[Sheepdog] sheepdog and RAID

[Ski Mountain]

Thanks for the reply.

While setting up RAID 0 would be a option, there must be more elegant solutions 
out there.  Just because one disk out of the whole array dies breaking the whole 
array and drastically changing the size of the sheepdog cluster.  

One problem I do see with starting many sheep daemons on the same server that 
has many disks is that (especially on small clusters) it is possible for all 
data for one or many Virtual Machine to be stored on one physical server.  


Would it be possible to do say
sheep /store_disk0  /store_disk1  /store_disk2  /store_disk3  /store_disk4  
/store_disk5 /store_disk6
So all mount points on a server would be handled by the same sheep using 
multiple threads

Add a mount point 
sheep -a /store_disk7

Remove a mount point
sheep -r /store_disk7

Hope that all makes sense, looking froward to seeing sheepdog grow and mature.  


________________________________
From: MORITA Kazutaka <morita.kazutaka at lab.ntt.co.jp>
To: Ski Mountain <ski_the_mountain at yahoo.com>
Cc: sheepdog at lists.wpkg.org
Sent: Thu, March 17, 2011 12:56:16 AM
Subject: Re: [Sheepdog] sheepdog and RAID

At Tue, 15 Mar 2011 20:42:46 -0700 (PDT),
Ski Mountain wrote:
> Sorry about prematurely sending the previous message by mistake.
> I have a question about setting up sheep dog.  Since sheepdog takes care of 
>data 
>
> replication and node addition and subtraction.  What do you do when you have 
> multiple disks in a machine in the cluster.  I assume you want to raid one the 

> OS, but what is best practices for the directory that stores the objects.  Is 
>is 
>
> best to start the sheep deamon on each mount point, or do you need to RAID the 

> multiple disks.  I figured that RAIDing the sheepdog disks would waste space 
> making multiple copies would be a waste since sheepdog is RAIDing the data.  So 
>
> would be best for sheepdog to take care of the data distribution and 
>replication 
>
> or what is the best method.  
> 
> $ collie cluster format --copies=3
> 
> Method stated in documentation
> $ sheep /store_dir
> 
> Or on each disk
> $ sheep /store_disk0
> $ sheep /store_disk1
> $ sheep /store_disk2
> $ sheep /store_disk3
> $ sheep /store_disk4
> $ sheep /store_disk5
> $ sheep /store_disk6
> 

I think using RAID 0 (no redundancy) is the easiest way.

Of course you can start a sheep daemon for each disk.  In that case,
please specify the different port number.  For example:

  $ sheep /store_disk0 -p 7000
  $ sheep /store_disk1 -p 7001
  $ sheep /store_disk2 -p 7002
  $ sheep /store_disk3 -p 7003
  $ sheep /store_disk4 -p 7004
  $ sheep /store_disk5 -p 7005
  $ sheep /store_disk6 -p 7006


Thanks,

Kazutaka



      
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.wpkg.org/pipermail/sheepdog/attachments/20110317/27f9128f/attachment-0003.html>