see e.g. http://localhost:8080/job/Builds/job/debs/job/deb-mozjs60-amber-phone-staging-aarch64/2/console

when a build fails like

I: unmounting dev/shm filesystem
I: unmounting proc filesystem
I: unmounting sys filesystem
I: Cleaning COW directory
I: forking: rm -rf /var/cache/pbuilder/build/cow.15260
gbp:error: 'git-pbuilder --jobs=auto -nc --build=full -sa' failed: it exited with 1
[Pipeline] echo
Script returned 1. Aborting build.

but then the upload step tries to pick up the pieces:

Running on Jenkins in /var/lib/jenkins/workspace/Builds/debs/deb-mozjs60-amber-phone-staging-aarch64@2
[Pipeline] {
[Pipeline] sh
...
+ + echo ./*.changes
wc -w
+ [ 1 -ne 1 ]
+ ls ./*.changes
ls: cannot access './*.changes': No such file or directory
+ changes_file=
[Pipeline] echo
Script returned 2. Aborting build.

and only this makes the pipeline stop. The pipeline should have stopped after the first build failure.