In a previous post I wrote about how I went about configuring a NTP server and setting NTP clients. A few days later, when speaking with my senior, the plan changed slightly and we weren’t configuring our own NTP server using Linux based OS, but get satellite feeds running our own LANTIME Stratum 1 Server o_O! In other words that previous post is now invalid haha!

For now I don’t know if I’ll be involved in that side of the project, I hope I am, but that’s for another day! Back to my actual point, because of the environment we run in our datacentres, our firewalls run with Virtual-Router Routing-Instances. A Virtual Router is similar than Cisco’s VRF concept however, with Juniper’s a Virtual Router is used for non-VPN related applications. So, I was asked to investigate if it was possible for our firewalls to be NTP clients to a NTP server via the master instance and also be able to act as a NTP server to attached clients within a routing-instance.

You can get more detail about Juniper’s Routing-Instance Types on their TechLibrary Routing Instances Overview

Although it sounds a bit of a mouthful, the topology itself is quite straightforward. The diagram below shows the topology I’ll be testing with:

As the diagram shows, I’ll be using a standalone Juniper SRX220h and two ESXi Ubuntu 14.04LTS servers. The SRX will be a NTP client of the NTP server (km-vm4) via the master inet.0 table. The second client km-vm1 will be located within the Routing-Instance “test” and will be using the SRX220 as its NTP server. Security policies will need to be defined, as the stateful functionalities of the SRX will still be in use. Without creating security between the zones all traffic will be dropped.

The Base configuration used in this topology is below. (I using this SRX for IPv6 testing as well, so ignore the vlan name lol)

Base Config⌗ 

set interfaces ge-0/0/0 enable
set interfaces ge-0/0/0 unit 0 family ethernet-switching port-mode access
set interfaces ge-0/0/0 unit 0 family ethernet-switching vlan members internal-v6
set interfaces ge-0/0/1 enable
set interfaces ge-0/0/1 unit 0 family ethernet-switching port-mode access
set interfaces ge-0/0/1 unit 0 family ethernet-switching vlan members ntp-server
set interfaces vlan unit 100 family inet address 192.168.1.1/24
set interfaces vlan unit 101 family inet address 192.168.2.1/24

set security policies from-zone vlan100 to-zone vlan101 policy allow-ntp match source-address any
set security policies from-zone vlan100 to-zone vlan101 policy allow-ntp match destination-address any
set security policies from-zone vlan100 to-zone vlan101 policy allow-ntp match application junos-ntp
set security policies from-zone vlan100 to-zone vlan101 policy allow-ntp then permit

set security policies from-zone vlan101 to-zone vlan100 policy allow-ntp match source-address any
set security policies from-zone vlan101 to-zone vlan100 policy allow-ntp match destination-address any
set security policies from-zone vlan101 to-zone vlan100 policy allow-ntp match application junos-ntp
set security policies from-zone vlan101 to-zone vlan100 policy allow-ntp then permit

set security zones security-zone vlan100 interfaces vlan.100 host-inbound-traffic system-services any-service
set security zones security-zone vlan101 interfaces vlan.101 host-inbound-traffic system-services any-service

set routing-instances test instance-type virtual-router
set routing-instances test interface vlan.100

set vlans internal-v6 vlan-id 100
set vlans internal-v6 l3-interface vlan.100
set vlans ntp-server vlan-id 101
set vlans ntp-server l3-interface vlan.101

With this setup, the overall goal of this testing is to see if it’s possible to advertise specific routes from the inet.0 table to another routing table, to allow end to end connectivity between routing instances.

This post will not show/explain how the stateful functionalities of the SRX series firewall works. That will be in another post :p

With all that talk and background out of the way… Let’s get cracking :)

Junos and NTP⌗

With the NTP server already configured, the SRX need to set as an NTP client. This configuration is done under system ntp stanza. We set the remote server, ntp version and preference. In addition, I set two other statements; one is optional and the other had to be set. The statements, boot-server and source-address, Juniper defines these statements as:

  • boot-server: When the router or switch boots, it immediately synchronizes with the boot server even if the NTP process is explicitly disabled or if the time difference between the client and the boot server exceeds the threshold value of 1000 seconds.
  • source-address: This is statement useful for controlling which source address NTP will use to access your network when it is either responding to an NTP client request from your network or when it itself is sending NTP requests to your network.

As described above, and in a nutshell, by adding the source-address this will allow other clients/devices to set an IP address that’s located on the SRX as their remote NTP server. Thus providing a /32 address that can be advertised. For my example the address 192.168.2.1, the gateway address for the ntp server, will be the address used as the source address.

marquk01@v6-testing# show system ntp 
boot-server 192.168.2.100;
server 192.168.2.100 version 4 prefer;
source-address 192.168.2.1;

We can verify the NTP association by using the command show ntp assocication.

marquk01@v6-testing# run show ntp associations                      
     remote           refid      st t when poll reach   delay   offset  jitter
==============================================================================
*192.168.2.100   43.255.190.16    3 - 148  256  377    2.313    0.748   0.063

With NTP verified on the SRX, we have to leak the NTP source address for the SRX 192.168.2.1 from the inet.0 table to the test.inet.0 table. As we can see, when we look at the routing instance’s routing table, we don’t have the route installed into the table:

marquk01@v6-testing# run show route table test.inet.0      

test.inet.0: 2 destinations, 2 routes (2 active, 0 holddown, 0 hidden)
+ = Active Route, - = Last Active, \* = Both

192.168.1.0/24     \*\[Direct/0\] 04:02:28
                    > via vlan.100
192.168.1.1/32     \*\[Local/0\] 06:31:31
                      Local via vlan.100

Routing Instance Explanation⌗

To get the route installed into the test.inet.0 table, we’ll need to leak the 192.168.2.1/32 route into the test.inet.0 table, and we’ll do this by creating two policy statements. For my example I’ve named them master and instance just for ease!

