Expose logs/metrics for types of build failures
In gitlab-com/gl-infra/production#6563 (closed) we had a very specific type of failure where cloning submodules failed under a specific set of conditions (customers who did not have the .git
extension on submodule paths).
Without directly looking at customer job logs, this error was difficult to diagnose as these types of errors are reported as script_error
.
It was brought up in https://gitlab.com/gitlab-com/gl-infra/reliability/-/issues/15415#note_871167144 that would be helpful to have error categories in logs or metrics that would have aided in the diagnosis of this problem as the change was rolled out.