How to get the exit status of a failed pipe in Bash
06/16/2017 • 2 minute readThe traditional method of accessing the exit status of a command in Bash is through the $? internal variable like so :
$ true
$ echo $?
0
But what if your first command fails and you pipe it to another command that will always succeed?
$ false
$ echo $?
1
$ false | tee log
$ echo $?
0
Thankfully, there are a few different ways to handle accessing the exit code for the correct command in the pipe.
Via $PIPESTATUS
The $PIPESTATUS array is made up of the exit status of each command in a pipe. Learn more on the Advanced Bash-Scripting Guide.
You can use:
$ false | tee log
$ echo "${PIPESTATUS[0]} ${PIPESTATUS[1]}"
1 0
You can also use ${PIPESTATUS[*]} or ${PIPESTATUS[@]}.
Note: $PIPESTATUS suffers from the usual issue of misbehaving when you pipe ls (which is why I never pipe ls when possible), so $PIPEFAIL may be a better option in those cases.
Via $PIPEFAIL
In Bash version 3 and beyond, you can use set -o pipefail to set the exit status to the status of the last failed command in a pipe instead of the last command in a pipe.
Via mispipe
The moreutils package includes a command named mispipe that has similar functionality to ${PIPESTATUS[0]}.
I don’t personally like installing extra software when I can just use Bash functionality, but not every user or OS has that option.
mispipe works by taking two arguments and returning the exit status of the first:
$ mispipe true false
$ echo $?
0
This has the advantage of only returning the exit status of the first command instead of triggering on any failure of the pipe sequence.
Conclusion
There are a variety of different ways to recover the exit status of a failed pipe in the sequence, and there are a lot more that I didn’t cover like creating your own named pipes with mkfifo or a tmp file with mktemp that stores the array of each exit status. These aren’t typically needed on a standard Linux/Bash install that you aren’t too worried about being POSIX compliant though.