| From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
|---|---|
| To: | decibel <decibel(at)decibel(dot)org> |
| Cc: | pgsql-hackers(at)postgresql(dot)org |
| Subject: | Re: Stats for inheritance trees |
| Date: | 2010-01-06 15:03:28 |
| Message-ID: | 1991.1262790208@sss.pgh.pa.us |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
decibel <decibel(at)decibel(dot)org> writes:
> On Dec 29, 2009, at 6:29 PM, Tom Lane wrote:
>> * when a tabstat message comes in, increment changes_since_analyze by
>> the sum of t_tuples_inserted + t_tuples_updated + t_tuples_deleted;
>>
>> * when an analyze report message comes in, reset changes_since_analyze
>> to zero.
> If that's being added, could we extend the concept to also keep a reltuples_delta column (name suggestions welcome!) that is = reltuples_delta + t_tuples_inserted - t_tuples_deleted, and then set reltuples_delta back to 0 after an analyze (or anything else that would reset reltuples)? That means you could use reltuples + reltuples_delta as a fairly accurate row count.
We already have a fairly accurate row count.
regards, tom lane
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Robert Haas | 2010-01-06 15:05:52 | Re: Status of plperl inter-sp calling |
| Previous Message | Zdenek Kotala | 2010-01-06 15:00:56 | Re: pg_migrator issues |