HowtoForge Forums | HowtoForge - Linux Howtos and Tutorials

HowtoForge Forums | HowtoForge - Linux Howtos and Tutorials (http://www.howtoforge.com/forums/index.php)
-   General (http://www.howtoforge.com/forums/forumdisplay.php?f=25)
-   -   Multiserver ISPConfig3 Failover, Loadbalance and High Avialability (http://www.howtoforge.com/forums/showthread.php?t=59690)

zenny 25th November 2012 15:57

Multiserver ISPConfig3 Failover, Loadbalance and High Avialability
 
I have read two of very nice tutorials here, viz.

1) http://www.howtoforge.com/installing...th-ispconfig-3 by Till

2) http://www.howtoforge.com/virtual-mu...th-ispconfig-3 by Michel

Both are very nice. Actually I am thinking of using a cluster (like 1 above) with split servers (like in 2) in openvz containers.

However, though the tutorial 1 by Till categorically states that it deals with "fail over, load balancing and high availability", but is it only me that I see no such things except syncing with unison /var/mail and /var/www?

It would be nice to if someone can exlpain how failover and loadbalancing is possible with two servers (either local or remote) which may look something like this (https://www.ovh.co.uk/dedicated_servers/ip_failover.xml) ?

Thanks in advance.

till 25th November 2012 17:00

Quote:

However, though the tutorial 1 by Till categorically states that it deals with "fail over, load balancing and high availability", but is it only me that I see no such things except syncing with unison /var/mail and /var/www?
Seems as if you did not read the full guide yet othwise you might have noticed that it contains also a synchronisation of the clinet databases trough mysql and it instructs you to install ispconfg as well in mirroe mode which does the configuration sync between the systems.

Quote:

It would be nice to if someone can exlpain how failover and loadbalancing is possible with two servers (either local or remote) which may look something like this (https://www.ovh.co.uk/dedicated_servers/ip_failover.xml) ?
The complete setup required for this kind of failover is described in guide 1). All configuration mirroring etc. is done by ispconfig, all you have to do in case of a afilure is to switch the IP address like described by ovh or put a loadbalancer in front.

onastvar 10th July 2013 17:42

ISPConfig3 Failover / ISPConfig3 Redundancy
 
Quote:

Originally Posted by till (Post 288703)
All you have to do in case of a failure is to switch the IP address like described by ovh or put a loadbalancer in front.

I have two physical servers at two different locations setup using this guide.

Could you please explain how to switch the IP?
Also, how put a loadbalancer in front, is there a guide to follow to set this up?

s1lv3r 12th July 2013 22:00

Quote:

Originally Posted by onastvar (Post 299641)
I have two physical servers at two different locations setup using this guide.

Could you please explain how to switch the IP?
Also, how put a loadbalancer in front, is there a guide to follow to set this up?

This is totally dependend on where you have your servers. As an example if you are with ovh you have to either rend a cisco hardware load-balancer or switch your fail-over-IP via the API if one servers fails. You could also use a vRack for this.

With some providers (mostly when they provide a private network for your servers only) it's also possible to simply use a gratuitous arp to take over the IP address (and the traffic) from another machine.

With the most providers I know, it's only possible to use fail-over-IPs between machines in the same datecenter due to routing limitations.

Either way, you got to see what options are offered by your provider for this case.

onastvar 15th July 2013 08:34

My servers are not in data centers, servers are in
two different countries, have 2 different ISPs.

How to manually switch IP (in case server 1 @ location 1 is down)
and I want to use server 2 until server 1 is back online?

I'm looking to find out what's the easiest/fastest way to use server 2
in case server 1 goes down (power outage/connection outage)?

Thanks!

s1lv3r 15th July 2013 10:28

hmmm ... that's not so easy in this case. If you don't have your own RIPE block which you can route yourself, your IPs are actually belonging to your provider and you are only renting them. It's very unlikely that your provider will allow you to route his IPs to a another provider.

I guess your only chance to switch your services from one server to another with this setup is to change your DNS to route to the other IP in case of failure. That's not really a great solution, because this means your services will be down until the dns propagation time is over, but the only solution I see in the moment.

onastvar 15th July 2013 23:09

I don't understand, in what environemt is
Installing A Web, Email & MySQL Database Cluster On Debian 6.0 With ISPConfig 3
tutorial helpful?

I have server 1 and server 2 (mirror of server 1) and
I cannot make it work just because I don't own IPs
but rent them.

Isn't there another solution, so server 2
could be used if server #1 is down?


Thanks!

Toucan 16th July 2013 00:59

You could try failover dns. Dns made easy offer this. Or you could to do this manually by changing the a record. Set a low ttl and traffic to your backup would be routed pretty quickly.

s1lv3r 16th July 2013 12:07

The tutorial is especially helpfull if:
  1. you are in a local network using local IP's,
  2. your provider gives a seperate network to you in which you control the routing,
  3. you have your own IPs which you can route yourself,
  4. your provider offers you any tool or method how to switch over IP adresses between servers (often sold seperately as "Fail-Over-IPs").

If none of the above applies to you, you are unable to switch IP's and left to DNS based fail-over only.


All times are GMT +2. The time now is 07:28.

Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2014, vBulletin Solutions, Inc.