IPSec Point to Point Tunnel Has Double Latency in One Direction
I've been trying to figure this out for a few days now, but I'm at a loss. Time to ask the experts.
I have two Debian Squeeze boxes that I'm creating a site to site vpn for.
Followed this how to: http://wiki.debian.org/IPsec
I can ping both sides of the internal interfaces, but pings from one side, have twice the latency of the other
side. I'm at a total loss about why this is occurring.
Site Office Network:
External Interface on eth0: 50.194.128.49/28
Internal Interface on eth1: 172.18.38.1/24
ip route add to 172.18.108.0/24 via 50.194.128.49 src 172.18.38.1
Site Branch Network:
External Interface on eth0: 71.33.229.33/28
Internal Interface on eth1: 172.18.108.1/24
ip route add to 172.18.38.0/24 via 71.33.229.33 src 172.18.108.1
Site Office /etc/racoon/racoon.conf:
path pre_shared_key "/etc/racoon/psk.txt";
path certificate "/etc/racoon/certs";
remote 71.33.229.33 {
exchange_mode main,aggressive;
proposal {
encryption_algorithm 3des;
hash_algorithm sha1;
authentication_method pre_shared_key;
dh_group 2;
}
}
sainfo address 172.18.38.0/24 any address 172.18.108.0/24 any {
pfs_group 2;
lifetime time 1 hour ;
encryption_algorithm 3des, blowfish 448, rijndael ;
authentication_algorithm hmac_sha1, hmac_md5 ;
compression_algorithm deflate ;
}
Site Branch /etc/racoon/racoon.conf:
path pre_shared_key "/etc/racoon/psk.txt";
path certificate "/etc/racoon/certs";
remote 50.194.128.49 {
exchange_mode main,aggressive;
proposal {
encryption_algorithm 3des;
hash_algorithm sha1;
authentication_method pre_shared_key;
dh_group 2;
}
}
sainfo address 172.18.108.0/24 any address 172.18.38.0/24 any {
pfs_group 2;
lifetime time 1 hour ;
encryption_algorithm 3des, blowfish 448, rijndael ;
authentication_algorithm hmac_sha1, hmac_md5 ;
compression_algorithm deflate ;
}
Site Office /etc/ipsec-tools.conf:
flush;
spdflush;
spdadd 172.18.38.0/24 172.18.108.0/24 any -P out ipsec
esp/tunnel/50.194.128.49-71.33.229.33/require;
spdadd 172.18.108.0/24 172.18.38.0/24 any -P in ipsec
esp/tunnel/71.33.229.33-50.194.128.49/require;
Site Branch /etc/ipsec-tools.conf:
flush;
spdflush;
spdadd 172.18.108.0/24 172.18.38.0/24 any -P out ipsec
esp/tunnel/71.33.229.33-50.194.128.49/require;
spdadd 172.18.38.0/24 172.18.108.0/24 any -P in ipsec
esp/tunnel/50.194.128.49-71.33.229.33/require;
Ping from Office to Branch External:
ping -c5 -n voipshinn
PING voipshinn (71.33.229.33) 56(84) bytes of data.
64 bytes from 71.33.229.33: icmp_req=1 ttl=52 time=70.4 ms
64 bytes from 71.33.229.33: icmp_req=2 ttl=52 time=70.7 ms
64 bytes from 71.33.229.33: icmp_req=3 ttl=52 time=84.5 ms
64 bytes from 71.33.229.33: icmp_req=4 ttl=52 time=70.6 ms
64 bytes from 71.33.229.33: icmp_req=5 ttl=52 time=69.8 ms
Ping from Office to Branch Iternal:
ping -c5 -n voipshinn-int
PING voipshinn-int (172.18.108.1) 56(84) bytes of data.
64 bytes from 172.18.108.1: icmp_req=1 ttl=64 time=84.3 ms
64 bytes from 172.18.108.1: icmp_req=2 ttl=64 time=85.1 ms
64 bytes from 172.18.108.1: icmp_req=3 ttl=64 time=77.8 ms
64 bytes from 172.18.108.1: icmp_req=4 ttl=64 time=78.2 ms
64 bytes from 172.18.108.1: icmp_req=5 ttl=64 time=79.0 ms
So about the same latency from the Office to the Branch over the VPN compared to pinging the external
interface directly.
Ping from Branch to Office External:
ping -c5 -n linuxgw
PING linuxgw (50.194.128.49) 56(84) bytes of data.
64 bytes from 50.194.128.49: icmp_req=1 ttl=51 time=70.9 ms
64 bytes from 50.194.128.49: icmp_req=2 ttl=51 time=71.6 ms
64 bytes from 50.194.128.49: icmp_req=3 ttl=51 time=70.4 ms
64 bytes from 50.194.128.49: icmp_req=4 ttl=51 time=70.2 ms
64 bytes from 50.194.128.49: icmp_req=5 ttl=51 time=69.4 ms
Ping from Branch to Office Internal:
ping -c5 -n linuxgw-int
PING linuxgw-int (172.18.38.1) 56(84) bytes of data.
64 bytes from 172.18.38.1: icmp_req=1 ttl=64 time=139 ms
64 bytes from 172.18.38.1: icmp_req=2 ttl=64 time=134 ms
64 bytes from 172.18.38.1: icmp_req=3 ttl=64 time=133 ms
64 bytes from 172.18.38.1: icmp_req=4 ttl=64 time=134 ms
64 bytes from 172.18.38.1: icmp_req=5 ttl=64 time=136 ms
Pretty much double the latency going over the VPN compared to just pinging the external interface directly.
I can find no reason why this is occurring. Anyone have any ideas why this is happening?
Chris
Reply to: