[00:02] ;_; [00:46] hmm [01:38] stgraber: drat. might be worth a bug against qemu and QEMU [01:38] thx === jaimef is now known as fsbot === fsbot is now known as jaimef [04:56] What is the command to print and update a logfile onscreen? I know there is watch but that oly works for a log such as dmesg as far as I kknow. [04:56] plan_1: tail -F ? [05:00] sarnold: That looks useful but I haven't seen it update yet do you know how often it should update? [05:01] plan_1: quite quickly; perhaps the slowness is the writing to the log file. [05:01] plan_1: writing to files via a standard output typically buffers 4096 bytes or so, thta might be a bunch of lines... [05:02] plan_1: e.g., grep's --line-buffered command line option will ask it to use line-buffering rather than block-buffering, so tail -f and tail -F and so forth respond more quickly. [05:07] sarnold: Thanks, looks like i have connections on 50440 [05:08] sarnold: But the firewall is not allowing them does that imply kernel level problems would you say? [05:08] awesome, I got it working from a friend of mine. Can someone confirm? [05:08] http://71.10.98.41:8000/ [05:08] Iapetus: woot! [05:09] sarnold: Well it is not set to allow them that is. [05:09] plan_1: you'd have to compare netstat -lnp output against the iptables -L output [05:10] can you reach it sarnold? [05:10] Iapetus: yeah, some anime screenshot and so forth :) [05:10] sick [05:11] fucking networking [05:15] sarnold: It doesnt show in netstat only iptraf. [05:16] sarnold: I usually do netstat -taupe netstat -lpn is showing unix sockets and I don't really know what all of that means. [05:17] plan_1: l just shows listening sockets.. perhaps it's an already-connected socket? [05:17] sarnold: If it is kernel level would it not also escape netstat as the evidence suggests? [05:18] plan_1: what problem are you trying to debug btw? :) [05:18] sarnold: Oh ok let me glance at the netstat man to display connected sockets. [05:19] sarnold: Unless you know off the top of your head what the netstat command would be. [05:23] sarnold: Reguardless of netstat I made the iptables list myself and know it is not allowing these connections showing in iptables. So what do you think kernel level problems? [05:26] sarnold: Still with me? [05:27] plan_1: I'm afraid I was never with you :/ I don't know what problem you're trying to solve. [05:27] sarnold: The problem is unsolicited network connections. [05:30] sarnold: You know the whole iptables not set to allow them bit. [05:30] sarnold: Is that not plain english? [05:31] plan_1: so, you've got network connections that you don't want? don't know which program is accepting them? not sure why iptables has let them through? or... I'm still not following. :) [05:32] sarnold: All of the above. [05:38] sarnold: How about this one why doesn't netstat show domain connections when using -taupe? [05:40] plan_1: what's a "domain connection"? [05:42] sarnold: If you check /etc/services it id udp port 53 DNS [05:45] Remember me highvoltage ? [05:45] plan_1: udp is connectionless; even when connect(2) is used on a UDP socket, it doesn't actually set up a connection, it just adds some filtering rules to the socket [05:45] plan_1: my netstat -anu output doesn't show any outgoing DNS, even though I'm sure my system is doing DNS once in a while ... [05:45] sarnold: That is what I thought more of a broadcasting but why the -u option in netstat then? [05:46] plan_1: just to show only udp [05:46] no need to see tcp if you're curious only about dns :) [05:46] sarnold: No I mean since udp is connectionless why is there a -u option since they will never show. [05:47] plan_1: ah! because listening sockets still show up [05:50] sarnold: Well there is no allowance for high port udp so there must be kernel level problems here. [05:52] I know it is not the Ubuntu way but is there somehow I may get a clean system I am willing to pay $100. [05:55] Remember me highvoltage ? [06:00] Remember me highvoltage ? [06:06] Remember me highvoltage ? [06:08] his connection is just timing out, over and over again... [06:08] Anyone know how to induce vomiting, I just ate some easter candy and it is making me sick that and the cans of dog food called beef stew. [06:09] Does the rest of the world get this canned protien textures mixed with preservatives? [06:10] Im in Missouri. [06:10] Is this type of stuff in europe also? [06:10] Or australia? [06:11] plan_1: eat burnt toast [06:13] histo: How much does it usually take you? [06:14] plan_1: huh? the charcoal will make you puke [06:14] I might try the finger throat method as I have no toast. [06:14] if you go to a hospital that's what they will give you charcoal [06:14] histo: I was asking how many slices. [06:14] plan_1: no idea never really tried just have heard about it. [06:17] finger throat method hurts too bad [06:21] cancel that idea, so what are you up to histo [06:35] who is using asterisk on ubuntu server? [06:35] For what [06:36] The crap is basically unusable [06:36] plan_1, IP PBX [06:37] linocisco: what would a person use asterisk for? like voip telephony or something? I guess I have to watch some videos [06:37] and assume you are talking to who on the other end? [06:37] akams AI [06:38] time to slay hells angels [06:39] ok koolhead17 im working on it [06:40] plan_1, ? [06:40] hoot n cold [06:40] if everyone else is lukewarm [06:44] fagboys walking around with "smart" phones [06:45] yelling like idiots [06:46] and a loud mouth slut following them [06:46] you want me to get cold as ice koolhead17 [06:53] "I killed all husnock everywhere" [06:55] The agents? [06:55] If they dont work for me they are agents of nothing. [06:56] Unpredictable. [06:57] plan_1: do you actually have anything ubuntu related to talk about? [06:57] Yeah but it gets ignored. [06:57] see how sarnold just kindof drifted off [07:01] hey bazhang thanks for writing the skeleton any chance I may get a clean kernel I have $100 [07:06] The LORD is my shepheard I shall now want. [07:07] plan_1, please stay on topic [07:07] * plan_1 laughes [07:16] histo, yes. for voip pbx === smb` is now known as smb === schmidtm_ is now known as schmidtm [09:54] Hello. I have some strange things happening with my Ubuntu Server. Since a few Days i am getting a huge load in "Connections through firewall" and " ipconntrack". Is there a way to see whats happening there? [10:15] I spy a linux format editor.... [10:47] Styler2go: use tcpdump to see network traffic [11:03] Is there a way to get the firewall to actually work? [11:19] plan1, sure, load the modules for it :) [11:28] ikonia, where? [11:36] evilnickveitch: he's wearing an evil disguise, and he maybe an ex-editor [11:45] what can i do with tcpdump? [11:47] Styler2go: dump tcp traffic. [11:48] Styler2go: Dump all network traffic actually. [11:51] zul: http://people.canonical.com/~jamespage/ca-updates/ [11:51] if you would be so kind :-) [11:52] jamespage: +1 (although not offically here yet) [11:52] zul, lol [11:52] okies [12:09] jamespage: penstack components in quantal queue.. 2013-02-21 .. those ones are good? [12:09] or totally superseeded by 2013-03-22? [12:10] Daviey, superceded by 2013-03-22 [12:10] Daviey, ugh - did you already accept those ones? [12:11] jamespage: no [12:14] jamespage: 2013-03-22 is good? [12:17] Daviey, yes - thats what I prepared last week [12:17] its the original srus with the security updates applied [12:18] right [12:20] jamespage: quantum (2013-03-01) is good? [12:21] Daviey, yep [12:21] thats a missing bit from the original SRU update that adam_g prepared [12:21] jamespage: except adam didn't add a bug reference to the changelog. *sigh*. I need that. [12:21] Daviey, bah - OK - please reject it and I'll sort that out [12:22] done [12:38] Hey everybody, I just installed a new motherboard asus p8z77-v in my server, and now I cant get the network up [12:38] Any hint on what to do? Please help me. [12:39] "Cannot find device eth0" [12:44] mkander_, sudo rm /etc/udev/rules.d/70-persistent-net.rules && sudo reboot [12:44] if that doesn't