[00:00] as a kernel parameter? [00:00] menu.lst? [00:01] paranoidi: just add it to the bootup [00:01] F6 if on the installer [00:01] it's already running system [00:01] i thought that was the old grub? [00:01] and yes, still running old grub [00:01] ah ok :) [00:02] \o/ [00:02] the new one looks much more confusing :) [00:02] paranoidi: just add it to the kernel line [00:02] RoyK: I'll try next time I boot, time for sleep soon [00:02] i never understood how you can go from easy understandable to such a crap as the new grub.... [00:03] yeah, I lost all desire to switch after reading about it for 10 minutes [00:04] one of the great points of linux for me was always the readable config files [00:05] anyway, this graphics mode is least of my worries, much worse is the fact that my crappy RocketRaid 2340 is not working anymore since some moron (me) accidentally uninstalled the kernel it was working with .. [00:05] I had about 10 different kernel versions so I thought to clean up a bit .. [00:05] well new grub has it too but totally complicated plus you have to run update-grub every time you change something... :( [00:06] \o/ for uninstalling stuff and wishing later one had just shot himself in the foot [00:06] I was under assumption that it would've been supported with latest kernel instead of that propietary crap but apparently not .. so I'm SOL [00:07] and I'm 99% sure it will not go into latest kernel [00:08] yes that happened to me with my old old motherboard. had a raid controller with a horrible binary blob driver. then came kernel 2.6 and they dropped support [00:08] * WinstonSmith was really pleased [00:09] I kind of feel bad for driver developers though since kernel api is changing so rapidly compared to windows .. [00:10] what's even worse this card has possibly a bug that non-raid drives (called "legacy") are corrupted in certain places of the disk [00:10] heh [00:10] and it's in the bios .. and latest is from 2007 === skrewler_ is now known as skrewler [00:50] SpamapS: hey another lxc q [00:50] SpamapS: can a 32 bit lucid be contained on a 64 bit natty ? [00:51] lifeless: it should work, since it is little more than a wrapper around the syscalls, and the 32-bit binaries work fine on 64-bit systems. [00:52] lifeless: I could see there being holes where somebody gets "cute" and thinks their on a 64-bit system and does the wrong thing in userland.. but that seems like a corner case that could be worked around. [00:53] lifeless: basically what works in chroots, works in lxc [00:55] lifeless: building one will take modifications to the templates it seems [00:55] arch=$(arch) [00:56] is this worth a bug, or will you JFDI fix it ? [00:56] I'm running 11.04 Server LTS. Has anyone seen the error when trying to restart nmbd? I had it working fine on my last install before I broke it and had to re-install (linux noobie here) [00:58] lifeless: its non trivial so a bug would be good, I'm not sure what options there are for passing arguments into the templates. [00:58] lxc-create ? [00:58] disorbtion1: you mean 10.04 LTS, or 11.04 .. 11.04 is not an LTS [00:58] bah [00:58] https://bugs.launchpad.net/ubuntu/+source/lxc ? [00:58] lifeless: yes [00:59] lifeless: /usr/lib/lxc/templates is where they live [01:00] yes sorry 10.04 LTS I was running 11.04 the first time (nonLTS) and everything was fine, blew it up and re-installed used LTS since I figured it would probably be more stable... [01:00] They seem a bit whacky to me... all the ubuntu ones should be nearly identical for 90% of the code.. probably just copy and pasted [01:00] disorbtion1: you can't restart something that isn't running [01:00] disorbtion1: service stop nmbd ; service start nmbd [01:00] SpamapS: https://bugs.launchpad.net/ubuntu/+source/lxc/+bug/798476 [01:00] Launchpad bug 798476 in lxc "it would be really useful to be able to make i386 lxc containers on a amd64 machine" [Undecided,New] [01:01] lifeless: indeed, I'll take a look at it tomorrow.. just about EOD here [01:01] cool [01:01] that arch=i386 is the only thing needed in the template? [01:01] (tomorrow meaning later on when the family is asleep and I "can't sleep") ;) [01:01] lifeless: yep [01:01] lifeless: so you could probably just make an lxc-lucid-i386 [01:02] which is probably what most people do [01:02] anyway, have to run.. good luck [01:02] it looks like it won't stay running. I start it and then go to restart and get the same error message... [01:03] also stop returns stop: unknown instance: [01:05] SpamapS: zomg [01:05] SpamapS: first line of man lxc-create makes me want to buy someone a grammar book [01:05] disorbtion1: I'd look in the logs, figure out why its failing [01:06] disorbtion1: does it get going at all? [01:15] remember total linux noob, had to look up how to view logs. which log should I be focusing on? [01:28] It was so simple once I looked in the logs, thanks for the help! [01:47] wonder if anyone has a solution. [01:47] i have a system that has an ssh port open to the public internet [01:48] and another system on the private network that it is connected to [01:48] i'd like to give someone access to the second system (not externally addressable), but ideally only after having acled through the first. [01:49] i know i can do this by letting them set up their ~/.config to use ProxyCommand and nc [01:49] however, that means [01:49] a.) they could "stop" at the middle system [01:49] b.) they have to deal with .ssh/config [01:50] i'm sure i could coble together some solution where a user on the public facing system had a shell of 'nc' or something, but it seems hacky and i don't want to do any work. [01:50] so i'm wondering if this is a problem that is already solved (and likely better than i woudl solve it) [02:05] smoser: Perhaps prepend command="ssh ..." to authorized_keys file on the first system? It could be on an account dedicated to them, or really any user so long as it was on the line that contained their public ssh key. [02:06] ah. yeah, maybe. [02:06] well done, mr hammond. [02:07] testing... [02:18] smoser: so, sorry for the question, but, why you should do that? [02:18] isn't easier to just let the uer log in into the second one? [02:19] instead of connect to first one and then connect to second one? [02:20] the first one is not connected to "public internet" [02:20] only private. [02:20] wait. [02:20] sorry [02:20] the system i want the user to get to exists only on private network. [02:20] the system i want them to hop through has only a single port open to internet [02:24] smoser: Yep, command= approach works, but you may need to fiddle with ssh options to get things to work cleanly. [02:24] For example, I needed to use command="ssh -p NNNN USERNAME@HOSTNAME" [02:25] ssh is freaking awesome [02:25] And, in the initial connect, I used "ssh -t -A" to force pseudo-tty allocation and to pass on my user agent. [02:28] hm.. i wouldnt have thoght you would have to use -t [02:31] Agent forwarding blows. [02:32] -oProxyCommand is where it's at [02:34] well, i think you use the combined trick [02:34] the command would be 'nc -q0 HOSTNAME', but then you probably have ot have the same username on both systems. [02:35] hm... maybe not though. have to play with it a bit. [02:40] Hola! Anybody want to field some krb5/nfs4 questions? [02:41] I'm dealing with an office full of wedged clients [02:48] smoser: If your authorized_keys command= uses "nc" that would be like telnet'ing to the remote ssh server. Your local ssh client would not be authenticating itself again once it's done with the first hop. [02:51] yeah. i realized that. [02:51] smoser: The first ssh could set up an ssh tunnel from the user's local computer through the gateway to the ssh port on the inner computer. Then they could initiate a second ssh to localhost that goes through the tunnel to the inner computer. This avoids problems with agent forwarding (which can be a security risk). [02:51] erichammond: you'll want to disable other port forwarding techniques and such [02:56] twb: You could assume that authenticating yourself on the gateway gives you authorization to connect to any port on the internal network. It's up to smoser's particular situation if this policy makes sense. [02:57] erichammond: in that case you don't need a command= [02:57] yeah, for me, they're in, they're in. [02:57] I assumed you were using command= to forcibly restrict the connecting user to ONLY hop from the bastion to somewhere specific [02:57] twb: You still need to authenticate yourself with the gateway and prevent command line access on that machine. [02:57] i want the command = because i dont want to have to set up .ssh/config [02:57] erichammond: ah, OK [02:57] and also, i dont really want them to be able to "stop" on the intermediate [02:57] (according to the original requirements) [02:57] i want only access to the second internal system. [02:58] erichammond: so broadly you want -fNL -NW [02:58] so the command= was nice. [02:58] *or -NW [02:58] And to enforce that usage in authorized_keys [02:59] twb: Can the gateway authorized_keys enforce -N on the original client? [03:01] Not sure, possibly command=/bin/sh [03:01] Er, /bin/false [03:03] Normally what I would do is either set up the bastion with normal users, or DNAT traffic to (say) 2022 from the bastion to the users' shell bastion [03:03] Or give them a VPN, with an endpoint terminating on the bastion [03:04] yikes, Just read "man authorized-keys" and it seems to imply that port-forwarding is allowed by default even when command= is used. Time to add "no-port-forwarding" to some servers. [03:04] I've got a sweet krb5/ldap/autofs/nfs4 stack.... except that it's broken. Anybody? [03:05] ahhh, our svn setup already includes: "no-port-forwarding,no-agent-forwarding,no-X11-forwarding,no-pty" [03:07] smoser: Might be a good idea to add those qualifiers to /root/.ssh/authorized_keys command= line in Ubuntu AMIs. [03:08] well... you're still getting in as root with the authorized keys [03:08] with legit authorized keys [03:08] being able to forward traffic isn't htat big of a deal [03:09] smoser: You have no idea what they do with the ubuntu user's ssh configuration after they start the instance. Might as well make the original ssh key have as little permission as possible over in that root@ account they forgot about and maybe never even knew allowed people in. [03:12] smoser: Submitting a bug. Would that be "cloud-init" or some other package? [03:12] cloud-init [03:25] bug 798505 [03:25] Launchpad bug 798505 in cloud-init "Tighten permissions on root@ ssh with EC2/UEC images" [Undecided,New] https://launchpad.net/bugs/798505 [03:28] ok, erichammond here is what i settled on. [03:28] pubhost has publicly open port X [03:28] user 'toprivhost' exists on pubhost [03:29] toprivhost .ssh/authorized_keys has the external user's public key in it with line like: [03:29] command="ssh ubuntu@privhost ssh-rsa AAAAB3NzaC1yc2EAAAAB.... thatuser@theirhost [03:30] command="ssh ubuntu@privhost" ssh-rsa AAAAB3NzaC1yc2EAAAAB.... thatuser@theirhost [03:31] i also generated a ssh private/public key as toprivhost@pubhost and put it in .ssh/id_rsa and .ssh/id_rsa.pub [03:31] to avoid agent forwarding? [03:31] toprivhost@pubhost's public key is copied to ubuntu@privhost:~/.ssh/authorized_keys [03:32] New bug: #798505 in cloud-init (main) "Tighten permissions on root@ ssh with EC2/UEC images" [Undecided,New] https://launchpad.net/bugs/798505 [03:32] and that .ssh/authorized_keys has a 'host=pubhost' option on the key [03:32] yeah, now we dont need ssh agent forwarding, and its all passwordless. [03:33] So you trust root users on pubhost with ssh access to that user on privhost, including somebody who gains root illegitimately. [03:33] i suppose, yeah. [03:34] but yo uwoudl do that too with ssh agent forwarding [03:34] smoser: good point. [03:35] and, ssh agent forwarding gives the root user access to any other hosts that like my agent. [03:36] Would be cool if the agent forwarding could be dropped after the second connection was made. [03:37] Hello all. Seeking assistance in finding procedure for updating /etc/motd w/o reboot: [03:37] distro: ubuntu server 10.04 [03:37] /etc/motd gets modified to the value of /etc/lsb-release:DISTRIB_DESCRIPTION. But this only takes effect after rebooting (more specifically, I think it's already modified before the reboot, when switching to RUNLEVEL 1). [03:37] Can /etc/motd be auto-modified w/o a reboot? [04:04] can anybody help with setting up a router in exim4? === devilinthedetail is now known as CrunchyChewie === medberry is now known as med_out === smb` is now known as smb [08:56] New bug: #798555 in cloud-init (main) "package grub-legacy-ec2 0.5.10-0ubuntu1.5 failed to install/upgrade: subprocess installed post-installation script returned error exit status 1" [Undecided,New] https://launchpad.net/bugs/798555 === marienz is now known as 92AAC5BT0 === 92AAC5BT0 is now known as marienz [10:21] New bug: #798587 in augeas (main) "Sync augeas 0.8.1-2 (main) from Debian unstable (main)" [Wishlist,Confirmed] https://launchpad.net/bugs/798587 [10:38] I'm installing 10.04 on a P4 with 2x2Tb SATA drives in RAID 1 config. The install goes fine until it tries to install grub where it fatally fails. Any idea why this'd happen? [10:52] kobrien: are the using 4kb blocks or 512b blocks? [10:52] block size is not specified for RAID 1 [10:52] ( mirror config ) [10:53] When grub fails, switch to vt4 and see what the error is [10:53] kobrien: not block size of the raid, block size of the disks' controllers [10:54] kobrien: 2TiB is the largest size that traditional BIOS/MBR systems can address, so some 2TB disks have adopted the new 4kB block, GPT scheme. [10:54] kobrien: Grub2 SHOULD handle those just fine, but maybe it's confusing the installer [10:55] twb: good tip to jump to vt. Not sure of the controllers block size. I'm not using GPT. I probably should. [10:55] kobrien: ah, I bet you chose manual partitioning, right? [10:55] yes [10:55] With GPT you need these stupid extra, special parittions to make it work [10:55] Without those grub will shit itself, which is probably what happened [10:55] I see. will a livecd of gparted do it? [10:56] What you should do is choose "automated partitioning (entire disk)", and see if it wants to create little grub partitions at the start [10:56] If it does, then blow away the root and boot partitions (but not the grub one), and reallocate the free space as you want [10:56] That's what I've done in the past and it worked [10:57] aha! It did want to make a grubbios one when I was exploring that menu [10:57] and I should mirror that partition too [10:57] ? [10:57] I have no idea [10:57] I left those the hell alone [10:58] GPT is retarded and OpenFirmware should've won, but worse is better :-( [10:58] heh, ok, well that's plenty info for another attempt at this. thanks [10:58] Oh, actually.. [10:58] yes? [10:59] It may be that you don't have 4kB blocks, but d-i is helpfully defaulting to that because it's 2TB [10:59] I got them to fix it so it would only do that for >=2TiB (1024s) instead of >=2TB (1000s), but that was post-lucid [11:00] So if you struggle enough you might be able to manage BIOS+MBR style [11:00] cool [11:00] ok, I'll give this a go [11:01] If you google for the model and make, it'll tell you if you have 4k or 512 blocks [11:01] will do [11:03] I'm trying to put in binary blob drivers and it would seem to be easiest if I had current kernel headers setup under /lib/modules/`uname -r`/build .. which is supposed to be often the case, why is it not on my stock ubuntu server kernel (10.04 - latest kernel) [11:04] do I need to install headers package for them to be in there? and which is it called? [11:06] paranoidi: you probably want linux-...