Backup Server

[anubis]

On Sat, 27 Dec 2003 03:30 am, samy lancher wrote:
> Hello all,
> I have a 4.5 FreeBSD server. It is our Email, web and database server. I
> would like to setup a backup server so that when the main server goes down
> the backup server takes over its job. Could some one please tell me the
> best way to setup a backup server and also suggest some good documentation.
>
> Thanks in advance,
> Naveen.
>
>
> ---------------------------------
> Do you Yahoo!?
> Protect your identity with Yahoo! Mail AddressGuard
> _______________________________________________
> freebsd-questions at freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-questions
> To unsubscribe, send any mail to
> "freebsd-questions-unsubscribe at freebsd.org"

I have had a bit of a look into this myself and this is my take on it.  I 
would like to hear of other people experiences too.

There are a number of things that you have to decide on first before you go 
any further. 

These are:
	budget
	how critical the system is to downtime
	how much data you are willing to lose
	how long are you willing to wait for the second system to kick in.
These will determine how you are going to build your system.  You will have to 
keep the answers in mind when you are looking at any solution.

What you seem to be looking for is a failover system.  There is a fair bit 
written about failover systems.  Googling will find you lots.  Make sure that 
you look up linux high availability and failover as well to get a broader 
view.  I have added some links below.

There is really 2 things that you are trying to do here.  Provide redundancy 
for the services and redundancy for the data.  The services are a bit easier 
and cheaper than the data.  The big problem is the data, especially 
databases.  Due to their nature they cant easily be copied while live.  

A solution to this is a SAN.  With lots of money it is easier as you can buy 
yourself a SAN and hook the two machines to it and host the data on the SAN.  
With some clever scripts from those HA sites when one machine goes down the 
other can take over and use the same data.  There are other solutions using a 
fancy Y shaped SCSI cable to a external drive array.  Others my be able to 
help here as I dont know about them.

The other alternative is 2 identical machines.
When you have 2 machines with the master storing data on its local drives it 
gets tricker.  This is where you have to decide on how much data you are 
willing to lose.  

As an example we have a bsd box that rsyncs our windows fileserver ever hour.  
Should windas go down we run a script on the workstations remapping our 
drives to the bsd box.  In this case we are prepared to lose up to an hours 
work.  We are also prepared to lose say 15-30 minutes of time mucking around.  

In your situation perhaps what you could do is upgrade to 5.1 and rsync 
snapshots of your data to the secondary machine.  You could use the failover 
setup as described on HA sites to fire up the services on the secondary 
machine and take over.  This should work as snapshots are supposed to capture 
an instant in time but I couldnt guarantee it until I tested it.  You would 
still be losing data as you could only snapshot data and transfer it in 
discrete intervals.

A handy thing that linux has that I dont think that freebsd has is drbd.  This 
is a block device that can mirror data across a network.  If freebsd had this 
it would be easy to make the second machine a true mirror of the first.  
I wonder if they are looking at a thing similar to this in the future.

Look here for some intersting reading

http://linux-ha.org/
http://www.drbd.org/
http://sporner.dnsalias.org/
http://failover.othello.ch/getting_started.html