Cisco AAA/Identity/Nac :: ASR 9010 Configuration To Connect To A Tacacs+ Server
Jun 10, 2013
We have an ASR 9010 with IOS XR, and we are making the configuration to connect to a tacacs+ server, this tacacs+ server works and is givins service to many other MPLS equipments. We have been following the guide:
Configuring AAA Services on
Cisco ASR 9000 Series Routers
but we have had a lot of troubles, in fact we have loose the administration of the box, at this moment the only lines that are in the ASR900 are: [code]
I see there is a similar post for Nexus 5000 to ACS 5.2. Identical symptoms. The supervisor crashed and switched to secondary. Is there a comparable field for ACS 4.1 that needs to have something in it? 2012 Apr 9 11:07:55 va-core02 %$ VDC-1 %$ %SYSMGR-2-SERVICE_CRASHED: Service "Tacacs Daemon" (PID 9390) hasn't caught signal 11 (core will be saved). 2012 Apr 9 11:07:55 va-core02 %$ VDC-1 %$ %SYSMGR SYSMGR_AUTOCOLLECT_TECH_SUPPORT_LOG: This supervisor will temporarily remain online in order to collect show tech-support. This behavior is configurable via 'system [no] auto-collect tech-support'.
Have you ever found the problem that if I set two tacacs server in my N7K and the primary tacacs server fail, won't switch over to another tacacs server.
I've been configured my device 6506-9 with TACACS+ server authentication: [code]
but when I tried to access the device only uses authentication local but not uses TACACs (with username/password defined) it can be an error in configuration? in the other devices of network this works properly, only it's wrong in Cat6506-E
Running ACS 5.1 appliance, and am seeing slow repsonse on TACACS authentications due to the ACS trying to reach overseas AD servers and failing. Is there any way to configure a /etc/host/ file locally on the ACS in order to force the appliance to use specific AD servers for authentication? As I understand the process currently, the ACS appliance will query the top-level domain and get a list of all the AD servers in DNS. In my case, this would include the AD servers overseas that we do not want to use.
I have both boxes heading off to ACS for TACACS lo gin authentication and for command authorization. When I first set things up everything works fine. I have a shell profile configured in ACS with Cisco-av-pair*shell:roles="network-admin" to set the network-admin role. I even have command sets configured to deny the use of configure terminal as I am using switch configuration profiles. Everything runs fine. User lo gins are authenticated by ACS and users have the correct command set applied to them.
The problem comes when I make a change to a shell profile in ACS. Even something as simple as changing the name of a shell profile causes the 5520's to crash as soon as I try to log on. If I unplug the management link so that the TACACS server is unavailable I can log on fine with the local admin user.
The NEXUS console reports this error. (amongst many others)
EDNAM-NEXUS-2 %$ VDC-1 %$ %SYSMGR-2-SERVICE_CRASHED: Service "Tacacs Daemon" (PID 4331) hasn't caught signal 11 (core will be saved).
A show system reset-reason shows:
EDNAM-NEXUS-2# sh system reset-reason ----- reset reason for Supervisor-module 1 (from Supervisor in slot 1) --- 1) At 389 usecs after Wed Jan 18 12:32:49 2012 Reason: Reset triggered due to HA policy of Reset Service: Tacacs Daemon hap reset Version: 5.0(3)N1(1c)
What is the procedure to delete the full startup-configuration of Cisco ASR 9010 with IOS-XR version 3.9.Tried the following but unable to do it:erase nvram, Commit Replce.
configure AAA (Radius server, access list) There are two devices An access point and cisco 881w. It is necessary to set up authentication through a radius server. You can configure detailed how to do this?
I have a couple of ACS 5.2 configured as active and backup and I am doing dot 1x authentication using these servers . I have configured the switch with the bellow configuration.
I have a Cisco ASA (8.2) setup with remote access for my users using Cisco VPN client. The authentication is passed off to my ACS 5.3 which then checks with AD. What I've done so far is create Access Policy rule where I define specifically the Location and NDG where the ASA is and then a DenyAllCommands command set. This should pass authentications just fine but this also gives those users the ability to remote connect directly into the ASA and login successfully. Even though there is a Deny Commands there I still would prefer they get Access Denied as a message. If I do a Deny Access on the ShellProfile then this stops the login authentication altogether.
I think i've got everything set up to authenticate against AD for Tacacs+ device logins. When i check the logs, i see:"24408 User authentication against Active Directory failed since user has entered the wrong password". This leads me to believe that it is checking AD correctly, however if i enter the password correctly for the same AD user, there is no log at all...no pass, no fail.
If i look at the Tacacs debugs on the switch, i see the following:May 25 10:55:07.927 CDT: TAC+: ver=192 id=874699084 received AUTHEN status = ERRORMay 25 10:55:09.932 CDT: TAC+: send abort reason=Unknown
Obviously the switch is communicating to ACS, and ACS is passing info back to the switch. ACS also appears to be communicating effectively with AD since it knows when i put in an incorrect password for the specific user.
I found that TACACS should be available for network access with ACS 5.2:(url) But when I'm trying to create Rule tu allow PPP authentication against TACACS server I get error.
I am running ACS 4.1.1.23 on a Microsoft server and I am trying to get TACACS to work with two Linux servers. The servers are capable of TACACS, are using port 49 and have the correct shared secret. I believe I do not have the devices configured properly on the ACS side. These 2 servers currently are using RADIUS and we are getting bit by the bug where the ACS application will start rejecting RADIUS authentication requests but still accept TACACS requests.
I am setting up reports for tacacs accounting on ACS 5.3. However, accounting only seems to work after entering enable mode on the switch. I would like to see all commands, even the enable command when in privlage 1 mode.
I can get it to authenticate. But I've read some posts on ACS 4.2 and authorization, but I don't find anything similar.I want to control down to what commands the authenticated user can run. I want the defintion to come from the ACS server, or at least control it from the ACS server. I want to minimize the changes on the JunOS side,but if it can't be easily done, I'll change the JunOS side.
ACS 5.3 configured with two rules, 1 rule for standard level 15 access for the Network Engineers and a 2nd rule to allow some limited access to switches: The limited access account has enough command set access to change the vlan on a switchport, so Configure Terminal, Interface FAx/x and switchport access vlan x.
Switch configuration:
aaa authentication login default group tacacs+ local aaa authentication enable default group tacacs+ enable aaa authorization console aaa authorization config-commands aaa authorization exec default group tacacs+ local aaa authorization commands 0 default group tacacs+ local aaa authorization commands 1 default group tacacs+ local aaa authorization commands 15 default group tacacs+ local aaa session-id common
Everything works well and the limited access users can only perform the commands i've setup.
Problem:The problem i've encountered is when one of the network engineers makes a change that would stop the device from being able to see the ACS server it stops allowing any commands to be typed in the router/switch. Additionally if you then connect to the device and login with the local username and password the device then waits for it to hit the TACACS server timeout for every command you enter. This is obviously very slow and painful for the engineer.
Question:Is there a way to set this up so the engineer logging in with full Level 15 access doesn't have to have each command authorized by the ACS server but still allow the limited access accounts to be able to make interface changes?
So far i managed my switches with TACACS+, however now i've to deploy 802.1X, requiring RADIUS only. For what i know, ACS (i'm using 4.2) allows to define a device using only TACACS or RADIUS, but not both. Do i am right? Or there is a way to define an AAA client to communicate with the same ACS using both the protocols?
Supposing i am right, i was then considering the following options: - configure all of the switches to use radius for any service (authentication, authorization etc ec) This simplifies the task, but i lose the TACACS+ services for the switches. Is this a big loss?
- configure the L3 switches to use a second Loopback, just for RADIUS services. This would allow to still use the TACACS+ but would require a new network just for the RADIUS service; furthermore L2 switches doesn't support two IP addresses and would require anyway a migration to RADIUS.
A considerable administrative overhead, in other words. I'm not willing to deploy a second RADIUS (ACS, Windows, whatever), in this moment.
The key point is this: reading around i see Cisco documentation recommending always to use TACACS+ for management, but in this situation is not possibile. In general, every time the device has a role of network admission (switch or access-point) RADIUS seems to be the protocol of choice. Moving to RADIUS would have some major drawback or only a change in the communication protocol? (I know the difference between TACACS+ and RADIUS: tcp vs udp, encryption of the whole packet vs encryption of only the password).
I want to setup two factor authentication via ACS 5.2 TACACS+ without having to use a token (such as that by RSA). Is there a way to do it?
More info:
Users from unconnected AD domains will be connecting to the routers and switches.There is a certificate server available to generate certificates.SSHv2 is the current login protocol.
Noticed tacacs authorization logs when you change password for a user ?? in authorization logs I can see the new password but same I can not see in accounting logs ? is it a normal behaviour ?? or do we need to do something to hide the password in authorization logs ?
For example if i type command username xyz priv 15 secret cisco 123
I see this command in accounting logs as uername xyz oriv 15 secret *** where as in tacacs authorization logs it shows username xyz priv 15 secret cisco 123
I am trying to access an ASA 5545 using TACACS+. I have the ASA configured as follows: aaa-server tacacs+ protocol tacacs+ aaa-server tacacs+ (inside) host 10.x.x.x [code]....
I have added the ASA in ACS with the correct IP and the correct key. When I try to test the authentication via test aaa-server authentication tacacs+ host 10.x.x.x username Cisco password Cisco, I get:
ERROR: Authentication Server not responding: No error.
I have several 2950 switches that I cannot get to work with TACACS. I'm using the same config for these that I am using for other cisco switches. [code]
I have ACS 5.2 and JUNOS 10.6.x I setup 2 classes eng-class and ops-class with read/write and read-only permission here is my configuration on JUNOS
set system login class eng-class idle-timeout 15 set system login class eng-class permissions all set system login user engineer full-name “Regional-Engineering” set system login user engineer uid 2001 set system login user engineer class eng-class set system login user engineer authentication plain-text-password xxxxxxx
[code]....
I have 2 separate Authorization policies for engineer and operator group.Result,
1. engineering group is working fine.
2. the operator group its not working im unable to login to device under this group "authentication failed" but on the ACS logs its successfully authenticated.
3. Web authentication is not also working for bot group.
we have a ACS server V4 installed on W2003 server ,when we make a telnet to an equipement on the wan the authentication pass on the first connexion ,but when we telent to a switch on the lan the first connxion fails and we need to retry to login .when i check the field attempt log on the ACS i dont find the field attempt.i find this issue in ALL switch on the LAN ,from the switch i can ping the the ACS server .this problem appear frequently?
I've set up my 5540 ASA to accounting commands on TACACS+.Every moviment done through ASDM is logged on TACACS+ by this form: cmd=perfmon interval 10.What does that mean?Why doesn't it record the exaclty command I'd issued?
After implementing TACACS, one of our routers takes about 8 seconds to response to any CLI command. We have no problems with other devices in the same location with the same AAA configuration. The router is talking to the ACS server (ACS 5.3) and the logs on the ACS server look normal for the router as well.
Having an issue with authenticating Juniper J Series and SRX devices with ACS 5.1 The devices can authenticate using TACACS to ACS 5.1 via the CLI (telnet / ssh connections) but cannot using the JWEB management page.Doing packet captures between the Juniper devices and the ACS 5.1 box shows the Authenticate phase passing, but it does not progress onto the Authorisation phase. There is nothing of interest in the ACS Logs (Even with the debugging levels turned right up) The same Access service is in use for both the CLI and GUI (JWEB).Using ACS 4.1, both CLI and JWEB authentication works.[URL]I'm thinking the issue is with ACS 5.0 / 5.1 and it maybe not liking the response from the Juniper (even though it should be the same mechanism)
I'm currently evaluating a scenario where AAA request are load balanced across multiple ACS 5.3 instances. The application delivery controller runs in L3 mode, which naturally causes the original packet's source IP address to be replaced by a randomly selected proxy address.As far as RADIUS is concerned, I can perfectly determine the originating NAS by means of a 'Device Filter' condition. Unfortunately, ACS seems to lack the possibility of achieving the same for TACACS+. According to the user manual, only the actual IP address from the received packet is taken into account. I've also come across the 'NAS-Address' attribute in the protocol dictionary, but it can't be used in a custom condition either.how to retrieve the initial device IP address from a TACACS+ request in order to use it for further policing?
I am having an issue with authorization on the Nexus 5548. Note: The tacacs configuration has and still works correctly with all non-Nexus gear.
Authentication succeeds, and initiatial authorization passes. However, all sh and config commands fail, though AAA Autho Config-Commands .... and Commands Default Group <Grp Name), are configured.
ACS generates the following error: 13025 Command failed to match a Permit rule. The Selected Command Set is DenyAllCommands. I created an AllowAll, but am unclear how to associate this with Access Policy.
i would like to use the ACS 5.3 as TACACS Proxy. Basically it works. But when checking the logs on the destination TACACS Server (ACS 4.2) i see that all requests (Source-NAs) came from the IP of the TACACS-Proxy. Not from the original source IP.
This is useless for my scenario, because on the destination TACACS Server the policies are built on the NetworkDevices Groups and AAA Clients = source IPs.
