- Freeradius-Users - lists.freeradius.org

preventing multiple authentication attempts for up to 2 minutes.
by Jim Whitescarver 18 Mar '16

18 Mar '16

I am using nlm_python plugin to trigger an out-of-band authentication and get a result typically completed in about 20 seconds but should not time out for two minutes. When I run radtest and a duplicate request comes in I get (0) Ignoring duplicate packet from client localhost port 32940 - ID: 2 due to unfinished request in component authenticate module python in the log. However, in an actual login attempt from a 3rd party system configured to use this radius instance I often see a second authentication attempt almost immediately started while the first is still in progress. How can I prevent this? I set idle timeouts in clients.conf and sites-available/default to 120. What else is needed? Thanks, Jim

3 4

understanding the process of setting up eap-tls server/client certs
by Michael Martinez 18 Mar '16

18 Mar '16

I'm working on setting up EAP-TLS so that the client (iPad) can be issued a client cert and use it to authenticate with Radius. I need some clarity on the process, particularly the roles of some of the different files generated and how to use them. 1. in order to generate the root ca, first I edit ca.cnf. It's straightforward except I don't understand the role of the "input" password. The "output" password I understand is for the private key - ca.key. 1.a. after editing ca.cnf, then i run make ca.pem. This uses openssl to run req to generate a self-signed root ca. Four files are generated: -- --- Michael Martinez http://www.michael--martinez.com

2 3

panic action in samba
by Alex Sharaz 18 Mar '16

18 Mar '16

Hi, I've been phasing in FR 3.0.11 on an Ubuntu 14.04 system with the default version of samba (4.1.6) as one of our outward facing (ORPS) servers. The intention is to migrate all of our 2.2.9 systems over to 3.0.11. Up till now this servers main task has been to proxy eduroam visitor auths to home institutions, but we do have a couple of monitoring systems performing EAP based health checks against the server which authenticates against our AD service using Samba. All this has been working fine. Yesterday, after running a few more auth tests from inside and outside our network using eapol_test, I reconfigured our inbound eduroam traffic (for York users visiting external sites) to use this server instead of one of our other 2.2.9 systems. While most auths "just worked" what we also saw were a flurry of The Samba 'panic action' script, /usr/share/samba/panic-action, was called for PID 22827 (/usr/sbin/winbindd). with corresponding auth failure messages within FR of the form Thu Mar 17 09:03:40 2016 : Auth: (66912) Login OK: [gw791(a)york.ac.uk] (from client roaming2.ja.net port 13 cli 54-E4-3A-1F-C7-8A via TLS tunnel) Thu Mar 17 09:03:41 2016 : Auth: (66916) Login OK: [gw791(a)york.ac.uk] (from client roaming2.ja.net port 13 cli 54-E4-3A-1F-C7-8A) Thu Mar 17 09:03:41 2016 : ERROR: (66920) mschap: ERROR: Program returned code (1) and output 'Account disabled (0xc0000072)' Thu Mar 17 09:03:41 2016 : Auth: (66920) Login incorrect (mschap: No NT-Domain was found in the User-Name): [erf504(a)york.ac.uk] (from client roaming2.ja.net port 13 cli 00-26-B0-04-2B-8A via TLS tunnel) Thu Mar 17 09:03:42 2016 : Info: (66923) eap_peap: The users session was previously rejected: returning reject (again.) Thu Mar 17 09:03:42 2016 : Info: (66923) eap_peap: This means you need to read the PREVIOUS messages in the debug output Thu Mar 17 09:03:42 2016 : Info: (66923) eap_peap: to find out the reason why the user was rejected Thu Mar 17 09:03:42 2016 : Info: (66923) eap_peap: Look for "reject" or "fail". Those earlier messages will tell you Thu Mar 17 09:03:42 2016 : Info: (66923) eap_peap: what went wrong, and how to fix the problem Thu Mar 17 09:03:42 2016 : Auth: (66923) Login incorrect (eap: Failed continuing EAP PEAP (25) session. EAP sub-module failed): [ erf504(a)york.ac.uk] (from client roaming2.ja.net port 13 cli 00-26-B0-04-2B-8A) Thu Mar 17 09:03:43 2016 : ERROR: (66928) mschap: ERROR: Program returned code (1) and output 'Logon failure (0xc000006d)' Thu Mar 17 09:03:43 2016 : Auth: (66928) Login incorrect (mschap: No NT-Domain was found in the User-Name): [clh553(a)york.ac.uk] (from client roaming2.ja.net port 13 cli A8-5B-78-74-D8-C9 via TLS tunnel) As shown above, we do have successful york realm auths happening as well. Before I start looking at this in more detail (additional samba logging) I was wondering if anyone has seen these (samba) messages before. Haven't touched the samba config, its whatever the default settings are. Rgds Alex

