| From: | rahulg <rahulgupta83(at)gmail(dot)com> |
|---|---|
| To: | pgsql-hackers(at)postgresql(dot)org |
| Subject: | Statistic Estimation in PostgreSQL |
| Date: | 2008-10-15 18:57:56 |
| Message-ID: | bf8efe22-eb2d-47e3-b92f-7ad4747198cb@n33g2000pri.googlegroups.com |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
I am facing problem in tracing in what events the selectivity
histogram in pg_statistic is stored/updated.
I went through the code in src/backend/commands/analyze.c and got to
see the code computing the histogram but when I tried to trace the
caller of analyze_rel or compute_index_stats, I find out that only
during Vaccum command, these functions are invoked. I am little
surprised as I did get to see PostgreSQL giving good selectivity
estimations despite me running Vaccuum command even once.
I did try to trace the control flow during table operations
(creation / insertion) but couldn't find any histogram updation
functions being invoked though I did get to know that some functions
related to catalog updations is being invoked.
Can anybody tell me in what all events is the pg_statistic table
exactly updated?
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Tom Lane | 2008-10-15 19:13:58 | Re: SlruPhysicalReadPage |
| Previous Message | Simon Riggs | 2008-10-15 18:37:55 | SlruPhysicalReadPage |