fix it, you lack driver support [12:45] mkander_: ifconfig .. it might not be called ethX [12:45] Ok Ill try that qman__ [12:45] Daviey: just "lo" in there [12:45] ok [12:46] wow looks like it worked :-) [12:47] YEEEhaa :) [12:47] thanks [12:53] jamespage/yolanda: https://code.launchpad.net/~zulcss/cinder/rc3/+merge/155723 [12:54] zul, no changes apart from version number? [12:54] yep [12:54] zul, approved [12:55] (thats my release that is) [13:07] hi [13:08] what is a good monitoring tool (for monitoring webserver/mysql/php-fcgi, networking) in CLI or really light application? htop is not so sufficient.. === kodapa_ is now known as kodapa === frankban_ is now known as frankban [13:13] stgraber: the lxc-start-ephemeral manpage says '-n name' is an option, but it's not inthe soruce [13:14] virtx: There are a few 'top-like' packages in the repositories for apache and mysql. It really all depends what sort of monitoring you are trying to do. [13:14] hallyn: ah yeah, I remember seeing that and forgot to fix it. Will do today [13:15] stgraber: cool, thanks :) [13:15] Pici, what do 'top-like' tools do? [13:16] apachetop says "It is modelled after the standard 'top' utility, and displays information such as the requests pers second, bytes per second and the most popular URLs displayed." [13:16] good, and for mysql? [13:17] for networking i'm using iftop, but vnstat seems better [13:17] vnstat is pretty nifty. [13:18] !info mytop [13:18] mytop (source: mytop): top like query monitor for MySQL. In component universe, is optional. Version 1.6-6 (quantal), package size 34 kB, installed size 152 kB [13:18] well, i try it [13:19] Pici: isn't there an apachetop tool for other webservers? [13:19] like nginx, lighttp [13:20] virtx: There might be. Try: apt-cache search nginx top or a similar query [13:21] jamespage: http://people.canonical.com/~chucks/ca/ [13:22] Pici: nothing :\ [13:28] hallyn: patch for -n/--name sent to lxc-devel [13:31] stgraber: cool, thx. getting ready to send a lxc-clone patch [13:50] zul, +1 [13:50] jamespage: thanks cinder ca right? [13:50] zul, yep [13:50] jamespage: k thanks [13:52] jamespage: did you promote the other ones this morning? [13:52] zul, everything pending is in proposed [13:52] http://reqorts.qa.ubuntu.com/reports/ubuntu-server/cloud-archive/grizzly_versions.html [13:52] jamespage: cool thanks [13:52] now we just need python-ceilometerclient [13:52] zul, I'll flush through to updates now [13:54] zul, OK - synced through [13:55] zul, so just pending ceph (which I just uploaded), cinder rc3 [13:55] and ceilometerclient when it arrives :-) [13:55] jamespage: ack === wedgwood_away is now known as wedgwood === wedgwood is now known as wedgwood_away === wedgwood_away is now known as wedgwood === wedgwood is now known as wedgwood_away [14:10] hallyn: hey, is it just me mis-reading the diff or do you end up with lxc.rootfs defined twice in the config? [14:12] stgraber: doesn't show up twice === wedgwood_away is now known as wedgwood [14:18] stgraber, hallyn re bug 1160360 following ogra_'s comment I tried FLASH_KERNEL_SKIP=true and update-initramfs works fine even with flash-kernel installed. Is it something that could be set by default in the template too ? [14:18] Launchpad bug 1160360 in lxc "flash-kernel failed in an armhf lxc container on ARM: /usr/sbin/flash-kernel: 214: /usr/sbin/flash-kernel: mkimage: not found" [Undecided,New] https://launchpad.net/bugs/1160360 [14:18] jibel: no, we don't do rootfs modifications from the templates [14:19] jibel: as we support copying the rootfs from an existing machine into a container [14:19] jibel: so it'd have to be a change in the flash-kernel package to check running-in-container and do the equivalent of FLASH_KERNEL_SKIP=true in that case [14:21] stgraber, okay [14:30] stgraber, jibel, patches accepted :) [14:33] when doing this: mkdir /long/path/name/newdir/ && touch /long/path/name/newdir/newfile is there a way to evade typing /long/path/name/newdir a 2nd time? [14:34] ogra_, checking