On Tue, Mar 29, 2011 at 09:50:05PM +0300, Ibrahim Harrani wrote: > Hi > > dspam=# ALTER TABLE dspam_token_data SET ( FILLFACTOR = 90 ); > ALTER TABLE > dspam=# CLUSTER dspam_token_data USING dspam_token_data_uid_key; > ERROR: index "dspam_token_data_uid_key" for table "dspam_token_data" > does not exist > dspam=# \d dspam_token_data > Table "public.dspam_token_data" > Column | Type | Modifiers > ---------------+---------+----------- > uid | integer | > token | bigint | > spam_hits | integer | > innocent_hits | integer | > last_hit | date | > Indexes: > "dspam_token_data_uid_token_key" UNIQUE, btree (uid, token) > > dspam=# CLUSTER dspam_token_data USING dspam_token_data_uid_token_key; > CLUSTER > dspam=# > > I think CLUSTER dspam_token_data; > enough for the further clustering? > > How often should I run CLUSTER? > > Thanks > Yes, once the CLUSTER index has been defined by a use, only a "CLUSTER table;" is needed. You should only need to cluster to maintain I/O performance. HOT should keep the table in good shape longer, but over time the amount of random I/O will increase and that is when you need to cluster. If you trend the I/O of your disks, it should be obvious when a cluster is needed.
Cheers, Ken ------------------------------------------------------------------------------ Enable your software for Intel(R) Active Management Technology to meet the growing manageability and security demands of your customers. Businesses are taking advantage of Intel(R) vPro (TM) technology - will your software be a part of the solution? Download the Intel(R) Manageability Checker today! http://p.sf.net/sfu/intel-dev2devmar _______________________________________________ Dspam-user mailing list Dspam-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dspam-user
