Date: Thu, 06 Dec 2001 23:22:00 -0700 From: Chris Fedde <chris@fedde.littleton.co.us> To: "Leif Neland" <leifn@neland.dk> Cc: "Patrick O'Reilly" <patrick@mip.co.za>, freebsd-questions@FreeBSD.ORG Subject: Re: Load balancing stuff (mainly samba) Message-ID: <200112070622.fB76M0N95888@fedde.littleton.co.us> In-Reply-To: <00b701c17a7f$ca784740$6d05a8c0@neland.dk>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, 1 Dec 2001 16:49:34 +0100 "Leif Neland" wrote:
+------------------
| > I have been asked to set up a mail server with a hot backup which
| could
| > take over should the first server fail. Does anyone have any
| > "real-world" experience using coda for this type of problem?
| >
| > Does coda even fit this problem? It is described as a replicated
| > network file system. That tells me that data would be kept safe by
| > replication, but there may still be a single point of failure,
| namely
| > the mail server itself which is simply making use of the file
| system.
| >
| > Any comments?
| >
| If you use two machines, and having mailservers (smtp and pop3) on
| both, I think that would work. Either the same machine as hosting the
| coda, or use two others for that.
+------------------
Coda might help in the mailboxes (use maildir format or some other
serialization mechanism) but it's not going to be a good
choice for the queue directory. Locking over any distributed file
system is always problematic.
+------------------
| You could write a script to change the ip of smtp and pop3 to point to
| the active server using ddns.
+------------------
This is better done through address adoption than DNS changes.
Simply bring up the IP of the failed service on an appropriate
alternate server. Just be sure that you move it back when the original
comes back up. Console warnings will help you remember to to do this.
FWIW. I think that automating fail-over is a waste of time. True
HA is hard and rarely works as you expect it to in a real emergency.
Make sure that your configuration boots cleanly, that your backups
work, your monitoring is good, and your operations procedures are well
ironed, before you spend time on automating fail-over.
--
Chris Fedde
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?200112070622.fB76M0N95888>