[00:19] this is quite nice [00:19] I'm home sick with a leg in a cast, boss sent me ane sms "are you logged onto the dhcp server?" replied "huh?" [00:21] he called me minute later, the dhcp servers had been down for half an hour or so, I checked /var/log/daemon.log, complaints about timesync between the two servers, installed and configurd ntp, restarted ntpd, dhcpd, done, about a minute work, and boss was like WTF did you do? [00:21] To which you replied "Fixed it - now where's my check?" [00:22] I'm called in sick, his only reply was 'well, roy, you've done your job today again :)' [00:22] I love debugging unices when I get to the point quickly [00:23] those windoze nerds just bladder and can't understand how something can be fixed that easily, or quickly [00:23] and WITHOUT A REBOOT [00:28] reminds me when I started virtualising windows... "what a terrible idea! How will users press the reset button now when its being slow?" (true story sadly) [00:29] :) [00:29] windoze is defective by design [00:30] yann2: did you ever get more info re 'kvm and memory usage'? [00:30] pmatulis, nope... [00:30] yann2: k [00:30] yann2: what was that about? [00:30] is a vm with postgres, still swapping when having plenty of filecache, still no idea, still thinking I should add more ram [00:31] google swappiness [00:31] thought about that, but didnt know what to set it to [00:31] linux will start swapping out things quite early by default [00:31] and a good value for that is quite disputed [00:31] it's a good thing to swap out crap you don't need in ram [00:32] and actually pmatulis told me not to, I think my original questoin was "what should i set swappiness to" :) [00:32] RoyK, when 60% of ram is used it starts to get me worried [00:32] default value is 60 [00:32] I usually set it to 100 [00:33] yann2: the point is that you want as much memory usable as possible, stuff hanging around eating memory is of no use [00:33] but when its claimed back it might take a lot of i/o, more than some filecache [00:34] plus on a postgres VM, filecache is probably the db files... and swap the postgres memory [00:34] not necessarily, swapped out doesn't mean released from memory [00:34] I meant used back, or read back :) [00:35] swapped out is an old term, these days things get swapped out and not released from memory [00:35] just with a postgres vm, you end up having postgres trying to read as much as possible into memory, and the filesystem doing the same for the same data [00:35] so that the memory contains the data, but can release it quickly, or release it from swap [00:36] I don't think postgres has any caching [00:36] it just relies on the filesystem buffers [00:37] yann2: don't care too much about how much is swapped out - care about i/o load [00:38] wonder what it uses 600M ram for then :( [00:38] if sar/sysstat/vmstat/something tells you wio is high, well, that's an indicator for need of more memory [00:38] RoyK, the thing is, it's a very sensible VM, if it runs out of memory it might kill my data [00:38] yann2: don't bother [00:38] and it would be good [00:39] I certainly dont want my VM to run out of ram... and monitoring the swap is the best way I know to monitor it [00:39] * RoyK guesses that last one should be 'no good' [00:39] no good yes :P [00:39] although its boring finance stuff :) [00:39] how is the i/o load of the vm? [00:40] has pikes [00:40] on average [00:40] apt-get install sysstat [00:40] enable it in /etc/default/sysstat [00:40] start it [00:40] pasted you a link in private [00:40] that will show you io load over time [00:40] with detailed stats [00:41] well, that's ok [00:41] some spikes, but overall just fine [00:42] maybe I should change my swap alerts levels [00:42] I really wouldn't worry about that [00:43] you're hardly using swap at all [00:43] got an alert at 40% :) [00:43] and the memory stats look good [00:44] disk i/o load is close to zero [00:44] so don't worry [00:44] like collectd? :) [00:45] well, add a truckload of memory and you might gain 10% performance [00:45] your system is not under heavy load [00:46] I know :) just concerned its sometimes swapping out 400MB when it has 500MB filecache [00:46] linux does that [00:46] it swaps out stuff not in use [00:46] so that the memory can be used for something useful [00:46] k [00:47] at swappiness = 60, being the default, it's quite moderate [00:48] at 100 it's a little more triggerhappy, better for most systems, at least those with low memory [00:48] low memory being < 200% needed of the apps [00:59] hi all [01:00] could anyone help me recovering grub2 on a software RAID? [01:01] after an upgrade from 8.04? [01:08] jdstrand: is bug 