mirror of https://github.com/postgres/postgres
The pending entry was not used when incrementing its data, directly manipulating the shared memory pointer, without even locking it. This could mean losing statistics under concurrent activity. The flush callback was a no-op. This code serves as a base template for extensions for the custom cumulative statistics, so let's be clean and use a pending entry for the incrementations, whose data is then flushed to the corresponding entry in the shared hashtable when all the stats are reported, in its own flush callback. Author: Sami Imseih <samimseih@gmail.com> Discussion: https://postgr.es/m/CAA5RZ0v0U0yhPbY+bqChomkPbyUrRQ3rQXnZf_SB-svDiQOpgQ@mail.gmail.com Backpatch-through: 18pull/247/head
parent
dd74e599b8
commit
6e8286f9a1
Loading…
Reference in new issue