You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The Wireguard role provided by OSISM cannot execute the “iptables” commands defined via the PostUp and PostDown directives because “iptables” is probably not part of the node image or the bootstrap installation process (anymore?).
In my specific case, the defined rules would also not help either, because the nodes of the SCS Hardware Landscape do not know a route to the IP addresses in the Wireguard VPN and so the predefined ACCEPT rule would not be enough, because the return direction of the packets to the Wireguard server could not be identified because of non existing routes on the nodes.
My suggestion would be to use “nft” to create the rules specifically for the VPN client IP instead of “iptables” and to use a SNAT/masequerading rule instead of an ACCEPT for every destination network.
(see also)
Using NAT would have the advantage that network clients using the Wireguard server of a node would have the same access options as clients initiating a connection directly on the node.
You also do not need routes on the nodes and you can have multiple VPN gateways with the same VPN Client ip ranges.
However, this method has the disadvantage that the VPN IPs are masked and therefore network-connections are less easy to assign to a VPN user.
References to existing reports
No response
Severity
low
Urgency
low
The text was updated successfully, but these errors were encountered:
OSISM release version
8.0.0
What's the problem?
The Wireguard role provided by OSISM cannot execute the “iptables” commands defined via the PostUp and PostDown directives because “iptables” is probably not part of the node image or the bootstrap installation process (anymore?).
In my specific case, the defined rules would also not help either, because the nodes of the SCS Hardware Landscape do not know a route to the IP addresses in the Wireguard VPN and so the predefined ACCEPT rule would not be enough, because the return direction of the packets to the Wireguard server could not be identified because of non existing routes on the nodes.
My suggestion would be to use “nft” to create the rules specifically for the VPN client IP instead of “iptables” and to use a SNAT/masequerading rule instead of an ACCEPT for every destination network.
(see also)
Using NAT would have the advantage that network clients using the Wireguard server of a node would have the same access options as clients initiating a connection directly on the node.
You also do not need routes on the nodes and you can have multiple VPN gateways with the same VPN Client ip ranges.
However, this method has the disadvantage that the VPN IPs are masked and therefore network-connections are less easy to assign to a VPN user.
References to existing reports
No response
Severity
low
Urgency
low
The text was updated successfully, but these errors were encountered: