I've frequently been bit by failures restarting my vpn on my Mac. The connection may appear to start up in Shimo (the menu icon will change) but it looks like vpnc can't connect.
Looking at the log, we see the connection gets stuck
ERROR: can't send packet: Can't assign requested address
This has plagued me for a while, and my only fix has been to restart the machine. I hate restarting my machine. So I finally made time to dig in and find a better solution
I noticed that I couldn't ping 199.12.34.56 (the address of the vpn server) either.
What appears to be happening is the route to the vpn server get stuck to an old address/interface.
netstat -nr
Routing tables
Internet:
Destination Gateway Flags Refs Use Netif Expire
default 192.168.1.1 UGSc 51 8 en8
127 127.0.0.1 UCS 0 0 lo0
127.0.0.1 127.0.0.1 UH 25 49984 lo0
169.254 link#13 UCS 0 0 en8
192.168.0/16 link#13 UCS 1 0 en8
192.168.1.1/32 link#13 UCS 1 0 en8
192.168.1.1 d4:ca:6d:dd:f1:a5 UHLWIir 52 0 en8 1188
192.168.1.13/32 link#13 UCS 0 0 en8
192.168.1.22 74:d0:2b:c8:df:f5 UHLWIi 1 12 en8 1159
199.12.34.56 172.28.4.1 UGHS 0 4 en0
172.28.4.1 gateway doesn't exist...
It turns out that is the gateway I was using before I moved to my current location. (and things work when I move back there) The solution
Deleting the route fixes the problem:
sudo route delete 199.12.34.56
and now ping and vpnc are happy.
I'm still not clear what is causing this, but it appears to only be happening when I switch both the address and the interface I'm using. Note that the above issue the route was on en0, but my main gateway was en8 (a USB ethernet dongle).
I reproduced the above situation when using vpnc on MacOS.
I have a script to kill the vpnc process and reconnect it. But it does not work if the connection is broken after the machine slept.
I found this page on Google and I delete the route to the vpn server, and the script got work again. Now I put the deleting route command into my script.