riotkit-org / tracexit

tracexit

Trace command till it Exits

Spawns process and tracks it till it runs, then stores its status code in a file. It's a simple CLI application just like tee, can be used together with tools like tee.

Input

tracexit does not take any commandline switches except --help / -h, everything is passed through to the child process. To use --help / -h it must be written as first argument, right after tracexit.

Environment variables:

• TRACEXIT_EXIT_CODE_PATH: File location where to store exit code of a process when it exits

Getting tracexit

Take a look at releases tab and pick a version suitable for your platform. We support Unix-like platforms, there is no support for Windows.

You can use eget as a 'package manager' to install tracexit

# for pre-release
eget --pre-release riotkit-org/tracexit --to /usr/local/bin/tracexit

# for latest stable release
eget riotkit-org/tracexit --to /usr/local/bin/tracexit

Usage examples

Export exit code to file "/tmp/exit-code" after process finishes.

export TRACEXIT_EXIT_CODE_PATH=/tmp/exit-code
tracexit mysqldump -u root -psomething | tee out.log

Set extra environment variables and pass to the process

Useful when spawning a process in an environment, that does not allow adjusting shell settings.

Use case: Kubernetes library in Python does not allow passing environment variables. To preserve full escaping possibility (spawning process as list of strings) and avoid invoking extra /bin/bash we use tracexit to set environment variables within a wrapper process.

tracexit env:SOME=thing mysqldump -u root -psomething

Appending paths:

tracexit env:PATH+kubectl=/opt/kubectl kubectl get pods -A

tracexit can concatenate variables - which means taking e.g. "PATH" real value and appending ":/opt/kubectl" in above example.

Use case

Invoking a process in a container, or in remote shell, but the library does not return exit code of executed command. tracexit will store exit code in a text file, so a next shell call can read it.

Example - usage with docker library in Python

docker library allows performing docker exec operations, but it does not track exit codes. Only command's output is returned. The process could be additionally detached, which also makes not possible to track its exit code.

# ...
container = client.containers.get('...')
exit_code, stream = container.exec_run(['tracexit', 'some-command'], stream=True)

# exit_code = None
# stream is of Generator type, after it ends we do not get exit code, only we know when output ends

Example - with kubectl

kubectl exec -it deployment/my-deployment-name -- tracexit env:PWD=/mnt/ some-command-here