While working on #28, I learned that CPython3.12 cannot build lxml < 5.0.0. See streamlink/streamlink#5336

Referencing this code in connect.py: by setting sock.settimeout(None), it becomes possible that a send_rpc call to a device hangs indefinitely if the device has gone into an unresponsive state without closing the connection. It also violates the expectation of the calling code that the timeout parameter on connect_ssh will put the socket in timeout mode. (I note as I am writing this that the docstring does say it's a timeout 'when connecting the socket, but it's non-obvious until entering this failure mode that this timeout is changed to None` after the connection is first established.

I'm not submitting a PR to fix only because it's not entirely clear to me what the intention was in setting this to None in the first place.

In "connect_ssh" if the calls to either "invoke_subsystem" or "open_session" fail for some reason, the transport is still connected and is never cleaned up, because we never create the SshSessionSock or the Session object.

In "connect_ssh" if the Session "init" call fails, the session object won't be returned, meaning there's no way to clean up the created thread / socket.

It would be useful to gate releases with testing against a real life NETCONF server. We could probably use a netopeer2-based Docker image to act as the server.

• Test SSH connections
• Test TLS connections

Add support for the eD25519Key type for hostkey / bkeys.

Add support for get-data.

get-data (and edit-data) are defined in https://datatracker.ietf.org/doc/html/rfc8526

The "get-data" command has a lot more options than other commands. Specifically:

• a datastore option must be specified (as defined in https://datatracker.ietf.org/doc/html/rfc8342).
• filtering, subtree-filtering or xpath-filtering.
• config-filter (boolean)
• origin-filters (or negated-origin-filters)
• max-depth
• with-origin

I don't know when we'll do this exactly, but this issue will be for tracking things that need to be done whenever we do the next major rev release:

• Remove the deprecated timeout parameters from the connect_* functions
• Ratchet python requirement up to 3.8 (or whatever the oldest non-EOL Python release is at the time)

connect_ssh currently doesn't allow for a ssh connection to pass in a hostkey to verify. netconf_client can pass this directly to paramiko when calling connect.

See Paramiko Transport constructor at:

https://github.com/paramiko/paramiko/blob/main/paramiko/transport.py#LL454-L457

The result of this is that netconf_client ssh connections that should time out are susceptible to hang forever instead.

Working on further investigation/resolution, putting this here for situational awareness.

Hi All,

am facing callhome issue only on windows machine but not on linux machine,

File "C:\Users\SL-BLR-10204\AppData\Local\Programs\Python\Python37-32\lib\site-packages\netconf_client\connect.py", line 171, in exit
self.server_socket.shutdown(socket.SHUT_RDWR)
OSError: [WinError 10057] A request to send or receive data was disallowed because the socket is not connected and (when sending on a datagram socket using a sendto call) no address was supplied

But, if I put line 171 from connect.py, in try & except its working fine on windows
can you please let me know whats the catch?

Hello

IPv6 and call-home seems to be very unstable. I can see that when I just do a "accept_one_tls()", Windows "netstat -a" does not display a listening IPv6 socket.

The "funny" thing is that there is a good chance that the connection is eventually accepted, so somehow the IPv6 socket is created.

Is there a way to tell the module to open both IPv4 and IPv6 listening sockets? Or is there a way to force it to do IPv6 only?

I see the following error message while using send_rpc() using python 3.6 version.

File "c:\python36\lib\site-packages\netconf_client\session.py", line 152, in frame_message_11
return header + msg + footer
TypeError: can't concat str to bytes