Re: Hardware failover webserver cluster
Denis thanks for your "nutshell explanation"
I will try the following:
SRV1 -----> Node 1
client --> |(shared IP) \/
| /\
SRV2 -----> Node 2
:
Node n
SRV1 and SRV2 "monitoring software" and nginx with different backends as
load balancer
Node1 - Node n are backend servers with the Web application and the
Database.
Can I use a VPS for SRV1/SRV2 or should I use dedicated hardware?
I know that can not be generally answerd but I think there are just
proxy's they donest need guaranteed I/O and work in most cased with the
RAM or Cache.
On 27.03.2014 18:21, Denis Witt wrote:
> On Thu, 27 Mar 2014 16:47:15 +0100
> basti <black.fledermaus@arcor.de> wrote:
>
>> perhaps that's a bit off topic here but can someone explain what I
>> need to build a hardware failover nginx cluster?
> I a nutshell:
>
> * (at least) two Servers
> * a monitoring software
> * a shared IP
> * something that will switch the shared IP in case of failures
>
> One way is to use Corosync and Pacemaker. They will bring so called
> Ressource Agents who are capable of monitoring your nginx-Daemon and
> your network connection. If the daemon fails, or your network
> connections seems to be shaky (or gone) the Shared-IP (Service-IP) will
> be transfered from one Server to the other. Before that all needed
> services will be started if they wasn't running before.
>
> So take a look at Corosync/Pacemaker.
>
> As you seem to be German, maybe this article I wrote will help you:
> https://www.21x9.org/neuer-server-pacemaker-corosync-und-salt-stack/
> It describes the installation of Corosync/Pacemaker using Saltstack (a
> configuration management tool). Leaving Saltstack aside it might give
> you a good overview about Corosync.
>
> Bye for now.
Reply to: