How can I capture exit code of run command and reuse in another job ?

[Constantin07]

Question

Body

Hi, I'm trying to use the exit code of a binary in order to decide to run next job or not and it doesn't work as expected.
Here is the code:

jobs:
  plan:
    name: Plan
    runs-on: ubuntu-latest
    outputs:
      exitcode: ${{ steps.plan.outputs.exitcode }}
    steps:
      - name: Plan
        id: plan
        run: |
          <some command here executed>
          echo "exitcode=$?" >> $GITHUB_OUTPUT
        continue-on-error: true

  test:
    name: Test
    runs-on: ubuntu-latest
    needs: [ plan ]
    steps:
      - run: echo ${{ needs.plan.outputs.exitcode }}

But the echo in the test job doesn't print any value:

Run echo 
  echo 
  shell: /usr/bin/bash -e {0}

What am I doing wrong ?

[airtower-luna]

By default the shell for run steps is configured to exit as soon as a command fails (that's the -e in the shell definition). So for any exit code != 0 the echo to set the output will never run.

You should be able to override that by setting the shell yourself, or (more obvious) disable the "exit on failure":

      - name: Plan
        id: plan
        run: |
          set +e
          <some command here executed>
          echo "exitcode=$?" >> $GITHUB_OUTPUT

Mind that this way the exit code of the last command will become the exit code of the step, so it won't be marked as failed. If you don't want that, you can carry the exit code over:

      - name: Plan
        id: plan
        run: |
          set +e
          <some command here executed>
          exitcode="$?"
          echo "exitcode=$exitcode" >> $GITHUB_OUTPUT
          exit "$exitcode"

[nicfv]

Randomly came across this and it was exactly what I needed. The set +e was what I was missing. I cannot find any documentation online - can you please explain what it does? Based on the behavior, it seems like it will allow scripts to continue sequentially running even after one fails.

[airtower-luna]

The full documentation is here: https://www.gnu.org/software/bash/manual/html_node/The-Set-Builtin.html

Summary for -e:

Exit immediately if a pipeline […] returns a non-zero status.

And the plus just disables that behavior (applies to all set options):

Using ‘+’ rather than ‘-’ causes these options to be turned off. The options can also be used upon invocation of the shell. The current set of options may be found in $-.

[piranna]

I'm not able to make it work. Instead a command, I'm calling a bash function I've defined before executing set +e, maybe that's the reason? And after getting the exit code I execute set -e to enable it again, in case it can affect.

[nicfv]

@piranna you would need to execute set +e first. I don't know if set -e is necessary.

[piranna]

I'm using set +e as first command in my script, don't know if just running it before when needed would be enough. Anyway I think my error was that I was testing -eq -1 meanwhile the exit code is 255, so it seems to work now... :-) Didn't tried to do back the set +e / set -e combo, though, so don't know if it works.

[Rayshard]

Another option that I've used is to trap EXIT at the start of the script. In this way, you can have multiple commands in the script, the script still exits when a command fails, and you get the exit code of the command.

- id: my-step 
  run: |
      trap 'echo "exit-code=$?" >> "$GITHUB_OUTPUT"' EXIT
      cmd-1
      cmd-2
      ...

[sohale]

trap
Thanks for saving lives