[RADIATOR] Load balancing RADIATOR with Cisco ACE

2012-05-10 Thread Janssen, G.H.C. (Gaston)
Hi,

We'd like to load balance RADIUS requests over several RADIATOR servers.
Therefor we will use an external hardware load balancer: a Cisco ACE (service 
module).
Is there anyone who has experience with this kind of combination, i.e RADIATOR 
 Cisco ACE.

Any (white) papers on this subject are welcome, either so any ACE configuration 
examples.

We are particulairy interested in field experiences in the combination Cisco 
ACE / RADIATOR.


(We already have taken notice of the Cisco configuration guide Configuring 
RADIUS Load Balancing
which in genaral describes it, but is not product specific (in this case 
RADIATOR)   :)

Regards,
Gaston

___
radiator mailing list
radiator@open.com.au
http://www.open.com.au/mailman/listinfo/radiator


Re: [RADIATOR] Load balancing RADIATOR with Cisco ACE

2012-05-10 Thread James
I've done it -- currently in production serving an environment with
over 80,000 users. No issues.

If you're load balancing TACACS+ you should enable stickiness so that
the session remains pinned to one Radiator server. If load balancing
simple RADIUS, just do a simple serverfarm and load balance with a
least connections or round robin LB algorithm.

Hope this helps.

-james

On Thu, May 10, 2012 at 5:15 AM, Janssen, G.H.C. (Gaston)
g.jans...@uci.ru.nl wrote:
 Hi,

 We'd like to load balance RADIUS requests over several RADIATOR servers.
 Therefor we will use an external hardware load balancer: a Cisco ACE (service 
 module).
 Is there anyone who has experience with this kind of combination, i.e 
 RADIATOR  Cisco ACE.

 Any (white) papers on this subject are welcome, either so any ACE 
 configuration examples.

 We are particulairy interested in field experiences in the combination Cisco 
 ACE / RADIATOR.


 (We already have taken notice of the Cisco configuration guide Configuring 
 RADIUS Load Balancing
 which in genaral describes it, but is not product specific (in this case 
 RADIATOR)   :)

 Regards,
 Gaston

 ___
 radiator mailing list
 radiator@open.com.au
 http://www.open.com.au/mailman/listinfo/radiator
___
radiator mailing list
radiator@open.com.au
http://www.open.com.au/mailman/listinfo/radiator


Re: [RADIATOR] Load balancing RADIATOR with Cisco ACE

2012-05-10 Thread Alexander Hartmaier
EAP and OTP also requires pinning which I personally would always use.

Am 2012-05-10 16:56, schrieb James:
 I've done it -- currently in production serving an environment with
 over 80,000 users. No issues.

 If you're load balancing TACACS+ you should enable stickiness so that
 the session remains pinned to one Radiator server. If load balancing
 simple RADIUS, just do a simple serverfarm and load balance with a
 least connections or round robin LB algorithm.

 Hope this helps.

 -james

 On Thu, May 10, 2012 at 5:15 AM, Janssen, G.H.C. (Gaston)
 g.jans...@uci.ru.nl wrote:
 Hi,

 We'd like to load balance RADIUS requests over several RADIATOR servers.
 Therefor we will use an external hardware load balancer: a Cisco ACE 
 (service module).
 Is there anyone who has experience with this kind of combination, i.e 
 RADIATOR  Cisco ACE.

 Any (white) papers on this subject are welcome, either so any ACE 
 configuration examples.

 We are particulairy interested in field experiences in the combination Cisco 
 ACE / RADIATOR.


