> Note that cluster on a randomly ordered large table can be
> prohibitively slow, and it might be better to schedule a
> short downtime to do the following (pseudo code)
> alter table tablename rename to old_tablename; create table
> tablename like old_tablename; insert into tablename select *
> from old_tablename order by clustered_col1, clustered_col2;
That sounds like a great idea if that saves time.
>> (creating and moving over FK references as needed.)
>> shared_buffers=160MB, effective_cache_size=1GB,
>> maintenance_work_mem=500MB, wal_buffers=16MB,
>> checkpoint_segments=100
> What's work_mem set to?
work_mem = 32MB
> What ubuntu? 64 or 32 bit?
Its a 32 bit. I dont know if 4GB files doesn't sound to small of a dump
for originally 350GB big db - nor why pg_restore fails...
> Have you got either a file
> system or a set of pg tools limited to 4Gig file size?
Not sure what is the problem on my server - I'm trying to figure out what
has pg_restore fail...