If you have two clusters, each on a private network, with master
nodes on both the private networks, and the public network(master nodes A and B), your floating license can be used as follows.
license server runs on master node A, and serves clients on private
network A.
license server runs on master node B, and serves clients on private
network B.
license server runs on public server C, and any public network
device, like master nodes A and B, can share the seats in the license.
So only the master nodes compile (along with any other public network
platform) and the slaves only run the executables.
See FAQ | PGI
for how one platform can borrow a seat for up to a week at a time.
The platform must be able to be on the network with the server before
being taken off the network to use the borrowed seat.
The cluster licenses typically are used on the public network, while
the executables created run on the private network clusters.
You borrow a license as a client. You run ‘lmborrow’ to talk to the
license service, in order to borrow a seat. Once successful, you can detach
from the network, and continue using the compilers for up to a week,
when the seat goes away and returns to the license service.
After you borrow, the rest of the users now share one less seat, until the
seat returns (by either borrow-timeout or the user returning the seat early).
What you cannot do is create a borrowed ‘license’ that can be sent to another
user somewhere to use on his machine.
Suppose I have a laptop A. The license is on Machine B.
So I ssh from laptop A to machine B and do
lmborrow pgroupd 21-may-2015
And then I disconnect from machine B and laptop A should have the license for that day.
But shouldn’t laptop A have some software on it that understands that license has been borrowed and to use that license. I mean how does laptop A understand that a license has been made available.