[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: server mirroring



This is a tough nut to crack.

There isn't just a single bullet that will fail over an entire system. 
You really have to look at applications that will need to fail over one
by one and determine how best to deal with these.

Some ideas to consider in your architecture:

* For web apps, you can rsync files from your staging environment to
your live environment as well as your failback box.  If your primary box
fails, you can bring your failback box online.

* If you need failover on web apps, certain J2EE containers will support
this, even to the level of failing over sessions to other machines in
the cluster.

* You can look at linux-ha.org for more ideas

* Think about load balancers and IP redirectors along the lines of F5's
big-IP or a Cisco LocalDirector, although you can easily replicate some
of the functionality that these products provide with a Debian box or
two.

* For spooling mail in the event of a primary machine failing, look at
adding additional MX records to your zone file.

* DNS is easy to handle... the O'Reilly DNS and Bind book is great for
learning about this.

* You do have a backup and recovery plan, don't you? ;)

But your question was about automatic, online mirroring of all file
changes... again, no silver bullet as far as I know of.  Depending on
your need, you could NFS mount filesystems off of a file server.  In the
system fails, mount the filesystems off of another box.  If the NFS box
fails, then you have problems.

For mirroring files, you should definitely look at rsync.  This is the
best solution to mirroring that I have seen.

If you are just talking about files, this is not a terribly difficult
problem, but when your apps have data in memory (i.e. session keys,
etc.) that you need to recover, this is remarkably trickier.

So there are some ideas.

phil.

Marcin Sochacki wrote:
> 
> Hi!
> 
> I would like to set up a secondary server with a complete mirror of my
> main production machine. In case the first one fails I could (manually)
> do a switchover. I don't need things like balancing, IP-takeover, etc.
> 
> The ideal solution should include automatic, online mirroring
> of all file changes from master server to the secondary one.
> 
> What is your suggestion to solve the problem?
> 
> Marcin
> 
> --
> To UNSUBSCRIBE, email to debian-isp-request@lists.debian.org
> with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

-- 

                                Whirlycott
                                Philip Jacob
                                phil@whirlycott.com
                                http://www.whirlycott.com/phil/



Reply to: