Nothing is using StrategyNotifyBgWriter() anymore
While browsing through some of the clock-sweep code I noticed that the
only place StrategyNotifyBgWriter() is called now is in
BackgroundWriterMain()[1]. Presumably this isn't what's desired. If
nothing else, it means the function's description comment is wrong, as
are comments in BackgroundWriterMain(). This isn't new; 9.2 shows the
same thing and that's when the function was added. I'm not sure what the
right fix here is, since ISTM joggling bgwriter for every call to
BufferAlloc() would be overkill.
1:
https://doxygen.postgresql.org/freelist_8c.html#aabbd7d3891afc1d8531c3871d08d4b28
Answering my own question... I now see that the wakeup does in fact happen in StrategyGetBuffer(). Sorry for the noise.
On 6/16/22, 5:32 PM, "Jim Nasby" <nasbyj@amazon.com> wrote:
While browsing through some of the clock-sweep code I noticed that the
only place StrategyNotifyBgWriter() is called now is in
BackgroundWriterMain()[1]. Presumably this isn't what's desired. If
nothing else, it means the function's description comment is wrong, as
are comments in BackgroundWriterMain(). This isn't new; 9.2 shows the
same thing and that's when the function was added. I'm not sure what the
right fix here is, since ISTM joggling bgwriter for every call to
BufferAlloc() would be overkill.
1:
https://doxygen.postgresql.org/freelist_8c.html#aabbd7d3891afc1d8531c3871d08d4b28