Hello Richard, > > I decreased the import time from >40 minutes to 32 minutes with changing the parameters to: > > > > shared_buffers = 450000 # kept same > > temp_buffers = 100000 # kept same > > > > work_mem = 32768 # now 32mb , before: 102400 = 100mb > > maintenance_work_mem = 2048000 # now 2.048mb, before: 32768 = 32mb > > > > Should I encrease these parameters? > > Nothing that should cause these problems. > > Anything unusual about your installation? Any oddities from ./configure > or similar? I've got no news regarding my installation problems, but I had to reduce the maintenance_work_mem to 256mb, as vacuumdb had this error: vacuumdb: vacuuming of database "adp_trtr" failed: ERROR: invalid memory alloc request size 2097151998 ( 2.GB was a bit to large, I guess ;-) ) But this shouldn't cause my problems, as you say. The installation is a standard solaris postgresql package - my admins say, I've got no priviledges to check this, further on , of course, I have no chance to recompile it, but at the moment it seems to be good to think about compiling the latest PG from sources ( http://www.postgresql.org/docs/8.2/interactive/install-procedure.html ). My system admins contacted the company from which we lent the system , to get more information, so I am waiting for this. Maybe the other alternative, RHEL + Postgres8.2.1 (off. RPMs for 64bit) is a good way to go. Thanks so long, Andre
