About

runp is a simple command line tool that runs (shell) commands in parallel to save time. It's easy to install since it's a single binary. It's been tested on Linux (amd64 and arm) and MacOS/darwin (amd64).

# file containing commands we want to run in parallel
$ cat cleanup.txt
kubectl delete all -l what=ckad
kubectl delete ing -l what=ckad
kubectl delete cm -l what=ckad
kubectl delete secret -l what=ckad

# run commands from the file in parallel
$ runp cleanup.txt
--> OK (1.05s): /bin/bash -c "kubectl delete secret -l what=ckad"
No resources found
--> OK (1.05s): /bin/bash -c "kubectl delete ing -l what=ckad"
No resources found
--> OK (1.08s): /bin/bash -c "kubectl delete cm -l what=ckad"
No resources found
--> OK (3.78s): /bin/bash -c "kubectl delete all -l what=ckad"
No resources found

# clone all github's organization repos (coming from stdin) in parallel
$ gh repo list $GHORG --limit 1000 | cut -f 1 | runp -p 'gh repo clone'

# check existing commits in all repos (coming from stdin) in parallel
$ ls | runp -p 'gitleaks detect --no-banner -s'

You might also like to see a related blog post.

Installation

Download the latest release to your bin folder (or some other folder on your PATH) and make it executable:

# Choose your system and architecture.
export SYS=linux  # or darwin
export ARCH=amd64 # or arm

# Download the binary and make it executable.
curl -L https://github.com/jreisinger/runp/releases/latest/download/runp-$SYS-$ARCH \
-o ~/bin/runp
chmod u+x ~/bin/runp

Description and usage examples

Commands (or parts of them) can be read from files or stdin and must be separated by newlines. Comments and empty lines are ignored.

You can use shell variables in the commands.

runp exit status is 0 if all commands exit with 0 (OK).

runp prints a progress bar and info about command's execution (OK/ERR, run time, command to run) to stderr. Otherwise stdin and stderr works as you would expect.

Compress files

find . -iname '*.txt' | runp -p 'gzip --best'

We used -p to add prefix string to the commands (filenames in this case)

Measure HTTP request + response time

export CURL="curl -w 'time_total:  %{time_total}\n' -o /dev/null -s "
for n in {1..10}; do echo $CURL https://yahoo.net; done | runp -q # 10 requests

Find open TCP ports

cat << EOF > /tmp/host-port.txt
localhost 22
localhost 80
localhost 81
127.0.0.1 443
127.0.0.1 444
scanme.nmap.org 22
scanme.nmap.org 443
EOF

cat /tmp/host-port.txt | runp -q -p 'nc -v -w2 -z' 2>&1 | egrep '(succeeded!|open)$'

We used -q to suppress output from runp itself. Then we redirect stderr to stdout since netcat prints its messages to stderr. This way we can grep netcat's messages.

Clone many repos

gh repo list <owner> --limit 1000 | while read -r repo _; do 
  echo gh repo clone "$repo"
done | runp

gh is the GitHub's CLI tool.

Development

Test and install (to ~/go/bin/):

make install

Test and build (cross-compile for multiple platforms):

# NOTE: don't forget to bump up the version in main.go!
make release