| From: | Andres Freund <andres(at)2ndquadrant(dot)com> |
|---|---|
| To: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
| Cc: | pgsql-hackers(at)postgresql(dot)org |
| Subject: | Re: (auto-)analyze causing bloat/load |
| Date: | 2014-10-18 18:33:57 |
| Message-ID: | 20141018183357.GH22660@awork2.anarazel.de |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On 2014-10-18 14:30:53 -0400, Tom Lane wrote:
> Andres Freund <andres(at)2ndquadrant(dot)com> writes:
> > Interestingly we already set PROC_IN_ANALYZE - but we never actually
> > look at it. I wonder what it'd take to ignore analyze in
> > GetSnapshotData()/GetOldestXmin()... At first sight it looks quite
> > doable to not have a snapshot around (or setup to be ignored) while
> > acquire_sample_rows() is running. As that's the expensive bit...
>
> Say what? How is ANALYZE supposed to know which rows are valid
> without a snapshot?
I haven't fully investigated it yet. But it's already not using it's own
snapshot to determine visibility. Instead it uses GetOldestXmin() +
HeapTupleSatisfiesVacuum().
Greetings,
Andres Freund
--
Andres Freund http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Tom Lane | 2014-10-18 18:54:12 | Re: initdb failure on RH 5.10 |
| Previous Message | Tom Lane | 2014-10-18 18:30:53 | Re: (auto-)analyze causing bloat/load |