[00:11] <ujjain> is there an ubuntu server netinstall?
[00:15] <SpamapS> ujjain: yes!
[00:15] <ujjain> SpamapS: mini.iso?
[00:16] <SpamapS> ujjain: https://help.ubuntu.com/community/Installation/Netboot
[00:16] <SpamapS> ujjain: also if you need to do more than 1 or 2, you can use Orchestra
[00:16] <SpamapS> ujjain: but thats only barely available in 11.04, and really slated for "awesomeness" in 11.10
[00:17] <ujjain> I don't use tftp.
[00:17] <ujjain> hehe, Ubuntu is realy getting pretty awesome even for servers
[00:17] <SpamapS> ujjain: if you can put physical media in, you may want the minimal.iso
[00:18] <ujjain> I'l l use the mini.iso, it's just a testserver anyway
[00:18] <ujjain> let's just hope I don't have a GUI
[00:20] <ujjain> k, that went wrong I think ^^
[00:25] <dhon_> hi all, does anyone know of a way to kill an active TCP socket? I'm trying to test what happens when a client connection drops out
[00:25] <SpamapS> dhon_: iptables can simulate it
[00:26] <dhon_> from the command line that is (not within the app)
[00:27] <dhon_> SpamapS: I haven't used iptables before, could you elaborate?
[00:29] <SpamapS> dhon_: well you could make it drop packets or refuse connections..
[00:30] <dhon_> SpamapS: do you know if it will apply to active connections, or only future ones?
[00:31] <SpamapS> dhon_: its very low level, you can add/remove rules
[00:32] <SpamapS> dhon_: what exactly are you trying to test though?
[00:33] <SpamapS> dhon_: sounds to me like what you want to do is something like   iptables -I INPUT -p tcp -s remote-host --port 44444 -j DROP
[00:33] <dhon_> I have a service that opens a TCP socket to a server (my service is a client here)
[00:33] <dhon_> after some long period, my services is not communicating with the server anymore
[00:34] <dhon_> I would like to test if the connection being dropped produces similar results, as it's not generating any helpful log messages
[00:35] <SpamapS> yeah, iptables (a firewall, btw) would be useful for that
[00:35] <dhon_> SpamapS: cool thanks, I'll read up on that iptables command
[00:35] <dhon_> SpamapS: sounds like what I want
[00:36] <dhon_> one more thing, will I need to reverse that command in the future to reallow connections again?
[00:43] <SpamapS> dhon_: yes
[00:44] <dhon_> just replace -A with -D?
[00:45] <SpamapS> yep
[00:45] <dhon_> oh sorry, not using -A
[00:45] <SpamapS> note that I used -I .. to put it at the top
[00:45] <dhon_> I see, thanks for the help SpamapS
[01:37] <tucemiux> hey what logs should I be checking on my server ?
[02:20] <joeylockie>  hello am seting up a server mostly for my family to use and am planing on instailing the server with a full desktop environments (which one yet i dont know) but in order to do this do I need to install any of the preedefind collections of software in the server installer? or is it possable to istall them (if I fiond out I need them) later?
[02:24] <Datz> joeylockie: the main difference between the server and desktop install is the kernel as far as I know.
[02:24] <Datz> So you may just want to go with the Desktop verison
[02:24] <Datz> that way you won't have to install the desktop environment
[02:25] <joeylockie> in the desktop am I ablo=e to add apahice and what not easyly?
[02:28] <Datz> yes
[02:28] <Datz> more easily if you are not familiar with the command line
[02:29] <joeylockie> ok I will look into that thanks
[02:35] <twb> While you *can* set up a desktop on your server, it is not recommended
[03:00] <ntoombs> qman__: you there?
[03:02] <ntoombs> does this forum help with routers?
[03:03] <ntoombs> i've configured my ubuntu server to connect to my router with a static ip address
[03:04] <ntoombs> however when i go into my router and check for attached devices i see my server but it has a different ip address
[03:04] <ntoombs> it's not even in the same range
[03:04] <ntoombs> my router is 192.168.2.1
[03:04] <ntoombs> and the device showingup is 192.168.254.115
[03:05] <ntoombs> i have my server set to 192.168.2.119
[03:05] <ntoombs> i did at one time however have my server's static ip address set to 192.168.254.115 awhile back but i don't know how that could have been stored anywhere on my server so i'm assuming its a problem with my router
[03:06] <ntoombs> or with my modem which is now in briged mode
[03:06] <ntoombs> does anyone think they could help with this problem?
[07:38] <jkup> Sup
[09:01] <jvargas> hello. a quick question, are there identified issues or considerations to run ubuntu servers under hyper-v virtualization platform?
[09:45] <jamespage> Daviey: please can you give me an opinion on bug 801038
[09:47] <Daviey> looking
[09:48] <jamespage> ta
[09:49] <Daviey> jamespage: I don't think this is something that will be addressed in Ubuntu.  I agree that libvirtd should check for an existing process, and exit gracefully.
[09:49] <Daviey> I would consider it wishlist and confirmed.
[09:50] <Daviey> I'm not sure i see the overwhelming concern tho :/
[09:50] <jamespage> that was my  thinking; it would be possible to put a check into the upstart config tho
[09:50] <jamespage> but I agree its a bit of an edge case
[09:50] <jamespage> marked as so
[09:54] <Daviey> jamespage: Yeah, just for reference checked to see what apache2 does in the same situation
[09:54] <Daviey> It is better handled.. as in the pid is tracked.
[09:55] <jamespage> It would make a nice server papercut some someone to work on
[09:57] <Daviey> jamespage: good thinking!
[11:10]  * RoyK += 0xc0ffee
[12:17] <Ursinha> good morning people
[12:36] <jamespage> morning Ursinha
[13:02] <sommer1> Woo, Friday!
[14:15] <hallyn> Daviey: regarding MIR for 'gpxe or ipxe' - note that ipxe is now in debian.  Can we just sync that?
[14:16] <hallyn> Daviey: also, I don't understand why I don't show up on http://status.ubuntu.com/ubuntu-oneiric/group/topic-oneiric-server-overview.html
[14:31] <Daviey> hallyn: are you in ~ubuntu-server?
[14:32] <hallyn> ?
[14:32] <hallyn> Daviey: what do you mean?
[14:32] <Daviey> hallyn: so gpxe or ipxe is tricky... I've left that one for now, because ipxe is a fork.. and a better one IMO... but we need to make sure.
[14:32] <Daviey> hallyn: the ~ubuntu-server launchpad group
[14:32] <hallyn> oh
[14:32] <hallyn> pretty sure
[14:33] <Daviey> hallyn: i just added you
[14:33] <Daviey> but that might not be the cause
[14:33] <hallyn> debian went with ipxe, and was hung a long time on the gpxe ITP, so it seems like we should follow them.
[14:33] <hallyn> mind you i've not looked at the source at all
[14:33] <Daviey> hallyn: so the -overview topic isn't yet complete.  I think it is still missing some blueprints.
[14:34] <Daviey> hallyn: I'd prefer ipxe i think... especially as it is already a build dep of xen in main! :)
[14:34] <Daviey> but zul added support for gpxe in cobbler.
[14:34] <Daviey> So zul probabyl needs to update to ipxe for that to proceed.
[14:34] <zul> Daviey: yeah probably
[14:34] <hallyn> since ipxe is a fork, that shouldn't be too hard to port right?
[14:34] <zul> nah
[14:35] <zul> vi s/ipxe/gpxe/g :)
[14:35] <hallyn> Daviey: anyway, my prodding on the overview page was just bc I don't see any place where I was assigned to MIRs.  Not that I mind, I'd like to get lxc working with cgroup-bin today :)
[14:36] <Daviey> hallyn: Hmm
[14:36] <hallyn> so if ipxe is in debian, does a simple sync request suffice, or do we need an MIR still to get it into ubuntu?
[14:37] <Daviey> hallyn: is http://status.ubuntu.com/ubuntu-oneiric/u/serge-hallyn.html of use?
[14:38] <hallyn> Daviey: yeah I recon
[14:38] <hallyn> feh
[14:38] <hallyn> (feh @ MIRs I see there :)
[14:39] <Daviey> hallyn: Are they ok?
[14:39] <hallyn> odd that i apparently am not on the team
[14:39] <Daviey> hallyn: you are not leaving Ubuntu are you? :)
[14:39] <hallyn> Daviey: not that i know of :)  i'll let you know if iget an email later
[14:41] <hallyn> Daviey: well, they are in sid it seems.  So I guess it should be less than an hour (i hope) which hopefully is less time than i'll have left after finishing with cgroups :)
[14:42] <hallyn> "I'm this close"  -- lol
[14:43] <hallyn> you know what'd be nice?  a nice beefy amd box with huge disk, so I can test kvm bugs in a nested kvm environment
[14:44] <Daviey> hallyn: That can probably happen
[14:44] <hallyn> ?
[14:44] <Daviey> hallyn: access to a big box.
[14:44] <hallyn> davieylabs?
[14:45] <Daviey> lol
[14:46] <hallyn> :)  really, i pay amazon and rackspace for hosting, but they don't have what i need.  you'd think someone would...
[14:46] <Daviey> hallyn: yes... someone probably does... but someone needs to check their pm's :)
[14:47] <hallyn> Daviey: trying a new client, and it errs the opposite way from quassel
[14:47] <hallyn> can't find th eright balance between too few and too many notifications
[14:47] <Daviey> lol
[15:17] <RoyK> rotfl - just went by the oracle.com store, and they're offering 'MySQL Standard Edition Subscription (1-4 socket server) (1 Year)' for just $2000 a year...
[15:19] <patdk-wk> much better than oracle pricing
[15:19]  * RoyK goes to Sahara and sets up a bazar selling sand
[15:35] <omidp> can anyone help me with this page ? https://help.ubuntu.com/community/Xen
[15:38] <slhsen> hi, I have a pc with 4  500G sata hard disk. can i install ubuntu-server on one of them and create a raid 5 array on other tree?
[15:38] <patdk-wk> if you wanted too
[15:38] <slhsen> by the way i'm talking about software raid
[15:38] <mncl-core> if you wanted to, you could indeed
[15:38] <patdk-wk> or you could make all 4 of them a raid5, and install ubuntu on them
[15:38] <mncl-core> or you could make all of them raid-10 and get that extra performance;
[15:39] <patdk-wk> mncl-core, not really
[15:39] <patdk-wk> 4 disk raid10 vs raid5 is really the same
[15:39] <mncl-core> patdk-wk: how so
[15:39] <slhsen> mncl-core, and loose some disk space right?
[15:39] <patdk-wk> raid5 will beat it in fact
[15:40] <mncl-core> patdk-wk: i would like to disagree with that statement. and in fact you also gain extra redundancy
[15:41] <|rt|> raid5 will likely be slower at writes...but otherwise I'd expect them to be similar in speed
[15:41] <patdk-wk> only extreemly small random writes
[15:42] <slhsen> ok another question, lets say I created a raid 5 array with all four disks. and my os became unbootable somehow, will i still be recover my data?
[15:42] <mncl-core> rt: true, but to state that RAID-5 will beat RAID-10 in fact, is completely false.
[15:42] <slhsen> *able to recover
[15:42] <patdk-wk> the statement that raid10 will beat raid5 is false also
[15:42] <|rt|> mncl-core: without context of a workload I don't think you can make claims that any raid level is faster than any other raid level
[15:43] <patdk-wk> slhsen, you recover it the same way you would recover anything else
[15:43] <mncl-core> rt: the context was stated that he is attempting to Software RAID, or RAID 4x 500GB Hard Disks
[15:43] <mncl-core> my statement was simple, that he could use RAID-10 and he would get extra performance.
[15:44] <mncl-core> which is technically accurate.
[15:44] <|rt|> given enough disks if your only task is serving files over a 1 or 2 GBit links then at some point it doesn't matter...from a performance point of view
[15:44] <mncl-core> rt: ahhh very true; but on a technical perspective, there would be a performance gain in other operations; e.g. installing packages, OS Operations, and so on..
[15:45] <patdk-wk> given my workload, reads and writes of 50-100mb files, raid10 makes no sense, it's slow
[15:45] <patdk-wk> given my database, yes, raid10
[15:45] <slhsen> patdk-wk, yeah but, if I save one disk for system, i think i can just install os again and my data will be there
[15:45] <patdk-wk> given an olap store, raid10 is kind of pointless
[15:45] <|rt|> mncl-core: but with 4 drives I think they are about equal...you'll start to see a larger difference as the number of drives increases....and if your workload is sufficiently random
[15:45] <patdk-wk> slhsen you could do that anyways
[15:46] <slhsen> how so
[15:46] <patdk-wk> just run the installer
[15:46] <patdk-wk> it should notice yo uhave raid setup
[15:46] <mncl-core> rt: i have benchmarked extensively RAID-10, true that workload must be considered and increase of disks. but it remains that even with 4x Disks the performance is increased.
