[01:10] Can someone explain to me how to read the instructions on a web page. Specifically, some lines of instructions place a # infront of the command. [01:11] example "#sudo vi /etc/samba/smb.conf" when I type it nothing happens. === RudyValencia- is now known as RudyValencia [01:11] # prefix is remark [01:12] Is it the same as commenting out a line of code? [01:12] yes [01:12] So then why do they tell me to use it? Did they make a typo (multiple times) on the tutorial? [01:13] probably there is a space between the # and the command, in a shell this usually means that you have to run the command as root (indicated by a # at the prompt [01:13] I'm trying to setup Samba and I get to where they say "do this: "#command" " and nothing happens. [01:13] so in this case, the # is not part of the command [01:15] now why you would need sudo when you alread have a root (#) prompt is beyond me ;) [01:22] jeiworth, after using the vi command to open a file, how do I exit? [01:25] type: ":q!" [01:26] if you are not familiar with vi i suggest you use nano instead [01:26] I have no idea what these commands are, I'm just trying to learn and follow the tutorials to set up everything [01:32] jeiworth, I am trying to explore my existing windows network, is Samba required? I get an unable to mount error when I try to browse. [01:35] for browsing you should just need samba-client iirc [01:35] and avahi [01:36] on a server i usually mount samba shares using cifs [01:36] I installed Samba [01:36] I'm in ubuntu desktop btw [01:38] and I'm confused. mind explaining? [01:41] hmm [01:42] well, samba is a bit tricky, anyway... try this: in your filemanager put this in the navigation bar: smb:// and hit enter [01:43] I dont have that option [01:43] I can't type in the location .............. [01:44] oh, there we go [01:46] okay, I got an error. Failed to retrieve list. [02:16] hi all.. I've found a project on bazaar.launchpad.net that I'd like to download and test (no deb it's a php web app). It has a decent tree structure.. How can I down load it easily.. as opposed to recreating the tree locally and downloading files one at a time.. there's got to be a better way.. so far wget doesn't like me.. Any help appriciated :) [02:21] izinucs: the project should provide a tar.gz which is just a compressed file that can be extracted into web server directory. [02:25] thesheff17: nope.. not on this one.. [02:26] izinucs: which project? [02:26] thesheff17: http://bazaar.launchpad.net/~mparic/openauctionlive/trunk/files/head:/ [02:37] izinucs: ok I got it...I wasn't too familiar with bzr command [02:38] first install bzr apt-get install bzr [02:38] then clone the repo: bzr clone http://bazaar.launchpad.net/~mparic/openauctionlive/trunk [02:41] thesheff17: interesting... I had to create a ssh key pair to do it.. [02:41] thesheff17: success! looks like I got it.. just another learning curve [02:42] izinucs: yea I have never used bzr but I assume it similar to git/svn etc. [02:42] izinucs: at least with cloning the repo [02:44] thesheff17: well your help was invalueble. I've been looking for days on how to do it. [02:47] thesheff17: If I delete the branch off my HD will it effect the online branch? It seems the directory I downloaded it in now gives me the option to "update" from the online branch.. I just don't want to wipe out the other guys work if I delete off my HD [02:48] izinucs: it won't break anything on the site. [02:49] izinucs: you would need commit access and then he would have to merge the branch. etc [02:49] thesheff17: ah.. ok..good to know [02:54] thesheff17: ok.. to copy the entire mess from me local to the machine I'm ssh'd into (also on my lan).. while in the ssh session on the remote machine I ... scp -r @IP:/home//path/to/files/* ?? but it doesn't work .. syntax error I believe [02:56] izinucs: scp -r * username@destServer:/folder/ [02:56] make sure you are in the dir you want to transfer [02:56] yep.. syntax.. thanks [02:58] thesheff17: that gives me a *: no such file or directory [02:59] that means the folder you are trying to send it to doesn't exist on the dest. [02:59] except I'm in it in the ssh session that I'm initiating the scp from [03:00] or maybe I should be local doing it [03:00] izinucs: you can unlimited logins to linux machines. [03:00] izinucs: are you on a server ed or do you have GUI? [03:01] not sure I understand that sentence [03:01] both.. I'm ssh'd into a server and on a laptop [03:01] with gui [03:01] izinucs: ctrl-t will just bring up another terminal tab [03:01] and ssh into the destination server and mkdir the folder you want [03:02] did that.. I'm actually on the server now using screen and irssi [03:02] izinucs: try the scp command again [03:02] from the server side or local side [03:03] izinucs: scp works both ways if the two servers can reach each other [03:03] izinucs: to push files is scp -r * user@ip:/folder/ [03:04] I think I got it.. I have to push it to /home on the server then cp to the /var/www directory and change owners.. [03:04] I can't scp directly to /var/www [03:04] izinucs: prob because you aren't using the user root to send the files [03:05] izinucs: but yes sending to the home dir and then moving/copying the files around and fixing permissions is safer.....I usually allow ssh root access. [03:06] I have in the past as well.. just haven't done that on this machine. [03:34] * Slyboots blinks [03:34] Uh.. how fast is Wifi G again? [03:35] Im pretty sure its faster than 2MB/sec [03:41] Slyboots, in practice, not really [03:42] the data rate is 54mpbs, but that doesn't account for interference and the fact that all associated devices divide that bandwidth up [03:42] Man; its more garbage than I though === bastidra1or is now known as bastidrazor [03:43] wired is always better [03:43] even 10baseT [03:43] mM [04:52] I just installed ubuntu server 10.04, i'm trying to set up a print server. I'm assuming I need to compile a driver for my HP P1006 printer. I followed the instructions at http://foo2xqx.rkkda.com/ .. I installed build-essential and tried to 'make', was told to install dc, installed and i'm getting an error at "zjsdecode.c: In function'decode':" then "zjsdecode.c:153: internal compiler [04:52] error: Aborted" any ideas? Thanks. [04:59] source codes has somekind of problems [05:00] jpj@bumba:~/Downloads$ apt-cache search foo2zjs [05:00] foo2zjs - Support for printing to ZjStream-based printers [05:01] why dont you use one from packet management? === JanC_ is now known as JanC [07:00] hey, guys! [07:01] need help [07:02] well, mind to ask your question ;) [07:03] i need to install dvd ubuntu without DVD-drive... i have USB 4gb hi speed... but for install i need delete some programms... but how? [07:04] delete from where? [07:04] from complect [07:05] sorry, i didn't get you. You want to install it from USB-stick, why you need to delete programs? [07:07] cause DVD version weights 4.3gb.... stick is 4gb... (exact 3.9) [07:09] "dvd ubuntu" sounds like a desktop CD [07:09] We support the server install media here [07:12] i'm sorry... [07:12] where can i ask about this situation? [07:13] so you are using desktop? [07:13] #ubuntu then [07:13] mark-kraevskijj: but i would buy bigget stick, or use alternative and install rest from internet [07:15] Or just install the ENTIRE THING from the network [07:15] yeah, desk.... [07:15] that's option also, but alternative installer gives you base system easily [07:16] yeah... you right) [07:17] http://archive.ubuntu.com/ubuntu/dists/hardy/main/installer-amd64/current/images/netboot/ boot.img.gz [07:17] i just wanted to delete some from distr... it'll become lighter...) [07:18] it's not that easy [07:18] but you could mount it to linux with loop option and remove deb-files, however that's quite stupid way [07:18] you waste lots of time [07:20] gobbe: or jigdo [07:21] But in that case you're still downloading the pool in advance [07:22] yes [07:22] hey guys.. i'm tryin to find soem Ubuntu Server package for a hosting server.. but apt doesnt find them.. what can i do [07:23] Vamps_: pick an alternative? [07:23] Vamps_: what package? [07:23] libsas12 pkg and sas12 [07:24] gobbe, how to mount? [07:24] i'm intalling 10.10 from the HOWTOForge ISPconfig install [07:25] mark-kraevskijj: with loop-option [07:26] how? [07:26] Vamps_: its libsasl2 [07:26] gobbe : libsasL2 lowercase ofc [07:28] i runned amazon instance with ubuntu 10.04, but sshd not in autoload. how to add sshd in autoload? [07:29] gobbe : thank you very much .. that solved the problem.. couldnt read it on the website as an L and not a 1 [07:29] is openssh-server install vlad? [07:29] openssh-server installed by default [07:30] sshd just not in autoload [07:30] update-rc.d ssh defaults [07:31] hi. can any one tell how to configure domain controller on Ubuntu server 10.04 (64bit) [07:31] thanks! [07:32] update-rc.d ssh enable [07:33] awanti, if by domain controller you mean a Microsoft Active Directory Domain Controller, that's only partially achievable [07:33] there is no drop-in replacement, yet [07:37] no actually i am going to install Ubuntu server 64 bit OS on my hp server machine. Now i want to create a domain, from that domain some windows 7 machines will be loging. [07:37] So i need your help to setup [07:39] that's what I meant [07:39] awanti, google : ubuntu bloke ldap [07:39] it's not currently possible, at least not completely [07:40] qman__, what do you mean ? [07:40] the software to do it is still in alpha last I checked, samba4 [07:40] qman__, active directory is just ldap samba & kerberos [07:40] and microsoft ldap schems [07:41] standard samba3 can only emulate NT/2003 level, and windows 7 is not going to like it one bit [07:43] ah i see now, looks like it will be a lot neater [07:44] qman__: hang on now, you can easily reconfigure Win7 to bind to a Samba3 domain. [07:44] it's a lot of work and I've not yet found a guide that actually works and covers all bases [07:44] i actually like using apache directory server instead of openLDAP [07:44] of course it's been a couple years since I last tried [07:45] but aside from samba4, I haven't seen anything to change my mind [07:45] all you need to do is google for windows 7 samba. First hit will be here: [07:45] http://wiki.samba.org/index.php/Windows7 [07:46] visit that website from the win7 client and follow the bugzilla link, it will make the necessary registry changes (2 keys) [07:46] you can't do group policy from samba3, but you can configure accounts through openldap [07:49] guide to setting up the Samba DC is here: https://help.ubuntu.com/10.04/serverguide/C/samba-dc.html [07:52] ugh.. ok one more question.. if i didnt put an X in the apache2 box for the phpmyadmin install.. does it use apache automatically or will i have to edit [07:53] apache is the default web server, so probably [07:53] unless it, for some reason, doesn't actually depend on a web server [07:53] qman: it gave 2 options apache2 and lithttpd [07:54] php5 will use mod_php instead of php5-cgi by default, though either would do [07:54] how can i check to see if mpa is set to apache2 after install [07:55] run aptitude install -s phpmyadmin [07:55] that will tell you what will be installed. On my machine it said that yes, it would install apache2 [07:56] mine said no packages to install [07:56] type aptitude show apache2 and that will tell you if it is installed ... you may have it already [07:57] i do know i installed apache2.. im tryin to figure out if phpmyadmin when it installed and config phpmyadmin.. i didnt X the apache2.. will it still bind to apache2 [07:57] ahh [07:58] how do i check that [07:58] well you can run sudo dpkg-reconfigure phpmyadmin to get the configuration prompts again [07:59] perfect let me try that first [07:59] reinstall DB or no [08:00] hmm probably not necessary [08:00] there we go.. got a different output then the first time.. thanks a7ndrew [08:00] a pleasure [08:07] hii guys... im connecting to an ubuntu machine using VNC... i want to get console access via vnc.. how can i do this? [08:08] twister004, use ssh [08:08] qman_.. i need to see the gui [08:08] it's a lot more secure and a lot less complicated [08:08] some connection changes [08:09] Im using ssh.. but I need gui access [08:10] well, that's not really an #ubuntu-server question, since ubuntu server doesn't have a GUI [08:10] someone may be able to help with that but it isn't me [08:10] sorry [08:12] thanks qman__ [08:19] man i dislike edit fstab for