[00:07] hi guys, can i use mirc/xchat with ubuntu server? [00:07] sure, if you're running X [00:07] or want to X forward it [00:08] Deeps: i'm not using X-server, thus not possible? [00:08] wel if you're not running X, how do you think a graphical client is going to be displayed? heh [00:09] you can X forward it from the server to your local X server if you wish [00:09] dou213: why dont you use a Terminal irc client? [00:10] jmedina: like? that's what i'm searching for i guess... didn't know how to put it :) [00:10] irssi [00:10] bitchx [00:10] you can login by ssh to the server and start chatting without problem, or you can use screen inside your ssh session, and when you are bored you can detatatch you session, and the next day re use your session [00:10] Deeps: sry for the inconvenience [00:10] I only know irssi [00:10] those are the 2 more popular terminal clients [00:11] ok Deeps and jmedinam thx i'll give it a try.. [00:11] jmedina, sry === dedi_away is now known as Dedi === Dedi is now known as dedi_away === dedi_away is now known as Dedi === Dedi is now known as dedi_away === dedi_away is now known as Dedi === Dedi is now known as dedi_away === dedi_away is now known as Dedi [02:07] Hello [02:18] will software raid acheave better speed if i make raid1 betwenn ide1 dev1 - ide2 dev1 or ide1 dev1 - ide1 dev2 (i have total four disks on other two i will setup lvm) === Dedi is now known as dedi_away === dedi_away is now known as Dedi === Dedi is now known as dedi_away === dedi_away is now known as Dedi === espacious_ is now known as asdas === asdas is now known as espacious === espacious is now known as asdaf === asdaf is now known as espacious [07:45] <_ruben> espacious: having 4 pata disks on 2 busses will kill your performance either way .. when just looking at the raid part, splitting the 2 members of the raid over the 2 busses sure is faster (shared busses suck) [07:57] _ruben glad somthing pointed me out that [07:57] infact my rebuild is extremly slow but i have raid1 on separate ide [07:58] fact is i have two separate ride1 one is (ide1-1,ide2-2) [07:58] other is ide1-2, ide2-1 [07:59] is doing both rebuilds on start it was fast enough now speed dropped to almos 0 [08:03] <_ruben> well, pata will always remain slow, especially with 2 devices on a single bus, you should buy 2 port ide controller and give each drive its own bus [08:05] _ruben u got it i use pata not sata i will buy it, i think its a good idea. [08:06] _ruben can u cca tell me what is the difference in speed on same and on separate ide? [08:08] another thing can i stop the rebuilding array proces? [08:08] since i have two running, i want first finish one and than do the second. === jmarsden_ is now known as jmarsden [08:12] Hiya! [08:12] hey Jeeves_ [08:13] good to see you here :) [08:18] :) [08:19] <_ruben> espacious: i dont know the exact comand by head, but with mdadm you can stop one of the (re)builds afaik [08:19] <_ruben> espacious: theoretically the speed gain is atleast a factor 2 [08:19] <_ruben> one bus, shared by 2 disks, each disk has 50% .. but there's quite a bit of overhead, so the gain would be even more [08:19] no realy 50% ? [08:20] that's sooo much! [08:20] i will buy the controller. [08:21] <_ruben> Jeeves_ returned from the dead ;) [08:21] _ruben u think i can remove the sdd1 from the array when is still syncing? [08:21] with the --remove command? [08:22] <_ruben> espacious: i think you can, you could stop the array first to be sure [08:26] ok thanks [08:27] moin [08:33] Morning [09:06] is eBox a recommended tool for server management? [09:07] ubuntu_: Recommended is a relative term [09:07] I'd say it depends on your experience level, desire to learn, and the type of server you're refering to [09:09] ropetin, my experience level is intermediate... I'm currently maintaining an office server since I have the most linux experience in the company, but I'm only a part-time administrator. [09:10] I'm happy to learn