  • Master: This statement allows all routes from routing instance test to be accepted into the master routing table.
  • Instance: This statement has two terms. Term 1 only allows the exact route 192.168.2.1/32 from the master instance to be accepted. Term 2 denies all other routes from master instance.
marquk01@v6-testing# show policy-options 
policy-statement master {
    from instance test;
    then accept;
}
policy-statement instance {
    term 1 {
        from {
            instance master;
            route-filter 192.168.2.1/32 exact;
        }
        then accept;
    }
    term 2 {
        then reject;
    }
}

Now we’ll need to import the relevant policy into each instance, under the routing-options stanza for the inet.0 table and routing-instance test routing-options for the test.inet.0 table.

Default Routing-Options⌗ 

marquk01@v6-testing# show routing-options 
static {
    route 0.0.0.0/0 {
        next-hop 10.1.0.1;
        no-readvertise;
    }
}
instance-import master;

Routing-Instance Routing-Options⌗ 

marquk01@v6-testing# show routing-instances test routing-options 
instance-import instance;

Having committed the policy statements, when we check both routing tables we can see the 192.168.1.0/24 route have been leaked into the inet.0 table, and that only the 192.168.2.1/32 route has been installed into the test.inet.0 table. All other routes have not been leaked.

Default Routing Table inet.0⌗ 

marquk01@v6-testing> show route table inet.0 

inet.0: 7 destinations, 7 routes (7 active, 0 holddown, 0 hidden)
+ = Active Route, - = Last Active, \* = Both

0.0.0.0/0          \*\[Static/5\] 1w2d 09:40:34
                    > to 10.1.0.1 via ge-0/0/7.0
10.1.0.0/24        \*\[Direct/0\] 1w2d 09:40:34
                    > via ge-0/0/7.0
10.1.0.158/32      \*\[Local/0\] 1w2d 09:40:39
                      Local via ge-0/0/7.0
192.168.1.0/24     \*\[Direct/0\] 2d 04:15:32
                    > via vlan.100
192.168.1.1/32     \*\[Local/0\] 2d 04:15:32
                      Local via vlan.100
192.168.2.0/24     \*\[Direct/0\] 2d 07:20:32
                    > via vlan.101
192.168.2.1/32     \*\[Local/0\] 3d 08:44:24
                      Local via vlan.101

Routing-Instance Routing Table test.inet.0⌗ 

marquk01@v6-testing> show route table test.inet.0   

test.inet.0: 3 destinations, 3 routes (3 active, 0 holddown, 0 hidden)
+ = Active Route, - = Last Active, \* = Both

192.168.1.0/24     \*\[Direct/0\] 09:32:22
                    > via vlan.100
192.168.1.1/32     \*\[Local/0\] 12:01:25
                      Local via vlan.100
192.168.2.1/32     \*\[Local/0\] 05:28:32
                      Local via vlan.101

Finally, on the client we’ll need to set a static route so that the host knows if it wants to get the 192.168.2.0/24 subnet it will need to use the gateway 192.168.1.1 via eth1

marquk01@km-vm1:~$ sudo route add -net 192.168.2.0/24 gw 192.168.1.1 dev eth1
marquk01@km-vm1:~$ ip route
default via 10.1.0.1 dev eth0 
10.1.0.0/24 dev eth0  proto kernel  scope link  src 10.1.0.137 
192.168.1.0/24 dev eth1  proto kernel  scope link  src 192.168.1.100 
192.168.2.0/24 via 192.168.1.1 dev eth1

Verification⌗

Once that route has been installed we can see that the host has now become a NTP client to the SRX by running the command ntpq -p. In addition on the SRX, we can see the flow session between inet.0 and test.inet.0 routing tables.

NTP Association⌗ 

marquk01@km-vm1:~$ ntpq -p
     remote           refid      st t when poll reach   delay   offset  jitter
==============================================================================
\*192.168.2.1     192.168.2.100    4 u   15   64    3    2.086   -0.094   0.109

SRX flow Session⌗ 

marquk01@v6-testing> show security flow session destination-prefix 192.168.2.1 
Session ID: 21218, Policy name: allow-ntp/4, Timeout: 40, Valid
  In: 192.168.1.100/123 --> 192.168.2.1/123;udp, If: vlan.100, Pkts: 7, Bytes: 532
  Out: 192.168.2.1/123 --> 192.168.1.100/123;udp, If: .local..5, Pkts: 7, Bytes: 532
Total sessions: 1

And with that we’ve been able to route between routing instances! The most important thing that I found whilst do this testing is that you need to remember to add the static route on the host or device that is directly connected to router with the routing-instance. This same setup would work with a switch in between the SRX and end host, just have the static route and you’ll be sorted. In addition to having the right security policies, as they’ll bite you in the bum as well :p

I’ve included the set commands that I used in my example below, if you wanted to give it a try for yourself :)

Set Commands⌗ 

set system ntp boot-server 192.168.2.100
set system ntp server 192.168.2.100 version 4
set system ntp server 192.168.2.100 prefer
set system ntp source-address 192.168.2.1

set routing-options instance-import p1

set policy-options policy-statement p1 from instance test
set policy-options policy-statement p1 then accept

set policy-options policy-statement p2 term 1 from instance master
set policy-options policy-statement p2 term 1 from route-filter 192.168.2.1/32 exact
set policy-options policy-statement p2 term 1 then accept
set policy-options policy-statement p2 term 2 then reject

set routing-instances test routing-options instance-import p2

Reference⌗

Share on LinkedIn
Share on Reddit