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
Right now the L4 proxy really only transfers the UDP/TCP payloads, and does not include the L4 headers. Thus it functions more like an application-level gateway.
Protocols that may include the original L3/L4 information in the application layer payload may fail. The obvious case is the authentication header in IPsec (see RFC 2402). RFC 3027 gives a pretty decent overview of some other problems when network addresses (and/or ports) are modified.
Will need to devise a better way to do the proxying in the future. For now, it seems to work with basic single-connection traffic that doesn't need a secondary channel/connection (e.g. works for SSH and HTTP, but difficult to get it to work with HTTPS).
The text was updated successfully, but these errors were encountered:
- Like the UDP version, the traffic is passed back along the path of
VNFs in reverse order, but does not re-enter the VNFs.
- Barring other improvements (erroneous error messages, stateful
close, passing of errors back through the chain, etc.), this
should commit should resolve the TCP/UDP case for #50.
- As noted in #51, however, we currently pass payloads without
headers, and proper tunneling that includes headers should be
addressed within that issue.
- Like the UDP version, the traffic is passed back along the path of
VNFs in reverse order, but does not re-enter the VNFs.
- Barring other improvements (erroneous error messages, stateful
close, passing of errors back through the chain, etc.), this
should commit should resolve the TCP/UDP case for #50.
- As noted in #51, however, we currently pass payloads without
headers, and proper tunneling that includes headers should be
addressed within that issue.
- Like the UDP version, the traffic is passed back along the path of
VNFs in reverse order, but does not re-enter the VNFs.
- Barring other improvements (erroneous error messages, stateful
close, passing of errors back through the chain, etc.), this
should commit should resolve the TCP/UDP case for #50.
- As noted in #51, however, we currently pass payloads without
headers, and proper tunneling that includes headers should be
addressed within that issue.
Right now the L4 proxy really only transfers the UDP/TCP payloads, and does not include the L4 headers. Thus it functions more like an application-level gateway.
Protocols that may include the original L3/L4 information in the application layer payload may fail. The obvious case is the authentication header in IPsec (see RFC 2402). RFC 3027 gives a pretty decent overview of some other problems when network addresses (and/or ports) are modified.
Will need to devise a better way to do the proxying in the future. For now, it seems to work with basic single-connection traffic that doesn't need a secondary channel/connection (e.g. works for SSH and HTTP, but difficult to get it to work with HTTPS).
The text was updated successfully, but these errors were encountered: