mirror of
https://github.com/neondatabase/neon.git
synced 2026-01-15 17:32:56 +00:00
fe615520dd
The issue is that get_vectored_reconstruct_data latency means something very different now with concurrent IO than what it did before, because all the time we spend on the data blocks is no longer part of the get_vectored_reconstruct_data().await wall clock time GET_RECONSTRUCT_DATA_TIME : all the 3 dashboards that use it are in my /personal/christian folder. I guess I'm free to break them 😄 https://github.com/search?q=repo%3Aneondatabase%2Fgrafana-dashboard-export%20pageserver_getpage_get_reconstruct_data_seconds&type=code RECONSTRUCT_TIME Used in a couple of dashboards I think nobody uses - Timeline Inspector - Sharding WAL streaming - Pageserver - walredo time throaway Vlad agrees with removing them for now. Maybe in the future we'll add some back pageserver_getpage_get_reconstruct_data_seconds -> pageserver_getpage_io_plan_seconds pageserver_getpage_reconstruct_data_seconds -> pageserver_getpage_io_execute_seconds