=== Nigel_ is now known as G [02:47] hello [02:49] just tried installing 13.04 with kvm, openvswitch and pci passthrough of a SAS controller using kvm managed mode [02:49] no luck on the last part tho === SpamapS_ is now known as SpamapS [04:00] morning [04:01] Is there a good way to clone ubuntu server instances? [04:01] moving from us servers to aus servers and provider doesn't have a way to clone instances across datacenters === hachre- is now known as hachre === Ursinha is now known as Ursinha-afk === Psi-Jack_ is now known as Psi-Jack [04:34] can the installer for Ubuntu Server 12.04 walk me through setting up a mirrored RAID? === thumper is now known as thumper-afk === Ursinha-afk is now known as Ursinha [06:27] I've got Ubuntu 12.04.2 and while startign drbd service I get "DRBD module version: 8.3.13 [06:27] userland version: 8.3.11 [06:27] you should upgrade your drbd tools!" multiple times. [06:34] I found out that variable DRBD_DONT_WARN_ON_VERSION_MISMATCH= might help on this, but how should I use it? === schmidtm_ is now known as schmidtm === smb` is now known as smb === Ursinha is now known as Ursinha-afk [07:30] I solved it. === jussi01 is now known as jussi [08:25] guzzlefry: yes, see serverguide advanced installation. === Ursinha-afk is now known as Ursinha === mld_ is now known as mld [10:33] I'm added myself to the www:data group which owns /var/www but i'm unable to change file permissions on a style sheet in a subdirectory of /var/www. i've checked and all directories in www are owned by www:data with g+w please help! [10:34] Um you can't have a colon in a group name - what do you actually mean? [10:35] lperry65: check the ownership on the files as well. [10:36] sorry should have said www-data [10:36] file permissions are -rwxrwxr-- [10:37] file is owned by www-data: www-data [10:37] Also, being in the group associated with a file does not grant you permissions to change permissions [10:37] Only the owning user can do that [10:37] ah i see, so I could edit the file but not change the permissions? ok [10:38] thank you :) [11:01] zul, http://pad.ubuntu.com/havana-2-problems [11:01] when you start :-) [11:19] jamespage: what if i never want to start? :) [11:19] * jamespage digs zul out of his hole [11:20] zul, hey! [11:20] ill take a look at cinder first...the nova one is easier [11:20] zul, we have a mismatched novnc/websockify causing the nova issue [11:20] zul, I think that is just a sync [11:20] jamespage: yeah cjwatson pointed that out last night [11:20] zul, the cinder issue is i386 only - but I've no idea bout greenlet [11:20] jamespage: ack [11:21] jamespage: i think its a sqlalchemy issue [11:21] zul, probably [11:21] * zul shakes his fist [11:25] zul, python-ceilometerclient is holding up the keystone client migration - looking at that now [11:25] er? [11:27] zul, pkg_resources.DistributionNotFound: python-keystoneclient>=0.2,<0.3 [11:27] jamespage: fudge [11:27] zul, python-keystoneclient - 1:0.3.1-0ubuntu1 [11:27] #bang [11:28] zul, trunk is OK - we might need to push a snapshot [11:28] nah just bump the requirements.txt file [11:29] zul, can you push your uploaded package changelog to the branch please [11:29] jamespage: sure [11:29] jamespage: lemme finish breakfast first [11:29] zul, sure [11:30] :-) [11:36] zul, nm - I see your branch - I'll merge in and update [11:36] cool [11:54] zul: https://code.launchpad.net/~james-page/python-ceilometerclient/fixup-keystone/+merge/176173 [11:54] jamespage: +1 [11:55] jamespage: just starting to fix cinder [11:55] zul, coolio [11:56] zul, taking a break for lunch [11:56] jamespage: ok hopefully i will have something cinder related for you after lunch === gary_poster|away is now known as gary_poster === gary_poster is now known as gary_poster|away === gary_poster|away is now known as gary_poster [12:34] Hi [12:34] when installing Ubuntu Server on a drive other then sda do I need to tell the master boot record to install to the drive (sde1) I have setup to mount at / or should installer do it automatically? [12:41] Beatstreet: it will ask you later, the default will be auto-detected and should be pre-set correctly to /dev/sde, but you can change it to something else, if you want to. [12:47] do I point it to /dev/sde1 (the partition) or /dev/sde (the drive)? [12:54] hi all [13:16] jamespage: we should be able to sync novnc === resno_ is now known as resno === medberry is now known as med_ [13:33] i'm updating a server