quota.. [08:22] ok, so im edit fstab with the following entry edit on the / dir ## the edit is ,usrjquota=quota.user,grpjquota=quota.group,jqfmt=vfsc0 .. when i mount -o remount / it fails.. why [08:24] wait.. is there supposed to be a $ at the end of the line after 0 0 [08:24] or is 1 the correct value [08:51] how do i get back to the *@* instead of the $ prompt [09:32] /etc/apache2/httpd.conf shouldn't be empty right ? [09:34] nvm [09:44] Greetings, has anyone had any experience using virtual box to set up an Ubuntu 10.04 LTS cluster for local testing? [10:12] I'm having trouble getting my ubuntu box to show up as a shared device on my mac [10:12] I'm able to connect to it with bonjour but its just not broadcasting I suppose. [10:15] Hi. What is the prefered ftp package? [10:17] New bug: #225398 in backuppc (main) "package backuppc 3.0.0-4ubuntu1 failed to install/upgrade: Unterprozess post-installation script gab den Fehlerwert 1 zur?ck (dup-of: 445154)" [Undecided,Confirmed] https://launchpad.net/bugs/225398 [10:21] Adman65: ssh [10:21] New bug: #702812 in backuppc (main) "Please merge backuppc 3.2.0-3 (main) from Debian unstable (main)" [Undecided,Confirmed] https://launchpad.net/bugs/702812 [10:22] Adman65: for read-only exports, HTTP is a better protocol; for uploads, SFTP is a better protocol. [10:22] ok, what do you recommend for sftp? [10:23] OpenSSH. === chrislabeard_ is now known as chrislabeard [10:37] hello, a short question regarding dovecot - what would be the best way to install dovecot 2.0 in ubuntu (10.4)? This version of ubuntu has only 1.2.9 in apt repositories, so is there any way to do it (I would prefer not building the source myself) [10:42] Neikius: This PPA has dovecot-2.0.3: https://launchpad.net/~kramerica/+archive/kra-dovecot (I have no experience with it) [10:43] i love driveby questions [10:43] :< [10:53] andreasf: just btw, does 10.10 have dovecot 2? I am contemplating how to upgrade so I can mantain this mess easily, but I *need* the 2.0 (dsync) [10:54] Neikius: I have no idea (I don't use dovecot). Have a look at http://packages.ubuntu.com/ [10:57] Version: 1:1.2.12-1ubuntu8 [10:57] in 10.10 [11:04] meh ... they stick to stable as crazy. rh has 2 for some time already I think :) means I will have to download from outside source. Thank you both. [11:06] Neikius: I was gonna check whohas(1), but it can't see the RHUN :-/ [11:06] Ahaha, it fails with -d fedora, too. [11:09] damn lol, even natty still has 1.2 [11:10] is 2.0 so unstable or what? its out for quite some time now already [11:11] anyway here -> http://wiki2.dovecot.org/PrebuiltBinaries link to xiaoka repository, but hey, I prefer to use the official ones no wonder why... [11:11] Neikius: Ubuntu probably just takes whatever Debian dishes out [11:11] They spend all their money on desktop wank [11:14] it IS rock stable on server ... but a bit far behind with features [11:19] If by "stable" you mean "regularly makes itself unbootable for bloody silly reasons" [11:30] ah anyway, I just noticed that xiaoka only had 32bit packages [11:30] I guess I will have to build it myself :( [11:30] Neikius: does it have source packages? [11:31] (.dsc, .orig.tar.xz, .debian.tar.xz) [11:32] it has [11:34] apt-get build-dep dovecot; apt-get install build-essential dpkg-dev devscripts; dget ; dpkg-source -x ; cd ; debuild [11:34] ...that'll give you a package built against your current system (i.e. release and arch) [11:37] since I have added the repo do I need to operate via urls to get the source and dsc? [11:37] If you added a deb-src line, then you can apt-get source it [11:38] ah [11:38] just so btw [11:38] I started getting an error about libapache2-mod-fcgid but haven't the time to check ... since some update this week [11:38] any clue? if not just ignore this for now [11:38] Sorry, I don't read minds [11:39] I've done aptitude safe-upgrade, since then I am having some weird error :) mah disregard this [11:40] hm [11:40] I can do 'apt-get source dovecot' [11:41] ah I will google it [11:44] I see the difference now, dovecot from ubuntu/debian has like ... 10 patches applied so this is the risk [11:48] That is called "integration" [11:48] it's why debian (or ubuntu) isn't just a big bag of upstream sources and an .ebuild [11:49] and why I can always do apt-get install dovecot-postfix and it will work I guess [11:50] yea I appreciate it, that is why I do use debian/ubuntu [11:50] but sometimes I need a new feature like this about dovecot and then its all custom onwards [11:51] If your fu is strong, you can import debian/patches into the new upstream [11:52] I'm guessing your not ready for that yet, grasshopper [11:53] time is all [11:53] some day maybe :) thank you for the help though, you have been most helpful from what I got [11:55] That's because I'm a Debian user :P [11:57] I almost died yesterday when I had to hack a gentoo arf [11:57] I guess debian makes people soft :) [11:57] I only deal with gentoo systems when they stop booting [11:58] oh that one was easy, just force mdadm to boot with 1 disk ... [11:58] the dude that made it locked out all ssh access to virtual machines and the host [11:58] I mean password access [11:59] Well, single-factor auth IS evil [11:59] it is, but when you get a custom-built machine with no manual and no access its just tedious [12:00] while the machine has to run in the background ofc [12:01] the root machine doesn't even have modules to mount snapshots in lvm [12:01] oh I missed debian so much :) === oCean_ is now known as oCean === lenios_ is now known as lenios [14:09] In /etc/network/interfaces, what is the difference between 'auto' and 'allow-hotplug'? I have read man 5 interfaces, however, it barely describes this. [14:24] tdn: iirc allow-hotplug will up the interface when detected. I've seen this fail at times, so I just use auto [14:25] <\sh> RoyK: allow-hotplug is debian, ubuntu doesn't have allow-hotplug ... read the bonding release notes from aeh karmic or lucid [14:25] k [14:26] <\sh> RoyK: /usr/share/doc/ifenslave-2.6/examples/two_hotplug_ethernet is a good example in lucid...karmic had a bug in the doc [14:46] \sh, ok [15:06] New bug: #702903 in tftp-hpa (main) "package tftpd-hpa 5.0-11ubuntu2 failed to install/upgrade: Pakiet jest w złym stanie - powinien zostać przeinstalowany przed usunięciem." [Undecided,New] https://launchpad.net/bugs/702903 [15:23] good day people! i hear webmin is a no-no now. any suggestions for a replacement? [15:29] WinstonSmith: not really [15:30] pmatulis, ok thanks. what is the equivalent of update-rc.d now? [15:31] i mean to enable/disable services? [15:32] hi, when updating apache "service apache2 start" does not work anymore, the reason is in /etc/init.d/, there is now a new script: apache2.1 while apache2 is zero bytes. why does this happen? (ubuntu 10.04). [15:33] 10.04 should have the apache2 script [15:35] it does, but it is only zero byte after the update, and a new one is there apache2.1 which does the job... [15:35] but this is a bug for sure. [15:35] well, report it :) [15:35] WinstonSmith: it's a little confusing i'm afraid, due to some services using upstart and some not [15:36] ok found it in launchpad already... some people were reporting it last year... [15:36] pmatulis, exactly my point. when i finally had the old System-V style figured out they changed :( [15:37] * RoyK installs native zfs [15:37] pmatulis, so how do i disable o service using upstart? [15:38] s/o service/a service/ [15:38] WinstonSmith: and to my knowledge update-rc.d is even not guaranteed to work with all sysvinit services [15:38] :O [15:38] * WinstonSmith thinks i actually hadnt figured out anything [15:38] WinstonSmith: upstartified services should have conf files under /etc/init [15:39] I think initctl is the upstart equivalent of update-rc.d [15:40] pmatulis, so if i just move the script out of the directory should i be safe? [15:41] genii-around, thanks that looks promising. there is no option to disable a job though [15:42] WinstonSmith: what release are you using anyway? [15:42] 10.10 [15:42] sorry should have supplied info right away [15:47] WinstonSmith: you can either rename the upstart file (ex: ssh.conf.OFF) or comment out the 'start on' line in ssh.conf [15:48] pmatulis, thanks! have a nice day & a better weekend! [15:49] WinstonSmith: in 11.04 you will be able to add a one-word line in the conf file: manual [15:49] pmatulis, sounds good [15:49] WinstonSmith: i prefer the file rename method as it's obvious what going on [15:49] *what's going on [15:50] pmatulis, yes with one glance of the dir you can discern the disabled ones [15:50] WinstonSmith: all very sub-optimal but... [15:51] pmatulis, yes well linux is a work-in-progress (i like that actually...always new things to learn ;) ) [15:59] WinstonSmith: I made a test script which just runs ls, added it with the old update-rc.d method. Was able to call it with new upstart method: sudo service ls-test and it lists / contents. But adding start/stop or so on as options give an error. Upstart has file /etc/init/rc-sysinit.conf which runs old-style scripts based on runlevels as before [16:00] with linux raid, can i mirror a portion of 2 disks? [16:00] aliverius: yes [16:00] aliverius: yep [16:00] with lvm or not? [16:02] genii-around, thank you for that info - will check that out. but it seems to me that upstart/system-v mix is far from ironed out [16:04] WinstonSmith_: I think because I didn't write a wrapper for ls which would accept start/stop/reload options is why [16:04] basically this is the layout i want: 1 mirrored portion of the disks will be the rootfs and a /home with sensitive data - raid 1, and what is left should be one logical volume that i dont care if they be lost. is it feasible? [16:05] also can i setup 1 disk for raid and lvm with ubuntu server and add the second disk at a later time? and just swap the drives whenever/if they fail? [16:06] genii-around, no i dont mean your test script just the mix of the 2 systems and not so much info around. did you try to stop pulse audio with "service pulseaudio stop" for example? and i searched a bit (well like 5 min) and did not find a easy solution to stop pulse === WinstonSmith_ is now known as WinstonSmith [16:08] WinstonSmith: I don't have any sound system running :) [16:09] genii-around, well just as an example. had this problem on my laptop [16:10] genii-around, i just mean before one could always rely on doing "/etc/init.d/servicename stop" and it WOULD stop :( [16:10] * RoyK thinks the move to upstart was a bit hasty [16:10] * WinstonSmith concurs [16:11] good old sysv scripts work [16:11] * genii-around makes more coffee [16:11] and one could easily write one [16:12] aliverius: you can't seriously think that if you lose one drive you can insert a 2nd blank one and boot up??? [16:13] aliverius: you should start with two drives. if you need to add the second half of the mirror later, that's doable, but not with lvm iirc [16:14] aliverius: adding one later will involve creating a broken MD mirror on the second drive, copying the data, and adding (or "replacing") the initial drive as the second half of the mirror [16:14] As soon as we get used to upstart they'll probably switch to systemd anyhow [16:15] genii-around, yes they like to keep us on the edge ^_^ [16:16] New bug: #665667 in euca2ools "couldn't launch an AMI bundled and uploaded using euca2ools" [Undecided,Fix committed] https://launchpad.net/bugs/665667 [16:18] * WinstonSmith wants a coffee too === NG_ is now known as ng_ [16:43] thank you RoyK. pmatulis that was a more productive answer :p [16:44] genii-around: that's funny but i don't think so [16:46] pmatulis: Because upstart is an inhouse project? [16:48] genii-around: not really, upstart is just a good product. it's introduction has been shaky is all ;) [16:50] * genii-around sips his coffee and ruminates [16:57] pmatulis: Some decent management tool for upstart would probably go a long ways [16:57] * genii-around ponders upstart-rc.d [17:06] i have another question. say my volume group consists of physical volumes in two different disks. then one and only logical volume is created. if one disk fails and so one pv is lost, how possible is it that the data in the good physical volume will survive? [17:06] i mean it will survive but will it be accessible? [17:07] retreivable? [17:08] no [17:09] if using a striped setup without redundancy, the whole volume will be lost [17:09] doesnt sound striped [17:09] the data will still