[00:36] I'm trying to change my hard drive out on my home server, which is just a single disk, and trying to go to a Raid 1, but I'm not understanding how I can do this. [00:36] Can anyone give me a few pointers? [00:37] Also to top it off, when the server was built, it appears that only 2 partitions were made, 1 being the root( / ) and 1 being the swap. Does that pose additional problems [00:39] p7ank5te7: create a new raid-1 with a missing drive on the other drive and copy the data over [00:39] you can't convert a normal partition to a raid-1 [00:40] See my problem is with the tutorials, they say I'm supposed to have root, boot, and swap partitions but I only have root and swap? [00:44] then create a small boot partition on the new drive [00:44] copy /boot there [00:45] use lvm for the rest [00:45] if you're new to linux, ask, it may take some turns to get it fixed right [00:46] or - just use partitions - one for /boot, one for the root, and one for swap [00:46] I am still a bit new to it, at least for the partitions and stuff, I have been learning slowly but surely. [00:46] then just use partitions for now [00:46] fdisk /dev/whatever [00:46] create a small boot partition, max 1GB [00:46] then the swap, then the root [00:47] swap should be on the start of the disk, since that's fastest [00:47] but start with the boot partition [00:48] then mount the lot in /mnt or smoewhere and rsync -avPHAx / /mnt [00:48] as in, mount the root on /mnt [00:48] create /mnt/boot [00:48] mount /boot there as well [00:49] and then do the rsync [00:49] then install grub on the new drive [00:49] try to reboot into it [00:50] RoyK, I appreciate it, so let me make sure I understand, create the partitions on the "Extra" drive, then rsync the content from the current to the new one, install grub, and try it? [00:51] p7ank5te7: erm [00:52] p7ank5te7: I think you'll need to do this slowly [00:52] p7ank5te7: first, create a partition for boot, swap and root [00:52] Sorry for being a complete idiot with this. [00:53] no, I was giving wrong advice [00:53] create these partitions as raid partitions [00:54] then create raid devices on them with something like "mdadm --create --level=1 --raid-devices=2 /dev/sdXY missing" [00:54] meaning you create a degraded raid on each of them [00:54] then create raid devices on them with something like "mdadm --create --level=1 --raid-devices=2 /dev/md0 /dev/sdXY missing" [00:54] etc [00:54] md0 md1 etc [00:54] one for each partition [00:55] then mount the root on /mnt or somewhere and create /mnt/boot, mount the boot partition there, and do the rsync [00:56] then install grub on the new system [00:56] So mount my current setup on mount then mount md0(boot) as /mnt/boot is that right? [01:00] p7ank5te7: yes [01:00] p7ank5te7: just make sure you mount the new raid on -mnt first [01:00] p7ank5te7: /mnt/even [01:01] p7ank5te7: /mnt even === Madkiss_ is now known as Madkiss [01:03] Ok. So I follow now, mount the raid root as "/mnt" then the raid boot as "/mnt/boot" === Ursinha-afk is now known as Ursinha [01:08] p7ank5te7: mount raid root on /mnt, mkdir /mnt/boot, mount raid boot on /mnt/boot, do the rsync [01:08] install grub [01:08] try to reboot into that - should work === heathjs is now known as heath [01:12] Let's see if it works === ejnahc_ is now known as ejnahc === phunyguy_ is now known as phunyguy [01:37] RoyK, i'm up to the Rsync part. As far as install grub, when I do that, just purge it and reinstall it and have it set to the secondary drive? === Sargun_ is now known as Sargun [02:03] p7ank5te7: install-grub /dev/blah [02:06] typically /dev/md0 [02:11] Will that work with grub 2? [02:14] yes [02:15] but gotta go - time's like 4:15 [02:35] Hi, I have added myself to the web group (www-data), when creating a directory the ownership goes to the user then the group (foobar:www..) is there a way to set it to the group then owner or what I would like is to keep it as the group but still permit me the user to add, delete and read files? === Guest49957 is now known as jrib [02:37] I have an emergency for work! I'm having problems mounting devices so I decided that I'll try wget or curl or something to retrieve the files from another location. [02:38] Are we able to use ubuntu one with server, or even dropbox? I just need a simple solution to get two small files on my ubuntu server vm [02:44] dropbox has a command interface so yes, n a quick check on google has given me to support what I thought http://rbgeek.wordpress.com/2012/08/19/how-to-install-and-configure-dropbox-on-ubuntu-server-12-04-lts/ [02:46] I assume ubuntu one would be the same but I havent checked [02:57] thanks amarcolino === roasted_ is now known as roasted [04:42] Can anyone answer a question? [04:44] Considering moving multiple services that are currently running on winblowz servers presently. Does anyone know if 12.10 or 13.04 server edition fully utilizes the RAM on a machine? [04:45] for some reason I am only seeing 8 gigs out of 16 on a box I installed it on... just seemed odd... wanted to see if anyone ran into the same issue or not... thanks for your time. === blkperl_ is now known as blkperl === DasIch_ is now known as DasIch [07:42] i dont see any /var/log/fail2ban.log . the iptables and service fail2ban status says that fail2ban is running fine.. any clue? === klaas- is now known as klaas [08:08] i would say then which distros are best for servers if wanted to be free and maintain our selfs? [10:23] ports can be forwarded by routers/switches and NATing can be done. how to do that if the thats a linux server instead? [10:24] what is the terminal command that goes through the folder / file sizes and let's you go through them like if it were explorer [10:24] i remember it starts with "n" ? [10:31] dassouki: "find" and "du -sh" :) [10:33] ya i use those two but there was a command app, that cycles through your folders and displays the info in a tree, so you can go deeper into a folder, etc === hodge is now known as Hodgestar === sygnous_ is now known as sygnous === Jever| is now known as Jevermeister === sygnous is now known as malikeye === malikeye is now known as sygnous [13:31] Hi, I have added myself to the web group (www-data), when creating a directory the ownership goes to the user then the group (foobar:www..) is there a way to set it to the group then owner or what I would like is to keep it as the group but still permit me the user to add, delete and read files? [14:30] amarcolino, www-data is the user the web server runs as and should not own any files; you should create a new group for website file ownership [14:31] take note that, out of the box, all files in /var/www are owned by root, not www-data [14:34] I've already changed ownership to the www directory as well as adding myself to the www-data group, however, the issue is when I add files or directories www-data gets second ownership, which is not helpful because some directories and files can't be auto-updated since the web group is meant to have full access, so I am wondering how I can solve this and if it is possible other then manually changing ownership? [14:41] anepanal1ptos: chmod +t /var/www [14:41] that should make group ownership stick [14:44] RoyK, will try that even knowing it was intended for someone else [14:45] heh - sorry ;) [14:46] chmod g+t, perhaps [15:03] RoyK, no good, if I were to create a test directory ownership would be my username and group would be www-data, however, I just thought that wouldn't be a problem if everytime something is created it would be given read and write permission by default, which it isn't [15:05] Royk, thanks I will work on this when I have more time [15:05] then the umask is wrong [15:07] Royk, how would I check the umask? [15:15] chat clients disconnect (after a while) when i switch to another Xorg server (by using ctrl+alt+F8). how does the former Xorg server ("at F7") tell that i switched away from it? i need to suppress that. [15:24] amarcolino: depends on how you upload things [17:52] In my home directroy in Ubunty server, I have a file "=" what is this file? [17:53] cat "=" gives 13.04 12.10 [17:56] bjrohan: That's not part of the distribution [17:57] seems a pretty sane guess that it's a mistake somewhere. something that's directing when its meant to be comparing [17:58] Hmm. I installed the distro fresh, installed KDE, and alfresco (with tomcat, sql etc) that is all [18:01] There's might be an error in the Alfresco installer that creates that file in $PWD [18:06] bjrohan: Did the installer exit successfully? [18:06] Yes [18:06] In that case, I'd remove the file and carry on [18:07] Ok. I am the owner of the file [18:07] If you ran the script as your user, you'll be owner of any files created by the script [18:10] What would the ownership look like if it was something that was created during the dist install? [18:12] Also, I encrypted the user home directories on install. Given this, can I using sudo 1.) see, or 2.) access files in another users encrypted directory? [18:14] Say the file and dir both do NOT have not permissions granted to anyone but the user [18:16] sudo ecryptfs-recover-private [18:22] jacobw: If I am understanding right, that is a way to recover encrypted data if one knows the passphrase (say a hdd craps out and you can no longer boot from it, use this command from a livecd to recover the date)? [18:23] That's right [18:24] bjrohan: ecyptfs volumes are mounted when the user logs in and unmounted when the user logs out [18:24] To get down to brass tacks, if users home directories are encrypted, I as the sys admin, can not use sudo to access their files b/c I don't have their passphrase, whereas if it is NOT encrypted, I can use sudo to access the files. [18:24] That's right [18:24] Nice, I understood it correctly :-) [18:26] bjrohan: You can see their files when they're logged in because you have root access to the filesystem and their encrypted data is mounted in the filesystem [18:27] bjrohan: When they're logged out, their encrypted data is just an encrypted file that you don't have the passphrase to [18:27] That makes sense :-) [18:28] Thank you for the info [18:28] bjrohan: However, there's the passphrase and the encryption key, the passphrase is just a wrapper around the encryption key. If you have the encryption key you can always decrypt the file [18:29] bjrohan: The encryption key cannot change after the file is created, if you log the encryption key when the ecryptfs is created, you can access it whenever you want [18:30] bjrohan: The passphrase is just a way of accessing the encrypted encryption key on disk with the login password [18:32] Very similar to GPG [18:32] It's the same scheme [19:00] what would cause a SSH session to hang during a write to a new ZFS tank? I just made a new tank with RAIDZ2, and now when I move anything larger than a 1Gb file, it locks up [19:09] Jeeves_Moss: zfs-fuse? [19:09] do apparmor profiles use "#" for comments? [19:10] this is my /etc/apparmor.d/libvirt/TEMPLATE: http://paste.scsys.co.uk/247709 [19:10] #include [19:10] RoyK, no, native [19:10] is that actually including anything? [19:10] RoyK, there dosen't look to be any disk activity [19:11] Jeeves_Moss: ubuntu doesn't have native zfs [19:11] apt-get install ubuntu-zfs [19:12] no such package [19:12] RoyK, and have they fixed the expand features so you can add disks to a pool to expand the sapace? [19:13] RoyK, ubuntu-zfs is already the newest version. [19:15] Jeeves_Moss: it's not in the repos === wizonesolutions_ is now known as wizonesolutions [19:16] RoyK, I literly just ran it [19:16] E: Unable to locate package ubuntu-zfs [19:16] that's on raring [19:16] and quantal [19:16] https://launchpad.net/~zfs-native/+archive/stable/ [19:17] yes, it's a ppa [19:17] yea [19:17] not official ubuntu [19:17] hummm [19:17] I'm wondering if I should roll the data out of the tank and fire up the fuse version === Jikan is now known as Jikai [19:19] I've been working with zfs for some years [19:19] on opensolaris/openindiana [19:20] and it's brilliant for its use, but I chose md for my home server [19:20] zfs lacks flexibility [19:21] how does zfs lack flexibility? [19:22] zfsonlinux works rather well on ubuntu [19:24] It depends what you need, most of the components of ZFS are available in other stacks [19:24] RAID/LVM is simple, COW isn't [19:27] ugh, this lockup is pissing me off [19:27] how can I track IO issues [19:28] Jeeves_Moss: iotop [19:28] Jeeves_Moss: Also, see IO wait/delay in top and other things [19:28] jacobw, shows nothing moving [19:30] Jeeves_Moss: Do you see anything in syslog/dmesg/kernlog? [19:30] jacobw, checking. I've tried moiving from USB to the root disk, and everything works. It just locks up when I try to write to the ZFS tank [19:32] jacobw, failed: use vmalloc= to increase size. === Jikai is now known as Jikan === lifeless_ is now known as lifeless [19:42] blkperl: you can't add more drives to a VDEV [19:43] blkperl: you can't rebalance a pool if more VDEVs are added [19:43] blkperl: really, it's just fixed - it's good for well-planned long-term storage, but not if you want it flexible [19:44] blkperl: and you can't change RAIDz levels either [19:44] nothing can be changed [19:45] yea. once it's built, you're kinda effed [19:45] this one is for a storage pool for VMs [19:46] zfs rocks - but it's not very flexible - that's my motto [19:46] It's flexible if you have a large array [19:46] no, it's not [19:46] you can add a new VDEV, ok, but it won't rebalance the raid [19:47] vdev being a block device? [19:47] LV/whatever the ZFS term is [19:47] that'll require the block pointer rewrite, which was posted as an idea some four year back [19:47] but never implemented [19:48] a vdev is a mirror or a raidz? [19:49] zpool create mirror d1 d2 mirror d3 d4 [19:49] makes two vdevs [19:49] I haven't heard the term vdev before [19:49] heh [19:49] then you don't know zfs [19:49] Well, I don't [19:50] Now Unity has crashed :| [19:50] zpool create asdf raidz2 d1 