search

GaloisInc / BESSPIN-CloudGFE

The AWS cloud deployment of the BESSPIN GFE platform.
Apache License 2.0
2 stars 2 forks source link

Connectal is flaky or something is not properly cleared #121

Closed rtadros125 closed 4 years ago

rtadros125 commented 4 years ago

If we run connectal more than a few times on the same instance, sometimes it starts misbehaving. Doesn't boot or the netword doesn't work well. Is there something that needs to be cleared between runs that we don't clear? Also, it seems the process never quite terminate cleanly. We use interactive spawning using python.pexpect, and sending ctrl-A x does not seem enough to kill the process.

jameyhicks commented 4 years ago

Will the researchers be using the serial console? If not then I don't think the ssith_aws_fpga process is listening for ctrl -A x and I recommend killing it with SIGTERM.

jrtc27 commented 4 years ago

The FETT tool uses the serial console to set up the network. (But I'm pretty sure the process will always be listening for that sequence; the configuration is just for which device is pulling out of the pipe at the other end of the input processing thread.)

andrew-bivin commented 4 years ago

Assigning folks that might have some insight on this issue

rtadros125 commented 4 years ago

Update: we figured out why Ctrl-A x didn't work. We were using \x01 x, but we should've been using \x01x instead. I will let you know if I encounter any flakiness after this update. I will close the issue, and re-open it later if I witness the problem again.