Fajar A. Nugraha wrote:
Glad to hear it :D Most of my changes are simple ones,
I've committed a number of the changes.
Yes. Which brings me to two other questions: 1) dhcp functionality is a good example that uses policy. It SHOULD be site specific, but currently all policies must be defined on policy.conf. So currently I must either: a. write the same block twice (as in your original sites-available/dhcp example. OR b. put all changes on policy.conf. OR c. Implement something like polcies directory (or, following the recent changes on modules, policies-available and policies-enabled).
The policies can go anywhere you want. If you're only running one DHCP server, you can put them in sites-enabled/dhcp
2) I need a place to store centralized configuration variable, where the same variable is used by policy, module, and sites. It can be: a. write the same value many times. Not good. b. edit radiusd.conf (or policy.conf) and place it there. Messy. c. create a new file included by radiusd.conf just before modules section. Works, but it still requires adding 1 line to radiusd.conf. d. Implement configs-available and configs-enabled directory, loaded just before modules section on radiusd.conf.
Yeah, that needs addressing in a general way for 3.0.
Current version of dhcp config code (with static IP, central config, and comment changes), diff against v2.1.x: https://github.com/fajarnugraha/freeradius-server/commit/fdff057
I've pulled in the changes && pushed updates back. The changes I made are much simpler than what you had. The extra policies you created were site-specific, and not suitable for general use. The MySQL example, and the dhcp_sqlippool module went in pretty much unchanged. The rest of the changes are much simpler, and should be easier to understand for most people. Please test them out.
As a side note, I noticed you closed pull request https://github.com/alandekok/freeradius-server/pull/40 , but it hasn't been merged in v2.1.x. Without that, debian package build still fails. Are you planning to commit an alternate fix? This is different btw from pull request #41, which you've applied (thanks for that).
It was pulled in, but not pushed back to github. I've done that now. Alan DeKok.