On Fri, 2006-10-13 at 13:52 -0400, Merlin Moncure wrote: > On 10/13/06, Tom Lane <tgl@xxxxxxxxxxxxx> wrote: > > Martijn van Oosterhout <kleptog@xxxxxxxxx> writes: > > > Is that really true? In theory block n+1 could be half a revolution > > > after block n, allowing you to commit two transactions per revolution. > > > > Not relevant, unless the prior transaction happened to end exactly at a > > does full page writes setting affect this? > No, full page writes only affects checkpoints. For a transaction to commit, some bits must hit permanent storage *somewhere*. If that location is in one general area on disk, you must either commit several transactions at once (see commit_delay), or you must wait until the next revolution to get back to that area of the disk. Regards, Jeff Davis
