Tim Wojtulewicz

This must be something related to the data you're receiving. We run an automated benchmark that runs for 3 days, and pauses for 2 (so it starts a new one...

@JustinAzoff was kind enough to give me a script that basically spams a cluster with DNS logging. I have it configured so each worker in a 4-worker cluster is sending...

I guess I'll need that `jeprof` data from you after all if you can get it working. There may be something that's not showing up because my testing is only...

Was there anything to do with this issue? Did we ever get to a root cause for the memory usage? Are you still seeing it?

Interestingly, I noticed we have a oss-fuzz report that has backtraces suspiciously similar to the ones reported above. Unfortunately oss-fuzz says the minimized test case that produced it is invalid,...

This whole github issue is great debugging information and I feel like we should capture it in the documentation somewhere.

@fox-ds Is (1) fixed already or is it still an open issue with the pcap you provided?

I think there's also a secondary issue here where if you define a second `event`, it only appears to do the type checking for the first of them.

If that's the case, it's been "broken" like this for quite a while.

As part of the work for fixing the Tag types, we're adding the concept of a "parent" enum type that contains all of the value from a bunch of child...