Andrew H Schwartz

Modified from 2called-chaos's response, this is my workaround that allows you to call binding.hpry (or any Object#hpry) that will (a) not clobber your console history, and (b) function otherwise identically...

@ekroon this still uses an implicit, single, default experiment class. This doesn't allow flexibility to, for example, run experiments that have different publish, raise, and enabled behaviors. I suggest @omkarmoghe's...

We are hitting what appears to be this issue, running ruby 3.3.1. I haven't tried to instrument additional logging on the fork hooks yet, though it sounds like from OP's...

After adding logging to the various available hooks, I have a bit more info to @PatrickTulskie and @nevans questions: Context: - Ruby 3.3.1p55 - Gems: resque 2.6.0, redis 4.8.1, rails...

Unfortunately it's inconsistent. Our system seems to go for hours without issue, then things start going south. And in the case of the worker I documented here, this is in...

I was able to get strace and rbspy working, and got another example to look at. Since it's a bit b uried, i'll highlight here the bit below about `friendly_errors`,...

rbspy for the parent process ``` $ sudo /home/runner/.cargo/bin/rbspy record -p 1 Time since start: 101s. Press Ctrl+C to stop. Summary of profiling data so far: % self % total...

Those are good suggestions to experiment; unfortunately I think I've run out of time to keep this condition going in our prod application and will have to resort to reverting...

We downgraded to Ruby 3.2 last week and haven't had any issues since. So FWIW, while telemetry may be closely involved here, but this does appear to be uniquely a...

I brought this up in #3184 and can add an additional perspective. In our case: - We are instrumenting a rails application, following [the documentation on how to do so](https://docs.datadoghq.com/tracing/trace_collection/automatic_instrumentation/dd_libraries/ruby/#rails)...