-headers and/or -source [11:07] yeah, just found out I don't have it .. it seems that there are few different kernel variations from each version though .. [11:08] generic, preempt, server [11:09] maybe the generic I have in now is good enough .. [11:10] paranoidi: you want the one that's for the kernel you're running [11:14] twb: yeah, I got that .. unsuprisingly the propietary crap does not want to go in as easily .. ;P [11:16] I recommed not using it [11:16] seems to be only choice .. [11:17] Loongsons are pretty open [11:17] you lost me [11:17] Loongson is a type of CPU [11:18] https://secure.wikimedia.org/wikipedia/en/wiki/Loongson [11:19] not a motherboard, 16-port sata controller .. [11:20] Yes, well [12:55] zul: Are you looking at doing an ipxe MIR? [13:16] New bug: #798663 in samba (main) "package samba-common 2:3.5.8~dfsg-1ubuntu2.2 failed to install/upgrade: subprocess installed post-installation script returned error exit status 10" [Undecided,New] https://launchpad.net/bugs/798663 [13:20] morning all [13:21] good localtime(); [13:28] Daviey: no [13:32] zul: it's a build dep of xen [13:33] oh....fu...fudge [13:34] heh [13:52] receiving netconsole with netcat doesn't make any line breaks. Is it possible to fix? [13:55] sorry, I was listening on syslog data :p [14:17] when i install guest additions on a new ubuntu guest, vbox complains that it cant find kernel headers. But I do have correct kernel headers installed. Anyone has seen this problem with the latest ubuntu-server machines? [14:20] nope [14:20] is anyone using a ubuntu server as a guest virtualbox machine here? [14:23] http://paste.pocoo.org/show/410202 [14:23] This is what I'm getting when I install guest additions...please see the complaint on "The headers for the current running kernel were not found..." [14:27] pythonirc101: there's an ubuntu package for vbox addons [14:27] should work [14:27] that is, works for me [14:27] RoyK: http://paste.pocoo.org/show/410204/ [14:27] so you install ubuntu vbox guest additions and not the one that comes with vbox? I'm using virtualbox 4.0.8 under the host, and ubuntu-server is installed as a guest [14:27] apt-get install virtualbox-ose-guest-x11 [14:28] apt-get install virtualbox-ose-guest-utils [14:28] x11 if using X [14:28] but for a server, you probably don't [14:29] RoyK: I did install ubuntu-desktop on my server (which is a guest) [14:29] apt-get install virtualbox-ose-guest-x11, then [14:30] is there a way to uninstall guest additions? [14:30] apt-get remove? [14:30] I installed the guest additions that came with virtualbox 4.0.8 [14:30] if that installer has an uninstaller, yes [14:30] See here: http://paste.pocoo.org/show/410204/ [14:31] but Ubuntu obviously can't remove something it hasn't installed [14:31] indeed [14:33] k - found the uninstaller [14:35] the x isnt still working quite right...the login screen is not centered for instance [14:35] + In seamless mode, its display becomes a small screen on the first monitor [14:36] Terminal fonts wont increase in size when i do ctrl+Shift++ [14:36] will reboot [14:36] and see [14:39] RoyK: Any other ideas? [14:56] Ursinha: o/ [14:57] pythonirc101: X and desktop questions aren't topic in here [14:57] Daviey: :) [15:05] what is the network gui on ubuntu-server? I want to switch off ip version 6 [15:08] pythonirc101: why would you want to switch it off? and there is generally no gui on aserver [15:08] because there is no need for it [15:08] if you can't manage a server in a way you are supposed to, maybe you should arrange for a real admin to do so [15:08] pythonirc101: normally there isn't a gui on a server [15:09] !guide [15:09] The Ubuntu server guide may be found at http://help.ubuntu.com/10.04/serverguide/C/ [15:24] how do i switch off ipv6 on ubuntu server? [15:26] pythonirc101: as a first, why? [15:27] RoyK: Because my system - administrator wants me to? [15:27] RoyK: I'm connected to someone else's local network [15:28] My other system is complaning that "System not running to use fully qualified hostnames" "Hostname xx.xxx.com is illegal" -- I think my /etc/hosts is fixed, and have no clue why the other system still complains [15:30] pythonirc101: http://kurl.no/v280 [15:34] Anybody want to help me debug my krb5/ldap/autofs/nfs4 stack? [15:34] I've got some wedged clients [15:42] hello there [15:42] what is the default mail server on ubuntu ? [15:44] zertyuio: postfix is well regarded [15:46] postfix seems too complicate to manage [15:46] anybody know where one goes for support on this krb5/nfs4 combo? [15:46] just for smtp [15:48] !ask | brianthelion [15:48] brianthelion: 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. :-) [15:48] zertyuio: well, you asked [15:49] I've never seen 'just smtp' in the real world. I think a simple smtpd doesn't exist, because smtp is never simple [15:50] let me explain you mail problem [15:50] Cool, thanks. My nfs4 clients are wedging when a logged-in user's Kerberos ticket expires. I assume this is not intended behavior. [15:50] Everything is on Natty [15:51] syslog: "Jun 17 07:44:48 carina kernel: [78074.876438] Error: state manager encountered RPCSEC_GSS session expired against NFSv4 server 192.168.0.2." [15:52] i m hosting a cms website on my personel server, where the form for contact detail working perfectly without since i unistalled postifx , submit form not working [15:53] what i have to do ? [15:54] I've got a bug report up: https://bugs.launchpad.net/ubuntu/+bug/794112 [15:54] Launchpad bug 794112 in ubuntu "Kerberos + LDAP + NFSv4 on Natty - Unable to recover unattended client" [Undecided,New] [15:54] yeah, that's me [15:55] Flat response so far, though [15:56] brianthelion: maybe try the same on the LTS release for comparison? [15:58] good idea, but probably not an option in this case [16:03] well this is annoying, I cannot stop ubuntu server with latest kernels to stay in the damn text mode, tried vga=normal, vga=711 (or so), nofb and vga16fb.modeset=0 === failover is now known as tarzxvf [16:34] paranoidi: for most installs, personally, I think the framebuffer is unnecessary and may lead to more problems than it fixes [16:34] RoyK: I agree, but I cannot seem to get rid of it .. [16:35] paranoidi: try vga=normal nomodeset [16:35] from https://wiki.ubuntu.com/FrameBuffer [16:38] I'll try it, maybe, seems that I need to rescue old kernel from old drive anyway since I cannot get my sata card to work with current ones [16:40] paranoidi: very often, new cards have the same chipset, but new PCI IDs [16:40] on which version are you? [16:42] RoyK: trying to get same crappy card work (RocketRaid 2340) [16:42] I don't think linux has any built in support for it [16:45] it might go in if I was skilled enough to tweak the opensource adaptation layer between kernel and binary driver of the card [16:45] paranoidi: use lspci and then lspci -vn to find its PCI ID [16:47] which part is PCI ID? [16:48] erm [16:48] 06:04.0 0100: 1103:2340 (rev 09) Subsystem: 11ab:11ab [16:48] would be my guess [16:49] seems the producer apparently released a driver in source, so it is, or was, possible to compile that manually [16:50] but since it never made it into kernel, the chances are prominent that the code sucked, and that it won't be compatible with current kernel versions [16:51] New bug: #798788 in nagios3 (main) "package nagios3-common 3.2.3-1ubuntu1.2 failed to install/upgrade: subprocess installed post-installation script returned error exit status 1" [Undecided,New] https://launchpad.net/bugs/798788 [16:51] RoyK: I don't thinkt here is full sources of the driver, it's just binary blob with opensource wrapper [16:52] anyway, i tried to compile it for latest kernels and it seemed to work fine .. but when loading it just stalls [16:52] that sucks even more :P [16:52] get a new controller, then [16:53] new one costs at least $500 .. [16:53] paranoidi: you get pretty cool controllers for even $100 on ebay [16:53] 16 ports ... [16:53] do you need 16 physical ports? why not a sas expander? [16:54] that might work, but those cost pretty much too .. and severely cut bandwidth? [16:55] with 6Gbps SAS, there'll be quite sufficient of bandwidth, especially for slow spinning drives [16:55] it probably works only in 3Gbps mode with sata 2 drives [16:55] or two of htese http://cgi.ebay.com/LSI-SAS3081E-R-8-Port-3Gb-s-SAS-RAID-Controller-/270765091146?pt=COMP_EN_Networking_Components&hash=item3f0adb694a [16:56] s/htese/these/ [16:56] that's actually pretty cheap [16:56] yep [16:57] and it's a good card [16:57] I have a few of those in OpenIndiana machines [16:57] although I would have to find new display card since it would use 2 slots compared to one [16:57] * RoyK prefers ZFS for storage [16:57] paranoidi: I thought this was a server...? [16:57] yes, but at home :) [16:58] still, a badass graphics card in a - server? [16:58] no, but I need some output since motherboard has none [16:58] dunno if I have any PCI cards laying around, anyway, not a huge problem [16:58] I was actually looking 16 port version from LSI earlier === med_out is now known as med === med is now known as medberry [16:59] LSI Logic SAS 9201-16i .. any word on that? [16:59] very good [16:59] 6Gbps SAS [17:00] I have a few of those as well [17:00] I could get that in some decent price from here (well, ~470$ still) [17:00] RoyK: it's supported with kernel without any problems whatsoever? [17:01] if on a budget, and if bandwidth isn't too important, this one is very stable http://cgi.ebay.com/Supermicro-8-Channel-Pcix-Sata-Adapter-Aoc-Sat2-Mv8-/270765097465?pt=LH_DefaultDomain_0&hash=item3f0adb81f9 [17:01] paranoidi: I haven't tried it on linux [17:01] I couldn't find verification if that card is supported on linux out of the box or not [17:05] paranoidi: You'll have to check - I haven't tried [17:05] paranoidi: using software RAID on linux for the storage? [17:06] if you haven't installed it yet, then I'd recommend looking at openindiana / ZFS [17:07] software raid [17:08] last time I checked ZFS couldn't be grown like software raid [17:08] it's not as flexible as linux software raid is, no [17:08] you can add VDEVs to ZFS [17:08] or you can replace drives with bigger ones [17:08] and when all drives in a VDEV are replaced, it'll grow (given you've set autoexpand=yes) [17:09] but you can't add drives to a RAIDz VDEV [17:09] right, and I start my packs with few drives and then expand as needed [17:09] ic [17:09] * RoyK just setup a server like that for his brother - two drives in RAID-5 [17:10] paranoidi: but then, do you really need 16 ports already? [17:10] no, but there's room for growth :D [17:10] just get an 8-port controller now, and get something bigger when you need it :P [17:11] but I have already more than 8 .. and I don't like to fiddle with the hardware too much [17:11] kk [17:11] >8 with data? [17:12] how big are these, btw? [17:12] 8x1TB and 3x2TB [17:12] k [17:12] if everything else fails I can ressurrect at least the 8 disk pack with some motherboard ports and old PCI sata card [17:17] heh [17:17] let's hope the old 2.6.31-14 copied from old drive works properly .. [17:17] I just went with 16 ports :) [17:17] and a 4 port controller [17:18] I'm hoping I like channel bonded sas [17:18] patdk-wk: what kind of write/read speeds are you getting? [17:18] if you're using them in raid [17:18] I haven't received it yet [17:18] patdk-wk: which controller? [17:19] an lsi 8port [17:19] 9201 or 6801? [17:19] sas2 one [17:20] 9201, then [17:22] I'm trying to setup a static ip box - ubuntu-server - for some reason, when i restart the box, i still get the dhcp ip, any ideas on how to fix this? [17:22] heh - http://linuxhcl.com/browse/search?offset=0&category=12 <-- doesn't even list LSI [17:22] I edited the / /etc/network/interfaces /etc/hosts and /etc/resolv.conf [17:22] changed the hostname [17:22] any other things i need to do? [17:22] if config still says my old dhcp ip [17:23] looks like 6261 [17:23] 9261 I mean [17:24] paranoidi, it will only give me about 1100MB/sec [17:24] but insane iops [17:24] I highly doubt that is the performance when writing to a file [17:25] why not? [17:25] patdk-wk: because the drives will be slow [17:26] Ì get about 100MB/s at tops with software raid-6 [17:26] they better be able to handle that [17:26] 70MB/s over samba [17:26] using what drives? [17:26] seems the LSI drivers are available as a download [17:26] binaries and source [17:27] hahaa, so same crap as with my rocketraid [17:27] well, except they might actually release updates [17:27] and with some luck the kernel would support it [17:27] only ever needed to download the binaries for userspace management of the raid [17:27] paranoidi: not really, it's opensource drivers AFAICS [17:28]