Kevin Goess wrote
> Can anybody help me understand what these statistics are suggesting,
> what's
> actually going on on this box/in postgresql? What is it writing to disk,
> and why? Is it just writing out new/changed rows, or what?
Not a clue on the statistics but most likely you are seeing checkpoint
activity.
At a simplified level all changes to the database are first persisted to
disk using WAL (write-ahead-logs). The changes are written to disk, into
WAL files, during commit via fsync. The original data files are not
affected immediately thus improving performance at the time of commit by
instead risking a prolonged delay in situations where an unclean shutdown
occurs. However, at some point the WAL files need to be removed and the
physical table files updated. This occurs during a checkpoint. A
checkpoint basically causes the on-disk files to become baselined to the
current reality so that only subsequent WAL files need be applied.
There is considerably more to this whole concept than I can go into off the
top of my head but in addition to looking at just I/O it would help to look,
simultaneously, at what processes are active.
HTH
David J.
--
View this message in context:
http://postgresql.1045698.n5.nabble.com/pgpgout-s-without-swapping-what-does-it-mean-tp5796346p5796355.html
Sent from the PostgreSQL - general mailing list archive at Nabble.com.
