Coming in late here, but I always thought the fact that the FPW happen
mostly at the start of the checkpoint, and the checkpoint writes/fsyncs
happen mostly in the first half of the checkpoint period was always
suboptimal, i.e. it would be nice of one of these was more active in the
second half of the checkpoint period. I assume that is what is being
discussed here.
Yes, this is the subject of the thread.
On the one end hand, whether is the first half or first quarter of first
tenth really depends on the actual load, so how much to rebalance depends
on that dynamic information. At the beginning there should be a short
spike for index pages which are quickly reused, and a longer spike about
data pages depending on the pattern of access and size of table.
On the other hand the rebalancing also depends on the measure chosen to
know about the overall progress, either WAL writing or time, and their
behavior is not the same, so this should be taken into account.
My conclusion is that there is no simple static fix to this issue, as
proposed in the submitted patch. The problem needs thinking and maths.
--
Fabien.
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
