Handle cases where a lot of tests fail due to a "systemic" problem

Sometimes, all tests in a job will start to fail because of a "systemic" problem, such as PG being overloaded etc. A good example is https://gitlab.com/gitlab-org/gitlab/-/jobs/4416118820 where a lot of tests fail with

        ActiveRecord::StatementInvalid:
          PG::ConnectionBad: PQsocket() can't get socket descriptor

That would probably result in many failure issues created for a transient problem that doesn't represent actual test failures.

Is there a way to detect these kind of failures?