3 3

Force PAP to use unix password instead of windows
by Angel L. Mateo 18 Mar '16

18 Mar '16

Hello, I have freeradius authenticating users with a ldap backend. In my ldap configuration I have: update { control:Password-With-Header += 'userPassword' control:NT-Password := 'sambaNTPassword' control:LM-Password := 'sambaLMPassword' ... } in the site configuration I have: authorize { files ldap pap } authenticate { Auth-Type PAP { pap } } With this configuration, I think that all authentications are done using the password in windows format (the one I have in sambaNTPassword and sambaLMPassword ldap attributes) instead of the "standard" unix password stored in the userPassword attribute. Is there any way to force the use the password in the userPassword attribute instead of NT/LM one? -- Angel L. Mateo Martínez Sección de Telemática Área de Tecnologías de la Información y las Comunicaciones Aplicadas (ATICA) http://www.um.es/atica Tfo: 868887590 Fax: 868888337

3 5

debug User-Profile
by MichaelLeung 18 Mar '16

18 Mar '16

hi list my freeradius version is 3.0.4 i have enabled ldap modules and the radius profile feature of it . and i need to check the user is in the speacific Ldap-Group, and assign the User-Profile which contain all radius Reply-Items in it . so when my NAS try to authenticate , i can only see radius -X responding : (0) Sending Access-Accept packet to host 10.1.1.13 port 1812, id=96, length=0 (0) User-Profile := 'cn=Device_Superior,ou=Admin,ou=Group,dc=gd,dc=abc,dc=com' it was not going to print out what reply item the User-Profile contained. and actually, i define the reply item as Huawei-Exec-Privilege := "15" it will give the highest admin right to the user belong to Group Device_Superior to Operate the Device . how can i debug the User-Profile?

1 4

Re: understanding the process of setting up eap-tls server/client certs
by Michael Martinez 17 Mar '16

17 Mar '16

I'm working on setting up EAP-TLS so that the client (iPad) can be issued a client cert and use it to authenticate with Radius. I need some clarity on the process, particularly the roles of some of the different files generated and how to use them. 1. in order to generate the root ca, first I edit ca.cnf. It's straightforward except I don't understand the role of the "input" password. The "output" password I understand is for the private key - ca.key. 1.a. after editing ca.cnf, then i run make ca.pem. This uses openssl to run req to generate a self-signed root ca. Four files are generated: * index: it's empty. I don't know what it's for * serial. *.ca.pem: the root ca * ca.key: the private key for the root ca 1.b. in mods-enabled/eap, designate the location of the ca.pem file in the "ca_file = " field 2. make the server certs. 2.a. edit server.cnf. Again, I don't know what the "input" password is for. The output password I assume is the password for the private key. 2.b. make server.pem generates a bunch of files. I'm assuming the ones we need are server.pem and server.key 2.c. in mods-enabled/eap do the following: * private key password = <the "output password I set in server.cnf"> * "private key file = " the location of the newly generated server.key * "certificate file = " <the location of the newly generated server.pem> 2.d. questions about the files generated by "make server.pem": * what's server.crt? what's server.pem? when do I use one versus the other? * 01.pem is identical to server.crt. why and where would I use 01.pem instead of server.crt? 3. Now make a client cert signed by the server cert. 3.a. edit client.cnf. Set "input" and "output" passwords to something unique to the client? Or does one or both of these need to match the password from the server cert?? 3.b. make client.pem. This generates a bunch of files I'm assuming the one we need to install the client is client.key (and client.pem?) 4. Now for the client. The README says we need to install ca.crt on the client. I assume also the client.key needs to be put on the client. 5. I'm using jradius to try to test this. In the "Keys" tab: * Client Certificate file: I need to put client.key here * Client Certificate password: the password that the private key was encrypted with * Root CA chain file: ca.crt * Root CA chain password: the "output password" from ca.cnf I'm sure I am bungling some of this. Any help is appreciated. -- --- Michael Martinez http://www.michael--martinez.com

