From: Jonas Oreland
Date: July 16 2004 2:35am
Subject: Re: question about disk usage
List-Archive: http://lists.mysql.com/cluster/147
Message-Id: <001501c46add$826154f0$0300a8c0@laptop>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit

> Hi all,
> 
> Sorry if this is a basic question, but it's unclear to me at this point 
> whether the cluster stores its data on the disk, or purely in memory, 
> and what the plans are for this in the future. Basically, the problem 
> I've got is this: I put my data in the cluster and stop/start the 
> cluster, and all my data's gone. I understand that ideally, one would 
> never have to stop the cluster, but realistically, what if there's a 
> power failure or other disaster? I can't have all the data just *poof* 
> away. And if I were able to run an ordinary backup server (say InnoDB) 
> to replicate what is in the cluster, then I really wouldnt need a 
> cluster at all. Now ... I notice that before I stop the DB nodes, the 
> data directory is very large -- is the data stored there, and is it in a 
> recoverable format, and if so, why doesn't my cluster restart with it's 
> last known data?

1) All data is stored in memory at all time.
2) Data is checkpoint:ed to disk every so often
    (LCP & GCP)
3) Data is restored during restart
    (Note: you should _not_ do "-i" if you want to restart from checkpoint)

4) Plan is to change 1, so that one can store more than one has RAM
    but other properties will remain the same.

/jonas