[15:47] <mncl-core> plus the extra redundancy aspect
[15:47] <mncl-core> should be considered
[15:47] <|rt|> on my server here (hardware raid with 12 drives) I did a ton of benchmarks trying to decide between raid 10 and raid 6...at the end of the day the difference was 1GB/s vs 700MB/s
[15:48] <mncl-core> rt: well i have several RAID-10 SANs for ESX Clusters and I also had a hard time deciding between RAID-6 and 10
[15:48] <|rt|> I decided to go with RAID6 and get the slightly better data protection and storage at the cost of the 300MB/s
[15:48] <mncl-core> but ultimately RAID-10 destroyed RAID-5 and 6
[15:48] <patdk-wk> the issue with raid5 has always been what happens when a disk is dead
[15:48] <patdk-wk> not how it performes when it's 100% functional
[15:48] <mncl-core> mind you i am using 24x WD Blacks
[15:49] <|rt|> yeah the difference in Mean Time between Data Loss of raid 5 and raid6 and raid10 is hugh
[15:49] <mncl-core> and real adaptec raid controllers
[15:49] <|rt|> huge
[15:50] <|rt|> RAID10 rebuild times are the fastest and a dead drive impacts it the least though
[15:51] <patdk-wk> if my san grows anymore
[15:52] <patdk-wk> I'm going switch it from raid10 to raid 51
[15:52] <patdk-wk> hmm, maybe that would be raid15 :)
[15:52] <patdk-wk> a raid5 of many raid1's
[15:52] <mncl-core> it should also be considered that the more drives you had, the greater probability for simultaneous drive failures.
[15:53] <patdk-wk> ya, raid15 or raid150
[15:55] <|rt|> some of the zfs based NAS/SAN solutions are pretty nice...like Nexenta
[15:55] <|rt|> ssd cache devices really speeds them up too
[15:56] <mncl-core> rt: cool find. will look into more, looks interesting. have been using openfiler for a while now
[15:57] <|rt|> a good number of the ex solaris devs work for Nexenta now
[15:58] <mncl-core> very cool, seems like it is well maintained and supports emulex targets
[15:59] <|rt|> some of the gluster stuff intrigues me too...but it's a very different approach
[15:59] <mncl-core> gluster is a very cool product
[16:00] <mncl-core> reliable, and when you mix it with an openqrm environment, things can get very interesting
[19:06] <zul> hggdh: hey did you get anywhere with bsdtar yet
[19:09] <tarrasch> I want my Ubuntu Server to always run a program (in my case a webserver), preferably as a seperate user. So whats the easiest way to add ever-running services?
[19:11] <patdk-wk> normally, the normal way, and let the service change uid itself
[19:11] <tarrasch> Basically the service should fire up by itself at startup, and I should be able to restart it when the server is up.
[19:11] <tarrasch> so i use my own user that starts a program that switches to another user and keeps running the program I intended?
[19:12] <patdk-wk> hmm, root starts programs at startup
[19:12] <tarrasch> oh ok
[19:12] <patdk-wk> and no, normally the program itself that you start, switchs itself
[19:13] <patdk-wk> if it doesn't, I might have issues with it's lack of security
[19:13] <tarrasch> I know my server already startsup apache, I guess I want something very similiar for myself only for my own webserver
[19:14] <patdk-wk> well, apache config file you set the user and group name
[19:14] <patdk-wk> once it gets started, it changes itself
[19:14] <patdk-wk> that is what your program should do
[19:14] <tarrasch> patdk-wk, you mean that the webserver Ive written (not me really) should at its initialization switch to a dedicated user?
[19:15] <patdk-wk> not at initialization, but normally after that
[19:15] <patdk-wk> cause normally they want to bind to port 80, or read ssl keys, that need root
[19:16] <tarrasch> hmm, I'm gonna run at port 3000+, that doesnt need root, no?
[19:16] <patdk-wk> no, ports <1024 need root
[19:16] <tarrasch> I'll let my apache-server reverseproxy to the other web-server running at 3000
[19:17] <tarrasch> And I've run the server "manually" myself succesfully as my own user (not root); and it works fine. Now I only need to "autorun" my program at startup etc.
