tac_plus-ng is a TACACS+ daemon. It provides networking components like routers and switches with authentication, authorisation and accounting services.
This version is a major rewrite of the original public Cisco source code and is in turn largely based on tac_plus, which comes with the same distribution. Key features include:
NAS specific device keys, prompts, enable passwords
Rule-based permission assignment
Flexible external back-ends for user profiles (e.g. via PERL scripts or C; LDAP (including ActiveDirectory), RADIUS and others are included)
Connection multiplexing (multiple concurrent NAS clients per process)
Session multiplexing (multiple concurrent sessions per connection, single-connection)
Scalable, no limit on users, clients or servers.
CLI context aware.
Full support for both IPv4 and IPv6
Implements and auto-detects HAProxy protocol 2.
Supports TLS
Compliant to RFC8907
Supports Linux VRFs
Supports (non-standard) SSH Public Key Authentication (see the Wiki for reference)
You can download the source code from the GitHub repository at https://github.com/MarcJHuber/event-driven-servers/. On-line documentation is available via https://projects.pro-bono-publico.de/event-driven-servers/doc/, too.
The following chapters utilize a couple of terms that may need further explanation:
Client or NAC | A Network Access Client, e.g. the source device of a ssh (or telnet) connection. |
Device or NAS or NAD | A Network Access Server or Device, e.g. a Cisco box, or any other client which makes TACACS+ authentication and authorization requests, or generates TACACS+ accounting packets. |
Daemon | A program which services network requests for authentication and authorization, verifies identities, grants or denies authorizations, and logs accounting records. |
AV pairs | Strings of text in the form attribute=value, sent between a NAS and a TACACS+ daemon as part of the TACACS+ protocol. |
Since a NAS is sometimes referred to as a server, and a daemon is also often referred to as a server, the term server has been avoided here in favor of the less ambiguous terms NAS and Daemon.
This section gives a brief and basic overview on how to run tac_plus-ng.
In earlier versions, tac_plus wasn't a standalone program but had to be invoked by spawnd. This has changed, as spawnd functionality is now part of the tac_plus binary. However, using a dedicated spawnd process is still possible and, more importantly, the spawnd configuration options and documentation remain valid.
tac_plus-ng may use auxiliary MAVIS back-end modules for authentication of users and authorization of users and hosts.
The only mandatory argument is the path to the configuration file:
tac_plus-ng [ -P ] [ -d level ] [ -i child_id ] configuration-file [ id ]
If the program was compiled with CURL support, configuration-file may be an URL.
Keep the -P option in mind - it is imperative that the configuration file supplied is syntactically correct, as the daemon won't start if there are any parsing errors.
The -d switch enables debugging. You most likely don't want to use this. Read the source if you need to.
The -i option is only honoured if the build-in spawnd functionality is used. In that case, it selects the configuration ID for tac_plus, while the optional last argument id sets the ID of the spawnd configuration section.
Both the master (that's the process running the spawnd code) and the child processes (running the tac_plus-ng code) intercept the SIGHUP signal:
The master process will restart upon reception of SIGHUP, re-reading the configuration file. The child processes will recognize that the master process is no longer available. It will continue to serve the existing connections and terminate when idle.
If SIGHUP is sent to a child process it will stop accepting new connections from its master process. It will continue to serve the existing connections and terminate when idle.
Sending SIGUSR1 to the master process will cause it to abandon existing child processes (these will continue to serve the existing connections only) and start new child processes.
Several level-triggered event mechanisms are supported. By default, the one best suited for your operating system will be used. However, you may set the environment variable IO_POLL_MECHANISM to select a specific one.
The following event mechanisms are supported (in order of preference):
port (Sun Solaris 10 and higher only, IO_POLL_MECHANISM=32)
kqueue (*BSD and Darwin only, IO_POLL_MECHANISM=1)
/dev/poll (Sun Solaris only, IO_POLL_MECHANISM=2)
epoll (Linux only, IO_POLL_MECHANISM=4)
poll (IO_POLL_MECHANISM=8)
select (IO_POLL_MECHANISM=16)
Environment variables can be set in the configuration file at top-level:
setenv IO_POLL_MECHANISM = 4
The daemon is configured using a text file. Let's have a look at a sample configuration first, before digging into the various configuration directives.
A single configuration file is sufficient for configuring quite everything: the spawnd connection broker, tac_plus-ng and the MAVIS authentication and authorization back-end.
The daemon supports shebang syntax. If the configuration file is executable and starts with
#!/usr/local/sbin/tac_plus-ng
then it can be started directly.
The first step is to configure the spawnd portion to tell the daemon the addresses and TCP ports to listen on and to, eventually pass realms:
id = spawnd { listen { port = 49 } listen { port = 4949 } listen { address = ::0 port = 4950 realm = customer1 } listen { address = 10.0.0.1 port = 4951 realm = customer2 } # listen { address = 10.0.0.1 port = 4951 realm = customer2 tls = yes } # # See the spawnd configuration guide for further configuration options. }
The thing that needs some explanation here is realms. A realm in tac_plus-ng summarizes a set of configuration options. Realms inherit configurations from their parent realm, including the parent ruleset, which will be evaluated if the local ruleset doesn't exist or doesn't return a verdict.
The default realm is internaly named default. Using realms is optional.
Now to the actual tac_plus-ng configuration which starts with
id = tac_plus-ng { # This is the top-level realm, actually.
The second line above starts a comment. Comments can appear anywhere in the configuration file, starting with the # character and extending to the end of the current line. Should you need to disable this special meaning of the # character, e.g. if you have a password containing a # character, simply enclose the string containing it within double quotes.
Typically, the next step is to define log destinations and tell the daemon to use them. This sample logs to disk, but other destinations (syslog, pipe) are available, too.
log authzlog { destination = /var/log/tac_plus/authz/%Y/%m/%d.log } log authclog { destination = /var/log/tac_plus/authc/%Y/%m/%d.log } log acctlog { destination = /var/log/tac_plus/acct/%Y/%m/%d.log } accounting log = acctlog authentication log = authclog authorization log = authzlog
Logs are interited to sub-realms and while sub-realms can define their own logging that won't override the parent realm definitions.
You can specify a retire limit to have the server auto-terminate and restart its worker processes:
retire limit = 1000
Then, there's the MAVIS part:
mavis module groups { resolve gids = yes resolve gids attribute = TACMEMBER groups filter = /^(guest|staff|ubuntu)$/ } mavis module external { exec = /usr/local/sbin/pammavis "pammavis" "-s" "sshd" } user backend = mavis login backend = mavis chpass pap backend = mavis
which defines interaction with external external back-ends.
You can define network objects for later use in ACLs:
net outThere { address = 100.65.3.1 address = 100.66.0.0/16 }
Networks can be hierarchic, too:
net all { net north { address = 100.67.0.0/16 } net south { address = 100.68.0.0/16 } }
Now, define device objects for your network access devices. Just like realms and networks these can be hierarchic:
device world { welcome banner = "\nHitherto shalt thou come, but no further. (Job 38.11)\n\n" key = QaWsEdRfTgY enable 15 = clear test address = ::/0 device south { address = 100.99.0.0/16 } device west { address = 100.100.0.0/16 } } device localhost { address = 127.0.0.1 welcome banner = "Welcome home\n" parent = world # for key and other definitions not set here } device rfc { address = 172.16.0.0/12 welcome banner = "Welcome private\n" key = labKey }
Now, define some profiles. These will be assigned to users later:
profile readwrite { script { if (service == shell) { if (cmd == "") set priv-lvl = 15 permit } } } profile getconfig { script { if (service == shell) { if (cmd == "") { set autocmd = "sho run" set priv-lvl = 15 permit } } } } profile engineering { script { if (service == shell) { if (cmd == "") { set priv-lvl = 7 permit } if (cmd =~ /^ping/) deny permit } } } profile guest { script { if (service == shell) { if (cmd == "") { set priv-lvl = 1 permit } } permit } }
Your can define groups to implement a role-based access control scheme ...
group admin { group north # "admin" is a member group south # of both } group engineering group guest
... and add users:
user demo { password login = clear demo member = engineering,admin } user readonly { password login = clear readonly member = guest }
You can optionally assign a profile to a user by either referencing or embedding it. This skips ruleset evaluation.
user demo1 { password login = clear demo profile = engineering } user demo2 { password login = clear demo profile { script { if (service == shell) { if (cmd == "") { set priv-lvl = 1 permit } } permit } } }
Finally, implement a rule-set to assign profiles to users. As mentioned a couple of lines above, the ruleset will not be evaluated for users with a profile pre-assigned, so if all your users have a profile pre-assigned you may as well skip that part:
ruleset { rule from-localhost { enabled = yes script { if (nas == localhost) { if (group == admin) { profile = admin permit } if (group == engineering ) { profile = engineering permit } } } } rule from-rfc { enabled = yes script { if (nas == rfc) { if (group == south) { profile = admin permit } if (group == engineering ) { profile = engineering permit } } } } } }
Configuration options include
global options
realms
devices
time specifications
profiles
groups
users
access lists
rules
Including Files | |
---|---|
Configuration files may refer to other configuration files: include = file will read and parse file. Shell wildcard patterns are expanded by glob(3). The include statement will be accepted virtually everywhere (but not in comments or textual strings). |
The global configuration section may contain the following configuration directives, plus the realm options detailed in the next section. realm confiurations at global level are implicitely assigned to the default realm and will be inherited by sub-realms.
A number of global limits and timeouts may be specified exclusively at global level:
retire limit = n
The particular daemon instance will terminate after processing n requests. The spawnd instance will spawn a new instance if necessary.
Default: unset
retire timeout = s
The particular daemon instance will terminate after s seconds. spawnd will spawn a new instance if necessary.
Default: unset
last-recently-used limit = n
A LRU limit can be set to prioritize new connections. If adding a connection exceeds a total of n connections then the least recently used connection will be closed. Set this somewhat lower than the users max parameter in the spawnd section to have any impact.
Default: unset
Time units | |
---|---|
Appending s, m, h or d to any timeout value will scale the value as expected. |
tac_plus-ng can make use of both static and dynamic (via c-ares) DNS entries. Configuration options at global (and realm) level are:
dns preload address address = hostname
Preload DNS cache with address-to-hostname mapping.
dns preload file = filename
Preload DNS cache with address-to-hostname mappings from filename (see your hosts(5) manpage for syntax).
Example:
dns preload address 1.2.3.4 = router.example.com dns preload file = /etc/hosts device router.example.com { # "address = 1.2.3.4" is implied key = mykey }
dns cache period = seconds
This option specifies the minimum DNS response caching time (default: 1800 seconds).
dns servers = "string"
This option specifies the servers to use. This string will be evaluated by ares_set_servers_ports_csv(3), please see the corresponding man page for details. The option isn't available if compiled without DNS support.
The following configuration options are available at global, realm, device and net level.
dns reverse-lookup [ nac | nas ] = ( yes | no )
This will perform a DNS reverse lookup on the NAC address, the NAS address or (if unspecified) both.
dns timeout = seconds
This option specifies the maximum amount of time to wait for a DNS response.
There are a couple of process-specific options available:
coredump directory = directory
Dump cores to directory. You really shouldn't need this.
Bascially, realms are containers to logically separate configuration sets. At top-level, there's the default realm (called default internally). Realms pass on most configurations (e.g. logging, users (if there are no users defined in that realm scope), groups, profiles) to their sub-realms.
Realm selection is intially based on spawnd configuration:
spawnd = { listen { port = 49 } # implied realm is "default" listen { port = 3939 } # implied realm is "default" listen { port = 4949 realm = realmOne } listen { port = 5959 realm = realmTwo } }
If VRFs are used and no realm is specified in the spawnd section, the daemon will try to use the VRF name as realm and fall back to the default realm if that "vrf realm" isn't defined.
A realm can be selected based on device address, too:
device myDevices { address = 10.1.23.0/24 target-realm = realmOne }
The syntax to use (and define) realms is
realm realmName { ... }
at top configuration level. Realms cover devices, users, groups, profiles, rulesets, timespecs, MAVIS configurations other configuration options.
The following options may be specified at realm level. This includes the default realm:
Logging options defined in the top-level default realm will be shared with sub-realms unless the sub-realm has its own logging configuration. The software provides logs for
Authentication
authentication log = log_destination
Authorization
authorization log = log_destination
Accounting
accounting log = log_destination
Connections
connection log = log_destination
Logs may be written to multiple destinations:
Valid log destinations are "named":
log mylog { destination = 169.254.0.23 # UDP syslog # or one of the following: # destination = [fe80::123:4567:89ab:cdef]:514 # IPv6 UDP, with non-standard UDP port # destination = "/tmp/x.log" # plain file, async writes # destination = ">/tmp/x.log" # plain file, sync writes # destination = "|my_script.sh" # script # destination = syslog # syslog(3) # syslog facility = MAIL # sets log facility syslog level = DEBUG # sets log level } authentication log = mylog accounting log = mylog authorization log = mylog
Syslog | |
---|---|
Logging non-session related output to syslogd(8) can be disabled using syslog default = deny |
Log destinations may contain strftime(3)-style character sequences, e.g.:
destination = /var/log/tac_plus/%Y/%m/%d.auth
to automate time-based log file switching. By default, the daemon will use your local time zone for time conversion. You can switch to a different one by using the time zone option (see below).
A couple of other configuration options that may be useful in log context include:
( authentication | authorization | accounting ) format = string
This defines the logging format. strftime(3) conversions are recognized. The following variables are resolved:
${cmd}, ${cmd,separator} | values of cmd= and cmd-arg= attribute-value pairs, separated by whitespace or separator. Will be mapped to ${args} if service isn't shell |
${args}, ${args,separator} | input attribute-value pairs (excluding service, separated by whitespace or separator |
${rargs}, ${rargs,separator} | output attribute value pairs, separated by whitespace or separator |
${device.address}, ${nas} [deprecated] | Device IP address |
${client}, ${nac} | Client IP address |
${user} | user name |
${user.original} | user name before any rewrite operations |
${profile} | profile assigned to user |
${service} | service type (e.g. shell) |
${result} | typically permit or deny |
${device.port}${port} [deprecated] | NAS port (console, tty, ...) |
${hint} | added/replaced for authorization, informal text for accounting |
${device.name}, ${host} [deprecated] | Device name of matching device declaration |
${client.dnsname}, ${nac-name} [deprecated] | Client DNS reverse mapping |
${device.dnsname}, ${nas-name} [deprecated] | Device DNS reverse mapping |
${msgid} | A message ID, perhaps suitable for RFC5424 logs. These are listed somewhere below. |
${type} | packet type (authen/author/acct) |
${accttype} | accounting type (start/stop/update) |
${priority} | syslog priority |
${action} | authentication info (e.g. pap login) |
${privlvl} | privilege level |
${authen-action} | login or chpass |
${authen-type} | Authentication packet type, e.g. AUTHEN/PASS, AUTHEN/FAIL |
${authen-service} | asciiascii/pap/chap/mschap/mschapv2 |
${authen-method} | krb5/line/enable/local/tacacs+/guest/radius/krb4/rcmd |
${rule} | Name of the matching rule. |
${label} | Ruleset label, if any. |
${config-file} | Configuration file name |
${config-line} | Configuration file line number |
${context} | Context variable (set via context = ...) |
${vrf} | Name of the current socket IPv4 vrf, supported on Linux (requires sysctl net.ipv4.tcp_l3mdev_accept=1) and possibly OpenBSD. |
${realm} | realm name |
${uid} | UID from PAM backend |
${gid} | GID from PAM backend |
${gids} | GIDs from PAM backend |
${home} | Home directory from PAM backend |
${shell} | Shell from PAM backend |
${dn} | Raw dn backend value, typically from LDAP |
${identity-source} | The IDENTITY_SOURCE backend value (the identitySourceName of the originating MAVIS module) |
${log-sequence} | Log sequence number |
${mavis.latency} | The milliseconds it took the MAVIS backend to answer a request |
${memberof} | Raw memberOf backend value, typically from LDAP |
${pid} | Process ID |
${server.name}, ${hostname} [deprecated] | Server host name |
${server.address} | Server address |
${server.port} | Server TCP port |
${session.id} | Session id |
${tls.conn.version} | TLS Connection Version (requires LibTLS or OpenSSL) |
${tls.conn.cipher} | TLS Connection Cipher (requires LibTLS or OpenSSL) |
${tls.peer.cert.issuer} | TLS Peer Certificate Issuer (requires LibTLS or OpenSSL) |
${tls.peer.cert.subject} | TLS Peer Certificate Subject (requires LibTLS or OpenSSL) |
${tls.conn.cipher.strength} | TLS Connection Cipher Strength (requires LibTLS or OpenSSL) |
${tls.peer.cn} | TLS peer certificate Common Name (requires LibTLS or OpenSSL) |
${tls.psk.identity} | TLS PSK identity (requires OpenSSL) |
The built-in defaults as of writing this are:
# Accounting to file/pipe: "%Y-%m-%d %H:%M:%S %z\t${nas}\t${user}\t${port}\t${nac}\t${accttype}\t${service}\t${cmd}\n" # Accounting to UDP syslog: "<${priority}>%Y-%m-%d %H:%M:%S %z ${hostname} ${nas}|${user}|${port}|${nac}|${accttype}|${service}|${cmd}" # Accounting to syslog(3): "${nas}|${user}|${port}|${nac}|${accttype}|${service}|${cmd}" # Authorization to file/pipe: "%Y-%m-%d %H:%M:%S %z\t${nas}\t${user}\t${port}\t${nac}\t${profile}\t${result}\t${service}\t${cmd}\n" # Authorization to UDP syslog: "<${priority}>%Y-%m-%d %H:%M:%S %z ${hostname} ${nas}|${user}|${port}|${nac}|${profile}|${result}|${service}|${cmd}" # Authorization to syslog(3): "${nas}|${user}|${port}|${nac}|${profile}|${result}|${service}|${cmd}" # Authentication to file/pipe: "%Y-%m-%d %H:%M:%S %z\t${nas}\t${user}\t${port}\t${nac}\t${action} ${hint}\n" # Authentication to UDP syslog: "<${priority}>%Y-%m-%d %H:%M:%S %z ${hostname} ${nas}|${user}|${port}|${nac}|${action} ${hint}" # Authentication to syslog(3): "${nas}|${user}|${port}|${nac}|${action} ${hint}" # Connections to file/pipe: "%Y-%m-%d %H:%M:%S %z\t${accttype}\t${nas}\t${tls.conn.version}\t${tls.peer.cert.issuer}\t${tls.peer.cert.subject}\n" # Connections to UDP syslog: "<${priority}>%Y-%m-%d %H:%M:%S %z ${hostname} ${accttype}|${nas}|${tls.conn.version}|${tls.peer.cert.issuer}|${tls.peer.cert.subject}" # Connections to syslog(3): "${accttype}|${nas}|${tls.conn.version}|${tls.peer.cert.issuer}|${tls.peer.cert.subject}"
Message ID | Description |
---|---|
AUTHZPASS | authorization succeeded |
AUTHZPASS-ADD | authorization succeeded, attribute-value-pairs were added |
AUTHZPASS-REPL | authorization succeeded, attribute-value-pairs were replaced |
AUTHZFAIL | authorization failed |
AUTHCFAIL | generic authentication failure |
AUTHCFAIL-ABORT | authentication was aborted |
AUTHCFAIL-BACKEND | the authentication backend failed |
AUTHCFAIL-BUG | authentication failed due some programming error |
AUTHCFAIL-DENY | authentication was denied |
AUTHCFAIL-WEAKPASSWORD | the password used didn't met minimum criteria |
AUTHCFAIL-ACL | access was denied due to ruleset or acl |
AUTHCFAIL-DENY-RETRY | the user tried the same wrong password once more |
AUTHCFAIL-PASSWORD-NOT_TEXT | the password isn't specified as clear-text |
AUTHCFAIL-BAD-CHALLENGE-LENGTH | the MSCHAP challenge length didn't match |
AUTHCFAIL-NOPASS | there's no passwort set for the user |
AUTHCPASS | authentication passed |
ACCT-START | accounting start |
ACCT-STOP | accounting stop |
ACCT-UNKNOWN | unknown (non-compliant) accounting data |
ACCT-UPDATE | accounting update/watchdog |
CONN-REJECT | connection was rejected |
CONN-START | connection was started |
CONN-STOP | connection was terminated |
time zone = time-zone
By default, the daemon uses your local system time zone to convert the internal system time to calendar time. This option sets the TZ environment variable to the time-zone argument. See your local tzset man page for details.
umask = mode
This sets the file creation mode mask. Example:
umask = 0640
All accounting records are written, as text, to the file (or command) specified with the accounting log directive.
Accounting records are text lines containing tab-separated fields. The first 6 fields are always the same. These are:
timestamp
NAS address
username
port
NAC address
record type
Following these, a variable number of fields are written, depending on the accounting record type. All are of the form attribute=value. There will always be a task_id field.
Attributes, as sent by the NAS, might be:
unknown service start_time port elapsed_time status priv_level cmd protocol cmd-arg bytes_in bytes_out paks_in paks_out address task_id callback-dialstring nocallback-verify callback-line callback-rotary
More may appear,. randomly..
Example records (lines wrapped for legibility) are thus:
1995-07-13 13:35:28 -0500 172.16.1.4 chein tty5 198.51.100.141 stop task_id=12028 service=exec port=5 elapsed_time=875 1995-07-13 13:37:04 -0500 172.16.1.4 lol tty18 198.51.100.129 stop task_id=11613 service=exec port=18 elapsed_time=909 1995-07-13 14:09:02 -0500 172.16.1.4 billw tty18 198.51.100.152 start task_id=17150 service=exec port=18 1995-07-13 14:09:02 -0500 172.16.1.4 billw tty18 198.51.100.152 start task_id=17150 service=exec port=18
Elapsed time is in seconds, and is the field most people are usually interested in.
The script tacspooflog-ng.pl (which comes bundled with this distribution, have a look at the tac_plus-ng/extra/ directory) may be used to make syslogd believe that logs come straight from your router, not from tac_plus-ng.
E.g., if your syslogd is listening on 127.0.0.1, you may try:
access log = "|exec sudo /path/to/tacspooflog-ng.pl 127.0.0.1"
This may be useful if you want to keep logs in a common place.
In contrast to the older tacspooflog.pl script tacspooflog-ng.pl will handle both IPv4 and IPv6 addresses and has more configuration options.
Please read tac_plus-ng/extra/tacspooflog-ng.README too, if you're thinking about using this script.
User messages, e.g. the Username prompt, can be customized, both at device and realm level:
message USERNAME = "Utilisateur: "
Supported messages and their defaults:
ID | Default value |
---|---|
ACCOUNT_EXPIRES | "This account will expire soon." |
BACKEND_FAILED | "Authentication backend failure." |
CHANGE_PASSWORD | "Please change your password." |
DENIED_BY_ACL | "Denied by ACL" |
ENABLE_PASSWORD | "Enable Password: " |
PASSWORD | "Password: " |
PASSWORD_ABORT | "Password change dialog aborted." |
PASSWORD_AGAIN | "Retype new password: " |
PASSWORD_CHANGE_DIALOG | "Entering password change dialog" |
PASSWORD_CHANGED | "Password change succeeded." |
PASSWORD_EXPIRED | "Password has expired." |
PASSWORD_EXPIRES | "Password will expire on %c." (fed to strftime(3)) |
PASSWORD_INCORRECT | "Password incorrect." |
PASSWORD_MINREQ | "Password doesn't meet minimum requirements." |
PASSWORD_NEW | "New password: " |
PASSWORD_NOMATCH | "Passwords do not match." |
PASSWORD_OLD | "Old password: " |
PERMISSION_DENIED | "Permission denied." |
RESPONSE | "Response: " |
RESPONSE_INCORRECT | "Response incorrect." |
USERNAME | "Username: " |
USER_ACCESS_VERIFICATION | "User Access Verification" |
A number of global limits and timeouts may be specified at realm and global level:
connection timeout = s
Terminate a connection to a NAS after an idle period of at least s seconds.
Default: 600
context timeout = s
Clears context cache entries after s seconds of inactivity. Default: 3600 seconds.
Default: 3600
This configuration will be accepted at realm level, too.
warning period = d
Set warning period for password expiry to d days.
Default: 14
max-rounds = n
This sets an upper limit on the number of packet exchanges per session. Default: 40, acceptable range is from 1 to 127.
password acl = acl
password acl may be used to perform simple compliance checks on user passwords. For example, to enforce a minimum password length of 6 characters you may try
acl password-compliance { if (password =~ /^....../) permit deny } password acl = password-compliance
Authentications using passwords that fail the check will be rejected.
password max-attempts = integer
The max-attempts parameter limits the number of Password: prompts per TACACS+ session at login. It currently defaults to 1, meaning that a typical login sequence with bad passwords would look like:
> telnet 10.0.0.2 Trying 10.0.0.2... Connected to 10.0.0.2. Escape character is '^]'. Welcome. Authorized Use Only. Username: admin Password: *** Password incorrect. Welcome. Authorized Use Only. Username: admin Password: **** Password incorrect. Welcome. Authorized Use Only. Username: admin Password: * Password incorrect. Connection closed by foreign host.
Using, for example,
password max-attempts = 3
(the actual default in earlier versions was 4) would change this dialog to:
> telnet 10.0.0.2 Trying 10.0.0.2... Connected to 10.0.0.2. Escape character is '^]'. Welcome. Authorized Use Only. Username: admin Password: *** Password incorrect. Password: **** Password incorrect. Password: ***** Password incorrect. Go away. Welcome. Authorized Use Only. Username:
It's at the NAS's discretion to restart the authentication dialog with a new TACACS+ session or to close the (Telnet/SSH/...) session to the user if TACACS+ authentication fails.
Thid directive can be used at device level, too.
anonymous-enable = ( permit | deny )
Several broken TACACS+ implementations send no or an invalid username in enable packets. Setting this option to deny tries to enforce user authentication before enabling. This option defaults to permit.
Alas, this may or may not work. In theory, the enable dialog should look somewhat like:
Router> enable Username: me Password: ******* Enable Password: ********** Router#
However, some implementations may resend the user password at the Enable Password: prompt. In that case you've got only two options: Either try
enable = login
at user profile level, which will omit the secondary password query and let the user enable with his login password, or permit anonymous enable (which is disabled by default) with
anonymous-enable = permit
in device context to use the enable passwords defined there.
augmented-enable = ( permit | deny )
For outdated TACACS+ client implementations that send $enable$ instead of the real username in an enable request, this will permit user specific authentication using a concatenation of username and login password, separated with a single space character:
> enable Password: myusername mypassword #
enable [ level ] = login needs to be set in the users' profile for this option to take effect.
Default: augmented-enable = deny
augmented-enable will only take effect if the NAS tries to authenticate a username matching the regex
^\$enab..\$$
(e.g.: $enable$, $enab15$). That matching criteria may be changed using an ACL:
acl custom_enable_acl { if (user =~ ^demo$) permit deny } enable user acl = custom_enable_acl
There are also experimental options for (non-standard) SSH public key authentication available. These may or may not supported by your vender:
ssh-key = public-ssh-key-in-OpenSSL-authorized_keys-format
Example: ssh-key = "AAAAB3NzO4S6C/SAu9E90P3n9dfbe3iNiK...STPC6V1fffa123OxmK3hhzwbl"
ssh-key-hash = ssh-key-in-OpenSSL-format
There's no use in specifying the hash if you've configured the public key, the daemon will care for that itself.
Example: ssh-key-hash = SHA256:kOkclqivcjludf/jdsfkyqpddffdk38U12+CkA8fBAC
These options are relevant for configuring the MAVIS user back-end:
pap password [ default ] = ( login | pap )
When set to login, the PAP password default for new users will be set to use the login password.
pap password mapping = ( login | pap )
When set to login, PAP authentication requests will be mapped to ASCII Login requests. You may wish to uses this for NEXUS devices.
May be overridden at device level.
user backend = mavis
Get user data from the MAVIS back-end. Without that directive, only locally defined users will be available and the MAVIS back-end may be used for authenticating known users (with password = mavis or simlar) only.
pap backend = mavis [ prefetch ]
Verify PAP passwords using the MAVIS back-end. This needs to be set to either mavis or prefetch in order to authenticate PAP requests using the MAVIS back-end. If unset, the PAP password from the users' profile will be used.
If prefetch is specified, the daemon will first retrieve the users' profile from the back-end and then authenticate the user based on information eventually found there.
This directive implies user backend = mavis.
login backend = mavis [ prefetch ] [ chalresp [ noecho ] ] [ chpass ]
Verify Login passwords using the MAVIS back-end. This needs to be set to either mavis or prefetch in order to authenticate login requests using the MAVIS back-end. If unset, the login password from the users' profile will be used.
If prefetch is specified, the daemon will first retrieve the users' profile from the back-end and then authenticate the user based on information eventually found there.
This directive implies user backend = mavis.
For use with OPIE-enabled MAVIS modules, add the chalresp keyword (and, optionally, add noecho, unless you want the typed-in response to display on the screen). Example:
login backend = mavis chalresp noecho
For non-local users, if the chpass attribute is set and the user provides an empty password at login, the user is given the option to change his password. This requires appropriate support in the MAVIS back-end modules.
mavis module module { ... }
Load MAVIS module module. See the MAVIS documentation for configuration guidance.
mavis path = path
Add path to the search-path for MAVIS modules.
mavis cache timeout = s
Cache MAVIS authentication data for s seconds. If s is set to a value smaller than 11, the dynamic user object is valid for the current TACACS+ session only. Default is 120 seconds.
mavis noauthcache
Disables password caching for MAVIS modules.
mavis user filter = acl
Query MAVIS user back-end only if acl matches. Defaults to:
acl __internal__username_acl__ { if (user =~ "[]<>/()|=[]+") deny permit } mavis user filter = __internal__username_acl__
TACACS+-over-TLS is not a standard. These features are experimental.
If compiled with OpenSSL or LibTLS support the following configuration options are available:
tls cert-file = cert-file
Specifies the public part of a TLS server certificate in PEM format.
tls key-file = key-file
Specifies the private part (the key) of a TLS server certificate in PEM format.
tls passphrase = passphrase
Specifies the optional passphrase to decrypt key-file.
tls accept expired = ( yes | no )
Accept expired certificates.
tls verify-depth = depth
Sets TLS verification depth.
tls cafile = cafile
Specifies a file with the CAs to use.
tls alpn = ALPN-Protocol-ID
There's currently no ALPN Protocol ID registered for TACACS-over-TLS, the official list is here: TLS Application-Layer Protocol Negotiation (ALPN) Protocol IDs.
tls auto-detect = ( yes | no )
Enable TLS auto-detection. Defaults to no.
If compiled with OpenSSL support, TLSv1.3 Preshared Keys and SNIs are supported:
tls psk = ( yes | no )
This enables PSK support at realm level.
PSK identity and key can be declared at device level:
tls psk id = myid tls psk key = 0123456789abcdef # in hex
tls sni = SNI
This adds SNI to the server name list of the current realm. A TLS connection requesting SNI will automatically be mapped to that realm.
Example:
id = spawnd { listen { port = 4949 realm = heck } listen { port = 4950 realm = heck tls = yes } spawn { instances min = 1 instances max = 32 } id = tac_plus-ng { ... realm heck { tls cert-file = /somewhere/tac-ca/server.tacacstest.crt tls key-file = /somewhere/tac-ca/server.key tls ca-file = /somewhere/tac-ca/ca.crt ... } } }
In realm context:
haproxy auto-detect = ( yes | no )
Enable HAProxy protocol v2 auto-detection. Defaults to no.
In spawnd listen context,
haproxy = ( yes | no )
will will tell tac_plus-ng to auto-detect that a connection is proxied via HAProxy protocol 2.
A suitable HAProxy configuration could look similar to:
frontend tacplus bind *:49 mode tcp default_backend backendtacplus backend backendtacplus balance source server tacserver1 127.0.0.1:4949 no-check send-proxy-v2
tls = ( yes | no )
will tell tac_plus-ng whether the connection is TLS encrypted.
vrf = ( vrf-name | vrf-number )
will tell spawnd listen to bind(2) to the requested VRF (vrf-name on Linux, vrf-number on OpenBSD).
Example:
id = spawnd { ... listen { port = 49 vrf = vrf-blue tls = true haproxy = true } .... }
Realms inherit quite some configuration from their parent realm:
Declaration of ... | is taken from parent realm ... |
---|---|
acl | if not found in current realm |
dns forward mapping | if not found in current realm |
group | if not found in current realm |
device (IP lookup) | if no device defined in current realm |
device (name lookup) | if not found in current realm |
log | always |
mavis module | if not set and no users defined in current realm |
network | if not found in current realm |
profile | if not found in current realm |
ruleset | if not set or undefined result in current realm |
timespec | if not found in current realm |
user | if not found in current realm |
Networks consist of IP addresses or other networks. They may overlap. Networks can be used in ACLs. The parent of a network may be set either implicitly (by defining it it parent context) or explicitly.
net home { address = 172.16.0.0/23 net dev { address = 172.16.0.15 } parent = ... }
The daemon will talk to known NAS addresses only. Connections from unknown addresses will be rejected.
If you want tac_plus-ng to encrypt its packets (and you almost certainly do want this, as there can be usernames and passwords contained in there), then you'll have to specify an (non-empty) encryption key. The identical key must also be configured on any NAS which communicates with tac_plus.
To specify a global key, use a statement similar to
device world4 { key = "your key here" address = 0.0.0.0/0 }
(where world is not a keyword, but just some arbitrary character string).
Double Quotes | |
---|---|
You only need double quotes on the daemon if your key contains spaces. Confusingly, even if your key does contain spaces, you should never use double quotes when you configure the matching key on the NAS. The daemon will reject connections from devices that have no encryption key defined. Double quotes within double-quoted strings may be escaped using the backslash character \ (which can be escaped by itself), e.g.: key = "quo\\te me\"." translates to the ASCII sequence quo\te me". |
Any CIDR range within a device definition needs to to be unique, and the most specific definition will match. The requirement for unambiguousness is quite simply based on the fact that certain device object attributes (key, prompt, enable passwords) may only exist once.
If compiled with TLS support, primary criteria for device object selection with TLS is no longer the NAS IP address but the certificate DNS SANs (OpenSSL only), the subject and/or the common name. E.g., CN=server.tacacstest.demo,OU=org,OU=local will check for device objects named CN=server.tacacstest.demo,OU=org,OU=local, OU=org,OU=local, OU=local and then for server.tacacstest.demo, tacacstest.demo and demo before falling back to IP based selection.
On the NAS, you also need to configure the same key. Do this by issuing the current variant of:
aaa new-model tacacs-server host 192.168.0.1 single-connection key your key here
The optional single-connection parameter specifies that multiple sessions may use the same TCP/IP connection to the server.
Generally, the syntax for device declarations conforms to
device name { key-value pairs }
The key-value pairs permitted in device sections of the configuration file are explained below.
key [warn] ( YYYY-MM-DD | s ) ] = string
This sets the key used for encrypting the communication between server and NAS. Multiple keys may be set, making key migration from one key to another pretty easy. If the warn keyword is specified, a warning message is logged when a NAS actually uses the key. Optionally, the warn keyword accepts a date argument that specifies when the warnings should start to appear in the logs.
During debugging, it may be convenient to temporarily switch off encryption by using an empty key:
key = ""
Be careful to remember to switch encryption back on again after you've finished debugging.
address = cidr
Adds the address range specified by cidr to the current device definition.
address file = file
Add the addresses from file to the current device definition. Shell wildcard patterns are expanded by glob(3).
single-connection ( may-close ) = ( yes | no )
This directive may be used to permit or deny the single-connection feature for a particular device object. The may-close keyword tells the daemon to close the connection if it's unused.
Caveat Emptor | |
---|---|
There's a slight chance that single-connection doesn't work as expected. The single-connection implementation in your router or even the one implemented in this daemon (or possibly both) may be buggy. If you're noticing weird AAA behaviour that can't be explained otherwise, then try disabling single-connection on the router. |
This configuration will be accepted at realm level, too.
parent = deviceName
This sets the the parent devices. Definitions not found in the current device will be looked up there, recursively.
device deviceName { DeviceAttr }
Devices can be defined in device context, too.
script { tacAction }
Scripts can be used in device context. These are run before AAA and mey be used to permit or deny access, or to rewrite usernames.
This configuration will be accepted at realm level (for the default host, too.
The connection timeout may be specified:
connection timeout = s
Terminate a connection to this NAS after an idle period of at least s seconds. Defaults to the global option.
The following authentication related directives are available at device object level:
pap password mapping = ( login | pap )
When set to login, PAP authentication requests will be mapped to ASCII Login requests. You may wish to uses this for NEXUS devices.
enable [ level ] = ( permit | deny | login | ( clear | crypt) password )
This directive may be used to set device specific enable passwords, to use the login password, or to permit (without password) or refuse any enable attempt. level defaults to 15.
Enable passwords specified at device level have a lower precedence as those defined at user or profile level.
Password Hashes | |
---|---|
You can use the openssl passwd utility to compute password hashes. |
You can enable via TACACS+ by configuring on the NAS:
aaa authentication enable default group tacacs+ enable
anonymous-enable = ( permit | deny )
Several broken TACACS+ implementations send no or an invalid username in enable packets. Setting this option to deny enforces user authentication before enabling. Setting this option here has precedence over the global option.
This configuration will be accepted at realm level, too.
augmented-enable = ( permit | deny )
For TACACS+ client implementations that send $enable$ instead of the real username in an enable request, this will permit user specific authentication using a concatenation of username and login password, separated with a single space character. Setting this option here has precedence over the global option.
enable [ level ] = login needs to be set in the users' profile for this option to take effect.
This configuration will be accepted at realm level, too.
password max-attempts = integer
The max-attempts parameter limits the number of Password: prompts per TACACS+ session at login. It currently defaults to 1.
This configuration will be accepted at realm level, too.
password expiry warning = number
A password expiry warning will be displayed to the user if less than number time is left. Appending s (that's the default) or any of m, h, d, w will scale the number up as expeced.
This configuration will be accepted at realm level, too.
The following authorization related directives are available at device object level:
permit if-authenticated = ( yes | no )
This will cause authorization for users unknown to the daemon to succeed (e.g. when logging in locally while the daemon is down or while initially configuring TACACS+ support and messing up).
This configuration will be accepted at realm level, too.
The daemon allows for various banners to be displayed to the user:
welcome banner ( fallback ) = string
motd banner = string
reject banner = string
The reject banner gets displayed in place of the welcome message if a connection was rejected by an access ACL defined at device, user or group level.
These configurations will be accepted at realm level, too.
failed authentication banner = string
The failed authentication banner gets displayed upon final failure of an authentication attempt.
message = string
The time when those texts get displayed largely depends on the actual login method:
Context | Directive | Telnet | SSHv1 | SSHv2 |
---|---|---|---|---|
device | welcome banner |
displayed before Username: |
not displayed |
displayed before Password: |
device | reject banner | displayed before closing connection | not displayed | not displayed |
device | motd banner | displayed after successful login | not displayed | displayed after successful login |
user or group | message |
displayed after motd banner |
not displayed |
displayed after motd banner |
Neither the motd banner nor a message defined in the users' profile will be displayed if hushlogin is set for the user.
Both banners and messages support the same conversions as logs, unless specified as user level.
Example:
device ... { ... welcome banner = "Welcome. Today is %A.\n" ... }
The directive
bug compatibility = value
may improve compatibility with clients that violate the TACACS+ protocol. Currently, the following bit values (yes, you can use bitwise OR here) are recognized:
Bit | Value | Description |
---|---|---|
0 | 1 | According to RFC8907 the data field should be ignored for ASCII authentications. Alas, IOS-XR puts the password exactly there. Set this if required. |
1 | 2 | Accept version 1 for authorization and accounting packets, seen with Palo Alto systems. |
2 | 4 | Accept key-based packet obfuscation for TLS (this violates draft-ietf-opsawg-tacacs-tls13-03.txt). |
3 | 8 | Accept TACACS+ payloads lower than advertized in the TACACS+ header. |
Example:
device ... { ... bug compatibility = 2 ... }
This configuration will be accepted at realm level, too.
For address based device lookups, the daemon looks for the most specific device definition. Values that aren't defined (if any) will be lookup up in the device's parent, which may be either set implicitely by defining a device in the context of it's parent device, or expliitely, using the parent statement.
device = customer1 { address = 10.0.0.0/8 key = "your key here" welcome banner = "\nHitherto shalt thou come, but no further. (Job 38.11)\n\n" enable 15 = clear whatever } device = test123 { address = 10.1.2.0/28 address = 10.12.1.30/28 address = 10.1.1.2 # key/banners/enable will be inherited from 10.0.0.0/8 by default, # unless you specify "inherit = no" address file = /some/path/test123.cidr welcome banner = "\nGo away.\n\n" }
timespec objects may be used for time based profile assignments. Both cron and Taylor-UUCP syntax are supported; see you local crontab(5) and/or UUCP man pages for details. Syntax:
timespec = timespec_name { "entry" [ ... ] }
Example:
# Working hours are from Mo-Fr from 9 to 16:59, and # on Saturdays from 9 to 12:59: timespec workinghours { "* 9-16 * * 1-5" # or: "* 9-16 * * Mon-Fri" "* 9-12 * * 6" # or: "* 9-12 * * Sat" } timespec sunday { "* * * * 0" } timespec example { Wk2305-0855,Sa,Su2305-1655 Wk0905-2255,Su1705-2255 Any }
Access Control Lists (or, more exactly, Access Control Scripts) are the main component of ruleset evaluation.
Scripts may currently be used for ACLs, in host and profile declaration scope and in rule sets. If a script in a hierarchy doesn't return a final verdict (these are permit and deny), other scripts in the hierarchy may be evaluated. Default evaluation order is
script-order host = bottom-up script-order realm = bottom-up script-order profile = bottom-up
but you may prefer to change that to top-down to have parent scripts executed first.
To provide an example for that: In
profile A { script { ... } profile B { script { ... } }
the script part from A will by default (bottom-up be evaluated, if the B script result isn't final.
In contrast, for
script-order profile = top-down profile A { script { ... } profile B { script { ... } }
the A part takes precedence and the B script will one be evaluated if the A result isn't final.
skip parent-script = yes may be used (at profile, host and realm level) to ignore scripts defined at a higher hierarchy level.
Scripting examples:
acl acl_name { tac_action ... }
Example:
acl myacl123 { if (nas == 1.2.3.4 || nac = SomeHostName || nac-dns =~ /\\.example\\.com$/) deny }
script = { tac_action ... }
Example:
profile tunnelAdmin { script { if (service == shell) { if (cmd == "") permit # required for shell startup if (cmd =~ /^(no\s)?shutdown\s/) permit if (cmd =~ /^interface Tunnel/) permit deny } } } user joe { password = ... member = ops } ruleset { rule opsRule { script { if (group == ops) profile = tunnelAdmin permit } } }
A script consists of a series of actions:
The actions return, permit and deny are final. At the end of a script, return is implied, at which the daemon continues processing the configured cmd statements in shell context) or standard ACLs (in ACL context). The assignment operations (context =, message =) do make sense in shell context only.
Setting the context variable makes sense in shell context only. See the example in the corresponding section.
Attribute-related directives are:
default attribute = ( permit | deny )
This directive specifies whether the daemon is to accept or reject unknown attributes sent by the NAS (default: deny).
( set | add | optional ) attribute = value
Defines mandatory and optional attribute-value pairs:
set unconditionally returns a mandatory AV pair to the NAS
optional returns a NAS-requested (and perhaps modified) optional AV pair to the NAS unless the attribute was already in the mandatory list
add returns an optional AV pair to the client even if the client didn't request it (and it was neither in the mandatory nor optional list)
set priv-lvl = 15
For a detailed description on mandatory and optional AV-pairs, see the "The Authorization Algorithm" section somewhere below.
Numbered Attributes | |
---|---|
A %%d added to an attribute will will result in a numbered attribute, starting to count at 1 (%%n would start counting at 0). For example, set route#%%d = "192.168.0.0 255.255.255.0 10.0.0.1" set route#%%d = "192.168.1.0 255.255.255.0 10.0.0.1" set route#%%d = "192.168.2.0 255.255.255.0 10.0.0.1" results in set route#1 = "192.168.0.0 255.255.255.0 10.0.0.1" set route#2 = "192.168.1.0 255.255.255.0 10.0.0.1" set route#3 = "192.168.2.0 255.255.255.0 10.0.0.1" |
Variables | |
---|---|
The same variables supported for logging can be used as attribute values, too. Example: set uid = "${uid}" |
return
Use the current service definition as-is. This stops the daemon from checking for the same service in the groups the current user (or group) is a member of.
Conditions:
Left-hand side | Operators | Right-hand side | Comment | |
---|---|---|---|---|
"string" | == != =~ !~ | String or REGEX | Log variable substitutions will apply | |
acl | == != | ACL object name | ||
arg[attr] | == != =~ !~ | String or REGEX | arg[protocol], arg[service], ... | |
authen-action | == != =~ !~ | String or REGEX | login, chpass | |
authen-method | == != =~ !~ | String or REGEX | login, enable, ppp | |
authen-service | == != =~ !~ | String or REGEX | none, line, enable, local, tacacs+ | |
authen-type | == != =~ !~ | String or REGEX | ascii, pap, chap, mschap, mschapv2, sshkey, sshcer | |
client | == != =~ !~ | Net object name or IP address or string. | Net incl. parents | |
client.name | == != | Net object name | Client remote address | |
client.address | == != =~ !~ | String or REGEX | Client remote address | |
client.dnsname | == != =~ !~ | Client DNS PTR record | ||
cmd | == != =~ !~ | String or REGEX | shell command line | |
context | == != =~ !~ | String or REGEX | current exec context | |
device | == != | Device (host) object name, net object name or device address | incl. parents | |
device.name | == != =~ !~ | Device (host) or net object name | incl. parents | |
device.address | == != =~ !~ | Device (host) address | ||
device.dnsname | == != =~ !~ | Device (host) DNS PTR record | ||
device.tag | == != =~ !~ | tagName | ||
device.tag | == != | user.tag | False if no match, true else. | |
dn | == != =~ !~ | String or REGEX | MAVIS dn attribute | |
identity-source | == != =~ !~ | String or REGEX | authenticating/authorizing MAVIS module | |
member | == != =~ !~ | String or REGEX | group membership | |
memberof | == != =~ !~ | String or REGEX | MAVIS memberOf attribute | |
nac | == != =~ !~ | [deprecated, use client] String or REGEX | net object name (incl. parents), client remote address (rem_addr) | |
nac-name | == != =~ !~ | [deprecated, use client.name] String or REGEX | client DNS PTR | |
nas | == != =~ !~ | [deprecated, use device] String or REGEX | device or net object name (incl. parents), matches NAC remote address (rem_addr) | |
nas-name | == != =~ !~ | [deprecated, use device.name] String or REGEX | NAS/NAD DNS PTR | |
password | == != =~ !~ | String or REGEX | session user password | |
port | == != =~ !~ | [deprecated, use device.port] String or REGEX | session port (vty02, console, ...) | |
priv-lvlp | == != =~ !~ | String or REGEX | session privilege level (0 ... 15) reported by the device | |
protocol | == != =~ !~ | String or REGEX | session protocol (ppp, ...) | |
realm | == != | Realm object name | ||
server.address | == != =~ !~ | Server address | ||
server.name | == != =~ !~ | Server host name | ||
server.port | == != =~ !~ | Server TCP port | ||
service | == != =~ !~ | String or REGEX | session service (shell, ...) | |
time | == != | Timespec object name | ||
tls.conn.cipher | == != =~ !~ | String or REGEX | TLS specific data | |
tls.conn.cipher.strength | == != =~ !~ | String or REGEX | TLS specific data | |
tls.conn.version | == != =~ !~ | String or REGEX | TLS specific data | |
tls.peer.cert.issuer | == != =~ !~ | String or REGEX | TLS specific data | |
tls.peer.cert.subject | == != =~ !~ | String or REGEX | TLS specific data | |
tls.peer.cn | == != =~ !~ | String or REGEX | TLS specific data | |
tls.psk.identity | == != =~ !~ | String or REGEX | TLS specific data | |
type | == != =~ !~ | String or REGEX | authen, author, acct | |
user | == != =~ !~ | String or REGEX | session user | |
user.tag | == != =~ !~ | tagName | ||
user.tag | == != | device.tag | False if no match, true else. |
Railroad diagrams for conditions:
cmd and context may be used in shell context only. tls_* conditions require libtls.
A script may refer to a rewrite profile defined at realm level to rewrite user names. For example, the following will map both admin and root to jane.doe, and convert all other usernames to lower-case:
rewrite rewriteRule { rewrite /^admin$/ jane.doe rewrite /^root$/ jane.doe rewrite /^.*$/ \L$0 } device ... { ... script { rewrite user = rewriteRule } ... }
You can limit the usage of a rewritten-to user with the rewritten-only directive, e.g.:
rewrite rewriteRule { rewrite /^.*$/ nopassword } user nopassword { password login = permit password pap = login member = ... rewritten-only }
The basic form of a user declarations is
user username { ... }
A user or group declaration may contain key-value pairs and service declarations.
The following declarations are valid in user context only:
alias = alternateUserName
Implement an alternate user name. This may be used multiple times.
password login [ fallback ] = ( ( clear | crypt ) password | mavis | permit | deny )
The login password authenticates shell log-ins to the server.
password login = crypt aFtFBT4e5muQE password login = clear Ci5c0
For the argument after crypt you may use whatever hashes your crypt(3) implementation supports.
If the mavis keyword is used instead, the password will be looked up via the MAVIS back-end. It will not be cached. This functionality may be useful if you want to authenticate at external systems, despite static user declarations in the configuration file.
If you're using password login = mavis, thefallback password will be used if there's a MAVIS backend error.
password pap [ fallback ] = ( ( clear | crypt ) password | login|mavis | permit | deny )
The pap authenticates PAP log-ins to the server. Just like with login, the password doesn't need to be in clear text, but may be hashed, or may be looked up via the MAVIS back-end. You can even map pap to login globally by configuring password pap = login in realm context.
If you're using password pap = mavis, thefallback password will be used if there's a MAVIS backend error.
password chap = ( clear password | permit | deny )
For CHAP authentication, a cleartext password is required.
password ms-chap = ( clear password | permit | deny )
For MS-CHAP authentication, a cleartext password is required.
password { ... }
This directive allows for nested specification of passwords. Example:
user marc { password { login = clear myLoginPassword pap = clear myPapPassword } }
enable [ level ] = ( permit | deny | login | ( clear | crypt ) password )
This directive may be used to set user specific enable passwords, to use the login password, or to permit (without password) or refuse any enable attempt. Enable secrets defined at user level have precedence over those defined at device level. level defaults to 15.
The default privilege level for an ordinary user on the NAS is usually 1. When a user enables, she can reset this level to a value between 0 and 15 by using the NAS enable command. If she doesn't specify a level, the default level she enables to is 15.
message = string
A message displayed to the user upon log-in.
hushlogin = ( yes | no )
Setting hushlogin to yes keeps the daemon from displaying motd and user messages upon login.
valid from = ( YYYY-MM-DD | s )
The user profile will be valid starting at the given date, which can be specified either in ISO8601 date format or as in seconds since January 1, 1970, UTC.
valid until = ( YYYY-MM-DD | s )
The user profile will be invalid after the given date.
member = groupOne[,groupTwo]*
This specifies group membership. A user can be a member of multiple groups and groups can be members of a parent group.
A user can be a member of multiple groups. A user that is a member of a group that comes with a parent group is a member of the latter, too. Group are defined using
group groupname { ... }
The following key-value pairs are valid for groups:
member = groupOne[,groupTwo]*
This specifies group membership.
parent = groupName
The parent of a group can be set explicitly.
group groupName { GroupAttr }
Groups may be parents of other groups.
Profiles are collections of services that can be assigned to users via the policy rule-set. Syntax is
profile profileName { profileAttr }
Also, an unnnamed profile may be configured in user context. This overrides rule evaluation and will be assigned unconditionally, e.g.:
user ... { ... profile { script { if (service == shell) set priv-lvl = 15 permit } } ... }
Assigning an existing profile to an user will also work:
user ... { ... profile = profileName{ ... }
Profiles are collections of services available to a user. A couple of configuration attributes are service specific and only valid in certain contexts:
SHELL (EXEC) Service
Shell startup should have an appropriate script definition
script { if (service == "shell" && cmd == "") permit }
defined. Valid configuration directive within the curly brackets are:
script { tacAction }
Commands can be permitted or denied using script syntax:
script { if (service == "shell" && cmd == "") permit if (cmd =~ /^write term/) deny if (cmd =~ /^configure /) deny permit }
profile SubProfileName { ... }
This defines a new profile that inherits most values from its parent profile.
parent = ParentProfileName
This sets ParentProfileName as parent profile. Parent profiles defined in parent realms are accepted, too.
Have a look at the authorization log in case you're unsure what commands and arguments the router actually sends for verification. E.g.,
Non-Shell Services
E.g. for PPP, protocol definitions may be used:
script { if (service == "ppp" && protocol == "ip") { set addr = 1.1.3.4 permit } }
The historical
default protocol = permit
will no longer be recognized but can be replaced with a simple
script { permit }
For a Juniper Networks-specific authorization service, use:
script { if (service == junos-exec) { set local-user-name = NOC # see the Junos documentation for more attributes } }
Likewise, for Raritan Dominion SX IP Console Servers:
script { if (service == dominionsx) { set port-list = "1 3 4 15" set user-type = administator # or operator, or observer } }
Quotes | |
---|---|
If your router expects double-quoted values (e.g. Cisco Nexus devices do), you can advise the parser to automatically add these: set shell:roles = "\"network-admin\"" and set shell:roles = '"network-admin"' } are equivalent, but the latter is more readable. |
MAVIS configuration is optional. You don't need it if you're content with user configuration in the main configuration file.
MAVIS back-ends may dynamically create user entries, based, e.g., on LDAP information.
For PAP and LOGIN,
pap backend = mavis login backend = mavis
in the global section delegate authentiation to the MAVIS sub-system. Statically defined users are still valid, and have a higher precedence.
By default, MAVIS user data will be cached for 120 seconds. You may change that period using
cache timeout = seconds
in the global configuration section.
Under certain circumstances you may wish to keep the user definitions in the plain text configuration file, but authenticate against some external system nevertheless, e.g. LDAP or RADIUS. To do so, just specify one of
login = mavis pap = mavis password = mavis
in the corresponding user definition.
User Authentication can be specified separately for PAP, CHAP, and normal logins. CHAP and global user authentication must be given in clear text.
The following assigns the user mary five different passwords for inbound and outbound CHAP, inbound PAP, outbound PAP, and normal login respectively:
user mary { password chap = clear "chap password" password pap = clear "inbound pap password" password login = crypt XQj4892fjk }
If
user backend = mavis
is configured in the global section, users not found in the configuration file will be looked up by the MAVIS back-end. You should consider using this option in conjuction with the more sophisticated back-ends (LDAP and ActiveDirectory, in particular), or whenever you're not willing to duplicate your pre-existing database user data to the configuration file. For users looked up by the MAVIS back-end,
pap backend = mavis
and/or
login backend = mavis
(again, in the global section of the configuration file) will cause PAP and/or Login authentication to be performed by the MAVIS back-end (e.g. by performing an LDAP bind), ignoring any corresponding password definitions in the users' profile.
If you just want the users defined in your configuration file to authenticate using the MAVIS back-end, simply set the corresponding PAP or Login password field to mavis (there's no need to add the user backend = mavis directive in this case):
user mary { login = mavis }
An entry of the form:
user lol { valid until = YYYY-MM-DD password login = clear "bite me" }
will cause the user profile to become invalid, starting after the valid until date. Valid date formats are both ISO8601 and the absolute number of seconds since 1970-01-01.
A expiry warning message is sent to the user when she logs in, by default starting at 14 days before the expiration date, but configurable via the warning period directive.
Complementary to profile expiry,
valid from = YYYY-MM-DD
activates a profile at the given date.
On the NAS, to configure login authentication, try
aaa new-model aaa authentication login default group tacacs+ local
(Alternatively, you can try a named authentication list instead of default. Please see the IOS documentation for details.)
Don't lock yourself out. | |
---|---|
As soon as you issue this command, you will no longer be able to create new logins to your NAS without a functioning TACACS+ daemon appropriately configured with usernames and password, so make sure you have this ready. As a safety measure while setting up, you should configure an enable secret and make it the last resort authentication method, so if your TACACS+ daemon fails to respond you will be able to use the NAS enable password to login. To do this, configure: aaa authentication login default group tacacs+ enable or, to if you have local accounts: aaa authentication login default group tacacs+ local If all else fails, and you find yourself locked out of the NAS due to a configuration problem, the section on recovering from lost passwords on Cisco's CCO web page will help you dig your way out. |
Authorization must be configured on both the NAS and the daemon to operate correctly. By default, the NAS will allow everything until you configure it to make authorization requests to the daemon.
On the daemon, the opposite is true: The daemon will, by default, deny authorization of anything that isn't explicitly permitted.
Authorization allows the daemon to deny commands and services outright, or to modify commands and services on a per-user basis. Authorization on the daemon is divided into two separate parts: commands and services.
Exec commands are those commands which are typed at a NAS exec prompt. When authorization is requested by the NAS, the entire command is sent to the tac_plus daemon for authorization.
Command authorization is configured by telling the ruleset to apply a profile to the user. See the Profile section for details.
Authorizing a single session can result in multiple requests being sent to the daemon. For example, in order to authorize a dialin PPP user for IP, the following authorization requests will be made from the NAS:
An initial authorization request to startup PPP from the exec, using the AV pairs service=ppp, protocol=ip, will be made (Note: this initial request will be omitted if you are autoselecting PPP, since you won't know the username yet).
This request is really done to find the address for dumb PPP (or SLIP) clients who can't do address negotiation. Instead, they expect you to tell them what address to use before PPP starts up, via a text message e.g. "Entering PPP. Your address is 1.2.3.4". They rely on parsing this address from the message to know their address.
Next, an authorization request is made from the PPP subsystem to see if PPP's LCP layer is authorized. LCP parameters can be set at this time (e.g. callback). This request contains the AV pairs service=ppp, protocol=lcp.
Next an authorization request to startup PPP's IPCP layer is made using the AV pairs service=ppp, protocol=ipcp. Any parameters returned by the daemon are cached.
Next, during PPP's address negotiation phase, each time the remote peer requests a specific address, if that address isn't in the cache obtained in step 3, a new authorization request is made to see if the peers requested address is allowable. This step can be repeated multiple times until both sides agree on the remote peer's address or until the NAS (or client) decide they're never going to agree and they shut down PPP instead.
Since we pretty much rely on having a username in authorization requests to decide which addresses etc. to hand out, it is important to know where the username for a PPP user comes from. There are generally 2 possible sources:
You force the user to authenticate by making her login to the exec and you use that login name in authorization requests. This username isn't propagated to PPP by default. To have this happen, you generally need to configure the if-needed method, e.g.
aaa authentication login default tacacs+ aaa authentication ppp default if-needed
Alternatively, you can run an authentication protocol, PAP or CHAP (CHAP is much preferred), to identify the user. You don't need an explicit login step if you do this (so it's the only possibility if you are using autoselect). This authentication gets done before you see the first LCP authorization request of course. Typically you configure this by doing:
aaa authentication ppp default tacacs+ int async 1 ppp authentication chap
If you omit either of these authentication schemes, you will start to see authorization requests in which the username is missing.
A list of AV pairs is placed in the daemon's configuration file in order to authorize services. The daemon compares each NAS AV pair to its configured AV pairs and either allows or denies the service. If the service is allowed, the daemon may add, change or delete AV pairs before returning them to the NAS, thereby restricting what the user is permitted to do.
The complete algorithm by which the daemon processes its configured AV pairs against the list the NAS sends, is given below.
Find the user (or group) entry for this service (and protocol), then for each AV pair sent from the NAS:
If the AV pair from the NAS is mandatory:
look for an exact attribute,value match in the user's mandatory list. If found, add the AV pair to the output.
If an exact match doesn't exist, look in the user's optional list for the first attribute match. If found, add the NAS AV pair to the output.
If no attribute match exists, deny the command if the default is to deny, or,
If the default is permit, add the NAS AV pair to the output.
If the AV pair from the NAS is optional:
look for an exact attribute,value match in the user's mandatory list. If found, add DAEMON's AV pair to output.
If not found, look for the first attribute match in the user's mandatory list. If found, add DAEMON's AV pair to output.
If no mandatory match exists, look for an exact attribute,value pair match among the daemon's optional AV pairs. If found add the DAEMON's matching AV pair to the output.
If no exact match exists, locate the first attribute match among the daemon's optional AV pairs. If found add the DAEMON's matching AV pair to the output.
If no match is found, delete the AV pair if the default is deny, or
If the default is permit add the NAS AV pair to the output.
After all AV pairs have been processed, for each mandatory DAEMON AV pair, if there is no attribute match already in the output list, add the AV pair (but add only ONE AV pair for each mandatory attribute).
After all AV pairs have been processed, for each optional unrequested DAEMON AV pair, if there is no attribute match already in the output list, add that AV pair (but add only ONE AV pair for each optional attribute).
The distribution comes with various MAVIS modules, of which the external module is probably the most interesting, as it interacts with simple Perl scripts to authenticate and authorize requests. You'll find sample scripts in the mavis/perl directory. Have a close look at them, as you may (or will) need to perform some trivial customizations to make them match your local environment.
You should really have a look at the MAVIS documentation. It gives examples for RADIUS and PAM authentication, too.
mavis_tacplus-ng_ldap.pl is an authentication/authorization back-end for the external module. It interfaces to various kinds of LDAP servers, e.g. OpenLDAP, Fedora DS and Active Directory. The server type is detected automatically. Its behaviour is controlled by a list of environmental variables:
Variable | Description | |
---|---|---|
LDAP_HOSTS |
Space-separated list of LDAP URLs or IP addresses or device names Examples: "ldap01 ldap02", "ldaps://ads01:636 ldaps://ads02:636" |
|
LDAP_SCOPE |
LDAP search scope for users (base, one, sub) Default: sub |
|
LDAP_SCOPE_GROUP |
LDAP search scope for groups (base, one, sub) Default: sub |
|
LDAP_BASE |
Base user search DN of your LDAP server Example: dc=example,dc=com |
|
LDAP_BASE_GROUP |
Base group search DN of your LDAP server Example: dc=example,dc=com |
|
LDAP_CONNECT_TIMEOUT |
Timeout for initital connect to remote LDAP server. Default: 1 (second). |
|
LDAP_FILTER |
LDAP search filter. Defaults:
|
|
LDAP_FILTER_GROUP |
LDAP search filter for groups. Default: "(&(objectclass=groupOfNames)(member=%s))" |
|
LDAP_USER |
User to use for LDAP bind if server doesn't permit anonymous searches. Default: unset |
|
LDAP_PASSWD |
Password for LDAP_USER Default: unset |
|
LDAP_MEMBEROF_REGEX |
Regular expression to derive group names from memberOf, Default: "^cn=([^,]+),.*" |
|
LDAP_TACMEMBER |
LDAP attribute to use for group membership (fallback only). Default: "tacMember" |
|
LDAP_TACMEMBER_MAP_OU |
Map organizational units from user DN to group membership. Default: unset |
|
USE_STARTTLS |
If set, the server is required to support start_tls. Default: unset |
Do not set this flag for LDAPS connections. |
FLAG_AUTHORIZE_ONLY |
Don't attempt to authenticate users. Default: unset |
|
TLS_OPTIONS |
Extra options for use with LDAPS or start_tls, in Perl hash syntax. See the Net::LDLAP documentation for details. Default: unset Example: "sslversion => 'tlsv1_3'" |
|
FLAG_AUTHORIZE_ONLY |
Don't attempt to authenticate users. Default: unset |
|
LDAP_NESTED_GROUP_DEPTH |
Limit nested group lookups to the given value. Unlimited if unset. Example: 1 |
ldapmavis-mt (mt stands for multi-threaded) basically evaluates the same envrionment variables as mavis_tacplus-ng_ldap.pl. It needs to be be invoked via the external-mt module and is suited for long-lasting authentication session, e.g. due to multi-factor authentication.
Example configuration for using Pluggable Authentication Modules:
id = spawnd { listen = { port = 49 } } id = tac_plus { mavis module groups { resolve gids = yes resolve gids attribute = TACMEMBER groups filter = /^(guest|staff)$/ } mavis module external { exec = /usr/local/sbin/pammavis "pammavis" "-s" "sshd" } user backend = mavis login backend = mavis device = global { address = 0.0.0.0/0 key = demo } profile staff { service shell { script { if (cmd == "") { set priv-lvl = 15 permit } } } profile guest { service shell { script { set priv-lvl = 15 if (cmd =~ ^/show /) permit deny } } } }
mavis_tacplus_passwd.pl authenticates against your local password database. Alas, to use this functionality, the script may have to run as root, as it needs access to the encrypted passwords. Primary and auxiliary UNIX group memberships will be mapped to TACACS+ groups.
mavis_tacplus_opie.pl is based on mavis_tacplus_passwd.pl, but uses OPIE one-time passwords for authentication.
mavis_tacplus_shadow.pl may be used to keep user passwords out of the tac_plusconfiguration file, enabling users to change their passwords via the password change dialog. Passwords are stored in an auxiliary, /etc/shadow-like ASCII file, one user per line:
username:encryptedPassword:lastChange:minAge:maxAge:reserved
lastChange is the number of days since 1970-01-01 when the password was last changed, and minAge and maxAge determine whether the password may/may not/needs to be changed. Setting lastChange to 0 enforces a password change upon first login.
Example shadow file:
marc:$1$q5/vUEsR$jVwHmEw8zAmgkjMShLBg/.:15218:0:99999: newuser:$1$pQtQsMuj$GKpIr5r2GNaZNfDfnCBtw.:0:0:99999: test:$1$pQtQsMuj$GKpIr5r2GNaZNfDfnCBtw.:15218:1:30:
Sample daemon configuration:
... id = tac_plus { ... mavis module external { setenv SHADOWFILE = /path/to/shadow # setenv FLAG_PWPOLICY=y # setenv ci=/usr/bin/ci # # There are more modern password hashes available via mkpasswd: # setenv MKPASSWD=/usr/bin/mkpasswd # setenv MKPASSWDMETHOD=yescrypt # exec = /usr/local/lib/mavis/mavis_tacplus_shadow.pl } ... login backend = mavis chpass ... user marc { login = mavis ... } ... } ...
mavis_tacplus_radius.pl authenticates against a RADIUS server. No authorization is done, unless the RADIUS_GROUP_ATTR environment variable is set (see below). This module may, for example, be useful if you have static user account definitions in the configuration file, but authentication passwords should be verified by RADIUS. Use the login = mavis or password = mavis statement in the user profile for this to work.
If the Authen::Radius Perl module is installed, the value of the RADIUS attribute specified by RADIUS_GROUP_ATTR will be used to create a TAC_MEMBER definition which uses the attribute value as group membership. E.g., an attribute value of Administrator would result in a
member = Administrator
declaration for the authenticated user, enabling authorization and omitting the need for static users in the configuration file.
Keep in mind that authorization will only work well if either
the tacplus_info_cache module is being used (it will cache authentication AV pairs locally, so subsequent authorizations should work fine unless you're switching to a tac_plus server running elsewhere).
or
single-connection is used and
mavis cache timeout is set to a sufficiently high value that covers the user's (expected) maximum login time.
Alternatively to mavis_tacplus_radius.pl the pamradius program may called by the external module. Results should be roughly equivalent.
## Use tacinfo_cache to cache authorization data to disk: mavis module tacinfo_cache { directory = /tmp/tacinfo } ## You can use either the Perl module ... #mavis module external { # exec = /usr/local/lib/mavis_tacplus_radius.pl # setenv RADIUS_HOST = 1.2.3.4:1812 # could add more devices here, comma-separated # setenv RADIUS_SECRET = "mysecret" # setenv RADIUS_GROUP_ATTR = Class # setenv RADIUS_PASSWORD_ATTR = Password # defaults to: User-Password # } ## ... or the freeradius-client based code: mavis module external { exec = /usr/local/sbin/radmavis "radmavis" "group_attribute=Class" "authserver=1.2.3.4:1812:mysecret" }
mavis_tacplus_sms.pl is a sample (skeleton) script to send One-Time Passwords via a SMS back-end.
If a back-end script fails due to an external problem (e.g. LDAP server unavailability), your router may or may not fall back to local authentication (if configured). Chances are that the fallback doesn't work. If you still want to be able to authenticate via TACACS+ in that case, you can do so with a non-MAVIS user which will only be valid in case of a back-end error:
... # set the time interval you want the user to be valid if the back-end fails: authentication fallback period = 60 # that's actually the default value ... # add a local user for emergencies: user = cisco { ... fallback-only ... }
To indicate that fallback mode is actually active, you may a display a different login prompt to your users:
device = ... { ... welcome banner = "Welcome\n" welcome banner fallback = "Welcome\nEmergency accounts are currently enabled.\n" ... }
Fallback can be enabled/disabled globally an on a per-device basis. Default is enabled.
authentication fallback = permit host ... { ... authentication fallback = deny ... }
When creating configuration files, it is convenient to check their syntax using the -P flag to tac_plus; e.g:
tac_plus -P config-file
will syntax check the configuration file and print any error messages on the terminal.
Trace (or debugging) options may be specified in global, device, user and group context. The current debugging level is a combination (read: OR) of all those. Generic syntax is:
debug = option ...
For example, getting command authorization to work in a predictable way can be tricky - the exact attributes the NAS sends to the daemon may depend on the IOS version, and may in general not match your expectations. If your regular expressions don't work, add
debug = REGEX
where appropriate, and the daemon may log some useful information to syslog.
Multiple trace options may be specified. Example:
debug = REGEX CMD
Trace options may be removed by prefixing them with-. Example:
debug = ALL -PARSE
The debugging options available are summarized in the following table:
Bit | Value | Name | Description |
---|---|---|---|
0 | 1 | PARSE | Configuration file parsing |
1 | 2 | AUTHOR | Authorization related |
2 | 4 | AUTHEN | Authentication related |
3 | 8 | ACCT | Accounting related |
4 | 16 | CONFIG | Configuration related |
5 | 32 | PACKET | Packet dump |
6 | 64 | HEX | Packet hex-dump |
7 | 128 | LOCK | File locking |
8 | 256 | REGEX | Regular expressions |
9 | 512 | ACL | Access Control Lists |
10 | 1024 | RADIUS | unused |
11 | 2048 | CMD | Command lookups |
12 | 4096 | BUFFER | Buffer handling |
13 | 8192 | PROC | Procedural traces |
14 | 16384 | NET | Network related |
15 | 32768 | PATH | File system path related |
16 | 65536 | CONTROL | Control connection related |
17 | 131072 | INDEX | Directory index related |
18 | 262144 | AV | Attribute-Value pair handling |
19 | 524288 | MAVIS | MAVIS related |
20 | 1048576 | DNS | DNS related |
21 | 2097152 | USERINPUT | Show user input (this may include passwords) |
31 | 2147483648 | NONE | Disable debugging |
Some of those debugging options are not used and trigger no output at all.
Debugging User Input | |
---|---|
The daemon will (starting with snapshot 202012051554) by default no longer outputs user input from authentication packets sent by the NAS. You can explicitly change this using the USERINPUT debug flag. Something like debug = ALL or using a numeric value will not work, it needs to be enabled explicitly, e.g.: debug = ALL USERINPUT Be prepared to see plain text user passwords if you enable this option. |
Is there a Graphical User Interface of any kind?
No, unless your favourite text editor does qualify.
I'm using the single-connection feature. How can I force my router to close the TCP connections to the TACACS+ server?
On IOS, show tcp brief will display the TCP connections. Search for the ones terminating at your server, and kill them using clear tcp tcb .... Example:
Router#sho tcp brief | incl 10.0.0.1.49 633BB794 10.0.0.2.17326 10.0.0.1.49 ESTAB 6287E4C4 10.0.0.2.24880 10.0.0.1.49 ESTAB Router#clear tcp tcb 633BB794 [confirm] [OK] Router#clear tcp tcb 6287E4C4 [confirm] [OK] Router#
Is there any way to avoid having clear text versions of the CHAP secrets in the configuration file?
CHAP requires that the server knows the cleartext password (or equivalently, something from which the server can generate the cleartext password). Note that this is part of the definition of CHAP, not just the whim of some Cisco engineer who drank too much coffee late one night.
If we encrypted the CHAP passwords in the database, then we'd need to keep a key around so that the server can decrypt them when CHAP needs them. So this only ends up being a slight obfuscation and not much more secure than the original scheme.
In extended TACACS, the CHAP secrets were separated from the password file because the password file may be a system password file and hence world readable. But with TACACS+'s native database, there is no such requirement, so we think the best solution is to read-protect the files. Note that this is the same problem that a Kerberos server has. If your security is compromised on the Kerberos server, then your database is wide open. Kerberos does encrypt the database, but if you want your server to automatically restart, then you end up having to "kstash" the key in a file anyway and you're back to the same security problem.
So storing the cleartext password on the security server is really an absolute requirement of the CHAP protocols, not something imposed by TACACS+.
With the scheme choosen for newer TACACS+ protocol revisions, the NAS sends the challenge information to the TACACS+ daemon and the daemon uses the cleartext password to generate the response and returns that.
The original TACACS+ protocol included specific protocol knowledge for CHAP. Please note that this version of the daemon implementation no longer supports SENDPASS, SENDAUTH and ARAP to comply to RFC8907.
However, the above doesn't apply to PAP. You can keep an inbound PAP password DES- or MD5-encrypted, since all you need to do with it is verify that the password the principal gave you is correct.
How is the typical login authentication sequence done?
NAS sends START packet to daemon
Daemon send GETUSER containing login prompt to NAS
NAS prompts user for username
NAS sends packet to daemon
Daemon sends GETPASS containing password prompt to the NAS
NAS prompts user for password
NAS sends packet to daemon
Daemon sends accept, reject or error to NAS
How do I limit the number of sessions a user can have?
With this version of the daemon you can't.
How can I configure time-outs on an interface via TACACS+?
Certain per-user/per-interface timeouts may be set by TACACS+ during authorization. As of 11.0, you can set an exec timeout. As of 11.1 you can also set an exec idle timeout.
There are currently no settable timeouts for PPP or SLIP sessions, but there is a workaround which applies to ASYNC PPP/SLIP idle timeouts started via exec sessions only: This workaround is to set an EXEC (idletime) timeout on an exec session which is later used to start up PPP or SLIP (either via a TACACS+ autocommand or via the user explicitly invoking PPP or SLIP). In this case, the exec idle timeout will correctly terminate an idle PPP or SLIP session. Note that this workaround cannot be used for sessions which autoselect PPP or SLIP.
An idle timeout terminates a connection when the interface is idle for a given period of time (this is equivalent to the "session-timeout" Cisco IOS configuration directive). The other timeouts are absolute. Of course, any timeouts set by TACACS+ apply only to the current connection.
profile ... { ... service shell { set idletime = 5 # disconnect lol if there is no traffic for 5 minutes set timeout = 60 # disconnect lol unconditionally after one hour ... } }
You also need to configure exec authorization on the NAS for the above timeouts, e.g.
aaa authorization exec default group tacacs+
Note that these timeouts only work for async lines, not for ISDN currently.
Note also that you cannot use the authorization if-authenticated option with these parameters, since that skips authorization if the user has successfully authenticated.
Can someone expand on the use of the optional keyword?
Most attributes are mandatory i.e. if the daemon sends them to the NAS, the NAS must obey them or deny the authorization. This is the default. It is possible to mark attributes as optional, in which case a NAS which cannot support the attribute is free to simply ignore it without causing the authorization to fail.
This was intended to be useful in cutover situations where you have multiple NASes running different versions of IOS, some of which support more attributes than others. If you make the new attributes optional, older NASes could ignore the optional attributes while new NASes could apply them. Note that this weakens your security a little, since you are no longer guaranteed that attributes are always applied on successful authorization, so it should be used judiciously.
What about MSCHAP?
The daemon comes with mschap support. Mschap is configured the same way as chap, only using the mschap keyword in place of the chap keyword.
MSCHAP requires DES support. Use the --with-ssl flag when configuring the package.
Marc Huber thinks that MSCHAP relevance is less than zero and expects it to be removed from the standard, as nobody uses it anyway.
While using a dedictated tac_plus-ng installation per tenant is certainly possible it lacks some elegance. There are other ways:
A single daemon can tell tenants apart by
device identity, either
by NAD IP address or
by certificate common name (currently irrelevant, as there's no NAD support)
realms, which are determined
by tacacs+ destination port or
by VRF (Linux, mostly)
Using the IP-based device identity should be sufficient for simple setups, but these don't scale and don't handle IP address conflicts. Options to cope with the latter involve realms. A realm is most basicly a text string the tcp listener (spawnd) assigns to a connection based on TCP destination port:
id = spawnd { ... listen { port = 49001 realm = customer1 } listen { port = 49002 realm = customer2 } ... }
In case VRFs aren't an option you can use HAProxy instances to transparently relay TACACS+ connections to tac_plus-ng:
id = spawnd { ... listen { port = 49001 realm = customer1 haproxy = yes } listen { port = 49002 realm = customer2 haproxy = yes } .... }
tac_plus-ng will then take the NAD IP from the HAProxy protocol v2 header.
Otherwise, the "listen" directive can be limite to your locally defined VRFs:
id = spawnd { ... listen { port = 49000 realm = customer1 vrf = blue } listen { port = 49000 realm = customer2 vrf = red } ... }
On Linux, if you set net.ipv4.tcp_l3mdev_accept=1, you can even get away with
id = spawnd { ... listen { port = 49000 } ... }
and the daemon will use the VRF name your clients did connect from as realm name.
The suggested setup for giving customers limited access to NADs is:
id = tac_plus-ng { mavis module external { your AD configuration goes here } profile ... { ... } realm customer1 { net custsrc { the IP ranges the end customer may log in from } rewrite normalizeCustomerAccount { rewrite /^.*$/ cust1-\L$0 } net custnet { the IP ranges the end customer may log in from } device customer1 { .... script { if (nac == custsrc) rewrite user = normalizeCustomerAccount } } ruleset { rule customer { if (nac == custnet) { if (member == ...) { profile = ... permit } deny } if (member == ...) profile = ... permit deny } }
In this example, you can easily share your LDAP (or AD) server between your own admin users and multiple tenants. The daemon will automatically prefix the customer accounts with a prefix and convert them to lower case. Note that the username rewriting happens using a script in device context. Rewriting won't work in scripts anywhere else.
The distribution includes the tactrace.pl Perl script. It's usually not installed automatically, due to some Perl dependencies that need to be met. It requires a couple of CPAN Perl modules, and a custom one. Your OS distribution might provide re-built packages for Net::IP and/or Net::TacacsPlus::Packet, so check for this first. For unavailable packages, you can do a manual install using cpan. Example for Ubuntu:
sudu apt install libnet-ip-perl sudo cpan install Net::TacacsPlus::Packet
Then cd to tac_plus-ng/perl and run make. tactrace.pl should now be ready to use.
Usage information:
$ This is a TACACS+ AAA validator for tac_plus-ng. Usage: ./tactrace.pl [ <Options> ] [ <attributes> ... ] attributes are authorization or accounting AV pairs, default is: "service=shell" "cmd*" Options: --help show this text --defaults=<file> read default settings from <file> --mode=<mode> authc, authz or acct [authz] --username=<username> username [ubuntu] --port=<port> port [vty0] --remote=<client ip> remote client ip [127.0.0.1] --key=<key> encryption key [demo] --realm=<realm> realm [default] --nad=<address> NAD (router/switch/...) IP address [127.0.0.1] --authentype=<type> authen_type [ascii] --authenmethod=<n> authen_method [tacacsplus] --authenservice=<n> authen_method [login] --exec=<path> executable path [/usr/local/sbin/tac_plus-ng] --conf=<config> configuration file [/usr/local/etc/tac_plus-ng.cfg] --id=<id> id for configuration selection [tac_plus-ng] For authc the password can be set either via the environment variable TACTRACEPASSWORD or the defaults file. Setting it via a CLI option isn't supported as the password would show up as clear text in the process listing.
Example:
# tactrace.pl --conf extra/tac_plus-ng.cfg-ads --user user01 127.0.0.1 ---<start packet>--- 127.0.0.1 session id: 00000001, data length: 46 127.0.0.1 AUTHOR, priv_lvl=0 127.0.0.1 authen_type=ascii (1) 127.0.0.1 authen_method=tacacs+ (6) 127.0.0.1 service=login (1) 127.0.0.1 user_len=6 port_len=4 rem_addr_len=9 arg_cnt=2 127.0.0.1 user (len: 6): user01 127.0.0.1 0000 75 73 65 72 30 31 user01 127.0.0.1 port (len: 4): vty0 127.0.0.1 0000 76 74 79 30 vty0 127.0.0.1 rem_addr (len: 9): 127.0.0.1 127.0.0.1 0000 31 32 37 2e 30 2e 30 2e 31 127.0.0. 1 127.0.0.1 arg[0] (len: 13): service=shell 127.0.0.1 0000 73 65 72 76 69 63 65 3d 73 68 65 6c 6c service= shell 127.0.0.1 arg[1] (len: 4): cmd* 127.0.0.1 0000 63 6d 64 2a cmd* 127.0.0.1 ---<end packet>--- 127.0.0.1 Start authorization request 127.0.0.1 looking for user user01 in MAVIS backend 127.0.0.1 user found by MAVIS backend, av pairs: MEMBEROF "CN=tacacs_admins,OU=Groups,DC=example,DC=local","CN=tacacs_readwrite,OU=Groups,DC=example,DC=local" USER user01 DN CN=user01,CN=Users,DC=example,DC=local IPADDR 127.0.0.1 SERVERIP 127.0.0.1 REALM default TACMEMBER "admins" 127.0.0.1 verdict for user user01 is ACK 127.0.0.1 user 'user01' found 127.0.0.1 evaluating ACL default#0 127.0.0.1 pcre2: '^CN=tacacs_admins,' <=> 'CN=tacacs_admins,OU=Groups,DC=example,DC=local' = 1 127.0.0.1 line 79: [memberof] <pcre-regex> '^CN=tacacs_admins,' => true 127.0.0.1 line 79: [profile] 'admins' 127.0.0.1 line 79: [permit] 127.0.0.1 ACL default#0: match 127.0.0.1 user01@127.0.0.1: ACL default#0: permit (profile: admins) 127.0.0.1 line 45: [service] = 'shell' => true 127.0.0.1 line 47: [cmd] = '' => true 127.0.0.1 line 47: [set] 'priv-lvl=15' 127.0.0.1 line 48: [permit] 127.0.0.1 nas:service=shell (passed thru) 127.0.0.1 nas:cmd* (passed thru) 127.0.0.1 nas:absent srv:priv-lvl=15 -> add priv-lvl=15 (k) 127.0.0.1 added 1 args 127.0.0.1 Writing AUTHOR/PASS_ADD size=30 127.0.0.1 ---<start packet>--- 127.0.0.1 session id: 00000001, data length: 18 127.0.0.1 AUTHOR/REPLY, status=1 (AUTHOR/PASS_ADD) 127.0.0.1 msg_len=0, data_len=0, arg_cnt=1 127.0.0.1 msg (len: 0): 127.0.0.1 data (len: 0): 127.0.0.1 arg[0] (len: 11): priv-lvl=15 127.0.0.1 0000 70 72 69 76 2d 6c 76 6c 3d 31 35 priv-lvl =15 127.0.0.1 ---<end packet>---
This documentation isn't well structured.
The examples given are too IPv4-centric. However, the daemon handles IPv6 just fine.
Some of the NAS configuration examples aren't recently tested. Refer to the IOS documentation for IOS configuration syntax guidance.
Please see the source for copyright and licensing information of individual files.
The following applies if the software was compiled with OpenSSL support:
This product includes software developed by the OpenSSL Project for use in the OpenSSL Toolkit (http://www.openssl.org/).
This product includes cryptographic software written by Eric Young (<eay@cryptsoft.com>
).
MD4 algorithm:
The software uses the RSA Data Security, Inc. MD4 Message-Digest Algorithm.
MD5 algorithm:
The software uses the RSA Data Security, Inc. MD5 Message-Digest Algorithm.
If the software was compiled with PCRE (Perl Compatible Regular Expressions) support, the following applies:
Regular expression support is provided by the PCRE library package, which is open source software, written by Philip Hazel, and copyright by the University of Cambridge, England. (ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre/).
The original tac_plus code (which this software and considerable parts of the documentation are based on) is distributed under the following license:
Copyright (c) 1995-1998 by Cisco systems, Inc.
Permission to use, copy, modify, and distribute this software for any purpose and without fee is hereby granted, provided that this copyright and permission notice appear on all copies of the software and supporting documentation, the name of Cisco Systems, Inc. not be used in advertising or publicity pertaining to distribution of the program without specific prior permission, and notice be given in supporting documentation that modification, copying and distribution is by permission of Cisco Systems, Inc.
Cisco Systems, Inc. makes no representations about the suitability of this software for any purpose. THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.
The code written by Marc Huber is distributed under the following license:
Copyright (C) 1999-2022 Marc Huber (<Marc.Huber@web.de>
). All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
The end-user documentation included with the redistribution, if any, must include the following acknowledgment:
This product includes software developed by Marc Huber (
<Marc.Huber@web.de>
).
Alternately, this acknowledgment may appear in the software itself, if and wherever such third-party acknowledgments normally appear.
THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL ITS AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.