High Availability SSH Server

[Mark Reynolds]

Hi,

I am trying to setup a highly available ssh server on Linux (probably
CentOS 5.4) , i.e. 2 nodes with a floating IP address, if one node
fails then the IP address will move over and sessions will continue
working. I am comfortable with using heartbeat and DRBD to to share
data and manage the floating IP address but wondered if there is any
way of having ssh share session / connection status across 2 nodes. I
may end up achieving this by running the ssh server in a VM (using
Linux and Xen) and migrating that instead but would rather avoid the
over heads.

I believe the current version of ssh from CentOS (and RedHat) is
openssh-4.3p2-36.el5.i386.rpm) but would be happy to compile a version
from source to meet my requirements.

I am aware that there is an appliance from ssh that can do this and
wondered if openssh was also capable of a high availability cluster
set up.

Any thoughts appreciated

Mark Reynolds

[Terry]

Post by Mark Reynolds
Hi,
I am trying to setup a highly available ssh server on Linux (probably
CentOS 5.4) , i.e. 2 nodes with a floating IP address, if one node
fails then the IP address will move over and sessions will continue
working. I am comfortable with using heartbeat and DRBD to to share
data and manage the floating IP address but wondered if there is any
way of having ssh share session / connection status across 2 nodes. I
may end up achieving this by running the ssh server in a VM (using
Linux and Xen) and migrating that instead but would rather avoid the
over heads.
I believe the current version of ssh from CentOS (and RedHat) is
openssh-4.3p2-36.el5.i386.rpm) but would be happy to compile a version
from source to meet my requirements.
I am aware that there is an appliance from ssh that can do this and
wondered if openssh was also capable of a high availability cluster
set up.
Any thoughts appreciated
Mark Reynolds

Linux clustering project should allow for the transition of an ssh
service between nodes using a single IP address. Your resources are
the IP and the SSH service. You may want to consider using a shared
disk and storing the SSH config and any keys centrally but not
required.

[Mark Reynolds]

Thank you for all your suggestions, I'll keep investigating.
Keepalived sounds particularly interesting.

Thanks again,

Mark

Keepalived?

Hi,
I am trying to s...

Linux clustering project should allow for the transition of an ssh
service between nodes using a single IP address.  Your resources are
the IP and the SSH service.  You may want to consider using a shared
disk and storing the SSH config and any keys centrally but not
required.