It seems that whenever we are doing a large operation -- eg, CREATE
INDEX on a sizable table, or VACUUM on a big database -- all other
postgres operations on that server stop dead. Attempts to do simple
operations on completely separate databases (but on the same server)
will hang (or perhaps they proceed extremely slowly) until after the large
operation is finished.
Is there any way around this limitation? Is this something that can
be addressed through parameter tweaking of the postmaster, or would it
be solved by having multiple CPUs on the database machine?
Boris Goldowsky