running-in-container in initramfs-hook/flash-kernel would be fine? [14:35] jibel, well, check the flash-kernel code ... it should be added everywhere where you find the env var [14:36] ogra_, okay, I'll send a patch. thanks [14:36] thanks too ! :) === LargePrime is now known as Guest62579 [15:18] hi, there is no reason for gstreamer and qt to be installed on my server? [15:18] (they were installed by default) [15:25] ? [15:27] ¿ [15:32] hi, there is no reason for gstreamer and qt to be installed on my server? RoyK << [15:33] and I can also remove samba? [15:34] savr: do you have little disk space? [15:34] no I just don't want junk installed [15:34] easier to debug problems [15:34] it doesn't matter [15:34] stuff not running doesn't make bugs [15:35] for example I had apache installed and it took me ages to figure out why tomcat wasn't working properly... [15:35] turned out apache was proxying tomcat automagically [15:35] savr: that's running processes [15:35] savr: not installed software [15:36] samba would also be running [15:36] then stop it, or just uninstall it [15:36] actually... apache2 package isn't installed by default... [15:36] samba isn't installed automatically. neither is apache [15:36] it is apache2-common or something like that... [15:36] that doesn't contain any daemosn [15:36] I'm using a packaged version of ubuntu to openvz [15:36] daemons, even [15:37] savr: worry about then if a problem arises [15:37] imho samba shouldn't be installed unless yo need windows connectivity [15:37] also it saves a ton of time upgrading [15:37] I don't [15:37] removing [15:39] * ogra_ wonders how you got it installed in the first place ... neither qt nor gstreamer nor samba are in an ubuntu server default install unless you actively select them [15:40] ogra_: 16:36 < savr> I'm using a packaged version of ubuntu to openvz [15:40] oh [15:40] missed that [15:40] scary [15:40] not sure if it is a good idea to use the ubuntu openvz template [15:41] smoser, About bug 1160543. Is that repeatable? In which case I might be inclined to build a test kernel with the anticipated upstream changes. ;) [15:41] Launchpad bug 1160543 in linux "kernel crash on EC2 raring" [High,Confirmed] https://launchpad.net/bugs/1160543 [15:41] smb, embarrasingly, the first i personally played with raring on ec2 was yesterday. [15:42] and i started an lxc container, stopped it and stareted and stopped [15:42] and it crashed. [15:42] savr: dpkg -l | pastebinit [15:42] (i thought i'd busted networking). i dont know much more about it. [15:43] I'm on mosh... need to ssh once the upgrade is finish === medberry is now known as med_ [15:47] jamespage: https://code.launchpad.net/~yolanda.robla/charms/precise/glance/ha-support/+merge/155771 [16:01] apt-get upgrade takes forever on openvz :/ [16:02] too many packages installed? slow host? [16:03] I'm the host [16:03] ssh into the mysql node... http://pastebin.com/iJ6azVa8 [16:06] ogra_: RoyK smoser smb [16:07] savr, dpkg/apt is sync-heavy [16:08] got gstreamer gtk can be remove too? [16:08] see 'unsafe-io' (man dpkg) or 'eatmydata' [16:10] apt-get purge -y openjdk-\* icedtea-\* icedtea6-\* apache\* x11-\* samba\* libgtk\* libgstreamer\* [16:10] that should clean it all up? [16:14] savr: pastebin dpkg -l ;) [16:14] I have [16:15] oh [16:15] there [16:15] smoser, Well ok, I think that it might be related to lxc's usage of mem cgroups. [16:15] mysql? [16:16] yeah that is the mysql node [16:16] so it needs mysql [16:16] ok [16:16] * RoyK isn't a big fan of mysql [16:19] so anything else I should get rid of? [16:21] not sure - a lot of libraries you might not need, though [16:21] how much does the root fs fill up now? [16:22] dunno [16:22] it is slowly still removing on two more nodes [16:22] it's really slow the server isn't doing anything other than running apt on a few nodes right now [16:23] savr: the 'deborphan' package can help find packages that aren't needed any more -- but be careful with its recommendations, it'll happily recommend a leaf package that you _want_ -- but with a careful hand driving it, very useful. :) [16:23] I've got 3 nodes running only apt right now. and that is all on the server === psivaa is now known as psivaa_afk [18:21] stgraber: I suppose you've seen https://www.youtube.com/watch?feature=player_embedded&v=wW9CAH9nSLs#at=291 already? [19:21] hallyn, sarnold: I think I finally understand bug 1157332. The logic in the package itself is correct for raring, but the problem is when you upgrade from precise or quantal to raring as the migration code is then triggered twice [19:21] Launchpad bug 1157332 in lxc "/etc/dnsmasq.d-available/lxc circular link " [High,Confirmed] https://launchpad.net/bugs/1157332 [19:22] stgraber: yay :) [19:22] stgraber: oh no, is that going to be the same for libvirt? [19:22] I'm confirming I can easily reproduce this with 12.10 => 13.04. If I can, then I'll add some code to fix the mess. It's easy enough to detect the circular link from the postinst, move everything back to normal and restart dnsmasq [19:22] sigh [19:23] thx :) [19:24] the good news being that in the worst case scenario we just need to fix raring + quantal. I don't think we need to SRU something to precise this time around === tobin_ is now known as tobin [19:24] stgraber: did it only trigger for me because I had upgraded precise->quantal->raring ? [19:25] sarnold: I certainly confirmed that precise->quantal->raring causes it, but I suspect just quantal->raring will too [19:25] stgraber: aha [19:29] sarnold: right, quantal->raring => broken links [19:30] alright, now to fix that mess :) [19:30] stgraber: good luck :) it sounds frustrating :) [19:31] sarnold: it's just frustrating because it's the 3rd time we have to fix it and we'll likely have to re-upload twice 3 different sources to get rid of it entirely [19:31] stgraber: oh _man_ :/ [19:31] libvirt, lxc and network-manager share the same bit of code and we have it in precise, quantal and raring [19:33] hallyn: hmm, so the problem is trivial to see but hard to fix ;) [19:33] hallyn: in short, we have the migration code in .maintscripts which tells dpkg to move the file if the previous version is older than X [19:34] hallyn: the problem being that we have 3 possible value of X depending on whether we're coming from precise, quantal or raring [19:34] hallyn: so our upgrade code is perfectly fine for users upgrading from initial raring to current raring, but will trigger twice (or more) for those coming from precise or quantal [19:35] hallyn: with the problem being that if we set that value to the oldest version with dnsmasq support (current SRU in precise), we may miss triggering for people upgrading from old quantal or old raring [19:35] in short, I think we should keep .maintscripts as it's, because it's not wrong. Then add some postinst code to detect double-migration and fix it [19:36] I'll test a fix here and attach to the bug report with details of my findinds, because the SRU team will want to know the details for sure (as it won't be good, clean packaging changes ;)) [19:37] sarnold: do you still have your broken dnsmasq setup or did you fix it? [19:40] stgraber: it's also frustrating because the libvirt one isn't even accepted itno -proposed yet :) [19:41] c'est la vie [19:41] hallyn: alright, I have a fix [19:42] stgraber: cool. maybe the fix will explain to me why we can't just use explicit package version comparisons in pre/postinst to fix it :) [19:44] hallyn: because we need explicit version comparison + series comparison [19:44] stgraber: I certainly don't _like_ my current dnsmasq setup -- I either have to type sec-precise-amd64.local _or_ I no longer get local LAN resolving. :/ I'm going to try a reboot after I fiddled with a few settings though [19:45] hallyn: as in, we need our migration code to trigger if the previous version is "<= X on precise" or "<= Y on precise" or "<= Z on raring" [19:45] hallyn: currently we just have it as "<= Z on raring" so anything that's lower than the original raring version will be migrated even if