I absolutely hate what I'm about to do, and hate when people do it to me in our communities and servers. I'm going to report an issue with the most vague information ever. :( (I don't have much to go off as we just had some major emergencies on our networks and everything was a war zone lol.)

Simply put, the last LP I been using has been built as of:
f556c75

Everything was great, however, somewhere since the above commit and the build I used (d8a7d8d), LP started locking up bungee at 100%.

I been reproducing it by:
(Easily reproduced 4 times)

• Starting Bungee
• Start a few back-end servers
• Restart one of the back-end servers once or twice
• Eventually bungee locks CPU right up at 100% and its rip from there.

Cannot reproduce on our testing environment with no players, can only reproduce with active players, thus sadly I can't "experiment with the builds in between to find exact culprit" as it takes a serious toll on our player base.

Rolled back down to the first commit linked above and I haven't been able to reproduce a single time (trying for over an hour now).

I really hate submitting this issue, so if it must be closed for insufficient info it's all good :)

This should be resolved in v4.1.100 as of commit 10c0efa

The summary is that instead of simply invalidating the cached data, it was being reloaded every time a change occurred.

If two changes occurred within the same instant, two reload tasks would be executed - and yeah, that explains the high CPU usage.