As a first step, it's fine to stop in rgw_bucket.cc. Eventually, we'll want a trace to cover each rados request, since those account for most of our time. But for your PR, I think it's better to get broad coverage of lots of different code paths, rather than going really deep into a few. On Wed, Jul 22, 2020 at 6:25 AM Abhinav Singh <singhabhinav0796(a)gmail.com> wrote: operation in a bucket, then upto what level should trace it so for the trace to be useful, for example after a bucket is created , then after that the major time is taken by `store_bucket_instane_info` and `store_bucket_entrypoint_info`, in the file `rgw_bucket.cc` so is it necessary to go deep inside this function(they are calling `RGWSI_Bucket_SObj` inside), or till rgw_bucket.cc is fine?