d2 d3 d3 d4 raidz2 d5 d6 d7 d8 d9 [19:50] two vdes [19:50] vdevs [19:51] (albeit bad syntax) [19:51] RoyK: ah ok, yep those things will probably be around for a long time [19:51] OK, so vdevs are PV in the LV group in ZFS terminology? [19:54] http://pthree.org/2012/12/04/zfs-administration-part-i-vdevs/ [19:54] :) [19:55] jacobw: really, I've been working with zfs for 3 years or so :P [19:56] OK [19:57] no, in LVM terminalogy, a PV is a disk, a VDEV is a group of disks set in a mirror or raidz-something [19:57] I see [19:57] better use MD if you want flexibility [19:58] chances for silent errors are rather low with small volumes, as in <20TB [19:58] or <100TB even [19:58] Agreed [19:59] Er, silent errors? As in write holes? [19:59] silent errors are when the drive doesn't report an error but delivers bad data [19:59] I see [20:00] whenever they get bp rewrite into zfs, it'll be fun [20:00] What do you mean by rebalance the RAID? [20:00] BP? [20:01] block pointer rewrite [20:02] if you add another drive to an md raid, it'll re-balance the data [20:03] if you add another vdev to zfs, it won't do much, just try to smear the data over the lot [20:03] rebalancing means moving the data over the existing vdevs to make performance better [20:03] I see === highvolt1ge is now known as highvoltage [20:05] What kind of performance difference does that make? [20:06] I didn't know mdraid did that [20:13] jacobw: it means new writes will only go to the new vdev [20:13] earlier it would make the zfs performance almost halt, but that's been fixed [20:16] I see [20:16] It sounds as if rebalancing is different the RAID and ZFS contexts [20:18] I'm trying to find references to it and hitting a lot articles about distributed filesystems like GlusterFS and proprietary SAN stuff [20:24] glusterfs comes on top of zfs or md [20:25] zfs is fine if you have a fixed set of disks and can plan things well [20:25] if you want to add more disks, it gets complicated [20:26] In Linux, how is it possible to merge/combine 3 internet connections (Dsl/fiber) (actually by eth1,2,3) and supply those to eth4. many client can be connected on a switch that is connected to eth4 of the server. The server does the load balancing and if one eth of 1,2,3, goes down. it shifts traffic to remaining 1,2. I have seen this in microtck router. How is it possible in Linux? [20:28] Quest: see http://www.lartc.org/ [20:28] Quest: it's not trivial, but quite doable [20:32] RoyK, hm. thats how microtick did it? [20:34] Quest: well, just try [20:34] where to start [20:34] read the docs, for a start [20:34] oh ok [20:56] is there a way to put a network bandwidth quota on each user when the network usage reaches above certain threshold? (something easy to use compared to ip tables) === heath is now known as data_scientist === data_scientist is now known as data_astronaut [21:16] Quest: Mikrotik RouterOS can bond interfaces using LACP, if that's what you want to do === data_astronaut is now known as heath [21:20] jacobw, what is lacp? [21:21] Quest: If you want to balance sessions over 3 subnets read those docs, and have a look at this, http://parkersamp.com/2010/03/howto-using-linux-as-a-simple-load-balancer-nat-router-firewall/ [21:23] Quest: Link Aggregation Control Protocol, it combines Ethernet ports in to channels, like port channel [21:23] hm\ [21:24] jacobw, i thought they used http://lartc.org/howto/lartc.rpdb.multiple-links.html [21:24] jacobw, and i think you have experience with mikrotick OS. its linux of their own? or theres some underlying os they use like centos or ubuntu? [21:25] Quest: I wasn't sure what you were talking about at first [21:26] jacobw, and i think you have experience with mikrotick OS. its linux of their own? or theres some underlying os they use like centos or ubuntu? [21:28] Quest: RouterOS is Linux based, but Mikrotik have replaced parts of the networking stack and included in house implementations of things like MPLS [21:28] Quest: If you look at the firewall interface, it's quite clearly iptables [21:29] hm [21:29] jacobw, i was wondering if it uses a known os underneath so i can use a package manager and install apps further in it? [21:30] Quest: No, that's certainly not possible :) [21:32] :( [21:33] Quest: RouterOS uses the Linux kernel only, it uses Mikrotik own userspace [21:34] hm. thanks [21:34] it will take quite long for me to do what mikrotik doesn, in pure linux and iptables [21:35] Quest: http://wiki.mikrotik.com/wiki/Manual:IP/Route#Multipath_.28ECMP.29_routes [21:37] Quest: The page you just linked from LARTC has everything you need for your scenario [21:41] jacobw, thanks! === bobbyz_ is now known as bobbyz