Hi. Last week our 60 GB database (per psql \l+) was (re-)replicated to
the DR site using SlonyI, and arrived 109 GB in size which caused a
problem as it filled up the filesystem on the DR server - we expected the
DR database to be the same size. Mystery.
Now just past weekend we upgraded our production server by pg_dump
and pg_restore, and again the database is 109 GB in size!
Most of our data is in a single table, which on the old server is 50 GB in
size and on the new server is 100 GB in size.
Could you please help us understand why a COPY of the data into a new
database (whether DR or the new server) results in different disk usage?
Somebody mentioned on the Slony users list that there is a kind of padding
that goes in that actually helps performance.
Is there a way to track disk usage MINUS the padding?
Thanks,
Aleksey