[19:18] <tarrasch> since my progrma is simple I won't need root. Is there a simplified process for setting up services that don't need any root?
[19:18] <patdk-wk> guess use su then to start it
[19:19] <patdk-wk> su some-user -s someprogram
[19:19] <patdk-wk> maybe?
[19:19] <patdk-wk> such a hack :)
[19:19] <tarrasch> Sounds great, though I still would like to 1. startup as the system boots up. 2. be able to restart it while running. Any ideas?
[19:21] <tarrasch> I've noticed the 'services' functionality, it seems good though a bit much work to setup something quite simple
[19:23] <tarrasch> 'su designated_user -s /my/web/server' Seems to work otherwise
[19:27] <x404x> How to replace a failed drive on a 3ware 9500 controller ?
[19:28] <x404x> I want it to recognize the drive on the same port the bad one uses
[19:36] <patdk-wk> disconnect old drive
[19:37] <patdk-wk> plug in new drive
[19:37] <patdk-wk> tell 9500 to use new drive as hotspare?
[19:46] <hggdh> zul: not yet, but I finished witht the kernels for this week; I will get back to it now
[19:47] <zul> hggdh: cool
[20:09] <qman__> I'm having a problem with SSH and key authentication
[20:09] <qman__> for the purpose of making tunnels, I have one user on an SSH server, and multiple clients, each with their own key to authenticate as said user
[20:10] <qman__> this worked great until recently when I started adding more clients, and it's rejecting my keys
[20:11] <qman__> what configuration do I need to make on the SSH server to allow more concurrent connections with different keys as the same user?
[20:15] <hggdh> zul: I do not see many options -- why was this link created, in the first place? It wasn't there before A1
[20:16] <zul> hggdh: no idea
[20:16] <hggdh> who would know?
[20:16] <zul> cjwatson: maybe
[20:17] <hggdh> s langasek says the link can be there, as far as ISO is concerned
[20:22] <qman__> ah, nevermind, it was a permission issue buried underneath that
[20:28] <Aison> funny, how do I install ubuntu server on a HP DataVault X510 ;)  there's no screen
[20:28] <Aison> maybe I should install the harddrive on a desktop machine and then plug it into the hp device?
[21:02] <maco> i just installed mysql-server on my lucid server. i gave it a nice long complicated root password, which i copied and pasted from keepass. then i tried to login with "mysql -u root -p" and pasted in the same nice long complicated root password for mysql...and it keeps giving me denied
[21:03] <mncl-core> of course
[21:03] <mncl-core> keypass is greate
[21:03] <maco> is it possible there is some parse-fail going on with the nice long complicated password?
[21:03] <mncl-core> but you cannot use special characters
[21:03] <maco> there we go
[21:03] <mncl-core> and you cannot use 25 characters' common mistake i've done many times before
[21:03] <maco> so how special are they allowed to be?
[21:03] <mncl-core> go with upper-lower-num 18 characters
[21:03] <maco> or is it alphanumeric only?
[21:04] <maco> ok
[21:04] <maco> thanks
[21:06] <maco> i frequently wish keepass had a "kinda special" and a "really special" checkbox, for those sites that allow @#$%^&*() but not ` " ' £
[21:37] <wmoxam> I can't seem to find xfs_freeze in Oneiric
[21:37] <wmoxam> is it moved to a package?
[21:38] <wmoxam> nm
[21:38] <wmoxam> found it
[21:44] <RoAkSoAx> zul: ping?
[21:46] <RoAkSoAx> SpamapS: you might be able to help me.. does it make any sense to tell dh_install init to install the script with defaults, but disabled? (so that daemon never starts on boot unless otherwise specified?)
[22:15] <SpamapS> RoAkSoAx: it should use some method *in the script* to be disabled
[22:16] <SpamapS> RoAkSoAx: like checking /etc/default/xxxx
[22:16] <SpamapS> RoAkSoAx: it should still run, but not start
[22:16]  * SpamapS prepares to leave for the airport
[22:41] <cjwatson> hggdh: what link is this?
[22:42] <cjwatson> oh, I see on #ubuntu-release, bah @ conversations across multiple channels
[23:55] <lcb> hi. any good phone answering system for 11.04 server? just in case someone knows...