Opentrons: bug: SSH via WiFi times out after several hours of solid connection

[Matt-Zwimpfer]

Overview

Able to establish a connection to the OT-2 via SSH over WiFi. However, after several hours of standby, the connection will be lost.

Steps to reproduce

1. Set up SSH connection and connect via SSH to the OT-2 with WiFi.
2. Keep connection on standby for several hours
3. SSH connection is lost

Current behavior

After several hours stand by, the ot2 will disconnect due to telnet port 22 giving a time out. The user is then unable to SSH to the OT-2 unless they connect to 48888 port in browser and open a terminal in the Jupyter Notebook, and type in netstat –a. Then SSH connection is able to be reestablished.

This was only succeeded with the netstat –acommand. Other commands (ls, pwd, ping <computer ip>) didn't work.

Expected behavior

The SSH connection shouldn't time out over an extended period of time.

[mcous]

SSH is not known for its ability to hold solid connections over a long period of time. Various server and client configurations (or good-old-fashioned network blips) can cause an SSH connection to go down, especially on the time scale of hours. This description...

The user is then unable to SSH to the OT-2 unless they connect to 48888 port in browser and open a terminal in the Jupyter Notebook, and type in netstat –a

...certainly makes it sound like the long-lived SSH connection is doing something like chewing through resources until something at the OS level messes up.

The OT-2 has both tmux and screen available so a user can keep a session running even while not actively connected over SSH. I think we should recommend that users use these tools if they're going to be running anything on the robot via terminal for really any amount of time over a few minutes

[mattwelch]

Bug triage determination: LHM