cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

how to get the socket information from servers via FNMS Agent

Dears,

i would like to know an insight on how we ware managing socket based licenses (eg RHEL Server ) 

As we the FNMS agent have the capability to know read the socket count ?

or do we need to opt for override option ?

 

Regards,

Junaid Vengadan

(1) Solution

@emtmeta - For the purposes of RHEL license calculations, the FlexNet Agent obtains the information that is required.  If you review the document found here:

https://www.redhat.com/cms/managed-files/li-rhel-subscription-guide-f13117wg-201807-en_0_0.pdf

Note the following definition.  Since RHEL is licensed by socket-pair, and a socket-pair is defined as 2 "occupied" socket, this implies that the license calculation is based on the physical processors that are installed in the existing sockets.  In other words, if you have empty sockets they are not included in the calculations.  Another way to think if this is that you only need to license your "used" sockets.

GLOSSARY
Guest: An instance of the software running in a virtual machine, which in turn is running on a hypervisor. In the Red Hat subscription model, a guest is associated with a physical system.
Physical node: A physical system on which you install or execute all or a portion of the software,
including, without limitation, a server, workstation, laptop, blade, or other physical system,
as applicable.
Socket: A central processing unit (CPU) socket on a motherboard.
Socket-pair: Up to two sockets where each is occupied by a CPU on a system. Two servers with a
single occupied socket on each must be entitled separately; therefore, you would purchase two subscriptions—one for each server.

Here is a discussion on the Red Hat community on the topic of Sockets vs. CPU

https://access.redhat.com/discussions/480953

 

Here is a follow up Community Post regarding how you can model your RHEL Licenses using FlexNet Manager:

https://community.flexera.com/t5/FlexNet-Manager-Knowledge-Base/Configuring-Red-Hat-Licenses-in-FlexNet-Manager/ta-p/5245

 

View solution in original post

(3) Replies

@emtmeta - For the purposes of RHEL license calculations, the FlexNet Agent obtains the information that is required.  If you review the document found here:

https://www.redhat.com/cms/managed-files/li-rhel-subscription-guide-f13117wg-201807-en_0_0.pdf

Note the following definition.  Since RHEL is licensed by socket-pair, and a socket-pair is defined as 2 "occupied" socket, this implies that the license calculation is based on the physical processors that are installed in the existing sockets.  In other words, if you have empty sockets they are not included in the calculations.  Another way to think if this is that you only need to license your "used" sockets.

GLOSSARY
Guest: An instance of the software running in a virtual machine, which in turn is running on a hypervisor. In the Red Hat subscription model, a guest is associated with a physical system.
Physical node: A physical system on which you install or execute all or a portion of the software,
including, without limitation, a server, workstation, laptop, blade, or other physical system,
as applicable.
Socket: A central processing unit (CPU) socket on a motherboard.
Socket-pair: Up to two sockets where each is occupied by a CPU on a system. Two servers with a
single occupied socket on each must be entitled separately; therefore, you would purchase two subscriptions—one for each server.

Here is a discussion on the Red Hat community on the topic of Sockets vs. CPU

https://access.redhat.com/discussions/480953

 

Here is a follow up Community Post regarding how you can model your RHEL Licenses using FlexNet Manager:

https://community.flexera.com/t5/FlexNet-Manager-Knowledge-Base/Configuring-Red-Hat-Licenses-in-FlexNet-Manager/ta-p/5245

 

@kclausen

Thanks a lot for the feedback 🙂

Hi,

Regarding one of your original questions:  No, the FNMS agent cannot gather socket data. In fact, most inventory solutions cannot detect sockets.

Technically an empty socket is like dead end of a cable. You can measure if there's a device at the end of the cable. But you don't really see if there's nothing.

If you still need socket data in the future, there are generally two typical approaches:

  1. Work under the assumption that all sockets are populated and simply copy the number of CPUs to sockets.
  2. Find a external source providing extra hardware details and enrich your system. Data Platform could be helpful with this.

Best regards,

Markward