be on the remaining disks, but you'll have a bit of an issue to retrieve it [17:10] aliverius: is this a mirrored setup, or striped? [17:11] neither. contiguous [17:11] then if you lose one pv, you lose it all [17:11] disks are too cheap these days to risk your data on non-redundant installs [17:12] so then better have two separate partitions, instead of one logical... [17:12] i want to get 2 disks. one part of them will be mirrored for safety [17:12] aliverius: that is, the data stored on the drive not malfunctioning will obviously be there, but you'll spend more time getting it out than you want to know [17:13] no tools to do that huh? [17:13] aliverius: even 2TB drives costs less than $100 these days [17:13] aliverius: mirroring is supported with md and lvm, you can use either [17:14] md also supports raid[56], which can be handy if you want more drives [17:15] youd think that JOBD in an lvm would allow one disk the survive one dying [17:16] JBOD [17:16] compdoc: well, he asked, and got an answer [17:16] something thats easy enough to test [17:16] raid5 isn't very handy with big disks [17:16] huh? [17:16] handy? [17:17] gobbe: why would that be? [17:17] RoyK: rebuild takes long, and causes lot's of disk io to rest of disks, and might cause more disk damages [17:17] that's why raid6 [17:17] for extra parity [17:17] I use raid 5 and four 2TB disks [17:17] well, one has five 2TB drives [17:18] rebuild takes a bit longer, sure, but it's still vastly better than striping [17:18] * RoyK uses zfs for that sort of things [17:18] RoyK: that's why you use both ;) [17:19] gobbe: I quite understand the raid5/6 thing - I have a box with 11x7-drive raidz2 at work [17:19] RoyK: i mean that the bigger disk, longer it takes to build and rebuild causes lot's of io to rest of disks, which might cause more damage [17:19] small raidz2 volumes, lots of data [17:19] but i don't say that you shouldnt use it [17:19] disks are quiet cheap so raid6 is better [17:20] gobbe: I can't really agree with the i/o part of this - i/o to one disk doesn't affect that to another [17:20] 3.5" drives are cheap. Im hoping 2.5" drives will become cheap [17:20] RoyK: well, rebuild reads all disks [17:20] obviously, yes [17:21] and risk that disks breaks is bigger when it's under heavy io [17:21] 2,5" drives are quite cheap too, the only thing being rather expensive these days are 10k or 15k drives, and SSDs, especially SLCs [17:21] yep [17:21] but you can't pack 2TB into a slim 2,5" drive [17:21] at least not yet [17:22] * RoyK wants a 2TB SSD for his laptop [17:22] don't get me wrong however, raid5 is ofcourse better than raid1, but due the fact of low disk price, i wouldn't go with raid5 anymore [17:23] one friend of mine has raid5 with three 512GB OCZ colossus's [17:23] :-) [17:24] I seen 1TB 2.5" drives, but they want a lot for them [17:25] the bad thing with large storage and most filesystems, is that they, the filesystems, mostly can't detect corrupted data [17:25] RoyK: that's why you run zfs [17:25] so use ext4 on 10TB and just wait for something funny to happen [17:25] I use a small drive to boot servers from, and raid for storage [17:25] gobbe: indeed :P [17:25] ext4 would cause something funny? [17:26] well, like ext3 or jfs [17:26] they cannot detect corrupted data [17:26] gobbe: http://pastebin.com/qRrQe0cQ [17:26] :P [17:26] and bigger the filesystem, better change that you find corrupted blocks [17:27] this is just 100TB [17:27] is there a native zfs for ubuntu? not fuse, I hope [17:27] RoyK: nice pool :-) [17:27] compdoc: I just installed native zfs for testing - seems to find my pool etc, but didn't mount - it's in early alpha [17:28] where did you get it? [17:28] http://zfsonlinux.org/ [17:28] I tried zfs-fuse, and write times were cut in half [17:28] write speed on zfs-fuse tend to suck, yes [17:28] KQstor promises native zfs [17:29] I liked zfs though - cool features [17:29] compdoc: installing openindiana isn't very hard, though :P [17:29] http://hub.opensolaris.org/bin/download/Community+Group+zfs/docs/zfslast.pdf [17:29] solaris has two good things, zfs and dtrace, but rest is quite crap :-) [17:29] not really [17:30]