what's needed, but preferably if it helps me maintain the server more effectively [09:11] the server itself is a file/print/backup/network server which I have direct access to [09:12] currently I maintain the server through a combination of commandline and GUI interfaces [09:13] ubuntu_: is it externally accessible or only on your LAN? [09:14] ropetin, at the moment it's only on the LAN [09:15] In which case, no worries, give eBox a go, see if you like it [09:16] ropetin, can there be difficulties if I'm using eBox and still tweaking the conf files manually? [09:16] I only used it one time, a while ago and didn't like it, so I can't say for 100%, but I'd think so, based on the way somethign like that works [09:18] thanks ropetin, I'll check it out and see if I like it [09:32] No worries, good luck [10:21] New bug: #299843 in open-iscsi (main) "armel build failure (package not yet in the archive)" [Medium,Triaged] https://launchpad.net/bugs/299843 [10:34] is there a good way to backup users and groups (incl. passwords)? [10:37] Shame eBox isn't installable from the standard repos at the moment. [10:38] backup the /etc/passwd, /etc/shadow and /etc/group files? [10:41] Deeps, so if I just restored those files to the new /etc directory, the users will be reinstated? [10:42] ubuntu_: Yes, although you might want to backup their home directories too. [10:43] Omahn_, yes, that would be part of the plan [10:43] ubuntu_: you also might wanna check that the system user uids and gids all match up to what your new system has too [10:43] e.g. if you've installed some services that run under their own user, e.g. named, postfix, mysql [10:44] better would be to just grab all the real user + group accounts (uid/gid >= 1000) and add them back in [10:45] Deeps, is there a way to restore users/groups selectively? [11:38] New bug: #299813 in xine-lib (universe) "armel builds failure (package not yet in the archive)" [High,Invalid] https://launchpad.net/bugs/299813 [11:53] where can i get help with software raid1? [12:02] New bug: #299866 in tomcat6 (main) "package tomcat6-admin 6.0.18-0ubuntu3 failed to install/upgrade: el subproceso post-installation script devolvi? el c?digo de salida de error 255" [Undecided,New] https://launchpad.net/bugs/299866 [12:11] espacious: Here, maybe. Just ask your question. [12:12] so i have 2x raid1 now it's rebuilding both arrays, and it became slow, almost 0. i want to pause one array or stop one to let first finish one. [12:13] at start was going ok normal rebuilding speed, then near the end it became very slow i rebooted did again the stuff but same happened = slow performance at the end. [12:13] also the console is much unresponsive [12:13] takes ages to make a command [12:14] take a look at that [12:14] http://pastebin.com/m66f9ed8c [12:16] shud i try "mdadm --stop /dev/mdx [12:16] ? [12:17] Which version of Ubuntu is this? [12:17] latest server ubuntu [12:17] 8.10 [12:17] anything interesting in dmesg? [12:18] cat /var/log/dmesg? [12:20] espacious: or just dmesg [12:21] i got slow respond on console [12:21] will paste it [12:22] do you have anything else heavily using the disk perhaps? [12:22] no only two raid's [12:22] its a new install almost. [12:23] just added apache and samba and raid. [12:23] apache is topped also samba [12:23] stopped* [12:24] last few lines of dmesg [12:24] http://pastebin.com/m23b9e410 [12:25] JBD: IO error reading journal superblock ??? [12:26] i think becouse i hard rebooted [12:31] piti if i mdadm --stop /dev/md1 i get [12:31] fail to stop array /dev/md1: Device or resource busy [12:32] Presumably because you have something mounted off /dev/md1 [12:32] md1 is lvm [12:32] should i unmount? [12:33] the rebuild array is in progres but it's froozen [12:33] If possible yes. I suspect you will also have to make the volume groups inactive. [12:35] ok will try. [12:37] disk led is up for hours now, console is very unresponsive. [12:37] i doubt its a fault disk. [12:38] Sounds like it might be a fault on the controller if you're sure the disk is healthy. [12:38] worked ok a couple of hours ago [12:38] so was not mounted [12:39] Hi, how can i use lvm with kvm? [12:42] Omahn_ how i make volume groups inactive? [12:44] umount /dev/fileserver/share ? [12:44] or lvremove ? [12:51] espacious: vgchange -a n [12:52] dows what? [12:52] does* [12:53] ok found the vgchange man [12:54] is safe to do that while an array is build? [12:55] espacious: This also doesn't look too good: # [12:55] [ 43.821632] attempt to access beyond end of device [12:56] celephais: What do you mean? [12:56] hmm. [12:56] soren, how can i install guest system on a logical volume? [12:57] i wil rebuild lvm as now i dont use entire disks and they are not same size but almost equal. [12:57] soren, or let a guest access a logical volume [12:57] celephais im sure u can find some how to's [12:58] i ran vgchange -a n to see. [12:58] espacious, with xen yes, but i can't find anything for kvm [12:59] espacious, and with vmbuilder i can use only files [12:59] celephais: Just like you would install it to any other file or device. [12:59] If you were using a file, you'd: kvm -hda /path/to/some/file. [12:59] If you were using an lv, you'd: kvm -hda /path/to/some/lv. Same. [13:00] soren, ok, is supported in vmbuilder? [13:00] root@kgsstore01:~# vgchange -a n [13:00] File descriptor 5 left open [13:00] for 5 min now.... [13:01] celephais: It's meant to be (it works for me), but i've heard about others who had less luck. The option, you're looking for is --raw. [13:01] soren, ok I'll try thank you [13:03] so stuck at vgchange... how can i repair that mess [13:05] im worried about the disks they are reading/writing now for a couple of houres non stop. [13:05] espacious: What does your partition table look like? [13:05] sda - sdd [13:06] a1 c1 raid1 normal ext3 [13:06] Ah, no. [13:06] what? [13:06] 0 [13:06] Whoops [13:06] I'd like to see your partition table. [13:06] fdisk -l /dev/sda, for isntance. [13:07] Actually, "sudo fdisk -l /dev/sd?" [13:07] now i can only wait but it is ... [13:07] sda1 / [13:07] That's not what I'm looking for. [13:07] ok will make output of fdisk [13:08] is the pc dont blow up, it's writing like an animal [13:08] if* [13:09] still stuck. [13:10] shuld i reboot and disct the second disk in the raid? [13:10] and then rebuild the whole again? [13:11] * soren thinks something is screwed [13:11] ...and I'm suspecting the partition table right now. [13:11] btw when i runed top it was ok.. [13:11] meaning? [13:11] ok will try to post out that fdisk [13:11] no proces was 100% cpu [13:12] Ah, no. There wouldn't be. [13:12] All the raid magic is in the kernel (and hence doesn't show up in top). [13:13] right. [13:14] espacious: Is this a server or a desktop box? [13:14] server. [13:14] Bummer. [13:15] u mean the motherboard etc... or the OS version? [13:15] Just the box, if it was just a desktop machine then a reboot might be a worthwhile. [13:16] its a normal pc its not a server ubuntu is server ver... [13:16] New bug: #299886 in mysql-dfsg-5.0 (main) "error processing /var/cache/apt/archives/mysql-common_5.0.22-0ubuntu6.06.11_all.deb (--unpack)" [Undecided,New] https://launchpad.net/bugs/299886 [13:17] still at vgchange should i ctrl+c? [13:17] Yeah, it should return immediately. [13:18] Is the filesystem ext3 on the logical volume? [13:18] all command have delay [13:19] yes on sda1 and ext3 is also on md1 (lvm) [13:19] Are these disks SATA or SCSI? [13:19] and both are rebuilded.. [13:19] pata [13:19] espacious: what do you mean when you say "md1 (lvm)"? Is it lvm over raid? [13:20] two disks in lvm group and in raid1 [13:20] md1 is usually raid. [13:20] oh. [13:20] i messed up sth? [13:20] * Nafallo makes a note to not comment without reading ;-) [13:21] neither ctrl+c has effect seems [13:21] espacious: so it's raid1 first, and then you created an lvm group using the raid device, that's it? [13:22] espacious: is / a logical volume too? [13:24] hmm.. i made first lvm on both disks and than put them in vlm group and in raid md1 [13:25] espacious: Might be easier if you could give us the output from the following commands, as root. [13:25] pvscan [13:25] vgscan [13:25] lvscan [13:25] mount [13:25] i know. [13:25] cat /proc/mdstat [13:25] dmesg [13:25] * soren still thinks fdisk -l /dev/hd? is what we want [13:25] Assuming the machine is responsive enough to produce the output :-) [13:25] waiting if it becomes responsive otherwire i will rebot [13:25] is not. [13:26] soren: The earlier mdstat pasted was referring to sd?, hence a little confusion when espacious mentioned PATA. [13:26] Omahn_: Why? And what's your point? :) [13:27] Very, very few disks (if any at all) are named hd? nowadays. [13:27] Err... [13:28] I mean, disks on very, very few controllers (if any at all) are name hd? nowadays. [13:28] Omahn_ they are normal 80 pin pata and are shown as sd** [13:28] * Omahn_ still has many machines with hd? disks. [13:28] I haven't had any for years and years. [13:29] * Omahn_ passes the willy waving award to soren [13:29] Even my ancient laptop (pre 2000, IIRC) has an sda disk. [13:29] Omahn_: :) [13:29] my lifebook too. [13:29] so ctrl+c stuck no nada nothing. [13:30] Switch to another console. [13:30] local? it won let me axx over lan [13:30] Yes, local. [13:30] brb [13:31] huh local console is full of errors! [13:31] DRDY ERR ata1.00 [13:31] error UNC [13:32] never seen that [13:32] Doesn't sound good.. [13:32] in fact ata2 not ata 1 [13:32] i see ! [13:32] damn. [13:32] so what is the procedure? [13:33] got the remote console.. [13:33] shutdown -r now? [13:33] Might as well. [13:34] since i presume it wont stop writing [13:34] what u suggest? [13:34] How have you got to this stage? Was this a RAID1 mirror that experienced a failed disk? [13:35] no i had raid1 on that i had my os i only created another raid LVM one but i had it before i only changed disks. there is almost no data on that lvm... [13:35] espacious: I'd start over. [13:36] ^^^^ [13:36] Omahn_: Error: "^^^" is not a valid command. [13:36] how. [13:36] espacious: How what? [13:37] i think the main disks should be ok. i can maybe just recreate raid and lvm again. [13:37] * soren mumbles some more about a partition table.. [13:38] espacious: I would recommend checking those disks before you go any further. [13:38] what exact output u wanna see fdisk -l ? [13:38] espacious: All of the output from "fdisk -l /dev/sda?" [13:38] Whoops [13:38] "fdisk -l /dev/sd?" I mean. [13:38] damn.;D [13:39] What now? [13:39] i entered the first one. [13:39] thamn this can take ages. [13:40] let me try local [13:40] "the first one"? [13:40] Oh, what I wrote first. [13:40] Right, sorry about that. [13:40] It shouldn't take long, though. [13:41] yes.shouldnt. [13:41] local console is not usable [13:41] outputs only errors cant login [13:42] also i noticed only sdc and sdb are involved in errors so is sth with the lvm disks not the main [13:42] aha fdisk output. [13:43] http://pastebin.com/m2baea8a6 [13:44] dont tell much [13:44] Was that the output from fdisk -l /dev/sd? [13:45] fdisk -l /dev/sda [13:45] Try again with 'fdisk -l /dev/sd?' without the quotes. [13:45] already in progres [13:46] ok was a bit faster [13:46] http://pastebin.com/m55f7e23d [13:47] Ok, now 'cat /proc/mdstat' [13:48] http://pastebin.com/m49d2a06b [13:48] read your mind:D [13:49] :-) [13:49] :D where i screwed up? [13:50] You didn't hot swap any of the disks did you? [13:50] no. [13:50] im also preaty sure there were both disks in md1 [13:50] when i booted up. [13:52] If you have lsscsi installed, the output from 'lsscsi' would be useful. [13:52] ... [13:53] if i could stop the rebuild somehow would help i think. [13:54] still no output. [13:55] I think a reboot would be a smart move. Something has gone badly wrong somewhere. It's extremely odd to see two RAID arrays, with different disks, fail