> Why 32 bits? Is that what your hardware is? The business started in 2005 and we have been using 32 bits since then. We have several machines, each with a remote replica databases (WAL shipping) configured and changing this to 64 bits is going to be a lot of work, let alone the down time of each server (pg_dump + pg_restore). But we will probably do this in the future after we finish some priorities. > That might be the problem. I think with 32 bits, you only 2GB of > address space available to any given process, and you just allowed > shared_buffers to grab all of it. The address space for 32 bits is 4Gb. We just tried to reach a balance in the configuration and it seems to be working (except for the ANALYZE command when the number of schemas/tables is huge). Some questions I have: 1) Is there any reason to run the ANALYZE command in a single transaction? 2) Is there any difference running the ANALYZE in the whole database or running it per schema, table by table? Thanks for all the help, Hugo -- View this message in context: http://postgresql.1045698.n5.nabble.com/Thousands-of-schemas-and-ANALYZE-goes-out-of-memory-tp5726198p5726351.html Sent from the PostgreSQL - general mailing list archive at Nabble.com. -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
