How to work around PERSISTENT_COMMAND_TIMEOUT

[jean-christophe manciot]

Hi there, 

Every ansible network command has a PERSISTENT_COMMAND_TIMEOUT which "controls the amount of time to wait for response from remote device before timing out persistent connection". 

Actually, when using the net_get module with a very large file, the command times out during the transfer, even though both sides are still exchanging SCP messages, meaning the transfer is still underway & active. 

Shouldn't it time out **only** when there is no more communication between both nodes for the PERSISTENT_COMMAND_TIMEOUT period of time?

- Either we are supposed to guess how long each transfer is going to last, depending on the size of the file, the network load, the CPU load on both sides & in between and so on and adjust  PERSISTENT_COMMAND_TIMEOUT accordingly

- or there is way to avoid the command to timeout during the transfer, while still timing out when the target does not answer anymore after some time

Any suggestion?

[jean-christophe manciot]

I have just found a mini workaround with ansible_command_timeout which can be used per command (instead of modifying the global PERSISTENT_COMMAND_TIMEOUT for all commands):

  vars:
        ansible_connection: network_cli
        ansible_network_os: nxos
        ansible_password: "{{ connections.ssh.password }}"
        ansible_user: "{{ connections.ssh.username }}"
        ansible_command_timeout: "{{ firmware.transfer_timeout }}"
  net_get:
...

However, if there is a real issue during the transfer so that the target does not answer any more, we have to wait until ansible_command_timeout times out before moving on, although the transfer issue may have happened at the beginning of the transfer.