Hi Folks,
I'm running strongSwan on Slackware with linux 4.18
This is the output from 'ip route show':
default via isp_router dev eth0 metric 1
10.0.0.0/8 via 10.68.63.3 dev eth1
10.68.63.0/26 dev eth1 proto kernel scope link src 10.68.63.62
127.0.0.0/8 dev lo scope link
172.16.0.0/12 via 10.68.63.3 dev eth1
192.168.0.0/16 via 10.68.63.3 dev eth1
and 'ip a s':
3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 [trimmed]
link/ether fc:4d:d4:2e:dc:6a brd ff:ff:ff:ff:ff:ff
inet 10.68.63.62/26 brd 10.68.63.63 scope global eth1
valid_lft forever preferred_lft forever
Connected to the eth1 there is also a 10.68.63.55 host:
the default gateway for this host is 10.68.63.62
This is the relevant configuration from stronSwan:
children {
networks {
local_ts = 10.68.63.54/31
remote_ts = 10.82.43.0/24
start_action = trap
}
}
The value trap installs a trap policy, which is going
to trigger the tunnel as soon as matching traffic will
be detected.
When I issue a ping from 10.68.63.55 host to a 10.82.43.0/24
network's host, I get this behaviour:
12:24:40.219185 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], proto ICMP
(1), length 84)
10.68.63.55 > 10.82.43.10: ICMP echo request, id 12199, seq 43, length 64
12:24:41.235072 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], proto ICMP
(1), length 84)
10.68.63.55 > 10.82.43.10: ICMP echo request, id 12199, seq 44, length 64
12:24:42.251095 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], proto ICMP
(1), length 84)
10.68.63.55 > 10.82.43.10: ICMP echo request, id 12199, seq 45, length 64
12:24:43.267007 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], proto ICMP
(1), length 84)
10.68.63.55 > 10.82.43.10: ICMP echo request, id 12199, seq 46, length 64
12:24:44.282948 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], proto ICMP
(1), length 84)
10.68.63.55 > 10.82.43.10: ICMP echo request, id 12199, seq 47, length 64
12:24:45.298873 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], proto ICMP
(1), length 84)
10.68.63.55 > 10.82.43.10: ICMP echo request, id 12199, seq 48, length 64
12:24:46.314840 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], proto ICMP
(1), length 84)
10.68.63.55 > 10.82.43.10: ICMP echo request, id 12199, seq 49, length 64
12:24:47.330767 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], proto ICMP
(1), length 84)
10.68.63.55 > 10.82.43.10: ICMP echo request, id 12199, seq 50, length 64
12:24:48.346708 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], proto ICMP
(1), length 84)
10.68.63.55 > 10.82.43.10: ICMP echo request, id 12199, seq 51, length 64
12:24:49.362627 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], proto ICMP
(1), length 84)
10.68.63.55 > 10.82.43.10: ICMP echo request, id 12199, seq 52, length 64
12:24:49.362686 IP (tos 0xc0, ttl 64, id 11631, offset 0, flags [none], proto
ICMP (1), length 112)
10.68.63.62 > 10.68.63.55: ICMP redirect 10.82.43.10 to host 10.68.63.3,
length 92
IP (tos 0x0, ttl 63, id 0, offset 0, flags [DF], proto ICMP (1), length
84)
10.68.63.55 > 10.82.43.10: ICMP echo request, id 12199, seq 52, length 64
12:24:50.379051 IP (tos 0x0, ttl 63, id 0, offset 0, flags [DF], proto ICMP
(1), length 84)
10.68.63.55 > 10.82.43.10: ICMP echo request, id 12199, seq 53, length 64
12:24:50.379113 IP (tos 0xc0, ttl 64, id 12190, offset 0, flags [none], proto
ICMP (1), length 112)
10.68.63.62 > 10.68.63.55: ICMP redirect 10.82.43.10 to host 10.68.63.3,
length 92
IP (tos 0x0, ttl 62, id 0, offset 0, flags [DF], proto ICMP (1), length
84)
10.68.63.55 > 10.82.43.10: ICMP echo request, id 12199, seq 53, length 64
12:24:51.398707 IP (tos 0x0, ttl 63, id 0, offset 0, flags [DF], proto ICMP
(1), length 84)
10.68.63.55 > 10.82.43.10: ICMP echo request, id 12199, seq 54, length 64
Linux is sending those icmp redirects *after* the
CHILD_SA has been established. When the CHILD_SA
is not yet established linux will not send any icmp
redirect.
Is this the expected behavior?
Obviously if I remove the following static route:
10.0.0.0/8 via 10.68.63.3 dev eth1
linux will not send anymore the icmp redirect.
Any feedback are welcome.
Thanks in advance
