While dealing with bloat on a table, myschema.mytable, we found that an associated index was also bloated. In order to correct the bloat we ran REINDEX myschema.myindex, expecting the bloat to be near zero after the REINDEX. The bloat on the table was reduced to zero. However, the index bloat remained high.
What would cause this bloat to remain after a REINDEX?
I have included the details of the index as well as data types below:
Postgres Version: 9.3.4
CREATE UNIQUE INDEX myindexON myschema.mytable USING btree (column1 varchar(255), column2 varchar(255))
Script running for index bloat: https://github.com/pgexperts/pgx_scripts/blob/master/bloat/index_bloat_check.sql (fixed sql syntax from current repo version)
Before output:
index_name │ cmnd │ bloat_pct │ bloat_mb │ index_mb │ table_mb │ index_scans │
myindex │ REINDEX INDEX schema.myindex │ 78 │ 9373.117 │ 11997.328│ 9146.984 │ 184882287 │
After output:
index_name │ cmnd │ bloat_pct │ bloat_mb │ index_mb │ table_mb │ index_scans │
myindex │ REINDEX INDEX schema.myindex │ 66 │ 5037.234 │ 7662.016 │ 9146.984 │ 184882287 │
Command run:
REINDEX INDEX schema.myindex
Dani Ames
Database Administrator | Stinger Ghaffarian Technologies (SGT)
Contractor to the U.S. Geological Survey (USGS)
Earth Resources Observation and Science (EROS) Center
dames@xxxxxxxx | Daytime: 605-594-2597
Earth Resources Observation and Science (EROS) Center
dames@xxxxxxxx | Daytime: 605-594-2597