1 0

RE: TTLS Issue after Ubuntu upgrade
by Arjan Sinnige 16 Mar '16

16 Mar '16

Alan, Just getting back on this as I was buzy last week. I was using the download from : ftp://ftp.freeradius.org/pub/freeradius/freeradius-server-2.2.9.tar.gz >> As openSSL version was changed I had to rebuild freeradius from source again. >> I stumbled into 2 issues during the build : >> - I had to disable ikev2 in debian/rules with --without-rlm_eap_ikev2 because it would not finish the build with it. >> - I had to disable iodbc in debian/rules with --without-rlm_sql_iodbc because it would not finish the build with it. >> As I am running LDAP as backend the second I could care less about but the first gave me a few frowns. > > That's weird. What were the build errors? > I just re-ran the build to give me the errors again. So for reference, here they are. I will be building from the version from github to see if all errors disappear. (INCL the length=5 error) === configuring in ./types/rlm_eap_ikev2 (/home/sinnige/freeradius2/freeradius-server-2.2.9/src/modules/rlm_eap/./types/rlm_eap_ikev2) configure: running /bin/bash ./configure '--prefix=/usr' '--disable-option-checking' '--build' 'x86_64-linux-gnu' '--exec-prefix=/usr' '--mandir=/usr/share/man' '--sysconfdir=/etc' '--libdir=/usr/lib/freeradius' '--datadir=/usr/share' '--localstatedir=/var' '--with-raddbdir=/etc/freeradius' '--with-logdir=/var/log/freeradius' '--enable-ltdl-install=no' '--enable-strict-dependencies' '--with-large-files' '--with-udpfromto' '--with-edir' '--enable-developer' '--without-rlm_eap_tnc' '--with-rlm_sql_postgresql_lib_dir=/usr/lib/x86_64-linux-gnu' '--with-rlm_sql_postgresql_include_dir=/usr/include/postgresql' '--without-rlm_sql_oracle' '--without-rlm_sql_unixodbc' '--with-system-libtool' '--with-system-libltdl' 'build_alias=x86_64-linux-gnu' --cache-file=../../../../../config.cache --srcdir=. configure: loading cache ../../../../../config.cache checking for gcc... (cached) gcc checking for C compiler default output file name... a.out checking whether the C compiler works... yes checking whether we are cross compiling... no checking for suffix of executables... checking for suffix of object files... (cached) o checking whether we are using the GNU C compiler... (cached) yes checking whether gcc accepts -g... (cached) yes checking for gcc option to accept ISO C89... (cached) none needed checking how to run the C preprocessor... (cached) gcc -E checking for grep that handles long lines and -e... (cached) /bin/grep checking for egrep... (cached) /bin/grep -E checking for ANSI C header files... (cached) yes checking for sys/types.h... (cached) yes checking for sys/stat.h... (cached) yes checking for stdlib.h... (cached) yes checking for string.h... (cached) yes checking for memory.h... (cached) yes checking for strings.h... (cached) yes checking for inttypes.h... (cached) yes checking for stdint.h... (cached) yes checking for unistd.h... (cached) yes checking for malloc.h... (cached) yes checking for ikev2_set_log_callback in -leap-ikev2... no checking EAPIKEv2/connector.h usability... no checking EAPIKEv2/connector.h presence... no checking for EAPIKEv2/connector.h... no configure: error: set --without-rlm_eap_ikev2 to disable it explicitly. configure: error: ./configure failed for ./types/rlm_eap_ikev2 configure: error: ./configure failed for src/modules/rlm_eap debian/rules:55: recipe for target 'config.status' failed make: *** [config.status] Error 1 dpkg-buildpackage: error: debian/rules build gave error exit status 2 === configuring in ./drivers/rlm_sql_iodbc (/home/sinnige/freeradius2/freeradius-server-2.2.9/src/modules/rlm_sql/./drivers/rlm_sql_iodbc) configure: running /bin/bash ./configure '--prefix=/usr' '--disable-option-checking' '--build' 'x86_64-linux-gnu' '--exec-prefix=/usr' '--mandir=/usr/share/man' '--sysconfdir=/etc' '--libdir=/usr/lib/freeradius' '--datadir=/usr/share' '--localstatedir=/var' '--with-raddbdir=/etc/freeradius' '--with-logdir=/var/log/freeradius' '--enable-ltdl-install=no' '--enable-strict-dependencies' '--with-large-files' '--with-udpfromto' '--with-edir' '--enable-developer' '--without-rlm_eap_tnc' '--without-rlm_eap_ikev2' '--with-rlm_sql_postgresql_lib_dir=/usr/lib/x86_64-linux-gnu' '--with-rlm_sql_postgresql_include_dir=/usr/include/postgresql' '--without-rlm_sql_oracle' '--without-rlm_sql_unixodbc' '--with-system-libtool' '--with-system-libltdl' 'build_alias=x86_64-linux-gnu' --cache-file=../../../../../config.cache --srcdir=. configure: loading cache ../../../../../config.cache checking for gcc... (cached) gcc checking for C compiler default output file name... a.out checking whether the C compiler works... yes checking whether we are cross compiling... no checking for suffix of executables... checking for suffix of object files... (cached) o checking whether we are using the GNU C compiler... (cached) yes checking whether gcc accepts -g... (cached) yes checking for gcc option to accept ISO C89... (cached) none needed checking for SQLConnect in -liodbc... yes checking for isql.h... no configure: error: set --without-rlm_sql_iodbc to disable it explicitly. configure: error: ./configure failed for ./drivers/rlm_sql_iodbc configure: error: ./configure failed for src/modules/rlm_sql debian/rules:55: recipe for target 'config.status' failed make: *** [config.status] Error 1 dpkg-buildpackage: error: debian/rules build gave error exit status 2

