Looking for How-To to setup a Fall-Back systeem for Directadmin system

HageHosting

Verified User
Joined
Sep 25, 2006
Messages
84
This is a question, and looking al over the internet for weeks. But stil no satisfied answer.
I am looking for a system or server setup, for a running directadmin webserver,
thats must be convert to a fall back server.
In short saying when one server crash the others are going further with almost no downtime.
I saw MooseFS and Cloudlinux, are al those system working on a cloud of different servers. And when a server crashed the others go further ?
 
I would like to study a Cluster of DirectAdmin Server, but this require some time, licenses and test server i dont have at the moment.

Im waiting for DA Staff to enable licenses i required for those study, but, will not take few time...

Regards
 
On what OS you liked to try ? CentOS with CLoudlinux or under MooseFS ? In these days with customers regarding good uptime there must be a solution:)
 
I saw MooseFS and Cloudlinux, are al those system working on a cloud of different servers.

MooseFS is a file system which operates across multiple servers, which may be in different datacenters (though obviously this will affect data access speed). It offers a degree of tolerance in that you can store multiple copies of each file (similar conceptually, I'm guessing, to RAID1), but it does nothing to manage multiple DirectAdmin servers to access the filesystem at the same time.

CloudLinux does not address what you're thinking of when you use the word Cloud; rather the word refers to the fact that the earlier use of the word referrs to Cloud meaning the Internet as a whole. CloudLinux is designed to create a cloud server (or in other words, a server on the Internet rather than one local to you) which is more secure than the standard CentOS/Red Hat Enterprise Linux distributions.

Jeff
 
Thanks for reply Jeff, but i wil be more in detail. I have now at this moment one server running with DirectAdmin under raid1.
And what the solution for me is a system with extern storage systeem for de databases and new servers in setup with redunancy with a slave and master server.
And when the master servers goes doen the slave server takes over.
And both servers with redunant power supply and a way that the backup servers takes over when the master crashed.
I dont know how big websites that does.
In short way

[INTERNET] <-----> [DA SERVER MASTER] <------> [DA SERVER SLAVE] <-----> [DATABASE SERVER IN KIND OF MooseFS filesystem] and when 1 slave possible mayby also 2 slaves?

For a good uptime guarantee
 
I do that, but it doesn't rely on DA. I run the free version of ESXi and do cross backups between servers at night. If something happens to one of the servers, then I just start the backup VM on the other.
 
Ok, here an update.

I did some test with iSCSI and ZFS file system (on a NAS4Free storage).

The problem as i did suppose is the sync between two servers, i mean, if both mount /home to the same iscsi device, each server dont see other server data.

So, first of all, i need to find a solution for this, im gonna read more about MooseFS to find out (and probably test it) if it match the requirements.

Any hint is pretty much appreciated.

Regards
 
I think, you must try FreeBSD + ZFS + HAST (high-avalibitily storage) + CARP (common arp) for microseconds fallback & premium failover, or if you want, try DRBD + Heartbeat in Linux, but I'm not shure about it's stability.

Tried FZHC with 2 virtual machines in XEN environment, it worked just great. Also, no license issues for both servers (master & replica) because both of them run (share) same IP (CARP) and serve same clients.
 
HI roman,

I'd tryed DRBD but with some issue, now i've switched to GlusterFS that allow simoultaneus mount from remote servers and sync using heartbeat for failover management.

Regards and thanks for your hint
 
In this setup single point of failure is storage itself :) Or you must replicate storage over gluster or HAST.
 
Back
Top