Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 10 Aug 2001 15:40:31 -0400 (EDT)
From:      Jim Durham <durham@w2xo.pgh.pa.us>
To:        Duncan Sayers <duncan@apdata.com.au>
Cc:        freebsd-questions@FreeBSD.ORG
Subject:   Re: Re[2]: backup server
Message-ID:  <Pine.BSF.4.21.0108101537440.95280-100000@w2xo.pgh.pa.us>
In-Reply-To: <Pine.BSF.4.05.10108101229440.63938-100000@snowflake.apdata.com.au>

next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, 10 Aug 2001, Duncan Sayers wrote:

> On Thu, 9 Aug 2001, Jim Durham wrote:
> 
> > On Sun, 5 Aug 2001, Alvin Sim wrote:
> > > Sunday, August 05, 2001, 02:51:00 AM, Christoph Sold wrote:
> > > 
> > > > Alvin Sim wrote:
> > > >> 
> > > >> i'm looking into implementing 2 servers for a dept. and am looking for
> > > >> ways to to "mirror" a server -- ie, if one fails, all clients will
> > > >> automagically connect to the second server -- ala NT PDC's.
> > > >> 
> > > >> what do i (basically) need to do/implement? any pointers to some
> > > >> relevant docs would be great. thanks.
> > > 
> > > > There is no such thing as the magic you describe. To get a little bit of
> > > > this ideal solution, you'd have to define
> > > 
> > > 
> > > > 1) which services this boxes have to provide
> > > Samba 2.2.x (user authentication), Squid (proxy), IPFW (& NATD?) for
> > > Internet access priviledges, Web-Caching, and maybe DNS. i'm sure there
> > > are going to be a couple more services but this is basically the
> > > basics/needed ones
> > > 
> > > > 2) what the least acceptable working level of that service is
> > > Samba and? DNS, since they need the user authentication for (domain)
> > > logons and DNS to resolve? i'm not sure if this is what you meant by
> > > 'least acceptable working level'
> > > 
> > > > 3) how to detect the failure and
> > > this is what i am looking into as well and i really can't answer this
> > > one. i dont know if anyone that have done a lot of years of
> > > administration knows when a server is going to make a boo-boo either.
> > > maybe there is this someone and maybe he'll give me some pointers in
> > > waht to lookout for :) but i really doubt it since there are basically a
> > > lot of probable cause for a server to go down
> > > 
> > > > 4) how to switch over that service safely to the second box.
> > > i suppose this is the subject line. how does an NT server works in a DC
> > > environment? basically, this is what i was thinking of doing but... how
> > > do you make freebsd do something similar? (ie, synchronizing all datas -
> > > if at all possible, which i doubt, at a certain interval time?) i dont
> > > know, frankly.
> > > 
> > > > each of those considerations is pretty complicated, and there is no such
> > > > thing as a standard definition. Thus you have either to invest some time
> > > > to think about it, or pay some amount of money to get somebody who does.
> > > 
> > 
> > I have also thought about this a bit at our company.
> > 
> > At lot depends on how "fancy" you want to get. A very workable system
> > would be to have a 2nd server on a different IP, and do a tar using an
> > "mtime" of 10 minutes (or whatever period) and output it through an NFS
> > mount to the other system and untar it. This would transfer only what had
> > recently changed to the other system.
> > 
> > You could write a "watchdog" script to monitor the primary server
> > from the secondary and initiate an ifconfig to the primary IP and
> > a restart of Samba, etc on the new IP.
> > 
> > Of course, there are many problems, like a partial failure of the
> > primary that would be either difficult to detect, or would leave
> > the primary running on it's IP and interfere with the new primary
> > (the old secondary). Anyone who had a file open on Samba would get
> > strange results!
> > 
> > The only way to ensure a seamless transition would be to write
> > everything to both servers all the time. I'm not aware of any
> > way to do this, but maybe others are.
> > 
> > Just some thoughts.
> 
> 
> 
> How about using rsync in a script run by cron at specified intervals to
> synchronise files from one server to another. To quote the man page:
> 
> "The rsync remote-update protocol allows rsync to transfer just the
> differences between two sets of files across the network link."
> 
> Then you could have a replicated server (give or take a couple of
> minutes of changes, depending on the sync interval), to which you could
> cut over to if your "main server" croaks  
> 
> -- 

I wasn't familiar with it, but I found it in ports and read the
package description. Sounds interesting, but if it is a cousin
of rcp, it's probably unsafe in a secure environment as you would
have to run portmap and rstatd. I'd feel better if it were using
scp instead of rcp 8-) .

-Jim



To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.21.0108101537440.95280-100000>