=== emma_ is now known as emma === _eB is now known as ebolorama === dendrobates is now known as dendro-afk [00:57] hi [00:58] i have http setup on /var/www----> how do i download the folder and file i have on it. on remote area, by accesing its public Ip.. [00:59] be able to view and download the file adn folder in it using url [01:00] anyone can suggest..? === freenode is now known as Guest73631 === crohakon is now known as crohakon|afk === crohakon|afk is now known as crohakon [02:37] hello. i've got a box running ubuntu-server. I've tried putting two different sata cards in it and it does not pick up either card. i do not even see in dmesg anything about the card. neither of the cards logos prompts during boot. this box did have an sata controller card in the past and used it just fine. any ideas? === dendro-afk is now known as dendrobates [02:53] hello. i've got a box running ubuntu-server. I've tried putting two different sata cards in it and it does not pick up either card. i do not even see in dmesg anything about the card. neither of the cards logos prompts during boot. this box did have an sata controller card in the past and used it just fine. any ideas? [02:54] CppIsWeird: does it show up in lspci? [02:55] benedikt, no [02:56] could i have disabled something in the bios a while ago? can options in the bios prevent sata cards from loading? [02:56] yes, its possible. [02:57] hmm, well now i have to wait a few hours... i decided to do a release upgrade [02:58] took my half an hour on my 10 year old server [03:02] hey all [03:02] anyone alive? [03:02] benedikt, i bet your internet is faster than mine === dendrobates is now known as dendro-afk [03:25] CppIsWeird, could also be a dead PCI slot, it happens [03:25] if you have a different slot, try it [03:26] tried them all === wbrb is now known as u0324 [04:27] is the -server kernel gone in karmic? [04:28] That's what I heard [04:31] For i386, yes. [04:31] For amd64 there is still a server kernel. [04:33] ah [04:35] micahg: though in i386 land, the generic-pae sort of serves the same purpose. [04:36] I thought the -server kernel had other tweaks besides pae [04:38] !pae [04:38] Sorry, I don't know anything about pae [04:39] micahg: you should be able to dget the amd64 .debs, dpkg -x them, and diff the .config.gzs [04:39] That ought to tell you exactly what is different [04:40] Oh, PAE = "I want to address all 8GiB but I can't use the native x86-64 build." [04:46] do i have to manually create a /Maildir for root? [04:47] i'm getting a bunch of permission denied errors on my mailserver [04:47] Nov 26 04:17:58 li127-132 deliver(root): chdir(/root) failed: Permission denied [04:47] they look similar to that one ^^^ [04:47] although i have my user account setup as an alias to root, so i thought they would redirect [04:48] Nov 26 04:16:57 li127-132 deliver(root): stat(/root/.dovecot.sieve) failed: Permission denied [04:49] does dovecot need the managesieve protocol enabled? i'm not sure really what it does, but i remember last time i setup a mailserver on my home server that the only protocol i needed, or ran, was imaps [04:57] 15:47 Nov 26 04:17:58 li127-132 deliver(root): chdir(/root) failed: Permission denied [04:57] billybigrigger1: who does dovecot run as? [04:58] If it isn't the superuser, it probably won't have access to user's home directories -- including root's. [05:06] can anybody give me a hand with vhosts? [05:12] What can we help with marcelcohrs [05:13] marcelcohrs, please keep chat in channel not PM's that way more people can help/learn. [05:14] do you know how i set the permissions to my account to have full access r &w at my www folder? [05:14] im connected via sshfs [05:15] twb: postfix and dovecot-auth are both run as root [05:15] Is /root o+rwx ? [05:15] manageseive-login, pop3-login, and imap-login are all run as dovecot [05:16] although, according to dovecot doc's, protocols = none in dovecot-postfix.conf = login's managed via dovecot-auth [05:16] Dessan: first thing, when i use sshfs to connect [05:16] doesn't seem to be true though [05:16] drwx------ 4 root root 4096 Nov 26 02:22 root [05:17] billybigrigger1: OK, so it looks like li127-132 (whatever that is) isn't running as root [05:17] where do you see that? [05:17] Either that or you have some kind of separate MAC in addition to the posix permissions [05:18] 15:57 15:47 Nov 26 04:17:58 li127-132 deliver(root): chdir(/root) failed: Permission denied [05:18] li127-132 is my hostname [05:18] given to me by the vps host for my node [05:18] Oh, linode [05:18] i changed my hostname to timmy though [05:19] sudo: unable to resolve host timmy [05:19] That was a dumb thing to do [05:19] could explain those messages, i get that after typing every command [05:19] why can't i change my servers hostname? [05:19] Because you haven't changed your server's hostname properly [05:20] edit /etc/hostname and then $ hostname -F /etc/hostname [05:20] You must edit /etc/hosts, /etc/hostname and reboot or run hostname -F /etc/hostname [05:20] ??? how else do you set it up? [05:20] oooh didn't edit hosts [05:20] Because sudo can't resolve "timmy", it won't let you run sudo at all [05:20] :P [05:21] there we go [05:21] messages gone [05:22] now i wonder if i should change dovecot-postfix.conf to protocols = imaps [05:22] instead of protocols = none [05:59] Does 8.04's d-i support the sshd option? [05:59] I thought it did, but I can't find it. [06:00] Oh, maybe I need to start d-i with "expert"? [06:00] "anna-install openssh-server" didn't work, either... [06:00] Grmph [06:11] does phpmyadmin not exist in jaunty anymore? [06:11] New bug: #488600 in php5 (main) "MySQLi Prepared Statements Not Working Correctly With UPDATEs" [Undecided,New] https://launchpad.net/bugs/488600 [06:11] E: Couldn't find package phpmyadmin [06:14] So, how can I make it so I can mount a directory on my server from my laptop? [06:14] NFS? [06:19] crohakon: any network filesystem will do [06:20] For ad-hoc use, I recommend SSHFS, though this isn't appropriate for heavyweight [06:21] billybigrigger1: it's in universe, because it receives no support. [06:21] billybigrigger1: universe was (stupidly) turned on by default in some earlier releases. [06:24] twb: That's a gross over-generalization. [06:25] There are ignored packages both in Main and Universe. [06:25] Certainly the odds are better in Main, but both are maintained. [06:25] twb: Also Universe is still on by default in new installs. [06:33] Okay, so I followed the server guide to install the NFS [06:33] But I cannot seem to mount it from my laptop [06:36] crohakon, by default, ubuntu desktop doesn't have the nfs-common package installed, which is needed to mount nfs shares [06:37] I didn't say it was unmaintained, I said it didn't receive support. [06:37] Granted, I may still be wrong. [06:37] But I was thinking of support contracts rather than maintenance of the package. [06:38] nvm I got it working [06:38] I also didn't realize universe on in 9.10. I thought it was off in 9.10, partly because that'd neatly explain billybigrigger1's problem. [06:39] crohakon: note that you shouldn't run NFS over an untrusted network (e.g. the internet). [06:40] They server is in my basement [06:40] =) [06:41] I just got tired of using FTP... making a change to a CSS file and uploading it. Now I can just open it directly and click the save button to update it. [06:42] hmmm [06:43] apparently I don't have permission to SAVE the file... [06:45] in exports I have: /srv *(rw,synv,no_subtree_check) [06:47] crohakon: your UIDs must be the same on both hosts [06:47] When using NFS, anyway [06:47] This is why NFS is typically used with LDAP or NIS. [06:48] twb: erm, sholdn't kerberos allow for different uids? [06:48] RoyK: I guess, if he's using krb [06:48] I assumed he was using bare NFSv3 [06:48] authentication with UIDs only is not really secure to put it mildly [06:49] RoyK: that's why I mentioned this ;-) 17:39 crohakon: note that you shouldn't run NFS over an untrusted network (e.g. the internet). [06:49] heh - yeah [06:49] sshfs over such networks works well [06:49] It is