There's an open PR to fix some mutex contention around the host_source and ring refresh concurrency that @nanassito contributed. See #1944.
Here's a mutex blocking profile taken from a production runtime sample under load.
I know that this doesn't account for all of the mutex contention, but it's still a decent percentage. Can we get some attention on that PR and have a maintainer take a look?
There's an open PR to fix some mutex contention around the host_source and ring refresh concurrency that @nanassito contributed. See #1944.
Here's a mutex blocking profile taken from a production runtime sample under load.
I know that this doesn't account for all of the mutex contention, but it's still a decent percentage. Can we get some attention on that PR and have a maintainer take a look?