Jobs command does not executed although I can run the command from the build machine

First issue here :slight_smile:

I would like to ask what could be the reason for my issue - where my setup is:

  • I have labgrid-client setup in my docker images and run that image in docker executor in CI. The purpose of the client is to upload (I use labgrid ssh - which uses ssh under the hood) the build images to the remote machine with scp. This is successfully executed.
  • Next, I try to upgrade the remote machine from the CI (labgrid ssh ‘command to upgrade’)
  • When the upgrade takes place, the uploaded image is deleted in my remote machine (the device reboots also)

What I see:

  • Copies the firmware to remote device successfully but does not run the upgrade command.
  • When I run the build with ssh enabled, I can run the same commands from the terminal without any issues
  • What could be the problem?

Output in circleci UI:
Shared connection to ****** closed.
connection lost (SSH error)
DEBUG: Stopping keepalive for *******

** Output when running command with ssh inside the build machine **
Shared connection to ****** closed.
connection lost (SSH error)
DEBUG: Stopping keepalive for ********
INFO: Socket already closed

What I don’t understand (or where I am stuck is why I see different behavior?)

Thank you for any kind of help :slight_smile:

Hello and thanks so much for reaching out and posting on Discuss!

Would it be possible to provide a build link to where you see this issue occurring?
Does the build fail before the step when running the upgrade command?
Are you able to rerun your build with SSH and attempt to run the upgrade command?

I look forward to hearing back from you!

Hi there, I have managed to fix the issue the other way. My problem was: I was running some test automation tool (labgrid which also has option to scp the files). As we have only one test device, I decided to fix that issue by using scp command directly from the shell. Only surprising issue was: I rerun the build with ssh and attempt the same steps (with same set of commands); and it runs without an issue (I did not dig around deeper, there should be something interesting going on because the remote device reboots). And we were thinking about if we use the remote runner setup; I asked for pricing around the same time; but never heard from anyone from CircleCI team :wink:


I am so glad to hear you were able to find a solution! Thanks for sending over the added info as well.

If you need assistance with pricing please feel free to submit a ticket using this portal: Submit a request – CircleCI Support Center

Happy building!