 (We already have taken notice of the Cisco configuration guide Configuring 
 RADIUS Load Balancing
 which in genaral describes it, but is not product specific (in this case 
 RADIATOR)   :)

 Regards,
 Gaston

 ___
 radiator mailing list
 radiator@open.com.au
 http://www.open.com.au/mailman/listinfo/radiator
 ___
 radiator mailing list
 radiator@open.com.au
 http://www.open.com.au/mailman/listinfo/radiator
 --
 Cheers, Alex


***
T-Systems Austria GesmbH Rennweg 97-99, 1030 Wien
Handelsgericht Wien, FN 79340b
***
Notice: This e-mail contains information that is confidential and may be 
privileged.
If you are not the intended recipient, please notify the sender and then
delete this e-mail immediately.
***
___
radiator mailing list
radiator@open.com.au
http://www.open.com.au/mailman/listinfo/radiator


Re: (RADIATOR) Load Balancing Radiator

2000-10-16 Thread Ron Hensley

In the main global section

BindAddress 10.0.0.1

Thats the one for the normal auth/accounting information to listen and
respond with.
Make it whichever ip bound to the nic, you want it to use and reload.

- Original Message -
From: "Chris" [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Monday, October 16, 2000 1:18 PM
Subject: (RADIATOR) Load Balancing Radiator



 I'm trying to load balance radiator across three seperate servers
 with an Extreme Summit 7i switch.  All servers respond correctly to
 requests out of the server farm.  However when put in the server farm they
 respond to the authentication request with the ethernet ip even though the
 request was sent to an ip on the loopback.  Because it is responding with
 a different ip than what the request was sent to, my portmasters are
 ignoring the response.  I noticed the 6.27.11 LocalAddress tag but seems
 to only work with AuthBy Radius.  Is there a way to have radiator respond
 with the ip that the request was sent to with AuthBy Unix?  The manual
 implies that this is default but it doesn't seem to be doing it. (perhaps
 because the address is on the loopback?)

 Has anyone run into the same problem?

 Here is my config:

 Foreground
 LogStdout   #THIS LINE IS FOR TESTING, OUTPUT GOES TO SCREEN
 LogDir  /var/log/radiator
 DbDir   /etc/raddb
 PidFile /var/run/radiusd.pid
 DictionaryFile  /etc/raddb/dictionary.livingston
 AuthPort1812
 AcctPort1813
 SnmpgetProg /usr/local/bin/snmpget
 Trace   4
 SocketQueueLength 10

 Client 1.2.3.4
   Secretx
   DefaultRealm  xxx
 /Client
 Client 2.3.4.5
   Secretx
   DefaultRealm  xxx
 /Client
 Client 3.4.5.6
   Secretx
 /Client
 Client 7.8.9.1
   Secretxx
 /Client
 Client DEFAULT
   Secretxx
   DupInterval   2
   NasType   Livingston
   SNMPCommunity frii
   LivingstonOffs22
   LivingstonHole1
 /Client

 AuthBy GROUP
   Identifier Frii
   AuthByPolicy ContinueWhileReject
   AuthBy SQL
 AuthSelect
 AccountingStopsOnly
 DBSource  x
 DBUsernamex
 DBAuthxx
 AcctSQLStatement insert into data values ('%n',%t,%{Acct
   /AuthBy
   AuthBy GROUP
 AuthByPolicy ContinueUntilReject
 AuthBy FILE
   Filename /etc/raddb/users-pop
 /AuthBy
 AuthBy FILE
   Filename /etc/raddb/users
 /AuthBy
   /AuthBy
 /AuthBy

 AuthBy UNIX
   Identifier FriiSystem
   Filename /etc/mypasswd
 /AuthBy

 SessionDatabase SQL
   Identifier FriiSessions
   DBSource  
   DBUsernamex
   DBAuthxx
   AddQuery replace into Sessions values.
   CountQuery select NASIDENTIFIER 
   DeleteQuery delete from Sessions where .
 /SessionDatabase

 Realm /realm1/i
   RewriteUsername   s/^([^@]+).*/$1/
   AuthBy Frii
   SessionDatabase FriiSessions
 /Realm
 Realm /realm2/i
   RewriteUsername   s/^([^@]+).*/$1/
   AuthBy Frii
   SessionDatabase FriiSessions
 /Realm
 Handler
   AuthBy Frii
   SessionDatabase FriiSessions
 /Handler

 Chris Bissell| Front Range Internet, Inc.
 [EMAIL PROTECTED]| www.frii.com [EMAIL PROTECTED]
 Technical Operations | 970-224-3668  800-935-6527


 ===
 Archive at http://www.starport.net/~radiator/
 Announcements on [EMAIL PROTECTED]
 To unsubscribe, email '[EMAIL PROTECTED]' with
 'unsubscribe radiator' in the body of the message.



===
Archive at http://www.starport.net/~radiator/
Announcements on [EMAIL PROTECTED]
To unsubscribe, email '[EMAIL PROTECTED]' with
'unsubscribe radiator' in the body of the message.



Re: (RADIATOR) Load Balancing Radiator

2000-10-16 Thread Chris


I tried this, so also to listen only on that ip, however this also did not
appear to work possibly because the ip is bound to the loopback (it has to
be bound to the loopback because of the method of load balancing the
Summit 7i is doing.

So when I did this, radiator only responded to requests on 1.2.3.4 (which
is configured on the loopback) but replied to those requests with the
ethernet ip.

I'm setting up a packet sniffer to confirm this wednesday AM so I don't
have to rely on lucent debug.

Chris

 In the main global section
 
 BindAddress 10.0.0.1
 
 Thats the one for the normal auth/accounting information to listen and
 respond with.
 Make it whichever ip bound to the nic, you want it to use and reload.
 
 - Original Message -
 From: "Chris" [EMAIL PROTECTED]
 To: [EMAIL PROTECTED]
 Sent: Monday, October 16, 2000 1:18 PM
 Subject: (RADIATOR) Load Balancing Radiator
 
 
 
  I'm trying to load balance radiator across three seperate servers
  with an Extreme Summit 7i switch.  All servers respond correctly to
  requests out of the server farm.  However when put in the server farm they
  respond to the authentication request with the ethernet ip even though the
  request was sent to an ip on the loopback.  Because it is responding with
  a different ip than what the request was sent to, my portmasters are
  ignoring the response.  I noticed the 6.27.11 LocalAddress tag but seems
  to only work with AuthBy Radius.  Is there a way to have radiator respond
  with the ip that the request was sent to with AuthBy Unix?  The manual
  implies that this is default but it doesn't seem to be doing it. (perhaps
  because the address is on the loopback?)
 
  Has anyone run into the same problem?
 
  Here is my config:
 
  Foreground
  LogStdout   #THIS LINE IS FOR TESTING, OUTPUT GOES TO SCREEN
  LogDir  /var/log/radiator
  DbDir   /etc/raddb
  PidFile /var/run/radiusd.pid
  DictionaryFile  /etc/raddb/dictionary.livingston
  AuthPort1812
  AcctPort1813
  SnmpgetProg /usr/local/bin/snmpget
  Trace   4
  SocketQueueLength 10
 
  Client 1.2.3.4
Secretx
DefaultRealm  xxx
  /Client
  Client 2.3.4.5
Secretx
DefaultRealm  xxx
  /Client
  Client 3.4.5.6
Secretx
  /Client
  Client 7.8.9.1
Secretxx
  /Client
  Client DEFAULT
Secretxx
DupInterval   2
NasType   Livingston
SNMPCommunity frii
LivingstonOffs22
LivingstonHole1
  /Client
 
  AuthBy GROUP
Identifier Frii
AuthByPolicy ContinueWhileReject
AuthBy SQL
  AuthSelect
  AccountingStopsOnly
  DBSource  x
  DBUsernamex
  DBAuthxx
  AcctSQLStatement insert into data values ('%n',%t,%{Acct
/AuthBy
AuthBy GROUP
  AuthByPolicy ContinueUntilReject
  AuthBy FILE
Filename /etc/raddb/users-pop
  /AuthBy
  AuthBy FILE
Filename /etc/raddb/users
  /AuthBy
/AuthBy
  /AuthBy
 
  AuthBy UNIX
Identifier FriiSystem
Filename /etc/mypasswd
  /AuthBy
 
  SessionDatabase SQL
Identifier FriiSessions
DBSource  
DBUsernamex
DBAuthxx
AddQuery replace into Sessions values.
CountQuery select NASIDENTIFIER 
DeleteQuery delete from Sessions where .
  /SessionDatabase
 
