Colleagues, I apologize if I didn’t understand something correctly and
if I’m writing irrational things.
Having looked at the hooks mechanism, I came to the conclusion that if
several clients are used on the computer (for example, it is a router),
then the hooks called will be launched in the ${HOOKS_DIR}/${HOOK}.d
directory when events occur in any logical direction. Even if these are
hooks for a different direction.
For example, when connecting to one server, you need to enable NAT on
the interface, and when connecting to another, NAT is not needed, but
you need to set the remapping of some ports and this needs to be
distinguished somehow.
In principle, you can find the address of the VPN collector in the
environment variables and navigate according to it. But there are a
number of ambiguities here. Firstly, the domain name is not transferred
to environment variables, but only the IP address resolved from it. And
it can change, sometimes without warning, that’s why DNS was invented.
Secondly, the domain name may also change and you will still have to
rewrite the script code with the hook.
I believe that the only reliable criterion may be the logical name of
the direction, which is set when the client starts. It is passed into
environment variables and lets scripts know that this is a job for them.
For example, --direct=DIRECT_NAME and $direct in the process environment
variables. Also, if the direction name is specified, then hooks can be
launched with the path ${HOOKS_DIR}/${HOOK}.d/${direct}/*, which will
allow you, in principle, not to launch hooks for other directions.
I'm only looking at a multi-client situation, but I figure the server
must have similar issues as well.
Please, colleagues, tell me if I am right in posing the question?
Ogogon.
_______________________________________________
openconnect-devel mailing list
openconnect-devel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/openconnect-devel
