Comment 7 for bug 2063353

I think I can now report we can reliably repeat this in a live environment.

I re-ran the aforementioned script in a freshly booted live environment (I had been reusing the same one all throughout the chronology of the last comment) and was able to reproduce the behavior, more or less. After 300 tests, I had amassed 103 failures, amounting to a 34.33% failure rate.

It seems you need to run `plasma-discover --backends packagekit-backend,snap-backend` at least 5-6 times before you are going to get a failure.