it already was through an SRU === sauce_ is now known as sauce [19:49] stgraber: bleh :) [20:07] hallyn: bug updated with patch. I had slangasek take a quick look to see that I wasn't insane and that it's the "best" way of fixing the mess and he seems to agree. [20:07] hallyn: I'll just do another precise -> quantal -> raring upgrade, then run my new postinst and see if that fixes it. If it does, I'll upload lxc to raring and quantal-proposed [20:07] stgraber: cyphermox will do network-manager I recon? [20:08] hallyn: I guess so. I'll open tasks for all 3 packages as usual and target to quantal + raring [20:08] I really hope it's the last time we need to go through that mess [20:08] stgraber: I'll copy it over to libvirt then, thanks. [20:08] stgraber: wait, so it doesn't apply to precise? [20:09] pretty sure I had a precise bug open for libvirt for the original [20:09] precise is broken but the broken bits won't ever be called, so it's fine [20:09] oh, right, so that's only for this bug. got it [20:09] * hallyn being silly === Jikan is now known as Jikai [20:21] hallyn: stgraber: what bug? [20:25] cyphermox: bug 1157332 [20:25] Launchpad bug 1157332 in network-manager "/etc/dnsmasq.d-available/lxc circular link " [High,Triaged] https://launchpad.net/bugs/1157332 [20:25] cyphermox: yet another fallout of the dnsmasq stuff :) [20:29] utlemming: hey, wondering how you get the cloudimage-rootfs label on the ubuntu cloud images. Is that code somewhere visible? [20:30] Its done via a custom code branch of Live-build. It's old and will get updated next cylce. But here it is: lp:~ubuntu-on-ec2/live-build/cloud-images/ [20:30] how to open port 80 in ubuntu? (i dont have any GUI) [20:31] Quest: man ufw [20:33] SpamapS, is the default firewall of ubuntu is iptables? and i need iptables command? [20:34] ufw is a convenience wrapper for iptables [20:34] Quest: ufw is the recommended CLI frontend for iptables [20:34] https://help.ubuntu.com/community/UFW [20:34] Quest: however by default on Ubuntu systems, we don't firewall anything, so just installing apache (or any web server) should just work without explicitly opening tcp/80 [20:35] can i do that directtly by iptables [20:35] utlemming: ty [20:38] * stgraber loves ephemeral containers + gigabit mirror, precise => quantal => raring all done in 2min [20:39] getting a nice 82MB/s from the local proxy+mirror [20:40] grr [20:41] * ogra_ is envious [20:41] ogra_: still on 2Mbps SDSL? or did you upgrade to something from this decade? :) [20:42] to lazy [20:42] and i wouldnt have anything to moan about [20:42] you know that if you continue like that cjwatson will have a better internet connection than you do, right? :) === Jikai is now known as Jikan [20:43] i might just upgrade to LTE then ... phones are stacking up around me recently [20:45] if you can get reasonable pricing for LTE, sure. I only enjoy reasonably priced LTE when I'm in Switzerland. Where I live in Canada I can't find anything with more than 6GB a month and even 6GB is around 70$ so way too expensive. [20:47] and so far I've only found a single ISP here that's crazy enough to sell me unmetered internet, the others still put limits at 500GB or similar and then I'd have to pay crazy expensive extras for the remaining 1.5TB or so... [20:47] wow [20:48] thats bad [20:49] I should really poke my ISP at some point and ask what percentage of their total IPv6 bandwith I'm using, because at 1.5TB a month, I'm sure it's a very significant chunk of their total IPv6 traffic (for a fairly small consumer ISP) ;) [20:56] hallyn: new LXC uploaded to quantal and raring [20:56] hallyn: I'll ask for a new backport to precise once quantal is moved to quantal-updates [20:58] hallyn: now to test your lxc-clone change and I'll be done with lxc for the day (and hopefully, the week ;)) [21:05] utlemming: do you know why walinuxagent got rejected? [21:05] stgraber: the lxc-start-ephemeral -n change ending up in yoru ppa will be helpful to me :) [21:06] hallyn: yeah, it should be there tomorrow. So far I just applied it to my local copy of the script for testing [21:50] qq I just accidentally deleted everything in a directory is there anyway to undelete what I just did? [21:53] Darkstar1: first, give up hope. They're probably gone. Second, the 'recover' package claims it can recover data using debugfs from ext2 filesystems. Maybe debugfs can help with ext3 as well. But, uh, I'd be surprised. [21:56] sarnold: Thanks. Hope is now crushed [22:01] I have a couple of spare hard drives laying around, and am experimenting RAIDing my boot drive. I'm wondering if I can safely shrink my existing boot partition to mirror a (much) smaller drive without screwing something up, and if I can do that in such a way that I can expand that into a RAID5, then RAID6, after RAID1? [22:01] The drive I am trying to mirror with is 10GB, the current boot partition is 55GB, and I have about 6GB of data on the boot partition. I'm worried there might be a problem with fragmented files and concatenation screwing something up. [22:09] I've been given the suggestion that I should make backups before attempting this, so if someone can give me information on how to implement a backup scheme (preferrably a Tower of Hanoi solution) in addition to RAID, that would be phenomenally helpful. [22:15] Googol30: rsnapshot is pretty keen [22:15] Googol30: I've got rsnapshot making houryly, daily, weekly, backups, between two hard drives on my laptop; it goes ~100 gigs of data in two minutes most times.. [22:16] is that at all related to rdiff-backup? [22:18] hallyn: I don't think so... [22:19] just wondering [22:19] rsnapshot uses raw rsync [22:23] sarnold: Does rsnapshot take advantage of "journaling" which I've heard the ext series of filesystems uses? I haven't quite looked into filesystem types or features recently. [22:24] Googol30: as far as I know, the journalling in ext3, ext4, and related filesystems happens entirely without the application being aware [22:24] Googol30: it isn't like using a filesystem snapshot, as offered by zfs or btrfs -- journalling is just a way to try to offer asynchronous speeds with synchronous safety. [22:25] lvm also gives snapshots and is more stable than zfs/btrfs [22:27] Googol30: ah yes, see xnox's comment. if you want to use snapshotting, consider using LVM snapshots (block-level) rather than filesystem level snapshots, as LVM snapshots are far better tested .. and feel 'simpler' (to me, anyway) [22:27] Googol30: I live fast and loose with my data -- _some_ backups are far better than none -- so I don't bother with snapshotting filesystems. I just want to have another handy copy of my data should one drive die. [22:28] sarnold: Which is why I'm in the process of RAIDing and implementing a backup scheme. [22:29] Googol30: because it appears you care more about your data than I do mine :D hehe [22:32] I have data which I can't afford to lose, and making manual backups when I feel like it is becoming tedious. Additionally, server availability is important as well, and since I have a couple hard drives laying around, I thought I might experiment with RAID. [22:34] About backups: are snapshots an acceptable form of backing up a disk? [22:36] heh? [22:36] snapshots != backup [22:36] if they where, they would be called backups [22:36] raid is not a backup [22:37] backups are what you use when you accidentally drop the server in a lake, under a forklift, when the building burns down [22:37] I'm aware that RAID is not a backup. It is simply for availability. And I was assuming snapshots are a form of backing up data. [22:37] raid is what you use to keep the system running while it has a disk failure [22:38] snapshots can be used like a, instant available history [22:38] .. and snapshots are what you do to get a more-consistent view of the data _while making a backup_ [22:38] Googol30: base dies -> all snapshots based on the base are dead as well. [22:38] Googol30: it's incremental "back-up" with stress on "incrementals" [22:39] reverse incremental [22:41] So what's the difference between snapshots and actual backups? And what do you suggest for offsite backups, as I don't feel like relying on other entities to keep my data safe, nor do I feel like dealing with setting an autonomous version of that up. [22:43] Correction: I don't feel like paying another company for something which I can do myself, just as reliably. [22:45] Googol30: a backup is a physically separate copy of the data; a snapshot is an "frozen" _view_ of the data in the past that some programs will see, while other programs are allowed to keep making changes to the data === Jikan is now known as Jikai [22:46] the issue is, someone doing something stupid to the disk will kill the snapshot and your data [22:46] Googol30: for "production" sorts of systems, you might do something to quiesce your databases and similar 'live' systems in a stable state, take a snapshot, then let them run again. then you run a backup tool on the _snapshot_ of the data, to make your physically separate copy of the bits. Once the backup is done, you do something to throw away the snapshot -- and only the 'new' data survives. [22:46] and you will need a backup [22:47] wow. re-reading my description, I'm afraid I only made things worse. heh. :( [22:47] likely issues, admin wipes a disk using dd [22:48] what was it we had a few days ago [22:48] someone relabeled their zfs disks [22:49] I'm fully aware of what humans can do to a disk or data, which is why I'm going through the process of making backups and RAID. [22:51] I don't believe raid fixs human at all [22:52] Additionally, I'm fully aware of what RAID does and what it is for. Instead of telling me the consequences of _not_ setting a backup scheme up, can you tell me _how_ to set something up? [22:52] you asked [22:56] Googol30: I found rsnapshot easy to install and run; the /etc/rsnapshot.conf had reasonable defaults already provided for many settings, and there's LVM integration options in case you want to use them. === wedgwood is now known as wedgwood_away [23:08] patdk-lap: To clear up doubt about my knowledge of computers, RAID is a system of physically seperate disks joined to form a single logical disk, which prevents against anomalies such as total disk failures, bad sectors, and human screw-ups such as pushing a disk off a table, because the information is spread out. You are correct in your repeated statements that RAID is not for data backups, while I am correct in my state [23:12] If you have any other doubts or questions, please ask and I'll clear things up. With all due respect, I don't like being called wrong when I _know_ that I'm not wrong. [23:13] sarnold: Would I go about installing rsnapshots through apt-get, or must it be downloaded from somewhere? [23:17] Googol30: rsnapshot is available for install as usual :) [23:21] And according to patdk-lap, snapshots aren't "real" backups. Is he correct here, or will snapshots work as some form of backing up data, or supplement backups somehow? [23:21] If he _is_ correct, however, what else would I need to do to actually _backup_ my data? [23:22] Googol30: heh, he's correct, snapshots are nothing like backups, since there is still only one copy of the data with a snapshot. [23:22] Googol30: snapshots are useful for ensuring data consistency while making a backup. [23:23] hei [23:24] sarnold: So snapshots aid in the production of backups, correct? What would I need to do in addition to making snapshots to backup that data? Simply copy that snapshot to another disk? [23:25] sarnold: Should I just install rsnapshot and read the man page? [23:26] Googol30: yes, the snapshots give you a consistent state to backup -- the snapshot would live for five or ten minutes while the backup is being made, then you'd collapse it away again [23:27] Googol30: you can backup the snapshot data however you wish, with rsnapshot or duplicity or whatever else... [23:27] (duplicity might be nice if you want off-site backups too.) [23:59] Hi all.. I have a question about ubuntu cloud images.