Pipeline Build Setup: /bin/sh: /usr/bin/mkfifo: not found

[ktomk]

Bitbucket Pipelines shows me in the very first line of the very first operation ("Build setup") the following error message:

/bin/sh: /usr/bin/mkfifo: not found

then this first operation finishes without error, the pipeline step is just NOT executed and next is Build teardown. Here the full log:

/bin/sh: /usr/bin/mkfifo: not found
+ umask 000

+ GIT_LFS_SKIP_SMUDGE=1 retry 6 git clone --branch="test" --depth 1 https://x-token-auth:$REPOSITORY_OAUTH_ACCESS_TOKEN@bitbucket.org/$BITBUCKET_REPO_FULL_NAME.git $BUILD_DIR
Cloning into '/opt/atlassian/pipelines/agent/build'...

+ git reset --hard 0ffb0fce5f9a57d9d6f5f71a2e8e7ca25f948ea8
HEAD is now at 0ffb0fc prepare release

+ git config user.name bitbucket-pipelines

+ git config user.email commits-noreply@bitbucket.org

+ git config push.default current

+ git config http.${BITBUCKET_GIT_HTTP_ORIGIN}.proxy http://localhost:29418/

+ git remote set-url origin http://bitbucket.org/$BITBUCKET_REPO_FULL_NAME

+ echo ".bitbucket/pipelines/generated" >> .git/info/exclude

+ chmod 777 $BUILD_DIR

Searching for test report files in directories named [test-results, failsafe-reports, test-reports, surefire-reports] down to a depth of 4
Finished scanning for test reports. Found 0 test report files.
Merged test suites, total number tests is 0, with 0 failures and 0 errors.

This looks like an error to me in the build setup as the setup is using something it does not provide.

Also I'm a bit puzzled as my pipeline is made red with no specific error information on the level of the pipeline.

This is how it looks like in a screenshot:

[ktomk · Accepted solution]

The error

/bin/sh: /usr/bin/mkfifo: not found

is caused when Atlassian Bitbucket Pipelines starts the build container (one build container runs for each pipeline step).

The command on run which causes the error is:

/bin/sh -c exit $( (/usr/bin/mkfifo /opt/atlassian/pipelines/agent/tmp/build_result ⤦
  ⤥&& /bin/cat /opt/atlassian/pipelines/agent/tmp/build_result) || /bin/echo 1)

Point in case with the "busybox" image from Docker Hub in use is, that there is no

/usr/bin/mkfifo

but

 /bin/mkfifo

A workaround is to modify the image adding a symbolic link:

 <<'EOD' docker build -t "pipelines:busybox" -
FROM busybox
RUN set -eux; \
  mkdir -p /usr/bin; \
  ln -s /bin/busybox /usr/bin/mkfifo;
EOD

 The new image can be used until busybox is supported out of the box (again?) on Atlassian Bitbucket Cloud.

[Daniil Penkin]

Hello @ktomk,

Welcome to the Community!

I'm actually not a Pipelines expert, but does the build use the default Docker image or a custom one? My guess is that you use a custom image which has an issue. If my assumption is correct, have you tried running your pipeline locally using the same image?

Cheers,
Daniil

[ktomk]

Dear @Daniil Penkin  ,

thank you for your warm welcome!

The problem seems to be that the image I use has the "mkfifo" utility at "/bin/mkfifo" but Bitbucket Pipelines expects it to be at "/usr/bin/mkfifo".

Changing the Docker image worked for me, if it's helpful: a symbolic link could do the trick.

It's "busybox" from the official images on Docker Hub which is quite popular.

So probably it should be "/bin/mkfifo" or just "mkfifo" in the first place (as some Linux containers might differ there but have it in the "$PATH") when the container is run.

While looking into it, I also ran over: Understanding the bin, sbin, usr/bin , usr/sbin split (Rob Landley; Dec 2010; Busybox mailinglist) which has some folklore about these directories.