just on my LAN... [06:50] RoyK: I suggested that, too. [06:50] crohakon: does the UID in question have write access to the dir? [06:50] as twb said, they must match on client and server [06:52] and setting up nis or ldap should be quite quick [06:56] I wouldn't bother to set up NIS/LDAP on a roaming laptop [06:56] nah [06:56] If it's a single-user network, just use usermod to make the UIDs of that one user match [06:56] usermod + find /home/alice -nouser -exec chown as root [06:56] Admittedly, I am a cowboy... [06:57] useradd -u x -g y [06:57] :P [06:58] usermod's too modern - easier to vi /etc/passwd / grop [06:58] s/grop/group/ [06:58] :) [07:00] I figured it out... I had the file permissions set wrong. [07:00] =) [07:00] Thanks for all the information though. [07:00] :) [07:02] There, now I don't have to mess with FTP =) [07:02] Beautiful [07:34] how to compile dhcp-3.0.5 with dhcp-3.0.5-ldap-patch ? [07:34] where to put he patch so that's it's taken into account ? [07:34] i can't find any readme about this [07:45] maxagaz: Are you creating a Ubuntu package, or compiling from tarball? [07:45] jmarsden, compiling from tarball [07:45] jmarsden, and i can't find any eradme file for this patch [07:46] If you do not know what it does and need a README for it, you probably should not be applying it! But if you *want* to apply it, use the patch command just as you would for any other patch. [07:47] man patch for lots of details, usually something like patch -p0 jmarsden, i know what it does... [07:47] Then why are you looking for a README?? [07:47] jmarsden, I didn't know the patch command to apply a patch [07:47] jmarsden, but I know it adds ldap support for dhcp3 [07:48] If you have never used patch in your life, I would suggest doing some basic tutorials before trying to compile and use patched services on your server! [07:48] jmarsden, before i apply the patch, where should i put the patch file, in dhcp-3.0.5 dir ? [07:49] Anywhere you like... patch -p0 Have you really never used a shell input redirect either?? [07:50] jmarsden, okay, but the command should be run in the dhcp-3.0.5 dir, right? [07:50] Yes. [07:50] thanks! [07:51] No problem... but please, take more time to learn about software development before patching critical services on your servers. [08:34] jmarsden, I'll try too, thank you very much [08:34] No problem :) [09:01] New bug: #488641 in ntp (main) "NTP client fails to install" [Undecided,New] https://launchpad.net/bugs/488641 [09:38] where is set the ip address of my vpn clients ? [09:39] how does openvpn attribute an ip ? [09:42] man openvpn and read the example in there ? [09:42] with --ifconfig-pool ? [09:45] Sure, if that's how you are using openvpn. === TeTeT_ is now known as TeTet === TeTet is now known as TeTeT [09:57] jmarsden, no, it's not... [09:57] i'm trying to understand the config someone else made [09:59] The most basic way would probably be openvpn --server NETWORK NETMASK which sets up that stuff sanely for you, unless you (or "someone else") are doing something 'clever' [10:03] jmarsden, everything is configured in /etc/openvpn/server.conf [10:03] vpn clients have addresses like 192.168.x.x [10:03] there are some "push" in the config [10:05] Well, there are three "private use" ranges in IPv4 -- a VPN could use any of them [10:05] Ok... server 192.168.101.0 255.255.255.0 [10:06] I guess it's this pool of addresses [10:06] That's what I said earlier... --server NETWORK NETMASK :) [10:07] In the config file the options drop the leading -- [10:07] As the man page says. [10:07] jmarsden, yes sorry it makes sense now [10:15] hello and good morning [10:53] What is the P-t-P address ? [10:54] Why can't I ping it ? [10:54] in openvpn [10:54] Do you have a route to it? [10:57] anyone uses LTSP? [10:57] twb, yes, i have a route to hit [10:57] acalvo, i do [10:57] !anyone [10:57] A large amount of the first questions asked in this channel start with "Does anyone/anybody..." Why not ask your next question (the real one) and find out? [10:58] twb: because if nobody uses that