696318 something you've ran across before? [01:08] Launchpad bug 696318 in libvirt "Starting VMs on qcow2 format with base images in level three fails" [Medium,Confirmed] https://launchpad.net/bugs/696318 [01:17] yann2: what disk format are you using with your plagued guest? [01:20] Anyone familiar with Supermicro ipmi on Ubuntu 10.04 or 10.10? [01:21] qcow2 pmatulis , on local disk [01:21] its not THAT plagued [01:22] sent you the graphs in pv [01:24] anyway its late here, I think I'm off to bed :) see you tomorrow, and thanks to both of you for your help [01:43] Hey, I have a strange issue with hostnames on ubuntu. When I send an e-mail there is a line: Received from damdamdam.org (ip6-localhost [127.0.0.1]) but my hosts file does not have damdamdam.org ... === PleXT is now known as PleXs_zZz [01:52] Frenk, mail is sent "from" the mailname you specified when configuring your mail server [01:54] qman__: I mean something else: the from adress is okay. But when I open all headers: Received from damdamdam.org (ip6-localhost [127.0.0.1]) by myrightdomain.org (Postfix) with ESMTP id 223 for ; Tue, 4 Jan 2011 02:32:41 +0100 (CET) [01:55] and I want to get rid of damdamdam.org [01:55] yes, this is what I'm talking about [01:55] that name is the system's mailname [01:55] what that name is has nothing to do with DNS, though you're supposed to choose a DNS accurate name [01:56] cat /etc/post*/main.cf | grep my >>> myhostname = myrightdomain.org [01:57] that is not the mailname [01:57] cat /etc/mailname [01:58] output is: static.MY-IP-ADRESS.clients.HOSTER.com [02:00] well, if it's coming from somewhere else, grep -R damdamdam.org /etc [02:00] might be the sender specifying that in headers too [02:10] qman__: I found smth but there are all domains listed that Im using. I just run grep for / ... waiting .. [02:27] anyone use Adaptec 2240900-R? [02:27] or Rosewill RC-218 [02:28] need more sata ports for my sw raid not sure which is best for ubuntu [02:34] I don't use hardware raid. [02:35] i dont either [02:35] i want to use sw raid [02:35] i just need more ports. [02:35] If you just need SATA sockets for your md RAID, just buy a $10 SATA controller card. [02:35] It shouldn't matter a damn as long as you can say "don't fakeraid", which most can -- and if the one you get can't, just buy another $10 card with a different controller. [02:36] the rosewill is the only option i found on newegg thats nonraid, 4xsataII, [02:36] pci-e 4x [02:36] They all CLAIM to be raid, but it's fakeraid [02:36] yes i know [02:36] don't worry about it [02:36] i was lookng at a 3ware card but thats way overkill [02:36] i dont need 300mb/sec [02:36] Just buy three $10 ones of different makes (and chipsets, if you can confirm that) [02:37] One of them will work and support "no raid" as an option [02:37] heh in my case i can only use 1. its on a atom ion2 [02:37] 1xpci-e [02:37] Yes, but buy three at once so you have a better chance of ONE of them working [02:37] LOL [02:37] oh its like that [02:38] thanks [02:53] I have a couple promise TX4s [02:53] the only features you might be interested in are NCQ and friends, for performance [02:53] everything else is irrelevant as long as it can operate in non-raid mode [02:55] NCQ is the noop call, right? [02:55] Oh, no, it's more like an interrupt protocol. Never mind [02:55] http://en.wikipedia.org/wiki/NCQ [03:00] k i do want ncq thanks [03:01] fubada: eh, depends if it's gonna be a database server or just host users' word documents :-P [03:05] if you have more than four drives, you're probably CPU and network limited anyway [03:07] twb ill surely have mysql or some db backend for services [03:07] i have 4 drives for md raid [03:07] and ssd for os [03:07] http://secure.newegg.com/WishList/PublicWishDetail.aspx?WishListNumber=12215494 [03:08] raid 10/0+1? [03:09] i was thinking raid5 [03:09] 4x2tb's [03:09] md raid [03:09] going to be CPU limited [03:09] really? dual core 1.8 [03:09] reads will be fast, but writes will be CPU limited [03:10] still pretty fast, but don't expect more than ~60MB/s [03:10] hm [03:11] it has to calculate the parity on the fly [03:12] raid 10 would not have that problem [03:12] can linux do sw raid10 [03:12] but gigabit ethernet will cap your throughput around 80MB/s [03:12] so unless that throughput is needed locally, not really worth it [03:13] yes [03:13] so what is the best option? a 3ware card runs 320$ [03:14] but i dont need that kinda read/write [03:14] not worth it, for that scale [03:14] raid 1? [03:14] if you need more throughput, spend more on a faster CPU and better drives, or just go raid 10 [03:14] raid 5 is too much cpu, im not familiar with raid10 [03:14] the drives are wd greens for low power [03:15] and low heat [03:15] green drives are the bottom of the barrel [03:15] they like to advertise low power but they're very slow and won't last under heavy loads [03:16] but again, it all depends on your needs [03:16] if it's just storing data backups, no problem [03:16] if you're hosting a high traffic database, get something else [03:16] htpc, home nas [03:16] backups [03:17] seed box for my torrents [03:17] 500/sec 24/7 [03:17] 500k/sec [03:17] you're going to need a faster CPU if you want good throughput with that going [03:18] encoding/decoding video and running torrents are going to put notable load on it already, plus a raid 5 [03:18] eh i dont think it'll be doing any encoding [03:18] ion2 is onboard for gfx [03:18] let me put it this way [03:18] :) [03:18] my file server, 8 disk raid 6, runs torrents [03:18] 2.2GHz single core CPU [03:19] my writes under normal circumstances cap around 35MB/s [03:19] when I start a bunch of torrents the server load exceeds 20 for several minutes [03:19] what.cd? [03:19] nah, 10-20 torrents [03:20] damn [03:20] and normal, seeding load is .10-.15 [03:20] that chip is dual core, so it is faster than mine, and raid 6 has twice the parity to calculate [03:20] but you're still looking at notable overhead [03:21] you're going to be waiting on it when you start big torrents, and your writes probably won't cap your gigabit [03:21] Mh can anyone explain this to me: https://help.ubuntu.com/10.04/serverguide/C/certificates-and-security.html there is some information: http://paste.ubuntu.com/550071/ What does this paragraph mean? Its my first SSL experience (Official SSL from a CA) - I mean eg. if Apache or Cyrus crashed, does it mean my watchdog cant start them again until I type in my password again for the cert? [03:21] got it [03:21] the goal was to build a lower power box, so i'll have to sacrifice on services [03:21] thats fine [03:21] ok, just so you're aware [03:22] appreciate the info thank you [03:22] it'll work fine, just a bit slower [03:22] i wanted to run xen on this lol [03:23] how do you make it so you can access your webpages internally like this: http://internal.ip.of.server/directory_of_webpage [03:23] qman__, can you think of a better 5400rpm drive [03:24] than the wd caviar green [03:24] i am currently using virtual hosts [03:24] I don't have suggestions there, I use 7200RPM drives [03:24] samsung HD103SJs, to be exact [03:24] nice [03:25] 72k's run too hot [03:25] 4 stacked will run really hot [03:25] in my experience, WD drives run a lot hotter than other bransd [03:25] though it has not affected longevity [03:25] 10C on average [03:26] the green drives may be a different story, I don't know because I don't use them, but for everything else that has been my experience [03:27] there's a samsung green drive [03:27] 5400 [03:27] its 20bucks cheaperr [03:27] SAMSUNG Spinpoint F4 HD204UI [03:28] one suggestion, make sure you configure SMART monitoring [03:28] I lost my first array when three drives failed in one day, and I already had to RMA one drive from this new array [03:29] 3 drives in one day [03:29] earthquake?! [03:29] no, just plain drive failure [03:29] those were seagate 7200.11 500GB [03:30] two an hour apart [03:30] i have 1 drive fault tolerance with raid10? [03:30] the third a few hours later [03:30] 4x2tb [03:30] any one drive, and in some cases two drives [03:30] cool [03:30] it depends on which two, though [03:30] right [03:31] are you using myth [03:31] mythtb [03:31] tv [03:31] no, my file server isn't also an htpc [03:31] :) [03:31] I tried to set up myth on a different machine but it wouldn't work with my graphics [03:31] had to install windows, unfortunately [03:31] im so enraged at this cablecard issue [03:31] encryption of hdmi, lack of cablecard support in linux [03:31] sdv [03:32] this makes mythtv not even worth it [03:32] it's not if you have encrypted qam channels [03:32] i want to fire my cablecompany now [03:32] I just want to buy the channels I use [03:32] and not the other shit [03:32] exactly [03:33] dream on ;) [03:33] im going to fire my cabletv [03:33] what do you watch anyhow? [03:33] a&e [03:33] nfl [03:33] history [03:33] natgeo [03:33] nfl you're set [03:33] espn [03:33] get an xbox [03:33] and downlod the espn app [03:33] :D [03:33] i have ps3 :P [03:34] does the ps3 have the espn app? ;) [03:34] i doubt it [03:34] anything that plays on espn [03:34] you can play on the xbox [03:34] i think you can stream hd nfl games for 100$/yr [03:34] in flash [03:34] what on a&e?> [03:34] horders and intervention [03:34] haha [03:34] other crap [03:34] ya, same [03:35] i really liked that show on amc [03:35] about meth... [03:35] doh whats the name [03:36] espn signed a deal with MS [03:36] no espn for ps3 [03:36] right [03:36] The_Tick, do you own a hauppage hd-pvr [03:36] nope [03:36] fubada: I have an xbox [03:36] and I'm almost off of cable [03:36] thats one way to have mythtv with cablecard [03:37] really? [03:37] but can only record/watch one channel [03:37] hrmm [03:37] yea it just records and encodes 1080i [03:37] so get more than one computer [03:37] and USB to your box [03:38] the media/communication companies in this country are filthy pigs. === WinstonSmith_ is now known as WinstonSmith === ^rifter^ is now known as rifter === cradek_ is now known as cradek === guampa is now known as guampa-off [06:24] oh dear - I think I've done something stupid. I have 320000 files in one directory (small css files).Am I reaching somehow some limits here on ext3? [06:24] just thinking of running out of inodes or something. [06:28] tom_: 'df -i' will show you your filesystem usage by inode count, I'd keep an eye on that as the first sign of things going funky [06:31] uff - still shows 26% free ... :-) but an ls | wc -l takes like 5 minutes. I guess I'm best of deleting the whole directory, they are just temp files. [06:32] I think what you'll start to run into first with that kinda scenario is the limits of shell globbing [06:32] tom_: by default GNU ls will sort records [06:33] tom_: there is an option to prevent this, which will make it emit results as they arrive from the directory listing syscalls [06:33] Alternatively, find -ls or just find [06:34] hello folks [06:34] Oh, and 320,000 files in a single dir will not be handled very well by ext[23] or fat -- possibly not by any block-oriented filesystem. Try an extent-oriented filesystem? [06:34] could someone help me to install grub to from cli? [06:34] stlsaint: "grub-install /dev/sdz", where sdz is your disk. [06:34] I don't event want them !! I think I must have screwed up in a php script. [06:34] stlsaint: if that doesn't work, you're on your own -- I use extlinux. [06:34] i installed server edition, but grub installed on the usb stick and not my master harddrive [06:35] but it should be safe to just delete that directory right? [06:35] tom_: find /tmp/foo -name 'php_??????.txt' -delete [06:35] tom_: yes, or just rm -r the parent dir [06:35] stlsaint: that is a known bug [06:35] stlsaint: I don't know what the workaround is, other than unplugging the USB key after you've loaded its kernel and ramdisk into memory. [06:36] oh man there has to be a better way than that! [06:36] stlsaint: it'll be fixed in natty (and maverick?) [06:36] thx for ur help twb shauno, that number just was astronomical [06:36] 10.04 [06:36] stlsaint: of course you can also fix things post-facto by booting a live image and using grub-install from there. [06:37] stlsaint: or you could install 8.04 normally, then upgrade it to 10.04 [06:41] this is a pretty big bug to let go unwarranted [06:41] smh [06:42] Er, it's not "unwarranted". [06:42] It has been fixed. [06:43] Since it does not affect the supported install media, I assume that the release policy didn't require it to be backported to stable releases. [06:44] twb: aye, i see what you mean but is it corrected in 10.04 as i see this being the most important as it is a LTS [06:45] stlsaint: AFAIK you're using an unsupported install medium, so if it breaks, you get to keep both pieces [06:47] so the only supported installs are cd? [06:47] i wouldnt understand how a net install would be supported... [06:47] but a usb is not... [06:47] AFAIK only CD installs are officially supported. [06:48] smh [06:49] FWIW a 10.04 netboot kernel and ramdisk can be bootstrapped from a USB key, then the USB key removed, and then the install completed from the network. This is how I perform installs when the PXE server is down. [06:50] aye [06:50] that is good tto know [06:51] That's these: http://archive.ubuntu.com/ubuntu/dists/{hardy,lucid,maverick}/main/installer-amd64/current/images/netboot/ubuntu-installer/amd64/{linux,initrd.gz} [06:52] Within that tree are boot.img and mini.iso that are basically the same content, but pre-rolled with a bootloader. [07:01] twb: i went with the re-install of grub from livecd (usb) [07:01] twb: thanks for your assistance [07:01] stlsaint: OK === sresu_ is now known as sresu [07:51] New bug: #697105 in apache2 (main) "Segfault on POST" [Undecided,New] https://launchpad.net/bugs/697105 [09:00] where i can discuss about ubuntu amazon or UCE images - any idea - is it the right room? [09:03] my log is filling up with collectd rrdtool -errors like these http://pastebin.com/Fm1LDc7V any experiences? [09:15] xampart: install ntp? [09:15] Hi, when I start the cman service in 10.10 I see these messages in daemon.log http://paste.ubuntu.com/550140/ [09:15] xampart: the first line looks like it's whinging because an epoch time counter somewhere went backwards [09:16] Or rather, stood still [09:16] xampart: what happened at Mon Jan 3 08:19:39 UTC 2011 ? [09:18] macno: udev is pissed because the rules file you just installed expects (presumably) an older kernel [09:18] macno: try copying it from /lib/udev/rules.d/ to /etc/udev/rules.d/ and removing the "misc/"'s [09:18] I upgraded from 10.4 to 10.10 [09:19] let me try [09:19] macno: note that I'm just guessing; I'm not familiar with "cman" nor 10.10 [09:32] twb: nothing special in syslog or messages [09:33] and ntp is installed [09:34] is it safe to use ubuntu-server as a desktop os? [09:34] does it have all the security stuff that ubuntu-desktop has? [09:37] no--name: yes, apart from the packages installed by default from the server cd will be different and the kernel installed will be the -server flavor [09:38] TREllis: is there anything I should install security wise that -server doesn't have installed by default? [09:41] no--name: the server install will have a very minimal package set, so by default you shouldn't have any ports open or anything. I'd start off with reading through the server guide, it has a fairly brief overview of things to look at securing, really depends on what services you want to run. https://help.ubuntu.com/10.10/serverguide/C/index.html [09:42] ok [09:43] so in the end it will be a lot more work than just going with -desktop? [09:44] no--name: "more work" would depend on what you want to use it for [09:44] ok [09:46] no--name: define "safe" [09:47] twb: as safe as ubuntu-desktop [09:48] It's safer. [09:48] Way safer. [09:48] It has almost nothing installed by default. [09:48] Less attack vectors. [09:48] no--name: IMO -server is not significantly LESS safe than -desktop; it may be MORE safe. [09:49] no--name: they differ primarily in the set of packages installed by default; they both back onto the same package repository. [09:49] twb: but I want to use it like I'd use -desktop [09:49] just because it has less bloat [09:50] argh, what is the package that provides the command "locate" ? [09:50] no--name: perhaps you should install from the "alternate" CD, choosing the "cli" option and then installing only what desktop packages you think you need. [09:50] jussi: mlocate (or locate or slocate, depending on vintage and preference). [09:51] twb: ok, what is the one that usually ships with maverick? [09:51] jussi: apt-file can tell you, in future. [09:51] jussi: mlocate, I expect [09:51] they didnt put it onto my server... :/ [09:51] twb: ok. I was just wondering if there were packages I'd need to add for security if I were to use it as a desktop OS [09:52] no--name: conventional wisdom is that installing packages makes a system LESS safe [09:52] ok [09:53] what concerned me is that update manager on -desktop comes up with all sorts of "security updates" but on -server it just comes up with kernel stuff. I guess that's because these security updates are for packages that aren't on -server? [09:54] no--name: that is simply because you have fewer packages installed. [09:54] ok, cool :) [09:54] no--name: security updates apply to both. [09:54] both what? [09:54] no--name: both -server and -desktop. [09:54] hmm [09:54] no--name: they back onto the same security repository. [09:55] why is it that when i ran the update manager on -server i only got kernel stuff? [09:55] 20:54 no--name: that is simply because you have fewer packages installed. [09:55] oh, right :/ [09:55] Well, it may also be because you screwed up something and updates aren't being received [09:55] hmm [09:55] it's too bad I don't have it installed right now [09:57] I'll install it on vbox and see how it goes. [10:17] Hi, please tell me if this is correct cp -r var/www/site1/folder var/www/site2/folder [10:25] twb: would recommend "install security updates automatically" or "manage system with landscape"? [10:25] would "no automatic updates" have update manager manage them isntead? [10:35] no--name: unless you have a support contract with canonical, there's no point in choosing landscape [10:36] ? [10:37] hmm, oh well. should i choose automatic or no automatic (seeing i will be using update manager) [10:38] No idea. I do not support desktops. [10:38] ok [10:38] well, i'll try no [10:38] last time i did install them automatically [10:39] please tell me if this is correct cp -r var/www/site1/folder var/www/site2/folder [10:39]