> I changed fsync to false. It took 8 minutes to restore the
> full database.
> That is 26 times faster than before. :-/ (aprox. 200 tps)
> With background writer it took 12 minutes. :-(
That seems reasonable.
> The funny thing is, I had a VMWARE emulation on the same
> Windows mashine, running Red Hat, with fsync turned on. It
> took also 8 minutes to finish.
> Probably the Linux code is better + VMWARE optimises (physical) disk
Vmware makes fsync() into a no-op. It will always cache the disk.
(This is vmware workstation. Their server products behave differntly, of
> It seems to me, I need 2 types of operating modes:
> - For bulk loading (database restore) : fsync=false
> - Normal operation fsync=true
Yes, fsync=false is very good for bulk loading *IFF* you can live with
data loss in case you get a crash during load.
> Am I right? How can I do it "elegantly"?
You'll need to edit postgresql.conf and restart the server for this.
> I Think, it should be a "performance tuning guide" in the
> (not just explaning the settings) Playing with the settings
> could be quite anoying.
There is some information on techdocs.postgresql.org you miht be
---------------------------(end of broadcast)---------------------------
TIP 5: Have you checked our extensive FAQ?