Multiple Radius Instances on the same machine
Hello all, I am interested in running more than one configuration including different authentification methods and different clients on one machine. I have Freeradius 1.1.7 in production, but I installed and tested Freeradius 2.0.5 because of the virtual servers. I want to run three separated radius instances on one radius server, for example: 1. instance with pam auth for internal clients (means clients list x) 2. instance with eap auth for external clients (means clients list y) 3. instance with pam auth for other external clients (means clients list z) The issue ist that the configurations of the virtual servers are separated but if one error occures in one instance configuration, the whole radius server can't be started. For example when I have a typing error in the config of the virtual server under sites-available like writing "ks pam" instead of "pam" then I can't start the whole server and none of the instances goes up. The problem is that I have many internal and external clients and the configs will be changed often from different people. I don't want that an error in the config for the external clients bothers the instance for internal clients. Furthermore I want to assign some resources to the different instances like max 5 server processes for instance "2" and max 15 server process for instance "1" and "3". But so far I understood it, this section is not in the config of the virtual servers but for the whole server. Therefore I have a few questions - can I run virtual servers without an error in one instance messing the whole server and the rest instances - is there a possibility to assign unequal resources to the separated instances like server processes, threads or number of requests. The goal is to limit the resources for external requests and give more resources to internal requests. - is possible to run several radius processes on one machine (not virtual servers) with different logs and configs. And how should I tell radiusd1 to use the configuration radusd1.conf and radiusd2 to use radiusd2.conf? Best regards, Irina
HRZ Konten wrote:
I am interested in running more than one configuration including different authentification methods and different clients on one machine.
That's possible.
The issue ist that the configurations of the virtual servers are separated but if one error occures in one instance configuration, the whole radius server can't be started.
$ man radiusd Read the description of the first option.
Furthermore I want to assign some resources to the different instances like max 5 server processes for instance "2" and max 15 server process for instance "1" and "3". But so far I understood it, this section is not in the config of the virtual servers but for the whole server.
Because it makes more sense there. This lets the server automatically balance load across threads... independent of which virtual server is being used. If you want to limit the number of server threads, run multiple configurations, and multiple server processes.
Therefore I have a few questions - can I run virtual servers without an error in one instance messing the whole server and the rest instances
Yes. You check the configuration before starting or re-starting the server. Don't re-start the server if the configuration is wrong
- is there a possibility to assign unequal resources to the separated instances like server processes, threads or number of requests. The goal is to limit the resources for external requests and give more resources to internal requests.
No. If resource limits are a problem, you should buy another machine. Realistically, resource limits won't be a problem on normal hardware.
- is possible to run several radius processes on one machine (not virtual servers) with different logs and configs. And how should I tell radiusd1 to use the configuration radusd1.conf and radiusd2 to use radiusd2.conf?
Yes. See "man radiusd". Look at the "-n" parameter. You can do *exactly* that. Alan DeKok.
participants (2)
-
Alan DeKok -
HRZ Konten