Back to Home / #xen / 2007 / 02 / Prev Day | Next Day
#xen IRC Logs for 2007-02-17

---Logopened Sat Feb 17 00:00:35 2007
00:59|-|symerian [] has quit [Remote host closed the connection]
01:03|-|symerian [] has joined #xen
01:36|-|Basic_py [] has quit [Quit: Leaving]
01:59|-|visik7 [] has joined #xen
02:12|-|mejlholm [~mejlholm@] has joined #xen
02:17|-|KriP [] has joined #xen
02:31|-|jerone_ [] has quit [Quit: jerone_]
02:39|-|visik7 [] has quit []
02:50|-|visik7 [] has joined #xen
03:00|-|DoberMann_ changed nick to DoberMann
03:09|-|KriP [] has quit [Read error: Connection reset by peer]
03:11|-|KriP [] has joined #xen
03:46|-|mejlholm [~mejlholm@] has quit [Quit: Leaving]
03:59|-|symerian [] has quit [Remote host closed the connection]
03:59|-|symerian_ [~logbot@] has joined #xen
04:07|-|sputhenp [~sputhenp@] has joined #xen
05:02|-|mejlholm [] has joined #xen
05:04|-|symerian_ [~logbot@] has quit [Write error: connection closed]
05:05|-|symerian [~logbot@] has joined #xen
05:05|-|mejlholm [] has quit []
05:09|-|mejlholm [] has joined #xen
05:46|-|cableman [] has joined #xen
05:52|-|mastermind [] has quit [Quit: Client exiting]
05:52|-|mastermind [] has joined #xen
05:54|-|pvanhoof [] has joined #xen
07:20|-|mejlholm [] has quit [Quit: Leaving]
07:25|-|mejlholm [] has joined #xen
07:39|-|kalden [] has joined #xen
08:11|-|symerian [~logbot@] has left #xen []
09:35|-|cableman [] has quit [Quit: Leaving]
09:39|-|aliguori [] has joined #xen
10:03|-|mejlholm [] has quit [Quit: Leaving]
11:22|-|cdub [] has quit [Ping timeout: 480 seconds]
11:44|-|trondaso [] has joined #xen
12:00|-|mejlholm [] has joined #xen
12:12|-|mejlholm [] has quit [Quit: Leaving]
12:21|-|sputhenp [~sputhenp@] has quit [Quit: Leaving]
12:34|-|Pivert [] has quit [Ping timeout: 480 seconds]
13:13|-|Pivert [] has joined #xen
13:30|-|trondaso [] has left #xen []
13:48<bluefox>Does Xen have load balancing built-in, or do I need a third party monitoring tool that looks for overloaded hardware nodes and executes a live migration?
13:51<brendan>it's not built in
13:51<bluefox>That's alright though, it's easy enough to monitor from the outside.
13:52<bluefox>Does XenSource supply software to do it?
13:52<bluefox>Or do I have to roll my own
13:52<riel>you'll have to roll your own
13:52<riel>I don't think anybody's done it yet
13:54<bluefox>it supplies instant live migration though so no big deal
13:56|-|ivan_ [] has quit [Quit: leaving]
13:57<bluefox>Looks like Xen can load balance on one hardware node though
13:58<bluefox>if I allocate 500MB of my 512 to Dom0, and use 200MB; and then load a 256MB DomU; it suddenly rips 256MB off my Dom0, because it's available.
14:28<waldi>read the main config
14:41<bluefox>I need a clustering network storage system though
14:41<bluefox>something that can distribute storage across hardware nodes in such a way that if a hardware node fails, nothing is lost
14:43<brendan>drbd might work, I guess
14:44<bluefox>nice, network Raid-1 basically.
14:45[~]bluefox was more thinking distributed with parity and N-levels of redundancy.
14:45<bluefox>I can't think of a way to do it asynchronously though
14:46<brendan>I don't know what does that
14:46<bluefox>I don't think anything does
14:47<bluefox>I'm thinking, if I have a server with 2000GB, another with 1200GB, another with 2500GB, it should be able to distribute data across them and manage redundancy so I can lose 1 host; if I have 15 servers, I should be able to tell it to be durable against loss of 3 hosts.
14:47<bluefox>This is going to lead to me actually designing something to do this isn't it :/
14:47<bluefox>Except I'll never implement it :/
14:47<brendan>good luck recovering if you lose the server with 2500GB
14:48[~]bluefox is not sure how to negotiate that
14:49<bluefox>Perhaps parity across 1200GB; then parity across 800; that leaves like 500 unprotected though.
14:50<bluefox>somehow it would have to ensure that data does not exist "on one host only" but rather prove that any given data has N paths each independent of each other for so many hosts etc etc etc
14:50<bluefox>things that someone who is actually smart would have to figure out ;/
14:53|-|rw23 [] has quit [Quit: Leaving]
15:17|-|aw [] has quit [Quit: Leaving]
15:37|-|KriP [] has quit [Read error: Connection reset by peer]
16:14|-|DoberMann changed nick to DoberMann[ZZZzzz]
16:49<bluefox>work in progress
17:05|-|aw [] has joined #xen
17:21|-|dendrite [] has joined #xen
17:21|-|hollisb [] has joined #xen
17:25<dendrite>Are there any docs that discuss xen specific vulnerabilities, e.g., breaking out of the vm, etc?
17:52<icblenke>bluefox: have you looked at Ceph, Glusters, GFarm, Cleversafe, or any other network distributed filesystem?
17:53<icblenke>drbd is great for failover pairs. drbd-0.8 lets you do master/master replication and live failover rather easily.
17:54<icblenke>They all boil down to FUSE based frontends to a "metadata" cluster and a "storage" cluster.
17:56<icblenke>we're looking at them now, evaluating them to see which is best for our needs.
17:57<icblenke>heading out for a while. I'd like to hear your evaluations. the last time I looked at Xen and ATAoE was over a year ago.
18:14<bluefox>No idea, I'm no expert in the field; I'm just absorbing this information as rapidly as I can
18:14<bluefox><-- brain is a relational database, associates everything with everything
18:17<murb>does ATAoE work with xen now days?
18:21<bluefox>murb: it's a block device exported over a network, looks like an IDE drive
18:22<bluefox>murb: ATAoE + NFS + Xen -> ATAoE cluster :)
18:22<murb>bluefox: i know what it is, before there were bugs which prevented it working with the xen network devices.
20:26<icblenke>there are still some irritating behaviors with ATAoE I'd love to hear if they've addressed: like having targets drop out causing the need to unload/reload the ataoe initiator to recover when the target is restored.
21:00|-|visik7 [] has quit [Read error: Connection reset by peer]
21:00|-|visik7 [] has joined #xen
22:08|-|hollisb [] has quit [Quit: Leaving]
22:32|-|cdub [] has joined #xen
22:58|-|VS_ChanLog [] has left #xen [Rotating Logs]
22:58|-|VS_ChanLog [] has joined #xen
23:26|-|DoberMann_ [] has joined #xen
23:28|-|DoberMann[ZZZzzz] [] has quit [Ping timeout: 480 seconds]
---Logclosed Sun Feb 18 00:00:23 2007