Category Archives: Networking

pfSense: Suricata Sync Causes XMLRPC Failures and carp backup events

Observed with pfSense 2.4.5p1 and Suricata 5.0.3 (and presumably older versions of both)

Once you enable Suricata config sync, any configuration changes take *ages* to save because Syncs basically start failing to complete – eventually falling through to timeouts.

You might start to see synchronise errors like this on the master (which will get flagged up as notifications):

/rc.filter_synchronize: A communications error occurred while attempting to call XMLRPC method host_firmware_version:

and/or

/suricata/suricata_logs_mgmt.php: A communications error occurred while attempting to call XMLRPC method exec_php:

and/or

/rc.filter_synchronize: New alert found: A communications error occurred while attempting to call XMLRPC method restore_config_section:

You might also notice that CARP is – to put it mildly – freaking out:

Carp backup event
Carp backup event
Carp backup event

And that OVPN / other packages likewise are having problems, stopping/starting/restarting because it thinks the WAN IP has changed as the CARP state flaps back and forth, with the system logging stuff like:

/rc.newwanip: rc.newwanip: Info: starting on ovpns1.
/rc.newwanip: rc.newwanip: on (IP address: ) (interface: []) (real interface: ovpns1).
/rc.newwanip: rc.newwanip called with empty interface.
/rc.newwanip: pfSense package system has detected an IP change or dynamic WAN reconnection - -> - Restarting packages.

The issue is likely that you have promiscuous mode enabled on your Suricata interfaces (because it is the *default* to enable it).

The kernel disabling and enabling promiscuous mode off and on as Suricata reloads during sync causes carnage with the sync TCP connection, CARP, and in turn, everything else.

Promiscuous mode should not be required if you are using Suricata in-line at layer 3 (i.e. on the firewall which is hosting your default gateway which is probably why CARP is running to begin with).

Simply disable promiscuous mode (at the very *least* on any interfaces you’re running CARP on which is probably all of them in an HA setup) and you’ll find things behave much better, and config syncs complete nice and fast again.

Configuring TACACS+ authentication and accounting on IOS 15

Just the bare minimum:

! you probably have this already, if you don't; you should read up on it first
aaa new-model

! use local users, and then all tacacs+ servers, to authenticate logins 
aaa authentication login default local group tacacs+ 

! give enable to tacacs+ users 
aaa authentication enable default group tacacs+ 

! send accounting records for when logins ('exec mode') begin and end 
aaa accounting exec default start-stop group tacacs+
 
! send accounting records for config commands 
aaa accounting commands 15 default stop-only group tacacs+ 

! send accounting records for outgoing connections made to other systems 
aaa accounting connection default start-stop group tacacs+ 

! send system event account records (reloads etc) 
aaa accounting system default start-stop group tacacs+ 

! OPTIONAL: On a router with multiple interfaces that could be chosen to
! reach the TACACS server it is best to specify one; we use Loopback addresses
! for iBGP peering, so it makes sense to use them here too
ip tacacs source-interface Loopback0 

! define at least one tacacs server with some friendly $SERVERNAME 
tacacs server $SERVERNAME
   ! Set the TACACS+ server's ipv4 $ADDRESS (or ipv6, adjust accordingly)
   address ipv4 $ADDRESS
   ! Set the encryption $KEY to match the key configured on the TACACS+ server for this device
   key $KEY
!

Now: BEFORE you log off, try to log in again and make sure you can still log in with your original local credentials.

If you can no longer login after making the above changes, you’ll need to fix that first before you disconnect to prevent you locking yourself out.