Paramiko issue and more verbosity

[Arzhel Younsi]

Hi!

I'm trying to gather more info before filing a ticket to you or Paramiko.

If I have two calls too close from each other to a juniper device (using the Ansible module), I get the error: "SSHException: Error reading SSH protocol banner" from Paramiko but with no other debug messages.

If I add a timeout of ~30s between the 2 calls (in my case a get_facts, then a install_config) it works fine.
I'd like to know how could I get more debug output from the underlying layers (py-junos-eznc and Paramiko)? And if anyone have an idea of what's going on.

Thanks!

[Rick Sherman]

Hi Arzhel,

You can get some more verbosity from Ansible using the option:

-vvv

That being said, if the Paramiko exception is bubbling up that means ncclient / PyEZ did not catch the exception.

Off the top of my head I suspect your device is limiting the number of TTY/SSH connections.  The Ansible modules open and close a NETCONF session when they are called, and the device may not be cleaning the session up fast enough.

You may be able to configure some logging on the device to diagnose the root.

http://www.juniper.net/documentation/en_US/junos14.2/topics/topic-map/netconf-traceoptions.html

-Rick

[Arzhel Younsi]

Thanks! I had a ssh and netconf rate-limit in addition to my playbook checking connectivity before doing the get_facts and install.

About the verbosity, -vvv only gives more ansible logs, I was wondering if it was possible to have the debug output of pyEZ and/or paramiko as well.