On Sun, Jan 25, 2015 at 8:54 AM, Mike Gabriel <mike.gabriel@das-netzwerkteam.de> wrote:
Hi Michael,
On So 25 Jan 2015 14:12:07 CET, Michael DePaulo wrote:
On Sun, Jan 25, 2015 at 3:31 AM, Kristof Hardy <kristof.hardy@catsanddogs.com> wrote:
Hi Mike,
thanks for the new release. I just updated and looks like I can't connect anymore.
I get an error on launching my Mate desktop: An unknown error occured during authentication! (I'm using pubiuc key authentication and I'm on Win8.1)
It worked fine with the previous version I used:
http://code.x2go.org/releases/binary-win32/pyhoca-gui/previews/0.5.0.0/PyHoc...
Let me know what I can do to get debug logs to you or something? I'm also on IRC if that's easier. (harkx)
Regards,
Kristof.
I observed the following:
If I configure PyHoca-GUI to use my password-protected key file (header below), I get the error: not a valid EC private key file!
If I configure PyHoca-GUI to "Discover SSH keys or use SSH agent for X2GO authentication", and then launch Pageant and load my PuTTY-formatted private key into Pageant, I get the error you got: An unknown error occurred during authentication
Did you do the latter?
Note that I can connect to this server using password authentication.
-Mike#2
It appears that the key file and pageant issues are separate issues. Let's focus on the latter.
if you launch pyhoca-gui with --libdebug from a terminal (msys Git terminal, or cygwin shell), then pyhoca-gui will throw the "unknown" exception on the command line.
If you launch it from the command-line or the start menu with the debug options, then it creates a log file such as this one: C:\Users\mike.DEPAULO\AppData\Roaming\pyhoca-gui\PyHoca-GUI.1516.log
It may well be, that the issue is solved in python-x2go 0.5.0.3. But not sure.
I am using the master branch of python-x2go, which is the unreleased version of 0.5.0.3.
Please report back what exact exception is thrown. Can I test pyhoca-gui on the WinBuilder machine?
See the log for the new version.
I have little experience with Python. Buy based on my Java experience, this bug looks like this is an incredibly simple regression in the paramiko library. (0.5.0.0-pre02 had paramiko 1.14.0, 0.5.0.3-20150124 has paramiko 1.15.2) https://github.com/paramiko/paramiko/compare/v1.14.0...v1.15.2#diff-9
I think line 221 should have "value" replaced with "descriptor", or lines 222 & 223 should have "descriptor" replaced with "value"
-Mike