On Thu, Jun 28, 2001 at 11:10:04AM -0400, Tom Lane wrote:
: If this is 7.1, another fairly likely possibility is that the fsync
: method being used for the WAL log needs to be changed on the Solaris
: box. The reason we have so many options is that some methods are way
: slower than others on certain platforms --- but we don't yet have
: anything in there to pick the right method for a given platform.
: Could you experiment with WAL_SYNC_METHOD on the Solaris system
: and see if your results change?
Hrmm, well, we're running it with fsync disabled, and according to the
docs, that should make the WAL_SYNC_METHOD irrelevant. We had
unacceptable performance with fsync enabled.
* Philip Molter
* DataFoundry.net
* http://www.datafoundry.net/
* philip@datafoundry.net