Sentry worker performance

Hi everyone,

We are setting up sentry in our company infrastructure. We have set up 8 sentry workers (on different host), each with 4 celery workers. However, with this setting, we can only process ~500 events per second. Is this expected, or there are some tuning tricks that are recommended to make sentry worker run faster?

I am really curious about what is the average throughput for a typical sentry worker?

For other settings, we use redis as cache, rabbitmq as broker, riak as kv store, and postgres as db (with pgbouncer). I can observe that there is really high CPU usages of sentry workers, and there is no I/O blocking, or db issue.

Thank you so much!


Can anyone share your experience please? We are really suffering from performance issue.