my question will get stuck in limbo [10:58] acalvo, if you have problem configuring it, you probably better ask in #ltsp === lionel__ is now known as lionel [10:59] twb, my config is this : [10:59] ifconfig-push 192.168.101.73 192.168.101.74 iroute x.x.1.0 255.255.255.0 [10:59] maxagaz: just to know your 2cents about it [11:00] acalvo, 2cents about what ? [11:00] the experience with the LTSP solution [11:01] acalvo, i don't unerstand [11:01] maxagaz: I mean if it suited your needs and you didn't experience any problem [11:02] acalvo, i donfigured it, but i never use it actually, it was just a test [11:02] oh, ok, thanks [11:02] I'll tell you one thing: don't try to deploy LTSP5 on OpenSuSE 10 [11:03] Or any LTSP on any SuSE [11:03] It's a ridiculous farce [11:08] hi, while dumping huge mysql database to smb share this occours in /var/log/syslog: CIFS VFS: Send error in SETFSUnixInfo = -11 [11:08] any ideas ? [11:09] hagedorn: CIFS is the network filesystem that Samba provides. [11:10] i know [11:10] maybe dump file too large [11:10] Google for SETFSUnixInfo = -11, then [11:10] maybe symlink problem [11:11] Are you saying that /var/log/syslog is mounted over CIFS? [11:11] Because that would be bad juju [11:12] twb: thanks for the advice [11:13] LTSP5 on Ubuntu is easy to deploy, because its LTSP's target platform [11:13] twb: no ,... syslog ist not mounted over cfis [11:13] only a backup dir [11:13] its a syslink problem at cifs [11:14] hagedorn: are both hosts running Ubuntu? [11:14] hagedorn: or is the CIFS server a Windows server? [11:14] no [11:14] hagedorn: if both are unix then symlinks should Just Work. [11:15] cifs server is manage trends stortrends server from ami [11:15] think its a redhat oder cent os [11:15] target server is 8.0.4.3 [11:15] Shrug [11:15] ITYM "8.04.3" [11:17] maybe i have to use nfs [11:19] Grr, 10.04's d-i puts grub on the wrong MBR, too [11:19] I hate grub so much [11:22] I guess LTSP can't handle windows machines, right? [11:57] New bug: #488712 in samba (main) "package samba-common 2:3.4.0-3ubuntu5.1 failed to install/upgrade: " [Undecided,New] https://launchpad.net/bugs/488712 [11:58] i'm trying to compile "dhcp-3.0.5" with a patch for ldap, but after "./configure", when I do "make", it returns this: make[2]: *** No rule to make target `-lssl', needed by `dhcpd'. Stop. [11:58] does someone knows why ? === mt-lith is now known as miket === miket is now known as michaelt === michaelt is now known as mt [12:06] acalvo: LTSP is just an OS. It boots on anything - it doesn't matter what's installed on the hard disk. [12:06] ok, thanks twb [12:07] acalvo: if you mean "can I use my Windows machine as an LTSP *server*", then that would depend if your Windows machine is a DHCP, TFTP and NFS server -- which it probably isn't. [12:07] no, my question was if there is something similar that boots a windows-compatible os [12:07] I suspect you could prepare LTSP disk images and serve them from some of Microsoft's high-end server OSs [12:08] acalvo: what does "windows-compatible" mean? [12:09] the LTSP idea is quite interesting, but I'd like to do that with windows OSes === mt is now known as mj-t [12:10] having an opensource server as PXE/DHCP/whatever is needed [12:17] acalvo: you mean that you want Windows thin clients? [12:17] something like that, yes [12:17] That is called "Microsoft Windows Terminal Server" [12:17] or any way to download a thin image from a server (thru pxe I guess) [12:18] twb: but that means you should have "Microsoft Windows Server XXXX", right? [12:18] No, TS is a server OS. [12:19] You license that instead of, say, Windows SBS. [12:19] in other words, you must pay for TS, right? [12:19] You must pay for a license (or violate copyright), yes. [12:20] If you want Windows, you're gonna have to pay for licensing. [12:21] Of course, once you have a running TS server, you can boot LTSP5 on your workstations and use them to connect to your TS server. [12:21] I'll try to find some way to auto download an image from a TFTP server [12:22] twb: ok [12:22] Note