Hi,
We're trying out the new "consumer" group protocols as the default for our schema registry cluster since we've had issues with leaders not getting elected quickly, and we observed that the CPU utilization shot up and stayed there. The memory foot print is similar to before, but the CPU utilization is about 10x.
Not sure why its thrashing, there are no errors, and we correctly observed brokers being assigned as the group coordinator.
Could you please take a look at it?
Thank you!
Hi,
We're trying out the new "consumer" group protocols as the default for our schema registry cluster since we've had issues with leaders not getting elected quickly, and we observed that the CPU utilization shot up and stayed there. The memory foot print is similar to before, but the CPU utilization is about 10x.
Not sure why its thrashing, there are no errors, and we correctly observed brokers being assigned as the group coordinator.
Could you please take a look at it?
Thank you!