Problem with back tunnel
Brought to you by:
ndwinton
Win32.
1. UDP tunneling does not work at all using back tunnel
(listenmode), all versions. Please test it on Win32.
2. In version 2.5.2 server wants to connect server back
multiple times, it's great! But client accepts ONLY FIRST
connection, ignoring all other. If I use 2.4.x as client,
multiple connections are ok, but there is a protocol
mismatch error. Seems somebody added a bug in 2.5.2?
Logged In: NO
there are more problems/bugs in 2.5.2 in listenmode:
3. if you have more tunnel/target pairs, only 1st connected
tunnel will function, later connects hang up ( server becomes
deaf ).
4. multiuse mode (simultaneous connects on same ports, which
is default) dont work (i think because of error #3 above).
5. if you close Listening Client , Server process will not exit
and stays in memory (in 2.5.1 server exists).
version 2.5.1 has NOT these problems (3,4,5)
/Happy User
PS.: seems authors abbandoned this project? ;(
Logged In: NO
I will add cfg files (is example of connecting to VNC server at
tunnel port 21) :
my Client setup (only important keywords, I left keyID
keywords):
server false
serverhost SERVERHOST_IP
serverport 21
listenmode true
tunnel 10000:127.0.0.1:5900
My Server setup (only important keywords, I left keyID
keywords):
server true
clienthost CLIENTHOST_IP
serverport 21
listenmode true
target 127.0.0.1:5900/tcp
this is default setup, all Timeout keywords are default.
the logs at Client side are:
1st time:
2005-06-28-10:33:57: accepted connection from server
2005-06-28-10:33:57: validating server IP address
2005-06-28-10:33:57: requesting protocol version 0x202
2005-06-28-10:33:57: accepted protocol version 0x202
2005-06-28-10:33:57: requesting TCP mode
after closing connection I get:
2005-06-28-10:34:05: connection closed
then I try to connect for 2nd time!
2005-06-28-10:34:12: accepted connection from server
2005-06-28-10:34:12: validating server IP address
2005-06-28-10:34:12: requesting protocol version 0x202
2005-06-28-10:35:00: ERROR: failed requesting protocol
version
the same thing what user koteroff figured out.
When I add TimeOut keywords onto Server cfg file, to have a
faster timeout reply, like this:
serverconnecttimeout 10 # seconds between connect retries
acceptconnecttimeout 10 # wait up for protocol exchange
targetconnecttimeout 10 # abort hung target connects
connectattempts 10 # connect retries before server exits
I will get in server log:
2005-06-28-10:28:23: initiating connection back to client at
CLIENTHOST-IP:21
2005-06-28-10:28:23: failed to connect back to client at
CLIENTHOST-IP:21: (Bad file descriptor)
(note the times are different because of different run)
/Happy User
Logged In: YES
user_id=83226
I believe this is fixed in 2.5.3 (just released). The
problem was one that only showed up on Windows when running
"detached".