11-02-2011 12:15 PM
Try moving your host-inbound-traffic configuration to the actual interface, rather than the zone.
I haven't gotten deep into this, but I just ran a real quick experiment, and I noticed that when I have this:
security-zone untrust {
host-inbound-traffic {
system-services {
ping;
}
}
interfaces {
ge-0/0/0.420 {
host-inbound-traffic {
system-services {
traceroute;
ssh;
http;
https;
}
protocols {
ospf;
}
}
}
}
}I am unable to ping the address of my ge-0/0/0.420 interface. However, if I change it to this:
security-zone untrust {
interfaces {
ge-0/0/0.420 {
host-inbound-traffic {
system-services {
ping;
traceroute;
ssh;
http;
https;
}
protocols {
ospf;
}
}
}
}
}I am then able to ping the interface.
I'm sure there's probably some blurb buried deep somehwere in a "You passed the JNCIS-SEC test, so you should know this!" document, but honestly, I've always attached my host-inbound-traffic directly to my interfaces, not to the zones, and I've never run into this issue before.
Try it, see if it works. ![]()
11-02-2011 07:13 PM
Host-inbound-traffic under a specific interface always overrides the zone configuration, but having it just configured in a zone Should Just Work™ just as well.
On an unrelated side note, there are extra services that are ONLY available when you configure under an interface - such as dhcp, bootp etc.
As for the original problem - I would disable "establish-tunnels immediately" on the SRX650. Cisco defaults to something equivalant to "establish tunnels on-traffic" (eg: something has to match the crypto-map before it will trigger IKE) and this can cause a synchronisation issue where both gateways are trying to initiate a tunnel to each other, and this WILL fail repeatedly. As a rule of thumb, I always set up SPOKES to connect back to the hub, with "establish-tunnels immediately" on the spoke side and no configuration on the hub.
Also I notice in your IKE Gateway configuration you specify a local-identity. This shouldn't be required in Main mode, so remove it from the configuration just in case it is interfering with anything.
11-03-2011 02:08 PM
dfex wrote:Host-inbound-traffic under a specific interface always overrides the zone configuration, but having it just configured in a zone Should Just Work™ just as well.
Agreed, and it's great when things really do just work, but even in my simple test, "Should Just Work", as we can see, didn't.
I was merely suggesting a couple simple tests to see if it had an effect, maybe it'll help, maybe it won't.
11-03-2011 02:45 PM
I took another look at this (tried your suggestion as well keithr - same result thou...) I have several interfaces on the srx and found out that traffic isn't entering the .455 interface but rather .15 (same physical interface though, ge-0/0/0) and they belong to the same security zone, untrust. When that became clear to me, I added a policy for untrust to untrust which permits ping, and voilà, the pings are suddenly answered... This might explain a lot I guess. I've just added "application any" to the untrust to untrust rule and will se if the VPN's act more stable than before.
Thanks for your suggestions!