1 0

rlm_ldap issue
by Alex Moen 15 Mar '16

15 Mar '16

Hi all, I am building a new radius server on Centos7 and configuring for LDAP auth. I have an older version running on a different server, and have brought over my configs and adjusted to the new version requirements (for instance, adding the ipaddr attribute to all the client stanzas). However, I seem to be having an issue with the rlm_ldap module, and the ldap config itself. BTW, ran all updates this morning, so everything should be at the latest versions (from the repositories, anyway). FreeRADIUS Version 3.0.4 is what is now installed. So, here's my ldap file from the old server (that works properly): --------------------------------------------- ldap rg600-1 { server = "10.255.255.40" identity = "cn=admin,o=ourorg" password = r!da2Lp basedn = "ou=administration,o=ourorg" filter = "(uid=%u)" ldap_connections_number = 5 timeout = 4 timelimit = 3 net_timeout = 1 tls { start_tls = no } access_attr = "uid" dictionary_mapping = ${confdir}/ldap.attrmap edir_account_policy_check = no groupname_attribute = gidnumber groupmembership_attribute = gidNumber } --------------------------------------------- I copied this file exactly to the new server. To test connectivity, I ran (on the new server): --------------------------------------------- ldapsearch -x -W -D "cn=admin,o=ourorg" -b "ou=administration,o=ourorg" objectclass=* --------------------------------------------- and, when asked, provided the r!da2Lp password, and successfully got the list of all objects... indicating that the connectivity, username, and password are perfectly correct. When I run radiusd -X -d /etc/raddb, I get the following end result of debug lines: --------------------------------------------- ... rlm_ldap (rg600-1): Opening additional connection (0) rlm_ldap (rg600-1): Connecting to 10.255.255.40:389 rlm_ldap (rg600-1): Waiting for bind result... rlm_ldap (rg600-1): Bind credentials incorrect: Invalid credentials rlm_ldap (rg600-1): Opening connection failed (0) rlm_ldap (rg600-1): Removing connection pool --------------------------------------------- indicating, I believe, a bad password or bad user ID. I have a feeling that the "!" in the password is causing issues. I have surrounded the password with single quotes, double quotes, ticks, and have escaped the ! with \, all to no avail. I cannot really change the password, as so many systems are using it that it is prohibitive. Anyone have any ideas? I'll supply any other required data... Thanks! Alex -- Alex Moen NSTII North Dakota Telephone Company 701-662-6481

4 8

Modules - dictionary
by Oscar Jofre 15 Mar '16

15 Mar '16

Hi all, I'm trying to understand how to configure radius Time Limit, Data download Limit, .. On a FreeRadius 3.0.10 using mysql and Mikrotik. First of all I wold like to know about to configure mikrotik dictionary (http://wiki.mikrotik.com/wiki/Manual:RADIUS_Client/vendor_dictionary) What I should do create ${prefix}/share/freeradius/dictionary/mikrotik.dictionary and add INCLUDE into raddb/dictionary ? or just add all lines into /etc/raddb/dictionary ? Where I should use the dictionary attributes on modules ? Then on mods-available there is sqlcounter with some configurations. Is there any other examples somewhere ? I need data limitation. And is it possible to create some group of users with diferents restrictions ? Thanks.

3 15

Re: Freeradius systemd radiusd.service fails
by Ibrahim Almahfooz 15 Mar '16

15 Mar '16

> What does "systemctl status radiusd.service" and "journalctl -xe" say? >It should tell you what it thinks is wrong. > >One possibility is that oracle depends on $ORACLE_HOME env variable, >which is present when you log in as root/normal user, but not >available when started by systemd. If that's the case, use Environment >or EnvironmentFile inside the systemd unit. Look at your existing >systemd unit files (e.g. ssh.service) for example. I really appreciate you help Fajar. It is solved by setting the enviroment variable on the systemd radiusd.service Thank you very much! On 15 March 2016 at 14:00, <freeradius-users-request(a)lists.freeradius.org> wrote: > Send Freeradius-Users mailing list submissions to > freeradius-users(a)lists.freeradius.org > > To subscribe or unsubscribe via the World Wide Web, visit > http://lists.freeradius.org/mailman/listinfo/freeradius-users > or, via email, send a message with subject or body 'help' to > freeradius-users-request(a)lists.freeradius.org > > You can reach the person managing the list at > freeradius-users-owner(a)lists.freeradius.org > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Freeradius-Users digest..." > > > Today's Topics: > > 1. Re: Freeradius systemd radiusd.service fails (Fajar A. Nugraha) > 2. > > Re: Freeradius systemd radiusd.service fails (Nathan Ward) > 3. Re: Freeradius systemd radiusd.service fails (Alan Buxey) > 4. Network Design - FreeRadius (Richard J Palmer) > 5. Re: Network Design - FreeRadius (Arran Cudbard-Bell) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Tue, 15 Mar 2016 15:44:04 +0700 > From: "Fajar A. Nugraha" <list(a)fajar.net> > To: FreeRadius users mailing list > <freeradius-users(a)lists.freeradius.org> > Subject: Re: Freeradius systemd radiusd.service fails > Message-ID: > <CAG1y0scxv33jrdkpdZPmDpJT++XPpww7Je= > e734Ch1tH8-r-fQ(a)mail.gmail.com> > Content-Type: text/plain; charset=UTF-8 > > On Tue, Mar 15, 2016 at 2:27 PM, Ibrahim Almahfooz > <ibrahim.nezar(a)gorannet.net> wrote: > > Hi everyone, > > > > We currently using freeradius version 3.0.11 and it is compiled to work > > with oracle db and everything is fine and smooth when running the service > > using radiusd -X or radiusd. > > > > Once we start the service using systemctl start radiusd.service or > > systemctl restart radiusd.service then we receive the below error: > > > > "Job for radiusd.service failed because the control process exited with > > error code. See "systemctl status radiusd.service" and "journalctl -xe" > for > > details." > > > > What does "systemctl status radiusd.service" and "journalctl -xe" say? > It should tell you what it thinks is wrong. > > One possibility is that oracle depends on $ORACLE_HOME env variable, > which is present when you log in as root/normal user, but not > available when started by systemd. If that's the case, use Environment > or EnvironmentFile inside the systemd unit. Look at your existing > systemd unit files (e.g. ssh.service) for example. > > -- > Fajar > > > ------------------------------ > > Message: 2 > Date: Tue, 15 Mar 2016 22:07:16 +1300 > From: Nathan Ward <lists+freeradius(a)daork.net> > To: FreeRadius users mailing list > <freeradius-users(a)lists.freeradius.org> > Subject: Re: Freeradius systemd radiusd.service fails > Message-ID: <DE682529-FA45-419F-82D1-4B4A1276E229(a)daork.net> > Content-Type: text/plain; charset=utf-8 > > > On 15/03/2016, at 21:44, Fajar A. Nugraha <list(a)fajar.net> wrote: > > > > On Tue, Mar 15, 2016 at 2:27 PM, Ibrahim Almahfooz > > <ibrahim.nezar(a)gorannet.net> wrote: > >> Hi everyone, > >> > >> We currently using freeradius version 3.0.11 and it is compiled to work > >> with oracle db and everything is fine and smooth when running the > service > >> using radiusd -X or radiusd. > >> > >> Once we start the service using systemctl start radiusd.service or > >> systemctl restart radiusd.service then we receive the below error: > >> > >> "Job for radiusd.service failed because the control process exited with > >> error code. See "systemctl status radiusd.service" and "journalctl -xe" > for > >> details." > > > > > > What does "systemctl status radiusd.service" and "journalctl -xe" say? > > It should tell you what it thinks is wrong. > > > > One possibility is that oracle depends on $ORACLE_HOME env variable, > > which is present when you log in as root/normal user, but not > > available when started by systemd. If that's the case, use Environment > > or EnvironmentFile inside the systemd unit. Look at your existing > > systemd unit files (e.g. ssh.service) for example. > > Another thing to check is your selinux config. Services started under > systemd can get more/less permissions than you expect. > > Watch /var/log/audit/audit.log when you start it. You might also try > disabling selinux (setenforce 0) temporarily to test. You should write a > policy that does what you want if you prove that selinux is the problem > though, rather than disabling it. It’s not too difficult. > > -- > Nathan Ward > > > > > ------------------------------ > > Message: 3 > Date: Tue, 15 Mar 2016 09:10:05 +0000 > From: Alan Buxey <A.L.M.Buxey(a)lboro.ac.uk> > To: FreeRadius users mailing list > <freeradius-users(a)lists.freeradius.org>, Ibrahim Almahfooz > <ibrahim.nezar(a)gorannet.net> > Subject: Re: Freeradius systemd radiusd.service fails > Message-ID: <6C7500E0-39B7-4BE4-BB03-6B8EAE15F2B1(a)lboro.ac.uk> > Content-Type: text/plain; charset="UTF-8" > > Read the output of those 2 commands it gives you (really, systemd does > try to help you! ) . Ensure that the permissions of all files in raddb > config are correct - owned by and readable by the user/gtoup that radiusd > runs as. Ensure that all log/other locations are the same. > If your module requires other environment variables ensure that those are > in the startup scripts. > > Finally. If using selinux et al , check those audit logs to see if the > system believes that the daemon has no right to connect to certain ports > etc and fix those issues (LOTS of people using postgres/mariadb etc so > those systems are pretty well covered but oracle is rare and distro > operators, for example, will not have provided patches/fixes/updates to > deal