with "apt-get update" and 2 packages are kept "back"...i know i can manually update that but why they arent done along with the rest? [13:33] StathisA: You'd need to use apt-get dist-upgrade [13:33] !dist-upgrade [13:33] 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. [13:33] ah thanks a lot! [13:39] jamespage: interesting... [13:39] * zul shakes his fist at sqlalchemy [13:41] * koolhead11 watches zul dancing :P [13:42] jamespage: fixed locally...need new greenlet [13:42] zul, most excellent! [13:42] zul, thought it might be something todo with that version bump [13:44] jamespage: i need to add a patch that bumps the sqlalchemy in the requirements.txt though [13:46] jamespage: http://pastebin.ubuntu.com/5900633/ [13:50] Would it be ok to ask configuration questions on this channel? [14:01] !ask [14:01] Please don't ask to ask a question, simply ask the question (all on ONE line and in the channel, so that others can read and follow it easily). If anyone knows the answer they will most likely reply. :-) See also !patience [14:05] jamespage: ping greenlet has been uploaded can you +1 https://code.launchpad.net/~zulcss/cinder/cinder-alchemy/+merge/176202 please === Guest50507 is now known as joe1234 [14:21] Anyone know how to configure forward commerce to work on apache? [14:22] jamespage: fyi horizon needed a new compressor version i think have it working now [14:27] Or if no one knows forward commerce, then how would one go about making a site accessible to the world rather than accessing it locally (this is for testing purposes BTW)? [14:42] Daviey: ping [14:44] zul: hey [14:44] Daviey: mind promoting heatclient https://bugs.launchpad.net/ubuntu/+source/python-heatclient/+bug/1203122 === caribou_ is now known as Caribou === BTCOxygen is now known as Guest20487 [15:46] jamespage: before you go https://code.launchpad.net/~zulcss/horizon/2013.2.b2/+merge/176230 [15:47] zul, not got just yet - looking now [15:47] jamespage: thanks [15:58] zul: + - Add python-neatclient [15:58] are you sure :-) [15:59] jamespage: but it was neat ;) [15:59] zul, I think some of the generated assets are also missing [15:59] jamespage: really? [16:00] Hi, how do I add the 12.04 repos in this link to sources.list?http://repo.zabbix.com/zabbix/2.0/ubuntu/ [16:00] how can you tell? === wxl_ is now known as wxl [16:33] would you guys recommend additional backups before updating linux-image-generic? (no fancy stuff on the server, just postfix/dovecot, mysql, apache, php , ruby on rails) [16:33] I recommend regular backups for anything you care about [16:39] zul, how are well looking now? [16:39] jamespage: pretty good...just need to work on ceilometer and heat [16:40] jamespage: for heat im just going to get it to build im not going to rip out the dbconfig-common parts (yet) [16:40] zul, ah - cinder is ftbfs [16:40] lemme look at that [16:41] jamespage: arrgh [16:49] zul, builds fine locally on i386 and amd64 [16:49] * jamespage hits the button of despair in LP [16:49] interesting [16:53] zul, meh - built this time [16:55] jamespage: lovely [16:55] Madkiss: howdy! [16:55] Madkiss: do the tests for libqb pass in debian? [16:56] jamespage: ill just upload horizon [16:59] Hey guys, I am running on 12.04 64bit. I want to install Apache 2.4.x which is the latest but 12.04 LTS repository provides 2.2.x version. Is there ready package for Ubuntu for apache 2.4.x? I don't want to recompile Apache if there is any security fix release. [17:00] No. If you install 2.4 on 12.04 you're on your own. [17:00] The Ubuntu packages will get security updates. [17:00] vedic: not in the official archive, 2.2.22 only [17:00] vedic: 2.4 apache transition involves a lot of different moving pieces. afaik there's nothing ready to download and try; 13.10 will be the first with apache 2.4, and backports feel unlikely. [17:01] pmatulis: 2.2.22 + patches. [17:01] zul, ack [17:01] ScottK: for sure, yeah [17:02] sarnold: I see. Then in that case, how do people who use production server manage security or bug fix release of apache? Do they remove source install and then recompile it? [17:02] vedic: most will be content to run apache 2.2 until 14.04 LTS is released. [17:02] Absolutely. [17:03] sarnold: Ah, thats long wait? [17:03] vedic: you would not believe how close it feels. :) [17:03] sarnold: any expected dates yet? [17:04] sarnold: LTS is coming....LTS is coming... [17:04] vedic: april 2014? :) [17:04] * zul shakes [17:04] zul: that's right out of some horror film, right? :) [17:04] sarnold: simpsons.. [17:04] sarnold: cant sleep...clown will eat me [17:04] lol [17:06] Waiting for 14.04. Supporting phones, tv etc [17:16] jamespage, i'm sure you know this.. [17:16] a .config script. [17:16] am i allowed to assume dependencies are installed ? === joseph is now known as joe1234 [17:28] No. Unpacked, not fully installed. [17:29] ScottK, so ... if i want to read a yaml config file. [17:29] i have to do that with sed and awk ? [17:29] that sucks. [17:30] (or some other 'essential' i guess) [17:30] Does the config run during package install (I was assuming that was the case)? === ogra_` is now known as ogra_ [17:40] ScottK, please excuse my grave ignorance. [17:40] thi sis something i've never really fully grokced [17:40] No problem. What are you trying to do? [17:40] the postinst does handle $1 == configure [17:40] and uses db-get [17:41] so is that a yes to "does the config run during a package install" [17:41] Yes, so you can only assume depends are unpacked, not fully installed. [17:45] so cannot use python then at that point ? [17:46] ScottK, i was wanting to read a yaml config file "correctly" 9not depending on formatting [17:46] like this [17:46] http://paste.ubuntu.com/5901317/ [17:47] hi. i want to make vpn server to connect to other sites via vpn (cisco, microsoft vpn..) can someone point me to the right direction for google? im a bit lost. basicly what i need is that i could connect to 5 vpns at same time? [17:49] anyone know when the next LTS release after 12.04 will be? [17:49] guessing 2014 [17:50] NomadJim: afaik the plan is still for 14.04 [17:50] cool [17:53] saban: some of the VPN implementations I've seen are very poor.. I wouldn't be at all surprised if at least one of them throws a complete fit with the others installed and running. [17:53] saban: ideally, you could just configure them all and have everything Just Work, but some people try to be too clever for their own good.. [17:56] smoser: I think you'd need to pre-depend on python-yaml. [17:56] sarnold: the only way that our nagios server can control sites is true vpn. everything else is blocked. but there are so many vpn (no site has the same vpn...) so i was tinking about dedicated "vpn server" to connect to all vpns and put in on esxi. im googleing about this and just cant find any good article about this :P [17:57] saban: hah, that's a clever idea. :) [17:59] can anyone recommend a good tool for scheduled backups to a ftp backup space? [17:59] people seem to like bacula [18:02] Ahti333: also investigate duplicity [18:02] Ahti333: here we are: http://duplicity.nongnu.org/ [18:03] bacula looks pretty complicated, i really only need something like "back everything except /tmp up to this ftp host, full backup every weekend, incremental on weekdays or sth like that [18:08] sarnold how does duplicity handle cross device directories? (/ is one volume, /home is another one) === Nafallo_ is now known as Nafallo [18:09] Ahti333: I've not set it up myself, it's still on my todo list :) hehe [18:09] oh okay :) it loos just like what i need though :D === Ursinha is now known as Ursinha-afk [18:10] hallyn: hi! does kvm work well for you on a saucy host? [18:11] hallyn: with the 3.10 kernel on my x201s, it hard locks my system as soon as I fire up a vm [18:11] 3.8.0-26 and earlier are fine [18:21] jdstrand: I'm using it nested - saucy vm is running kvm with no problems [18:22] (my saucy laptop can't very well run kvm - or anything - without overheating and shutting down so i haven't tested that yet) [18:22] hallyn: well, it totally hard locks here [18:22] I have to use 3.8 [18:22] jdstrand: so you boot a 3.8 kernel, and run newest qemu-kvm, and it works? [18:22] yes [18:23] please do file a bug (kernel also affecting qemu...) no idea what htat would be [18:23] hallyn: I also downgraded qemu-kvm and tried with the 3.10 kernel, and same issue [18:24] Subject: commit f8f559422b (KVM: MMU: fast invalidate all mmio sptes) causes hang [18:24] (on lkml this weekend) [18:24] the problem is that it crashes so hard I don't get a trace. I do get NUL in the kern.log though [18:24] so at least there's that :\ [18:26] hm, that thread says it's amd-only - you don't roll low-rent like that [18:26] hallyn: this is a host lockup btw, not guest [18:26] heh, this is an i7 [18:26] wowzers [18:27] hallyn: also, fyi, appears 3.10 doesn't support nested=1 [18:27] ??? [18:27] I had to remove that from /etc/modprobe.d/qemu* to load the kvm_intel module [18:28] weird, modinfo /lib/modules/3.10.0-4-generic/kernel/arch/x86/kvm/kvm-intel.ko says it is supported [18:29] jdstrand: any chance you have some weird case of module not matching kernel? [18:29] not sure how that would happen [18:29] I updated today [18:29] (note the host lockup issue happened with 3.10.0-2 too) [18:30] let me reboot into 3.10.0-4 so I can report the lockup and see what it says again [18:32] hallyn: [18:32] $ cat /proc/version_signature [18:32] Ubuntu 3.10.0-4.13-generic 3.10.1 [18:32] $ sudo modprobe kvm_intel [18:32] ERROR: could not insert 'kvm_intel': Unknown symbol in module, or unknown parameter (see dmesg) [18:33] 3.10.2 is out btw [18:33] [ 106.814382] kvm_intel: Unknown parameter `nested' [18:34] DYSW: ah, in 3.10.0-5.14 I see "rebase to v3.10.2" [18:34] * jdstrand tries [18:36] jdstrand: nested= definately still exists in arch/x86/kvm/vmx.c... [18:37] yeah, modinfo says it is supported too [18:37] filename: /lib/modules/3.10.0-4-generic/kernel/arch/x86/kvm/kvm-intel.ko [18:37] parm: nested:bool [18:37] I have an older kernel though, let me update [18:37] Hey all, can someone help me troubleshoot a fresh install of Precise amd64 that won't boot after a kernel upgrade? [18:38] Additionally, can someone help me with why my grub is completely fscked up when displaying? [18:38] DWSR: on hardware or vps ? did you use a non standard kernel and updated grub when done ? [18:39] DYSW: I did a dist-upgrade off of a fresh install, so it would have been using the standard ubuntu kernel. [18:39] And I assume that a dist-upgrade deals with all of the neccessary changes to GRUB. [18:39] sure it would do that itself [18:39] So basically. [18:40] If its on hardware, a log could help [18:40] lol. [18:40] I guess that whole "I can't boot" thing doesn't really mean much to you? [18:40] Xen, vps can be killed with apt-get dist-upgrade so, its important we know if its virtual or not [18:41] lol. it does, but ubuntu boot menu has the option of using an earlier kernel or recue etc. from there you can get in and see the logs [18:41] I can't actually SEE the options. [18:41] Sec, I'll post a pic. [18:42] But first thing i would do, is check iso hashes, reburn and try, or reinstall to usb and boot. [18:42] DWSR: I think some releases require you to hold down left shift or something while booting to get grub menu :/ [18:42] I have the grub menu. [18:42] Again, I'll show you in a second with a picture. [18:42] Since, a thousand words and all that. [18:43] https://www.dropbox.com/sc/labf5h6z4evioxj/OZ6G0MKNRp [18:43] I will note that the BIOS menu displays correctly. [18:44] And that Ubuntu didn't have display issues when I was in the install originally, and the LiveUSB didn't have an issue either. [18:44] So, NOT HARDWARE RELATEd. [18:44] Your screen is off. I would try all the options. the one with (mode) in it first [18:44] No, it's not. [18:45] DWSR: hah, that's a good one, never seen that.. [18:45] well, I haven't seen that since CRT days.. [18:45] i cant see any words on the screen just (mode) at the ned of a line, all to the left is black. [18:45] Assuming that the hardware is at fault is quite frankly, ignorant. [18:45] The panel is completely fine and displays correctly EVERYTHING ELSE from this computer AND from another. [18:45] just looks like the screen is not centered. Doesnt matter. just try all the 4 options. It wont take much time [18:45] The panel is fine. [18:46] Ok, getting a bunch of spew regarding my drives. [18:46] not hardware related, your setup is fine, but the screen can still be offcentered [18:46] Which is a change from before. [18:46] There we go, it's mounting /. [18:46] Ok, that (mode) line was the recovery option. [18:47] i know [18:47] So I'm in the recovery menu. [18:47] It wouldn't actually make it here before. [18:47] So, yay progress. [18:47] see hope [18:47] I guess first order of business is unbreak system. [18:47] Should I just try the dpkg option? [18:48] Not sure, i have never needed to use the recovery option. [18:48] welp, I'm trying a fsck first. [18:49] And after that I'll try the dpkg option. [18:49] The reboot was graceful, but I've done a couple of hard poweroffs (presumably) in the middle of the boot process, so the FS might be asploded. [18:49] And the fsck just....froze? [18:50] how long should a 300GB fsck take? [18:50] also, can I just say that fsck is such a hilarious shortening. [18:53] DWSR: Depends on the filesystem, and even more on the content, it can take several hours. [18:54] bekks: Mostly empty space. [18:54] It's kind of fine though [18:55] DWSR: Then just let the fsck run until it finishes. [18:55] I sent a ^C and now the system boots just fine. [18:55] You cancelled the fsck run. [18:55] yeah. [18:56] So you have to run it again. [18:56] You're assuming, of course, that it's required. [18:56] It is, in fact. [18:57] Why? [18:57] The system's booted currently. [18:57] Because your filesystem still may be corrupted. [18:58] Yeah, acceptable risk until I figure out my display issues. [18:58] Booting does not tell you anything about the health status of a filesystem. The only thing thats obvious now is, that it isnt corrupted that severe that it wont boot anymore. [18:58] Yep. [19:00] Let's just take it as read that I understand and accept the risks, ok? [19:05] I'm now having display issues in GRUB and Ubuntu. The panel I'm connected to is a 1080p panel, but doesn't appear to be detected as such [19:09] DWSR: search for 'gfxmode' in the grub info documentation; you might be able to force a graphics mode that does work with your display [19:11] sarnold: What about inside of Ubuntu? === JanC_ is now known as JanC [19:14] DWSR: that'll require fiddling with X configuration, either xrandr or Xorg.conf... [19:14] nox. [19:15] sarnold: No X installed. [19:16] DWSR: ah, then the kernel's Documentation/kernel-parameters.txt file may have some guidance, the vga= or video= options may help [19:17] sarnold: The panel is detecting the video signal as 1400x1050. [19:20] DWSR: oh, yes, the kernel does fancier things these days.. you can also try adding 'nomodeset' to the kernel command line. Or, force the kernel mode setting as described https://wiki.archlinux.org/index.php/Kernel_Mode_Setting [19:24] Patrickdk you around? http://pastebin.com/YPA7X0Ym [19:25] dunno what I'm looking at [19:27] Some data during another "blip". [19:27] Run queue spikes, load is low, cpus accross the board drop in usage.... [19:28] Even systat had problems collecting the data, as you can see by the timestamps. One is 39 SECONDS late [19:28] It's like the machine decides to take a lil breather [19:28] jsonperl, note the cpu usage during those spikes? [19:29] whree cpu usage is normally spread out, with >10% usage over many cores [19:29] in the spikes, it's limited to one core [19:29] Then it's all in the single dig [19:29] yea [19:29] defently a locking/mutex issue [19:29] Could you explain w a small example? [19:30] something is holding a lock around stuff, while it works, and is keeping it for an extended period of time [19:30] where normally it wouldn't [19:30] At the os level [19:30] don't know [19:30] What types of things lock like that? [19:30] but since you said you have like 10-20 of the program running [19:30] I would have to say so [19:30] Yep 14 [19:30] me2 [19:31] It's definitely not overheating? That wouldn't act like that? [19:31] jsonperl: No. [19:31] oiwait doesn't go up, so not a disk [19:31] Thermal throttling acts much differently. [19:32] From a hardware perspective, it just clocks down the processor until the processor stops overheating. [19:33] Which means that you wouldn't see a sudden shift from 4 cores to 1. [19:33] Advantages to using an LVM for a single disk? [19:34] many :) [19:34] dynamic repartition, snapshotting [19:34] reasons to not use it, the same :) [19:34] I've had very very good experience with LVM [19:34] even with hardware raid [19:34] lvm can be ok, but it can get in the way also [19:34] Ick hardware raid. [19:35] well, they haven't started making bbwc for soft-raid yet [19:36] Unless you're dropping a few bills for good cards, you might as well be using soft-raid. [19:36] HW raid gets a bad rap from crap onboard controllers [19:37] onboard controllers is not hw-raid [19:37] Anyone in here want to buy 8x4GB DDR2 ECC RAM? [19:37] I really haven't a clue where I would list something like this. [19:37] ebay [19:37] Yea I can vouch for ebay [19:38] patdk-wk: No, but HW raid requires controllers. Hence why HW-raid gets the bad rep. [19:38] heh? [19:38] s/requires controllers/requires good controllers/ [19:39] You don't get good controllers for <$100. [19:39] hardware raid gets a bad rap, cause of the people making it keep producing incompatable firmware and randomly adding/removing features from the firmware [19:39] I'm talking >$700 [19:39] patdk-wk: Never been so flush with cash as to be playing with that kind of hardware. [19:39] crap raid cards are <$400 [19:40] patdk-wk: No, crap is <$200. Cheap is <$400. [19:40] <$100 has to be software raid [19:40] I thought hardware raid mostly exsisted because more computers couldnt handle the load of a software raid but over the eyears as power increase software became more reliable [19:40] hardware raid exists for two reasons [19:40] I think I got LSI's in the 300 range and they've been completely fine [19:40] Beatstreet: That's certainly been part of it. Hardware, done right, is better. [19:40] you need it to boot [19:40] or you need it for bbwc [19:41] patdk-wk: You can boot from softraid. =\ [19:41] DWSR, you can? on every os? on every motherboard? [19:41] patdk-wk: Can you think of an example that you could demonstrate server level locking [19:41] what happens when grub gets screwed up on your drive? [19:41] I agree that hardware done right is better but the money involved to get it right is a lot [19:41] the bios booted the first disk and won't try the second [19:42] system won't boot [19:42] patdk-wk: can is binary. It doesn't have to work in every case, only 1 to disprove your statement. [19:42] patdk-wk: I never claimed it was a good idea. Merely that it was possible. [19:42] yes, but hardware raid is known to work :) [19:42] software is more of a, make sure you test it good [19:43] patdk-wk: Back to my, it does work, but isn't recommended. [19:43] Also, ~$200 LSIs are great for home use. [19:43] Cheap != crap. [19:43] $200 lsi's don't support bbwc [19:44] You don't need BBWC if you configure your RAID correctly. [19:44] yes you do [19:44] Won't argue it's better to have, but you don't need it. [19:44] No, you don't. [19:44] anyone set up or maintain ipsec tunnels between ubuntu and pfsense? [19:45] Better? Yes. Neccessary, no. [19:45] if I wasn't going use a bbwc, I would just use softraid [19:45] then I have complete flexability over moving my raid to any other system [19:45] patdk-wk: That's a personal choice. You can configure LSI cards to always flush cache. [19:46] It's not neccessary to have a WC at all. [19:46] the lsi card your talking about has no cache [19:46] you need a real raid card for cache, and those support bbwc [19:46] patdk-wk: Correct. I'm talking about disk cache. [19:46] patdk-wk: You seem to be confusing neccessary with optimal. [19:47] These things work, and work just fine, in the manner in which I've specified. [19:47] Whether or not performance levels are to your personal definition of satisfactory are a COMPLETELY different subject. [19:47] not with the latency requirements I need, therefor required :) [19:48] patdk-wk: Since you mentioned $700 cards, that's probably true. [19:48] But just remember "This doesn't work" is a very binary statement. [19:48] And that this is a channel for server support and that covers all manner of use cases. [19:49] For example, like home servers such as the one I'm currently reinstalling because, for the love of all things digital, I don't understand why dist-upgrade should BREAK an installation. [19:49] it doesn't [19:49] It does. [19:49] Because I just had it happy. [19:49] happen* [19:49] no, something else broke it [19:49] dist-upgrade just caused you to notice it [19:50] running a RAID 5 on 3 drives. Is it better to create a small partition on each drive and RAID that for an OS partition then just putting the OS on a partition on the first drive? I've lost sda before and box wont boot with OS RAID'd across all 3 [19:50] Oh god no. [19:50] You should get a small drive and have /boot on it. [19:50] Even use a flash drive if that's all you have laying around. [19:50] RAID, in whatever form you're using it, should always have access to the entire disk. [19:51] flash drive would be a really slow running OS would it not? [19:51] Hence why I said /boot. [19:51] I guess I need to read up on the /boot thing [19:52] Beatstreet: Basically, you can have your boot partition (/boot) installed on a seperate device from your root partition. [19:52] Meaning you can pull fun shenanigans like make your system unbootable unless you have your flash drive in. [19:52] so the /boot on the flash drive just boots it and it not accessed after that? [19:53] Beatstreet: Not unless you're changing the contents of it for some reason. [19:53] Which would be only during a kernel upgrade, or a GRUB upgrade, or something similar. [19:53]