[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: SSH channels



Phil Shafer wrote:
Martin Bjorklund writes:
In draft-ietf-netconf-ssh-05,txt, section 5 implies that multiple
channels are not supported (at least not very cleanly), since it
explicitly states that when an agent receives a <close-session>, it
shall terminate the SSH session and the TCP connection.

I don't see this.  Section 3 of draft-ietf-netconf-ssh-06.txt reads:

   After the ssh-connection service is established, the client will open
   a channel of type "session", which will result in an SSH session.

So the "SSH session" refers to the SSH channel of type "session".

Then in Section 5, we have:

   ... When the agent processes a <close-session>
   command, the agent shall respond and terminate the SSH session.  The
   agent MUST NOT process any RPC commands received on the current
   session after the <close-session> command.

If you see the use of netconf over a distinct channel as a distinct
netconf session, then the close-session on one would not affect the
other.  You get this for free with openssh, since the ssh daemon spawns
the subsystem as a child process, cleaning up when all children have
been reaped.  If you start two netconf subsystems, sshd will continue
until both have died.

The other reading would mean that <close-session> would need to kill
the parent sshd.  Even so, if the only thing blocking this is the close
RPC, we could fix the close RPC.


Fair enough.
Like almost every thing else in this protocol,
there are usually multiple passages to cite that may be
interpreted in particular ways that support many viewpoints
on the same complex problem.

(Like another book I know of, but that's out of scope ;-)



Thanks,
 Phil

Andy

--
to unsubscribe send a message to netconf-request@ops.ietf.org with
the word 'unsubscribe' in a single line as the message text body.
archive: <http://ops.ietf.org/lists/netconf/>