Comment 1 for bug 1354318

Quick notice: whoopsie-upload-all was never designed to be called in parallel from an upstart job. It was a script that we want to run at the end of merge proposals etc. in the CI machinery to upload all crashes that occurred during the test run. For that reason it must wait until everything has been uploaded, otherwise the testbed will be torn down and the .crash is lost.

For upstart we could call this with a --timeout=0, so that it behaves asynchronously. If we rely on the upstart job in the CI machinery now, we instead need a new script which does the "wait until everything is uploaded".

Adding flock()ing to the .crash files in process_report() sounds fine to me.