  Realm /realm1/i
RewriteUsername   s/^([^@]+).*/$1/
AuthBy Frii
SessionDatabase FriiSessions
  /Realm
  Realm /realm2/i
RewriteUsername   s/^([^@]+).*/$1/
AuthBy Frii
SessionDatabase FriiSessions
  /Realm
  Handler
AuthBy Frii
SessionDatabase FriiSessions
  /Handler
 
  Chris Bissell| Front Range Internet, Inc.
  [EMAIL PROTECTED]| www.frii.com [EMAIL PROTECTED]
  Technical Operations | 970-224-3668  800-935-6527
 
 
  ===
  Archive at http://www.starport.net/~radiator/
  Announcements on [EMAIL PROTECTED]
  To unsubscribe, email '[EMAIL PROTECTED]' with
  'unsubscribe radiator' in the body of the message.
 
 

Chris Bissell| Front Range Internet, Inc.
[EMAIL PROTECTED]| www.frii.com [EMAIL PROTECTED]
Technical Operations | 970-224-3668  800-935-6527


===
Archive at http://www.starport.net/~radiator/
Announcements on [EMAIL PROTECTED]
To unsubscribe, email '[EMAIL PROTECTED]' with
'unsubscribe radiator' in the body of the message.



Re: (RADIATOR) Load Balancing Radiator

2000-10-16 Thread Ron Hensley

That is odd. I didnt mention it, but I also use load balancing, though with
a Linux Server doing the clustering
rather then a layer 2 switch. Same concept though, it intercepts the packets
destined for the radius server
ip address, and redirects them to the cluster nodes, who have the ips bound
as loopback addresses, so
that they will not respond to ARP broadcasts and interfere with the cluster
server doings its job.

Anyways, the BindAddress is working on my 3 Suns, Solaris 2.6 and 7.0, when
using the loopback, clustered
address. The only other time I had the problem like that, is when my NAS
servers were speaking to
the radius servers, by way of a different ip address then the replies were
coming back from, as you surmised.
However on every flavor of radius ive used, using a localaddress or
bindaddress to force the issue has solved it.

Heh sounds like a packet sniffer is the only way to go, as well as trace 4
logs on Radiator and any debug
logs your NASs can produce.

- Original Message -
From: "Chris" [EMAIL PROTECTED]
To: "Ron Hensley" [EMAIL PROTECTED]
Cc: [EMAIL PROTECTED]
Sent: Monday, October 16, 2000 5:21 PM
Subject: Re: (RADIATOR) Load Balancing Radiator



 I tried this, so also to listen only on that ip, however this also did not
 appear to work possibly because the ip is bound to the loopback (it has to
 be bound to the loopback because of the method of load balancing the
 Summit 7i is doing.

 So when I did this, radiator only responded to requests on 1.2.3.4 (which
 is configured on the loopback) but replied to those requests with the
 ethernet ip.

 I'm setting up a packet sniffer to confirm this wednesday AM so I don't
 have to rely on lucent debug.

 Chris

  In the main global section
 
  BindAddress 10.0.0.1
 
  Thats the one for the normal auth/accounting information to listen and
  respond with.
  Make it whichever ip bound to the nic, you want it to use and reload.
 
  - Original Message -
  From: "Chris" [EMAIL PROTECTED]
  To: [EMAIL PROTECTED]
  Sent: Monday, October 16, 2000 1:18 PM
  Subject: (RADIATOR) Load Balancing Radiator
 
 
  
   I'm trying to load balance radiator across three seperate servers
   with an Extreme Summit 7i switch.  All servers respond correctly to
   requests out of the server farm.  However when put in the server farm
they
   respond to the authentication request with the ethernet ip even though
the
   request was sent to an ip on the loopback.  Because it is responding
with
   a different ip than what the request was sent to, my portmasters are
   ignoring the response.  I noticed the 6.27.11 LocalAddress tag but
seems
   to only work with AuthBy Radius.  Is there a way to have radiator
respond
   with the ip that the request was sent to with AuthBy Unix?  The manual
   implies that this is default but it doesn't seem to be doing it.
(perhaps
   because the address is on the loopback?)
  