with it) > > alan > > ------------------------------ > > Message: 4 > Date: Tue, 15 Mar 2016 10:32:00 +0000 > From: Richard J Palmer <richard(a)merula.net> > To: FreeRadius users mailing list > <freeradius-users(a)lists.freeradius.org> > Subject: Network Design - FreeRadius > Message-ID: <56e7e4a0.1078.c84.9b5(a)merula.net> > Content-Type: text/plain; charset=us-ascii; format=flowed > > > Hi > > Currently I have a NAS device in a hosting centre and the SQL Server > with the auth and accounting data in another data centre. > > There's a 1G link between the two and the ping time between them is > ~9ms > > In time I am hoping to move the SQL server closer to the main NAS, but > for now would the best option be to place the FR server closer to the > NAS (with a higher latency to the SQL server) OR place the FR serve > closer to the SQL box ? > > Currently I have the FR server near the NAS and the SQL server remote > - but wanted to check this was 'sensible' > > Moving to FR 3 (from an old server with 2.12) I am seeing more log > messages like: > > Tue Mar 15 10:26:24 2016 : Error: (27774) Ignoring duplicate packet > from client mer-fire11 port 3799 - ID: 181 due to unfinished request > in component post-auth module sql > > and > > > Tue Mar 15 10:27:12 2016 : Error: (27785) Discarding duplicate request > from client mer-fire11 port 3799 - ID: 158 due to delayed response > > I have my DBA checking the radius server to see if anything can be > tuned here - but I wanted to ensure we had the best possible setup > here - We are using MSSQL as the backend here - via UnixODBC/freetds > > Thanks in advance > > Richard > > > ------------------------------ > > Message: 5 > Date: Tue, 15 Mar 2016 10:54:03 +0000 > From: Arran Cudbard-Bell <a.cudbardb(a)freeradius.org> > To: FreeRadius users mailing list > <freeradius-users(a)lists.freeradius.org> > Subject: Re: Network Design - FreeRadius > Message-ID: <493E2773-63FC-4B6E-87F3-2F2E48A6DBAF(a)freeradius.org> > Content-Type: text/plain; charset="utf-8" > > > > On 15 Mar 2016, at 10:32, Richard J Palmer <richard(a)merula.net> wrote: > > > > > > Hi > > > > Currently I have a NAS device in a hosting centre and the SQL Server > with the auth and accounting data in another data centre. > > > > There's a 1G link between the two and the ping time between them is ~9ms > > > > In time I am hoping to move the SQL server closer to the main NAS, but > for now would the best option be to place the FR server closer to the NAS > (with a higher latency to the SQL server) OR place the FR serve closer to > the SQL box ? > > FR closer to SQL. Latency between NAS and FR doesn't prevent requests > being served, whereas latency between FR and SQL does (because of FR's > blocking architecture). There's also likely to be more chatter between FR > and SQL. RADIUS is a pretty simple protocol, just request/response. > > You likely won't see a performance (or latency) difference until you put > the system under load. > > > > > Currently I have the FR server near the NAS and the SQL server remote - > but wanted to check this was 'sensible' > > > > Moving to FR 3 (from an old server with 2.12) I am seeing more log > messages like: > > > > Tue Mar 15 10:26:24 2016 : Error: (27774) Ignoring duplicate packet from > client mer-fire11 port 3799 - ID: 181 due to unfinished request in > component post-auth module sql > > Yeah, SQL DB being slow. > > -Arran > > Arran Cudbard-Bell <a.cudbardb(a)freeradius.org> > FreeRADIUS development team > > FD31 3077 42EC 7FCD 32FE 5EE2 56CF 27F9 30A8 CAA2 > >

1 0