How to investigate install.sh errors?

Lumberyard Discussion
#1

From GameLift console:

image
image1748×311 17.9 KB

it is clear the install.sh script failed, but there is no log to download.

$ aws gamelift describe-instances --fleet-id fleet-9930ecdd-bbc2-456a-8f04-b8bdc9b8d8ab
gives me no instances, so I can’t ssh them to check the log file written by install.sh

How can I investigate why my install.sh script failed?

Thank you

#2

I could be wrong, but your instance should remain in your fleet for some time after failures so you can debug. I would ensure you are setting the right region in your describe-instances call ie does list-fleets show your fleet.

Secondly, to debug install.sh failures:

  • Validate it locally
  • Open the fleet debug ssh port on fleet creation and remote into the instance as soon as its attached to the fleet.
    • Run the install.sh on the box again after it fails to see if its todo with missing things or poor pathing.
    • Grab the installer logs from the instance

I will pro-actively reach out to the GameLift service team so they can log at that fleet (but I need to know the region of the fleet as well)

#3

Never mind, I already deleted the fleet.

I worked this out making the install.sh script to not block on errors.

Yes the region was the right one and I couldn’t open any ssh to nothing due the fleet had no instances; more than this validating the install script.sh locally is not a final solution… if it fails due to host configuration or libraries differences…

Just be warned about the difficulties to investigate such cases: in case of install.sh error is not possible to investigate the problem because:
a) no log can be downloaded from the web console
b) no instance is created

Cheers