[14:20] what the hell...stupid selinux [14:20] centos totally fsck'd my practical final exam [14:21] i spent the whole time trying to figure out why ssh with keys to a centos vm wasn't working [14:21] turned out to be selinux related [14:21] i guess ssh-copy-id from ubuntu to centos didn't set the appropriate context on the authorized keys file [14:22] luckily, i gave them 2 classes to do the exam...and it's mostly automated by scripts [14:22] but still [14:22] stupid selinux [15:26] * _bbb_ shakes fist at selinux [15:27] <_bbb_> setenforce 0 ftw [15:50] * thafreak always feels bad just turning it off... [15:50] like saying, this condom sucks, and throw it away mid coitis [15:51] thafreak, discretely though, of course :) [15:51] totally asking for herpes...or getting back door'd... [15:51] wait...i think i'm off topic [15:52] or am i... [15:52] hah [15:56] <_bbb_> so apparantly [15:56] <_bbb_> i have genital herpes [15:57] that's okay, everyone else does too. [15:57] just don't tell anyone. [15:59] server herpes...i think it's symplex qq [16:00] so you guys know about puppet? [16:00] i figure i should be using it... or even start learning about it. [16:01] it's one of those things where, i think i'm a failure as a small sysadmin [16:01] but some of me is like.. eff it.. for craps sake [16:34] i played with puppet [16:34] it's not bad [16:35] never setup a puppet master though...just made single recipes [16:36] i liked the puppet syntax better than chef [16:37] depends on what you want to do i guess...so many tools with lots of overlap [16:37] i also really like fabric [16:37] <_bbb_> i like turtles [16:38] * canthus13 meant to go to a puppet presentation at ILF... [16:38] I fell asleep instead. :P [16:38] sleep is for the weak [16:38] * thafreak looks at canthus13 [16:40] * canthus13 pokes at thafreak. [16:40] ouch [16:40] * canthus13 was saving energy for the afterparty. :) [16:40] i mean, pfft, that didn't hurt [16:40] oic [16:40] who was there? any entertainment...or just booze? [16:42] Just booze... and a hovercopter thing. [16:43] the main entertainment was the hackerspace with lockpicking competitions. [16:45] and...you pick any locks [16:45] Yep. :) [16:47] you win anything? [16:47] Nope. [16:47] * canthus13 can handle simple padlocks but that's about it. [18:27] ah up in cleveland? [18:28] thafreak, bro, thanks for the response on puppet :) [18:28] np [18:28] if you can stand ruby code, chef seems to be more popular [18:29] if you can't stand to look at ruby...puppet seems easier to swallow... [18:29] and if you start with just a single puppet manifest file, it's actually not too hard to jump in [18:30] or, if you like python, fabric is cool...basically write code that gets run over an ssh connection to a remote machine [18:30] or :) [18:30] you can use fabric to put puppet manifests on remote machine and then run it [18:32] but anyway...for whatever reason the hotness for the "agile" "devops" type sysadmins seems to be chef [18:35] Anyone try 12.04 lately? server or xfce? (i'll avoid asking about unity, as that's justa can of worms) [18:35] or lubuntu [18:37] installing 12.04 server tomorrow night on two machines and have 12.04 unity on 4 machines now so far so good they fixed a lot of the unity bugs [18:37] cool [18:38] toddc: they got rid of unity? [18:39] YAY!! they updated the qemu-kvm to support librados...sweet [18:39] i was worried for a while it wouldn't make it in [18:39] FYI, raddos is neat... [18:39] well there is a gnome 2 option but I have been using unity for a while and got used to it [18:40] * thafreak tries to change subject off unity [18:40] thafreak, damn bro, that is some crazy information [18:40] thafreak, what is librados? i love kvm [18:40] goddamn i gotta google this s hit [18:40] i wasted...erm spent, a number of hours...erm weeks...trying to figure out which one i'd like to use best [18:41] so, you hear of ceph? [18:41] never [18:41] ceph is a distributed filesystem in the main line kernel [18:41] oh awesome [18:41] like andrew [18:42] how it works, is it uses object stores, that store dumb blobs of data [18:42] AFS [18:42] i dunno the underlying nature of AFS...but from the user point of view, they're probably similar [18:43] well, ceph isn't really ready for prime time yet...BUT, the dumb blob storage part is there [18:43] one thing they built ontop of it is RBD images...which are like NBD (network block devices) [18:44] but the RBD is broken up into chunks, and striped across your cluster of dumb blob storage machines [18:45] so, you can give a VM a disk device that is an RBD drive, and it's data is striped across a bunch of storage nodes for speed/redundancy [18:46] at work, we have 4 kvm nodes, and right now, they use a central NAS for storing disk images...well, I can't reboot the NAS now, not without taking EVERYTHING down [18:46] so i've been scouring the net looking for a more distributed solution, and these rbd images seem like they might be the ticket [18:47] where do you keep up with this information? [18:47] is there a magazine i should be reading? [18:47] so I'm excited 12.04's kvm has support for rbd...finally something for me to be excited about for 12.04 :) [18:47] uh...nope...i spend LOTS of time on google honestly [18:48] wow this is great [18:48] probably 1/3 to 1/2 my day realistically [18:48] very little time to actually work on anything :) [18:48] yea, RnD :) [18:51] so what kind of network infrastructure would be internal to that file system? would it have to be fiber? [18:53] no [18:53] but most of the people seriously building distributed file systems are always talking about 10gig-e [18:53] so gigabit copper is fine [18:53] ah ok [18:54] shit :) [18:54] yeah, for most normal people, gigabit is still faster than most disk drives [18:54] gigabit is faster than most drive interfaces. :P [18:54] and would there be a redundant switch(es) between this? [18:55] could be i suppose [18:55] i guess, which is the next step in total redundancy? [18:55] not sure how multipathing would work like that though [18:56] i don't know if there is such a thing...there's always trade offs [18:56] true redundancy, you'd need multiple geographically separate locations... [18:56] and sync'ing between them usually makes things slow... [18:57] distributed filesystems like this are more meant to scale up in size and speed more than be really redundant [18:57] aaah.. i see.. i was headed in the wrong direction [18:57] making it possible to store petabyte size stuff [18:58] so... do you back up this monstrosity? or is it it's own backup.. you hope atleast? [18:58] now maybe you could hook something like async drbd to your rbd image and sync it to a remote location too... [18:58] is the filesytem taking snapshots like zfs? [18:58] for me, i'm going to make the central NAS be the backup [18:59] well, the storage nodes that store objects, they use an underlying file system [18:59] and they recommend btrfs currently [18:59] but ext4 would work too [18:59] yea, btrfs last time i looked wasn't ready for anything [18:59] not sure where you would do snapshotting at... [19:00] it works pretty good in some situations...hopefully it will be ready soonish [19:00] yea, what your talking about is way over my head :) it makes me excited though [19:00] me too... [19:01] there's also something called sheepdog, fyi, if you have more than one kvm server [19:01] i'm still stuck in user application land.. not so much infrastructure at this point. [19:01] i haven't dug into that too deeply though [19:01] i have two kvm servers [19:01] i will look it up [19:02] i guess it's not uber fast, but more focused on reliable replication of block devices between nodes... [19:02] so you can live migrate between any of your N nodes [19:02] wow nice [19:03] i'll hopefully check it out and this rbd stuff soon [19:03] that stuff is very exciting. i might wait a bit till it becomes polished a little, i hope that doesn't make you hate me : [19:03] see which works best [19:03] nah [19:03] i trust the LTS stuff [19:03] so if it's in there, i'll use it [19:03] if I have to build things from source...i tend to avoid it sadly [19:03] so i understand [19:04] 8.04 was a bit of a kvm pain :) [19:04] the rbd stuff is in this LTS...and i think sheepdog is too...so i'm assuming they felt it polished enough [19:04] awesome, i will try it then [19:04] i used xen still on 8.04 [19:05] actually, I still have one 8.04 xen box running....need to migrate the vm's off [19:05] yea, it wasnb't really that long ago [19:05] i wish i had more machines to test on [19:30] skellat: Your podcast was in the Ubuntu Weekly News. [19:32] :-) [20:00] Cheri703: You happen to see something like this http://www.webupd8.org/2012/04/valvle-works-on-steam-for-linux.html ? [20:02] Unit193: Funny. [20:02] canthus13: I don't care, but what's-his-name might. [20:41] i've used steam in wine...but i would probably do alot more pc gaming if they actually did port steam to linux === ronnoc_ is now known as ronnoc [21:42] I'm alive >.> [21:42] This is news? ;0 [21:42] Welcome back. [21:42] It should be ;P [23:21] Visig0th: That's a rather lame visigoth raid. :/ [23:23] Sorry, was having client issues.. [23:25] I shoud've known better than trying to experiment with xchat again after being away from it for so long and having been used to irssi. [23:25] Its fixed now, so no more jumping in and out. [23:26] And of course hello, btw. I'm Bradley. I generally go by simply Visigoth, that is without the "0". I'm in SW Ohio.