Обсуждение: pgStatTuple
Hi all, I am using v. 8.0.1, using a Windows 2003 server platform. I want to use pgStatTuple to analyze performance problems when visiting user sites; it is a good tool. I know how to set it up; but I couldn't find an interpretation of what it generates as a result, so I can be sure of interpreting its output correctly and fully. The output is: table_len(integer) in what size unit? pages? tuple_count(integer) how many t tuple_len(integer) length of average tuple size? tuple_percent (float) ?? dead_tuple_count(integer) tuples that contain dead rcds that could be freed dead_tuple_len (integer) how much space (in bytes?) is dead dead_tuple_percent (float) % of dead space vs entire space for object free_space (integer) how much space (in bytes?) is already allocated and free free_percent (integer) % of free space vs entire space for object Can anyone help? May also require definition of 'tuple' when used in a PostgreSQL sense, also. Thanks for any help you can give.
On Fri, Feb 24, 2006 at 05:03:58PM -0500, Lane Van Ingen wrote: > I want to use pgStatTuple to analyze performance problems when visiting user > sites; it is a good tool. I know how to set it up; but I couldn't find an > interpretation of what it generates as a result, so I can be sure of > interpreting its output correctly and fully. Have you seen README.pgstattuple? Here's an excerpt: table_len -- physical table length in bytes tuple_count -- number of live tuples tuple_len -- total tuples length in bytes tuple_percent -- live tuples in % dead_tuple_len -- total dead tuples length in bytes dead_tuple_percent -- dead tuples in % free_space -- free space in bytes free_percent -- free space in % It can be educational to create a table to play with and watch pgstattuple's output change depending on what you do (insert, update, delete, vacuum, etc.). > Can anyone help? May also require definition of 'tuple' when used in a > PostgreSQL sense, also. Tuple, row, and record are generally used synonymously; tuple seems to be the more common term in discussions of theory. If that's too broad a generalization then maybe somebody will provide a more precise explanation. -- Michael Fuhr
No, I didn't see the README file. Thanks so much for the info, Michael! -----Original Message----- From: Michael Fuhr [mailto:mike@fuhr.org] Sent: Friday, February 24, 2006 6:13 PM To: Lane Van Ingen Cc: pgsql-admin@postgresql.org Subject: Re: [ADMIN] pgStatTuple On Fri, Feb 24, 2006 at 05:03:58PM -0500, Lane Van Ingen wrote: > I want to use pgStatTuple to analyze performance problems when visiting user > sites; it is a good tool. I know how to set it up; but I couldn't find an > interpretation of what it generates as a result, so I can be sure of > interpreting its output correctly and fully. Have you seen README.pgstattuple? Here's an excerpt: table_len -- physical table length in bytes tuple_count -- number of live tuples tuple_len -- total tuples length in bytes tuple_percent -- live tuples in % dead_tuple_len -- total dead tuples length in bytes dead_tuple_percent -- dead tuples in % free_space -- free space in bytes free_percent -- free space in % It can be educational to create a table to play with and watch pgstattuple's output change depending on what you do (insert, update, delete, vacuum, etc.). > Can anyone help? May also require definition of 'tuple' when used in a > PostgreSQL sense, also. Tuple, row, and record are generally used synonymously; tuple seems to be the more common term in discussions of theory. If that's too broad a generalization then maybe somebody will provide a more precise explanation. -- Michael Fuhr