We're seeing an insignificant amount of time being spent waiting to write to respCh inside cluster's Put and it seems like this can easily be eliminated by making the respCh have a buffer. Since the cluster is limited by the speed of spin, this is causing hundreds of goroutines to be backed up waiting to write to callCh.
We're seeing an insignificant amount of time being spent waiting to write to
respCh
inside cluster'sPut
and it seems like this can easily be eliminated by making the respCh have a buffer. Since the cluster is limited by the speed ofspin
, this is causing hundreds of goroutines to be backed up waiting to write tocallCh
.