- Freeradius-Users - lists.freeradius.org

Freeradius3.0.3, Oracle database and UnixODBC
by Puzzel 06 Feb '15

06 Feb '15

Hi, I'm trying to integrate Freeradius and Oracle database. I'm using UnixODBC for this. My Oracle client and ISQL is working (connecting to oracle database) but when i try to start radius i have logs: rlm_sql_unixodbc: 01000 [unixODBC][Driver Manager]Can't open lib '/usr/lib/oracle/11.2/client/lib/libsqora.so.11.1' : ????: undefined symbol: get_vtable rlm_sql_unixodbc: SQL down 08003 [unixODBC][Driver Manager]Connnection does not exist rlm_sql_unixodbc: Can't allocate the statement rlm_sql_unixodbc: Socket destructor called, closing socket rlm_sql (oracle_eduroam): Opening connection failed (0) rlm_sql (oracle_eduroam): Removing connection pool This configuration works fine on FR2.

4 7

Multiple authentication for same user.
by Krishnan V R 06 Feb '15

06 Feb '15

Hi, I have freeradius and coova chilli(version 1.3.0) running in one machine and webserver running in a other machine. For the same user, Is it possible to give separate access for webserver and separate access for the internet. i.e I need a mechanism of two types of authentication for the same user. one authentication for the webserver access and the other authentication for the internet access. Thanks in advance for any suggestions. Thanks Krishnan

3 4

Re: What else should radmin do?
by Rui Ribeiro 05 Feb '15

05 Feb '15

Hi Alan, It would be interesting to make radmin work as non-root. At least in Debian 7 and 8, it has not work since 2.2.2 at least; I am using 3.0.7 from git atm. Despite the non privileged user being in the correct groups, and the socket being rw for the correct group, radmin blocks, and never gives back the prompt. . rt_sigaction(SIGPIPE, {SIG_IGN, [PIPE], SA_RESTORER|SA_RESTART, 0x7f0e60614180}, {SIG_DFL, [], 0}, 8) = 0 socket(PF_LOCAL, SOCK_STREAM, 0) = 3 connect(3, {sa_family=AF_LOCAL, sun_path="/var/run/freeradius/freeradius.sock"}, 37) = 0 read(3, "", 8) = 0 read(3, "", 8) = 0 read(3, "", 8) = 0 read(3, "", 8) = 0 read(3, "", 8) = 0 read(3, "", 8) = 0 read(3, "", 8) = 0 read(3, "", 8) = 0 read(3, "", 8) = 0 read(3, "", 8) = 0 read(3, "", 8) = 0 read(3, "", 8) = 0 read(3, "", 8) = 0 read(3, "", 8) = 0 radius2:~$ id -Gn rui rui adm sudo audio plugdev radius freerad radius2:/var/run/freeradius$ ls -la total 4 drwxr-xr-x 2 freerad freerad 80 Feb 3 20:14 . drwxr-xr-x 18 root root 640 Jan 27 17:49 .. -rw-r--r-- 1 freerad freerad 6 Feb 3 20:14 freeradius.pid srw-rw---- 1 freerad freerad 0 Feb 3 20:14 freeradius.sock -- Regards, -- Rui Ribeiro Senior Sysadm ISCTE-IUL https://www.linkedin.com/pub/rui-ribeiro/16/ab8/434

3 3

Custom attributes
by Flamur Rogova 04 Feb '15

04 Feb '15

Hi, I want to use custom attribute of integer type, to hold user's total amount of consumed bytes, something like this ATTRIBUTE Total-Used-Bytes 3003 integer What is bit width of custom integer attributes, 32 or 64 bit ? I need it to store 64 counters. Br/Flamur

2 1

What else should radmin do?
by Alan DeKok 04 Feb '15

04 Feb '15

I’ve pushed some changes to radmin, which will be in the 3.0.7 release. The changes make radmin work a little better, in that it now makes a distinction between STDOUT and STDERR. Also, when a command fails, radmin will exit with an error: $ radmin -e “help” … $ echo $? 0 Versus: $ radmin -e “no such command” … $ echo $? 1 These changes were made possible by updating the underlying framework used by radmin and the server. The result of these framework changes is that we can now exchange *anything* safely between the server and radmin. So… what would be useful to exchange? We can’t do configuration changes, as the server can’t write to /etc/raddb. But we could do test packets, for example. $ radmin “send packet X pretending to be from NAS 192.0.2.1” And get a response. This means it would be possible to perform better tests on live systems. Any other ideas? Alan DeKok.

3 6

Re: Disconnect message!
by Kos 04 Feb '15

04 Feb '15

I found this awesome Java API(https://github.com/GideonLeGrange/mikrotik-java) and make possible to disconnect the user through. Thanks all for your answers.

1 0

Expiration Feature / Session-Timeout Problem
by Sven 04 Feb '15

04 Feb '15

Hello, i try to use the Expiration Feature to get control over the Password lifetime of our Users. The idea was to set a date maybe 90 Day from last Passwordchange. This worked well, Users got denied but ive tested this feature only with a few Days ahead from now. Now i found out that if i set this to more than a Month from now that some of our Devices rejects the Access while i can see in the Radius Log that they where supplied with a Access-Accept. I guess that some of our older Devices got Problems with the releated Session-Timeout value which is in the Access-Accept reply too. Is there a way to not send Session-Timeout Data to these or all Devices ? In my setup i dont think that i need this because my goal is not to Disconnect a User, only to block him from logging in. Maybe there is a better solution to get this "feature" :) Im not sure if the Expiration Feature is the right thing for this task. Kind Regards, Sven

3 3

Stored Procedure
by Robert Graham 03 Feb '15

03 Feb '15

How do you execute a stored proc against a sql sever to authenticate a user. What do you do with the response to either accept or reject the user? We are trying to authenticate the user by returning a 1 for yes or 0 for no. The stored proc is executed by the update control but the User Password is not being added, its only adding the User-name and leaving the Userpassword field blank. Tmp-String-0 := "%{sql:EXEC [IsUserAuthenticated] ('%{User-Name}','%{User-Password}'" } if (Tmp-String-0 =~ /1/) update control { &Auth-Type := Accept }

2 1

freeradius memory leak after use detail file
by Yanrui Hu 03 Feb '15

03 Feb '15

free radius 2.2.6 it works well on production system and one day we change to not write accouting data to database directly, instead, we write it to detail disk file and sync to db latter. after this change we found the memory consumption increase day by day. with valgrind I test it in lab environment with same configuration. here is the output: anybody can help if this is leak and if newest 3.2.x already fixed it? valgrind --leak-check=full --time-stamp=yes --show-leak-kinds=all --log-file=/tmp/valgrind.log /usr/sbin/freeradius ==00:00:00:07.586 59730== LEAK SUMMARY: ==00:00:00:07.586 59730== definitely lost: 125 bytes in 3 blocks ==00:00:00:07.586 59730== indirectly lost: 480 bytes in 20 blocks ==00:00:00:07.586 59730== possibly lost: 0 bytes in 0 blocks ==00:00:00:07.586 59730== still reachable: 6,173,476 bytes in 70,774 blocks ==00:00:00:07.586 59730== suppressed: 0 bytes in 0 blocks ==00:00:00:07.586 59730== ==00:00:00:07.586 59730== For counts of detected and suppressed errors, rerun with: -v ==00:00:00:07.586 59730== Use --track-origins=yes to see where uninitialised values come from ==00:00:00:07.586 59730== ERROR SUMMARY: 41122 errors from 100 contexts (suppressed: 0 from 0) ==00:00:03:34.727 59824== Thread 6: ==00:00:00:07.569 59730== 160 bytes in 4 blocks are still reachable in loss record 891 of 1,273 ==00:00:00:07.569 59730== at 0x4C2CFA7: malloc (vg_replace_malloc.c:296) ==00:00:00:07.569 59730== by 0x4E48CA5: rbtree_insertnode (rbtree.c:270) ==00:00:00:07.569 59730== by 0x4E48EB8: rbtree_insert (rbtree.c:299) ==00:00:00:07.569 59730== by 0x40D829: cf_item_add (conffile.c:531) ==00:00:00:07.569 59730== by 0x40D829: cf_item_add (conffile.c:507) ==00:00:00:07.569 59730== by 0x4104AA: cf_item_parse (conffile.c:1085) ==00:00:00:07.569 59730== by 0x410890: cf_section_parse (conffile.c:1197) ==00:00:00:07.569 59730== by 0xF1DE5DE: detail_instantiate (rlm_detail.c:132) ==00:00:00:07.569 59730== by 0x41D38B: find_module_instance (modules.c:620) ==00:00:00:07.569 59730== by 0x41FC63: do_compile_modsingle (modcall.c:1966) ==00:00:00:07.569 59730== by 0x41C69F: load_component_section (modules.c:908) ==00:00:00:07.569 59730== by 0x41CBE1: load_byserver (modules.c:1109) ==00:00:00:07.569 59730== by 0x41DA6C: virtual_servers_load (modules.c:1273) ==00:00:00:07.564 59730== 56 bytes in 1 blocks are still reachable in loss record 792 of 1,273 ==00:00:00:07.564 59730== at 0x4C2CFA7: malloc (vg_replace_malloc.c:296) ==00:00:00:07.564 59730== by 0x424B18: rad_malloc (util.c:347) ==00:00:00:07.564 59730== by 0xF1DE577: detail_instantiate (rlm_detail.c:126) ==00:00:00:07.564 59730== by 0x41D38B: find_module_instance (modules.c:620) ==00:00:00:07.564 59730== by 0x41FC63: do_compile_modsingle (modcall.c:1966) ==00:00:00:07.564 59730== by 0x41C69F: load_component_section (modules.c:908) ==00:00:00:07.564 59730== by 0x41CBE1: load_byserver (modules.c:1109) ==00:00:00:07.564 59730== by 0x41DB14: virtual_servers_load (modules.c:1244) ==00:00:00:07.564 59730== by 0x41E0B7: setup_modules (modules.c:1568) ==00:00:00:07.564 59730== by 0x4097C2: main (radiusd.c:289) -- Best Regards, Yanrui Hu -- *This email may contain or reference confidential information and is intended only for the individual to whom it is addressed. Please refrain from distributing, disclosing or copying this email and the information contained within unless you are the intended recipient. If you received this email in error, please notify us at legal(a)appannie.com <legal(a)appannie.com>** immediately and remove it from your system.*

3 2

Is it possible to have different TLS configurations per EAP type?
by Louis Munro 02 Feb '15

02 Feb '15

Hello, I have a set of FR 2.2.5 servers which are currently handling PEAP connections. We now want to add EAP-TLS authentication, which is pretty straightforward. The only stumbling block I have is that the EAP-TLS client certs are signed by a root CA different from the one that signed the server certificate currently used for PEAP authentication. While that is not a problem in itself it makes client devices configuration more troublesome than I would like because it forces the user to install two root CA certs. So I am trying to write a configuration that would support two eap configurations. Instantiating two modules is not the problem, it’s more a matter of finding how to select a configuration to be used while maintaining backwards compatibility for the existing PEAP clients. Assuming the following modules: eap { […] tls { certdir = ${confdir}/certs cadir = ${confdir}/certs private_key_file =${certdir}/server.key certificate_file = ${certdir}/server.crt CA_file = ${certdir}/ca_PEAP.crt dh_file = ${certdir}/dh random_file = /dev/urandom cipher_list = "DEFAULT" make_cert_command = "${certdir}/bootstrap” […] } eap eap-tls { [..] tls { certdir = ${confdir}/certs cadir = ${confdir}/certs private_key_file = ${certdir}/ca.key private_key_password = whatever certificate_file = ${certdir}/ca.pem dh_file = ${certdir}/dh random_file = /dev/urandom cipher_list = "DEFAULT" make_cert_command = "${certdir}/bootstrap" CA_file = ${certdir}/ca.pem […] } This does not seem to work: authorize { # assuming an instance called eap-tls eap-tls { updated = return } # else eap { ok = return } […] } And since things like EAP-Type are detected by the eap module itself I also can’t seem to do if (EAP-Type && (EAP-Type == EAP-TLS)) { eap-tls { ok = return } } else { eap { ok = return } } Which leaves me with a chicken and egg type of situation. Is this doable in FR 2? Am I just missing the obvious? Thank you for your help. -- Louis Munro lmunro(a)inverse.ca :: www.inverse.ca +1.514.447.4918 x125 :: +1 (866) 353-6153 x125 Inverse inc. :: Leaders behind SOGo (www.sogo.nu) and PacketFence (www.packetfence.org)

3 3