having a volume api service and endpoint, no necessarily a functioning volume server. one sec
[21:38] <jdstrand> adam_g: ok, thanks
[21:41] <adam_g> jdstrand: try registering another service in keystone of type 'volume', and register a set of endpoints associated with the service, with all 3 URLs pointing to: http://localhost:8776/v1/$(tenant_id)s
[21:41] <adam_g> nova-api should be listening on 8776 already
[21:42] <jdstrand> yes, 8776 has a listener
[21:47] <jdstrand> adam_g: \o/
[21:48]  * jdstrand updates wiki
[21:48] <adam_g> jdstrand: nice
[21:49] <jdstrand> adam_g: thanks! I never would have gotten there today :)
[21:49] <adam_g> jdstrand: if you really want to provide a volume service, you can initialize a LVM volume group named 'nova-volumes' and install nova-volume, i think thats all thats needed
[21:50]  * jdstrand nods
[21:59] <cornfeed> hello hello, I have added a custom PPA source, and for some reason apt-get update is not pulling in the newest ones...is there a  cache I need to flush or something? I am lost. if i use chrome to browse the source I added in sources.list then I can see that the package has been updated to a newer 3.0.1 but my system wont upgrade from 2.3.3
[22:01] <patdk-lap> apt-get would automatically upgrade
[22:01] <cornfeed> thats what I thought, but its not
[22:01] <patdk-lap> unless it's being held back (missing dependency) or maybe they named it different, so it's a totally different package
[22:01] <patdk-lap> cornfeed, revise what you said
[22:01] <cornfeed> let me put up a dpaste so you dont think i am crazy
[22:01] <cornfeed> brb
[22:01] <patdk-lap> it will, something else is wrong :)
[22:02] <cornfeed> :-D
[22:02] <cornfeed> i really hope so
[22:03] <patdk-lap> all we really need to see is
[22:03] <patdk-lap> dpkg -l chrome
[22:03] <patdk-lap> apt-cache show chrome
[22:06] <cornfeed> hahaha, yes nvm
[22:06] <cornfeed> i am just silly
[22:06] <cornfeed> thanks
[22:06] <patdk-lap> chromium vs chrome?
[22:07] <cornfeed> no its a zentyal package (ebox formerly)
[22:08] <cornfeed> i will put a dpaste so you see what I was hung up on
[22:08] <cornfeed> http://dpaste.org/IE6qe/
[22:08] <cornfeed> in apt-cache show, it says the version is 3.0.1 and 2.3.3 :-/
[22:09] <cornfeed> oh shit i get it now, apt is very different than portage
[22:10] <patdk-lap> well, apt-cache shows what is available
[22:10] <patdk-lap> not what is installed
[22:10] <patdk-lap> dpkg shows what is installed, 3.0.1 :)
[22:11] <cornfeed> its listing all packages available in the repo where dpkg is what is installed
[22:11] <cornfeed> yeah i was just typing that :-P
[22:11] <patdk-lap> yep
[22:11] <cornfeed> thanks for the help man, no better lessons than feeling retarded
[22:11] <cornfeed> :-P
[22:19] <jdstrand> adam_g: so, there should be any problem with me using 'keystone service-create --name nova --type volume ...' and 'keystone service-create --name nova --type compute ...' should there? notice that --name is the same
[22:20] <jdstrand> the description is different
[22:20] <jdstrand> meh, I'll just change it
[22:21] <adam_g> jdstrand: IIRC, the name can be the same, type needs to be unique. might as well just keep them distinct, tho
[22:21] <jdstrand> yeah
[22:37] <jmedina> Hi there, anyone have experience with iscsi and multipath in 12.04?
[22:37] <patdk-lap> iscsi yes, multipath yes, multipath on linux no
[22:46] <jmedina> patdk-lap: thanks, I having problemas with a IPSAN DELL Powervault MD3200i
[22:54] <jdstrand> zul: ok, one last question for the quantum MIR
[22:58] <zul> jdstrand: okies
[23:00] <zul> jdstrand: ill fix that up with the rootwrap with the cisco plugin and niciria plugin im not sure if it uses rootwrap or not
[23:01] <jdstrand> jdstrand: yeah, there were no comments in the plugin .ini files for those two, so not sure if upstream has support for it yet. let me know what you find out
[23:02] <jdstrand> zul: I'm heading out now-- feel free to let me know on irc or the bug
[23:02] <zul> jdstrand: yeah i have mentally head out already ill comment on the bug
[23:02] <jdstrand> thanks
[23:25] <patdk-lap> jmedina, iscsi multipath seems to just work
[23:25] <patdk-lap> nothing needed to make it work, atleast it works here, 9 way multipath
[23:28] <jmedina> patdk-lap: I just removed multipath because I think it is a problem with iscsi initiator
[23:28] <patdk-lap> hmm?
[23:29] <patdk-lap> I installed multipath, logged into my iscsi target, and it just all came up
[23:29] <patdk-lap> and new multipath devices in /dev/mapper
[23:29] <patdk-lap> lots and lots of /dev/sd* devices
[23:29] <jmedina> I have to paths to the target
[23:29] <patdk-lap> I have 9 paths :)
[23:29] <patdk-lap> 9 paths, two luns
[23:30] <jmedina> the problem is when loggint to the second target
[23:30] <patdk-lap> I was able to read and write fine
[23:30] <patdk-lap> you sure the dell unit is setup for multipath?
[23:44] <jmedina> patdk-lap: yeap, the problem occurs when I use both RAID controllers in the SAN
[23:44] <jmedina> I just changed to using both ports on the first controller, I'm going to test with this