Koozali.org: home of the SME Server

Legacy Forums => General Discussion (Legacy) => Topic started by: Christopher on February 16, 2005, 09:16:56 AM

Title: SME Primary & Secondary Servers
Post by: Christopher on February 16, 2005, 09:16:56 AM

Hi all,

I would like to create a system where I have an SME server which is the "Primary" server, then a 'secondary' SME server, that normally is simply an RSYNC backup of the Primary server, but in the event of the primary server going down, the secondary server can be cut in quickly, perhaps by a quick name change or reboot, so the primary can be repaired 'offline'.

This is as well as my normal backup routine.

The setup makes perfect sense to me, but I can't find any examples of anyone doing, so i figure, either I'm searching for the wrong thing or it doesn't make sense to anyone else!

The real issue is that I support systems remotely, and I need a setup which will not be down until I get there to restore a CD backup.

Does anyone have any experience to share, I would love to do a HowTo if there isn't already one out there.

Many Thanks

Chris

Title: SME Primary & Secondary Servers
Post by: irian on February 16, 2005, 05:13:04 PM

Hi Christopher,

I tried this once, but couldn't get the machines to "talk" with each other smoothly.
Since I didn't have any time left I installed a second server with Fedora Core 2, that just stores the backups of the data-files using rsync.

If you or anyone else gets this working, can you please write up a how to. I would be very interested.

Quote

The real issue is that I support systems remotely, and I need a setup which will not be down until I get there to restore a CD backup.

That was the reason for me to try and get this to work too.

Irian

Title: SME Primary & Secondary Servers
Post by: jfarschman on February 16, 2005, 05:23:03 PM

Chris,

  I don't believe anyone has done a system like this.  There has been some work done by Stephen Noble on rsync.  His work is quite good and echoes the best practices outlined a scrounge.org

http://www.scrounge.org/linux/rsync.html

  I did this sort of work for Sun/Cobalt RaQ servers a couple of years ago.  It was a bit tricky to get started.  Mostly, it was about knowing what file 'NOT TO SYNC' and puttin them in the exclude list.

  If you are interested, I can attach my scripts and notes here for the RaQ servers.  They may help get you started.

Title: SME Primary & Secondary Servers
Post by: Christopher on February 17, 2005, 04:26:03 PM

Hi Jay,

Your scripts would be a big help,

thanks.

Regards,

Chris.

Title: SME Primary & Secondary Servers
Post by: jfarschman on February 17, 2005, 05:14:31 PM

Okay...

  Here is something I wrote a while ago explain the methods used on the RaQ Servers:

http://www.hitechsavvy.com/modules.php?op=modload&name=News&file=article

This examples are of an older script.  I have a new script here, but it's not configured to run a nightly backup.

http://www.hitechsavvy.com/downloads/rsync_script.tar.gz

Note: my backups happen on a second interface in RFC 1918 space.  Drop me an email if you want to chat about this.  jay@hitechsavvy.com

Title: SME Primary & Secondary Servers
Post by: hanscees on February 19, 2005, 10:10:43 PM

Setting this up may be rather complex.

Another way to do this is to "clone" your system to another hard-disk in the same machine once a day. If that hard-disk is in a drive-bay all one has to do is get it out, stick it in a prepared machine and boot.
You can train the local secretary to do that.

Not exactly what you want, but it might work.

another thing you can do is make a mondo backup on a cd.
You can let a second server boot with that cd and install it in 15 minutes. But you have to be there really to do that.




hc

Title: SME Primary & Secondary Servers
Post by: stiperstones on February 20, 2005, 01:49:09 PM

Try this link maybe the sort of ting you are looking for

http://sme.swerts-knudsen.dk/index.html?frame=http%3A//sme.swerts-knudsen.dk/howtos/howto_14.htm

Title: SME Primary & Secondary Servers
Post by: Damian on February 27, 2005, 08:46:52 PM

Chris,
I've set up automatic failover on Linux machines for corporates using shared NFS for the data store (and once shared SCSI storage for a linux cluster running oracle 9i). I used heartbeat which is a seriously cool package. There was some talk in the heartbeat forums about using DRDB to replicate the data on write  across both servers so that when you failover you have everything available to the slave. Cool system and if both machines are within network distance, not too hard to set up.
Haven't tried it with SME although I don't see why it souldn't work. Could look into it if you like.
Damian

Title: SME Primary & Secondary Servers
Post by: del on February 28, 2005, 05:29:24 AM

This maybe of interest:
http://www.opens.com.br/documentacao/HA/
Regards,
Del :-)

Title: SME Primary & Secondary Servers
Post by: jfarschman on February 28, 2005, 06:31:37 PM

This has got to be one of the coolest forums I've ever participate in.

Thank you guys for pointing out all kinds of cool resources.

 :hammer: