In bug 1296728 BenWa did some work to fix the breakpad-mac-update-symbols scripts to work with modern OS X update packages. We still need to get it running on a regular basis. I think this should mostly just involve creating a taskcluster hook that runs this script as a task on a daily basis: https://docs.taskcluster.net/reference/core/hooks/api-docs

I added some bits to run this as a taskcluster hook. Here's my latest attempt at running it: https://tools.taskcluster.net/task-group-inspector/#/crTM9LAiQx-Xo9vFdJaOGQ/Nox5NcBOT3K9O56SyAftkQ?_k=ic9na1 It hasn't run successfully yet, it needs some tweaking.

I spent some more time fiddling with this. It was running out of disk space, so I changed the worker type it was running on. Also, I had been using a fork of reposado because I had some changes so we didn't have to download every single software update Apple offers (there are a lot, and most of them are not useful here), but that meant we were missing the upstream changes to add the catalog URLs for 10.12, so we weren't picking up the latest updates. I made some more changes to handle that and it seems to be working, although it hasn't finished its latest run yet. The hook is defined here: https://tools.taskcluster.net/hooks/#project-socorro/fetch-mac-update-symbols The task defined in the hook just clones this repo and runs run.sh: https://github.com/luser/breakpad-mac-update-symbols That runs the run-taskcluster.py script, which schedules two other tasks to do the real work from the {fetch,upload}-task.json files in the repository. The latest run is here, it found a bunch of symbols from macOS 10.12.4 that are currently missing from crash reports: https://tools.taskcluster.net/task-group-inspector/#/Wd_6Gau4T6KO2EcGeosu0Q?_k=1esud1 I asked Socorro to reprocess all crashes with `libsystem_kernel` in the signature, so that should take care of most of them.

Also just in case anyone but me actually looks at the task status: the symbol upload task claims to have failed, but in reality the upload was successful, the script just timed out waiting for a response from the server. I'm going to call this FIXED since this is configured to run on a daily basis now.

I reloaded one of those libsystem_kernel crashes and it has a good stack now: https://crash-stats.mozilla.com/report/index/4a19e768-74ca-425a-b29b-71c730170427#tab-details