that netbooting isn't the same as a thin client -- LTSP is really the latter, but it has branched out into the former. [12:23] AFAIK there is no way to netboot Windows -- and if there is, it's probably not legal. [12:23] well, if only the community I managed will understand and swith to any form of GPL OS [12:23] will be perfect [12:23] but, while stucked with windows, everything is painful [12:23] FOSS code is plenty painful, too. [12:24] yes, but at least it lets you learn, understand and personalize [12:24] windows just let's you "run" (however it does) [12:25] I sure don't let my users understand or personalize their system. Removing that functionality was one of the reasons DoJ got me to replace their existing Windows systems :-) [12:25] But I know what you mean :-) [12:27] I didn't mean to let the users do whatever they want (but they'll try however...), but for the IT dept to create a good OS which covers all the necessities [12:27] stick to the topic, I think I'll end up setting up a PXE server with TFTP, and deploy a new image once every week [12:28] the only thing is how to change the hostname of the windows machine... [12:40] If you think you can deploy a new SOE every week, you are batshit insane. [12:40] Or you are a student of the "agile development" school [12:40] (That being is a strict subset of the former class.) [12:56] Hey guys. Does anyone know how I can tell which ethernet ports have an ethernet wire connected from the terminal? [12:58] jetole: ethtool or mii-tool [12:59] yeah I was looking at the ethtool and ip man page but not sure what I am looking for [13:00] oh, the --test option maybe [13:00] nope [13:05] Just run it without args? [13:05] ethtool: bad command line argument(s) [13:05] For more information run ethtool -h [13:06] thats what happens if I run it without args [13:23] mdz: Hi. About your comments on https://blueprints.edge.launchpad.net/ubuntu/+spec/server-lucid-automated-testing [13:23] mdz: Good point, I hadn't thought of that. [13:23] mdz: I was thinking we could do that in a PPA? [13:24] mdz: So instead of splitting it out and integrating into checkbox, we rely on the PPA's for this and just upload each package we want to test each day with no source changes. [13:25] The PPA's would even e-mail me (or whomever owns the PPA) about failures. [13:30] soren: If it's just to detect build failures, surely that would be better done locally - rather than using a PPA? [13:31] Daviey: Why? [13:32] soren: that abstract basically says "we gonna run tests". Shouldn't a blueprint describe a change (e.g. "we're gonna test foo and bar from now on"). [13:32] Daviey: What would I gain by spending time setting up a build environment when Launchpad gives me one for free? [13:33] soren: PPA is a limited resource, building packages just to see if they compile seems very wasteful.. not to mention slow. [13:33] twb: Meh. There's already too much duplicated information (work items in the status whiteboard and in the actual spec on the wiki). [13:33] Building a local build server isn't that difficult. [13:33] Daviey: I know. I have two of them. That's not really the point :) === simplexi1 is now known as simplexio [13:34] soren: fair enough (re. "meh"). [13:34] soren: well can you elaborate on your point? [13:34] Daviey: I could also build all my other packages locally and publish them somewhere. [13:34] Daviey: Yet I don't. [13:35] Daviey: Perhaps you can elaborate on the "limited resource" thing? [13:35] soren: oh; I didn't realize the page you linked to wasn't the spec itself. [13:36] soren: It's the same reason you build a package locally before putting it in a PPA.. it's slow, and i would have thought most people test build locally before making public. [13:36] Daviey: I'm not sure why it's more or less wasteful to run it on Launchpad instead of a local machine. Wasteful in terms of what? [13:37] You should farm off the compute effort to BOINC! [13:37] * twb ducks [13:37] Unless i'm mistaken, the point of a PPA is making it easy to share things.. Using it as part of an automated test framework to look for FTBFS, seems to me, not what it was designed for. [13:37] Daviey: I test build locally for several reasons: [13:38] a) I'm already actively work on the package in question. Building the package locally is simply a natural step in working on the package. [13:38] b) There's no point bothering the buildd's with needless builds (which a broken build would be). [13:39] Neither applies in this case. [13:39] My screen goes black after an unknown period of time. If I start typing something, the screen shows the text that was previously there. I would like to disable this, can anyone tell me how? [13:39] a) This will happen daily, without humans being involved. [13:39] furom: is this on the tty, or in X? [13:39] twb: this is in the tty/shell. [13:39] b) The exact /purpose/ of this upload is to detect failures that crept in, so it's everything /but/ pointless to have this build run. [13:40] furom: first of all, check your cabling is secure [13:40] It console blanking. It's always been that way. [13:40] soren: I don't agree, but ho hum :) [13:40] twb: I run this in a virtualbox as part of my development environment. [13:40] Daviey: Which part do you disagree with, specifically? [13:40] furom: oh, that'll probably be a virtualization issue, then [13:41] furom: the virtual machine somehow isn't seeing that you're typing all the time. [13:41] twb: Er.. No. It's just regular console blanking. Linux has done this for ages. [13:41] (if not always) [13:41] furom: or you simply aren't typing, so the tty's screen saver (i.e. blanker) enables. [13:41] furom: the setterm manpage describes how to disable the tty screensaver. [13:42] twb: odd.... I've never ran across this issue before.... I figured it was just a screen saver. I spend quite a bit of time not typing in the shell which I know that causes it to go black. [13:42] twb: setterm? [13:42] furom: setterm. [13:42] twb: ohhh manual page for it... I see. [13:42] setterm! [13:42] twb: thanks. [13:43] You may wish to put the appropriate command in /etc/profile.d/no-blanker or in ~/.profile or similar. [13:44] I appreciate that a lot. I think the issue is solved. [13:45] Daviey: AFAICT, whether Canonical has to throw resources (hardware and man hours for setting up and maintaining) at a completely separate build infrastructure for this or just add another machine to the PPA buildd pool seems like a reasonably easy choice. [13:46] Presumably the choice is ultimately Canonical's anyway, since it's their kit? [13:46] Indeed. [13:47] Vaguely apropos: yay for dh_auto_test! It taught me about tricking buildds into running upstream's tests [13:47] I wasn't thinking much about where the ressources were thrown. I was really soliciting feedback on the architecture. [13:48] (Unless they're smart enough to include "nocheck"...) [13:48] I see this on every machine with a 3ware RAID card (read cache disabled). Where can that cache be enabled? [13:48] [sda] Write cache: enabled, read cache: disabled, doesn't support DPO or FUA [13:48] twb: Wow. I had never heard of that. [13:49] soren: The fact that PPA can't easily do snapshots, makes it suboptimal IMO. [13:49] soren: it means that now Darcs' functional tests get run on armel, which GHC (the Haskell compiler) doesn't even officially support! Whee! [13:49] alvin: IIRC I see that on every machine I've ever had. [13:49] Daviey: Snapshots? [13:50] soren: Archive snapshots, we did talk about this in that session :) [13:50] twb: without 3Ware controllers? Because other machines are showing the cache as enabled. [13:50] alvin: I can't affore 3ware [13:50] Daviey: Hm... Sorry, I don't remember this at all. :( [13:51] alvin: I've had some IBM and HP kit cross my bench, but I think they had fakeraid. [13:51] twb: In that case, I think your cache shouldn't be disabled too. You have the same problem. [13:51] Both times $boss was too stupid to buy the bloody RAID5 chip [13:51] fakeraid doesn't count. [13:51] alvin: I have that on a 3ware box aswell - [ 10.907602] sd 0:0:0:0: [sda] Write cache: disabled, read cache: disabled, doesn't support DPO or FUA [13:51] [ 1.212404] sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA [13:51] On my laptop :) [13:52] aha, I think soren's laptop has it right. Cache should be enabled. [13:52] alvin: well, in almost all cases I am using md raid. If there's fakeraid or hardware raid underneath, it's off or set up as one-disk RAID0 arrays. [13:53] Are you saying that fakeraid disables read cache? [13:53] Buggered if I know [13:54] alvin: I've just checked another 3ware box and got, [sda] Write cache: enabled, read cache: disabled, doesn't support DPO or FUA [13:54] I just assumed that it was crap gear and didn't care -- I'm mainly dealing with glorified mail/fileservers. [13:54] Daviey: our experience matches mine. How's the performance? [13:54] [ 2.988006] sd 1:0:0:0: [sda] Write cache: disabled, read cache: enabled, doesn't support DPO or FUA [13:54] great [13:54] I get that on my Eee PC 701, which is an SDD without any md raid or lvm [13:55] alvin: throughput has never been a problem, i always use 3ware when i want RAID5 and care about throughput. [13:55] ok, the question is: how can that cache be enabled. 3Ware tools, or an ubuntu setting. [13:56] [ 21.587035] sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA [13:56] ...that's a generic no-hw-raid, no-fakeraid SATA disk. [13:56] I guess I was only remembering the DPO/FUA part [13:56] pass.. i find it odd that i have two servers - both with different defaults. [13:56] Daviey: I did have a lot of performance issues with 3Ware. There are a lot of bug reports and forum posts about that floating around. (e.g. bug 113532) [13:56] Launchpad bug 113532 in linux-source-2.6.20 "3ware kernel driver 2.6.20 - 2.6.24 disk write performance extremely slow." [Undecided,Invalid] https://launchpad.net/bugs/113532 [13:57] twb: so it is right. It's only disabled on 3Ware controllers [13:57] I certainly would not bother with 3ware unless I needed enough throughput to justify all the extra babysitting and tuning [13:57] alvin: well, you could see the WRITE cache was disabled on my SDD. [13:58] I can't help thinking that if throughput is that important, I'll just hand the problem over to a big iron shop. [13:58] twb: Oh, right. I only looked at the read cache. Is your performance ok? (= is the cache important? Am I missing something here? In my expeerience: disabled cache = horrible performance) [13:59] alvin: buggered if I know -- the most intensive things I deal with are tomcat, clamav, omscan (scalix), postgres, etc. And none of those run on my Eee PC [13:59] The write speed on my SDD isn't fantastic -- if I queue up a lot of writes it's noticably slower than a SATA disk. [14:00] But I expected SDD to have a crap write profile [14:00] Daviey: I'm sorry that I seem to have forgotten about this snapshotting thing. It really doesn't ring a bell. Can you elaborate a little bit to jog my memory? [14:01] twb: I have no experience with SDD. Too 'new' for me ;-) If I need performance, there are 15.000rpm SAS drives too. [14:02] alvin: oh, I should add that I have also deliberately pissed about with e.g. my ext2 filesystem's commit interval on the SDD box [14:03] alvin: that probably affects write speed [14:04] twb: I would like to know what happens if you enbale the write cache (if possible), and mount with 'noatime' offcourse. [14:04] alvin: already using noatime, I dunno how to enable it [14:04] FSVO it = write cache [14:07] how can i create gpt disks during the installation of ubuntu-server 9.10? [14:09] twb: Me neither. Probably with hdparm. i'll read the manual [14:10] acemo: You can't? What is going wrong? (I use lvm instead) [14:12] twb: Setting your write cache is: $ sudo hdparm -W 1 /dev/sdX [14:12] But the read cache... nowhere to be found [14:12] Now to install hdparm... [14:14] Yes, it's not in the minimal installation (and that's not bad) [14:26] Hi, I am interested if anyone knows advantages of using OpenNebula over Eucalyptus [14:30] n8ature: I'd love to chat about that, but I'm on my wait out for an errand. [14:30]