John Barham

I had this exact problem so I created a Django management command to do bulk deletions of failed jobs. The code gist is [here](https://gist.github.com/jbarham/de89f85e900bf5f3adce3f0e4d65c5d9).

Sure you can delete 50 jobs at a time from the web UI. But if you have hundreds or thousands of failed jobs to delete that becomes very tedious.

I experienced this exact issue. Thanks to advice from a colleague, the cause seemed to be that my `/etc/machine-id` was empty. Fix was to run `sudo systemd-machine-id-setup`. But ISTM to...

Hah, you beat me to it. Here's another approach that wraps the cache map in a thread-safe container: https://github.com/jbarham/Key-Value-Polyglot/blob/master/memg.go