• If you are still using CentOS 7.9, it's time to convert to Alma 8 with the free centos2alma tool by Plesk or Plesk Migrator. Please let us know your experiences or concerns in this thread:
    CentOS2Alma discussion

2 linux server and LOAD BALANCING & HIGH AVAILABILITY

C

Chaneque

Guest
Hello, I have 2 servers with redhat Enterprise IS. I need have LOAD BALANCING & HIGH AVAILABILITYY in both servers and to handle plesk. Some idea, or where to begin to investigate.

thanks
 
Plesk Expand...

PSA itself is not built for clustering without alot of work. ART is the only person I know that has pulled it off, but then again, hes a creator of PSA.. Two server load balancing would actually require three servers. Search the forum.
 
Plesk does not imposes additional problems to a HA project, but it does not make the life easier. I recommend you to consider a shared disk project like DRBD as its extremely complicated to implement and does not involve large costs.
 
One interesting project perlboy pointed me to the other day is openssi.org, which abstracts away multiple physical servers into one virtual server. 2 systems probably wouldnt be enough for something like that (4 is probably a good start), but in terms of integration with psa, I set up an 8 node system running rh9/GFS and PSA installed just like a normal box. Nothing special required, I just installed psa like normal.
 
One thing I have been toying with is since I am virtualizing all my hosting servers to VMWare GSX server, I can share a physical disk between the servers per their clustering how-tos. That along with multiple interfaces that can talk over their own internal swich, can make for a nice HA cluster. Only thing I'd worry about is getting MySql to sync nicely, its something I haven't messed with yet. :confused:
 
yeah I've done a lot of clusters that way, where the physical servers each have a standalone OS and then you glue the whole thing together with LDAP, NFS and sqlrelay. Its always been a major hassle to set up and maintain. It is doable though, you've just got to run a really tight ship on the way each node is set up.
 
Back
Top