like this. [13:55] * Omahn_ still suspects controller [13:55] * Omahn_ has to head off now [13:56] thanks for helping. [13:56] Didn't really help much, but no problem :-) [14:10] on server 8.06 where do i specify the name servers -- resolve.conf says not to edit it [14:11] either in resolv.conf, or as args in /etc/network/interfaces on the network interface [14:11] i reboted the server now how can i stop the rebuild that has again started? [14:11] for now the console is responsive. [14:11] mdadm --stop /dev/md1 dont help [14:12] thanks deeps [14:12] dns-nameservers arg in interfaces [14:12] see interfaces(5) manpage [14:12] i lie, it's not in there lol [14:13] (in the manpage, that is) [14:15] ok i managed to stop the array [14:15] mdadm --manage /dev/md0 --fail /dev/sdb1 [14:15] what shoul i do now? [14:16] cool [14:51] Hey i'm getting frequent kernel panics that all seem to be indicating that something to do with iptables is causing them. I'm not sure how to investigate further to try to figure out what's causing them though. Any ideas? [14:52] 8.04 with all updates installed [14:53] kernel version? [14:58] 2.6.24-21-server [15:06] soren: question about iscsi merge - why do you start iscsi before networking? [15:10] Koon: what should be done about bug 286828? [15:10] Launchpad bug 286828 in samba "Access to samba<3.0.26 shares using CIFS is broken on 8.10" [Undecided,Confirmed] https://launchpad.net/bugs/286828 [15:10] Koon: it seems that we've found what the problem is. However I stil wonder why the cifs client broke in the first place [15:11] mathiaz: I don't. [15:12] mathiaz: I wanted to check if dapper was affected [15:12] mathiaz: I start it before S40networking, yes. [15:12] soren: hm - the iscsi init script is started at S25 [15:12] mathiaz: also I'm just not sure saying it's a server-side problem will "solve" the issue for those who can't upgrade (closed NAS boxen) [15:13] smbfs was apparently ignoring this bug quite well. [15:13] soren: while networking is at S40 [15:13] Koon: exactly. [15:13] mathiaz: Yes... [15:13] mathiaz: But the network is brought up before that. [15:14] udev discovers a nic -> it's configured. [15:14] soren: right :D [15:14] mathiaz: but I also undersand steve french's pov: as upstream he shouldn't consider the bug on his side [15:15] mathiaz: that leaves us in the middle, and our choice to use CIFS where SMBFS was used before [15:15] soren: even in the case of interfaces setup in bonding? [15:15] mathiaz: No. [15:16] Koon: well... backward compatibility could be taken into account. [15:16] mathiaz: so my plan was: 1/check if dapper samba server is affected 2/try to see if we can workaround this in CIFS [15:16] mathiaz: but S40 is way too late for something that might be providing your block devices for all your filesystems. [15:17] mathiaz: I'm just not sure we can achieve 2/ without Steve's help :) [15:17] Koon: dapper is probably affected. [15:17] Koon: I haven't tested though. [15:17] Koon: OTOH hardy is not [15:18] soren: right - could the iscsi block devices be considered as network filesystems? [15:18] soren: it seems that the pre-condition for iscsi to work correctly is to have the network up and running [15:18] No. [15:19] Sorry for being so terse... I'm trying to look after my daughter at the same time, and she's rather unhappy. [15:19] mathiaz: all the reporters use the 3.0.23-3.0.25 range, and I find it funny noone has reported the bug using dappers' samba server [15:19] soren: ok - we can resume this conversation later [15:20] that's why I wanted to check (and identify precisely which upstream bug causes this) [15:23] Koon: right. Even if dapper is not affected, I'm still not convinced that we can close to bug with a answer - please upgrade your server [15:23] mathiaz: I agree with you. [15:24] Koon: it may worth asking steve opinion if adding a workaround this would be possible [15:25] mathiaz: sure, want me to do it ? I wanted to identify the bug more precisely before I did that. [15:25] Koon: sure. Seems that your plan is good - first try dapper [15:25] Koon: and then get in touch with steve [15:26] Koon: to see if a workaround could be added to the code. [15:26] yep. [15:28] Koon: awesome - thanks [15:47] mathiaz: Yes, network needs to be up and running for iscsi to work. However, "network being up and running might not mean what you think". [15:48] * soren is bad at placing quotation marks.. [15:48] I'll try again: [15:48] "network being up and running" might not mean what you think. [15:49] If you've booted off of iscsi (some BIOS'es support this), the network is up and running even before you boot the kernel. [15:49] So you might not even have to do anything to get it to work. [15:50] mathiaz: S25 is about the right place for open-iscsi. We just need to make sure that bonded interfaces are up at that point as well. [15:51] I proposed a patch for that a long time ago. [15:52] IIRC, I'm still awaiting feedback on it. [15:52] The gist of it: [15:53] Err... [15:53] * soren was looking at the wrong file. [15:54] soren: ok - another solution that debian tried was to use ifup.d script [15:54] * soren doesn't like it [15:55] mandriva had something in fstab specifying that the iscsi mount point used network, so it's mounting would only be attempted later after network was up [15:55] soren: it was reverted later [15:55] _netdev IIRC [15:55] * ahasenack jumped into the middle of the discussion, and now jumps out [15:56] soren: and they mount _netdev filesystems [15:56] ahasenack: Right. I don't like it. [15:56] ahasenack: right - I think that's what debian is doing for now. [15:56] soren: It also seems that the use case of root on iscsi is separate from non-root on iscsi [15:57] mathiaz: IIRC, yes. [15:57] mathiaz: What I changed in ifenslave-2.6 was: [15:58] Instead of the bond0 interface listing its slaves, the slaves would refer to the bond0 interface. [15:58] The if-up.d would check if the bond0 was "complete" and bring it up when all the slaves were available. [15:59] I.e. when udev discovers the last one. [15:59] ...as opposed to S40, which is way too late to be useful. [16:00] soren: ok. === kirkland` is now known as kirkland [16:32] soren: other question related to iscsi - why is iscsi shutdown at K25? [16:32] soren: that is way before S40umountfs [16:32] mathiaz: That's a bug, I believe. [16:32] soren: ok. [16:33] The road of iscsi zen is paved with mysteries and surprises. Beware. [16:33] soren: right - debian added a new shutdown script [16:33] soren: umountiscsi.sh at K80 in rc6 [16:34] soren: that's before S20sensigs [16:35] soren: it seems that iscsi devices should be umounted around the same time as S31umountnfs [16:36] If you can find a way to identify iscsi mounts, that's the rigth place to unmount them probably. [16:36] soren: http://paste.ubuntu.com/74398/ [16:37] soren: that's what is used to umount iscsi devices [16:38] mathiaz: At a glance, that seems to at least attempt to do the right thing. Let's try it. [17:05] soren i screwed up again or there is sth wrong [17:05] i recreated raid and lvm [17:06] i saved the sda1 the drive is ok raid was rebuild [17:06] but when trying to rebuild raid with lvm [17:06] http://pastebin.com/m312628e5 [17:07] where im making the mistaake now i made also xfs not ext3..is maybe that the disks are not exactli the asme? [17:08] It's hard to say. I don't know what you've done. You're just showing me the result. [17:10] i folowed one tutorial now. [17:10] but quite as i setuped earlier [17:12] That doesn't help me much :) [17:13] i know can u just point me a bit [17:13] or tell me what to show u [17:19] espacious_: How did you set it up? [17:19] first i make fdisk /dev/sdb and sdc type Raid linux auto [17:20] than i added one disk [17:20] to the xxx_vg [17:20] made xxx_lv [17:20] mkfs xfs [17:20] hello there ! [17:21] addet the other disk /dev/sdc [17:21] an started the array was going ok till reached about 50% [17:21]