   Has anyone run into the same problem?
  
   Here is my config:
  
   Foreground
   LogStdout   #THIS LINE IS FOR TESTING, OUTPUT GOES TO SCREEN
   LogDir  /var/log/radiator
   DbDir   /etc/raddb
   PidFile /var/run/radiusd.pid
   DictionaryFile  /etc/raddb/dictionary.livingston
   AuthPort1812
   AcctPort1813
   SnmpgetProg /usr/local/bin/snmpget
   Trace   4
   SocketQueueLength 10
  
   Client 1.2.3.4
 Secretx
 DefaultRealm  xxx
   /Client
   Client 2.3.4.5
 Secretx
 DefaultRealm  xxx
   /Client
   Client 3.4.5.6
 Secretx
   /Client
   Client 7.8.9.1
 Secretxx
   /Client
   Client DEFAULT
 Secretxx
 DupInterval   2
 NasType   Livingston
 SNMPCommunity frii
 LivingstonOffs22
 LivingstonHole1
   /Client
  
   AuthBy GROUP
 Identifier Frii
 AuthByPolicy ContinueWhileReject
 AuthBy SQL
   AuthSelect
   AccountingStopsOnly
   DBSource  x
   DBUsernamex
   DBAuthxx
   AcctSQLStatement insert into data values ('%n',%t,%{Acct
 /AuthBy
 AuthBy GROUP
   AuthByPolicy ContinueUntilReject
   AuthBy FILE
 Filename /etc/raddb/users-pop
   /AuthBy
   AuthBy FILE
 Filename /etc/raddb/users
   /AuthBy
 /AuthBy
   /AuthBy
  
   AuthBy UNIX
 Identifier FriiSystem
 Filename /etc/mypasswd
   /AuthBy
  
   SessionDatabase SQL
 Identifier FriiSessions
 DBSource  
 DBUsernamex
 DBAuthxx
 AddQuery replace into Sessions values.
 CountQuery select NASIDENTIFIER 
 DeleteQuery delete from Sessions where .
   /SessionDatabase
  
   Realm /realm1/i
 RewriteUsername   s/^([^@]+).*/$1/
 AuthBy Frii
 SessionDatabase FriiSessions
   /Realm
   Realm /realm2/

Re: (RADIATOR) Load Balancing Radiator

2000-10-16 Thread Hugh Irvine


Hello Chris -

On Tue, 17 Oct 2000, Chris wrote:
 I'm trying to load balance radiator across three seperate servers
 with an Extreme Summit 7i switch.  All servers respond correctly to
 requests out of the server farm.  However when put in the server farm they
 respond to the authentication request with the ethernet ip even though the
 request was sent to an ip on the loopback.  Because it is responding with
 a different ip than what the request was sent to, my portmasters are
 ignoring the response.  I noticed the 6.27.11 LocalAddress tag but seems
 to only work with AuthBy Radius.  Is there a way to have radiator respond
 with the ip that the request was sent to with AuthBy Unix?  The manual
 implies that this is default but it doesn't seem to be doing it. (perhaps
 because the address is on the loopback?)
 

You should use the "BindAddress" global parameter to set the address to your
loopback. If the outbound packet has a different IP address, I would suspect
that it is the operating system that is using the ethernet source IP address
rather than Radiator. What system are you running on?

hth

Hugh


-- 
Radiator: the most portable, flexible and configurable RADIUS server 
anywhere. SQL, proxy, DBM, files, LDAP, NIS+, password, NT, Emerald, 
Platypus, Freeside, Interbiller, TACACS+, PAM, external, etc, etc.
Available on Unix, Linux, FreeBSD, Windows 95/98/2000, NT, MacOS X.



===
Archive at http://www.starport.net/~radiator/
Announcements on [EMAIL PROTECTED]
To unsubscribe, email '[EMAIL PROTECTED]' with
'unsubscribe radiator' in the body of the message.