Horton

Re: crash-safe visibility map, take five

Robert Haas <robertmhaas@gmail.com>

View thread
From: Robert Haas <robertmhaas@gmail.com>
To: Jeff Davis <pgsql@j-davis.com>
Cc: Noah Misch <noah@leadboat.com>, Pavan Deolasee <pavan.deolasee@gmail.com>, Merlin Moncure <mmoncure@gmail.com>, pgsql-hackers@postgresql.org
Date: 2011-06-23T22:18:50Z
Lists: pgsql-hackers 
On Wed, Jun 22, 2011 at 10:40 PM, Jeff Davis <pgsql@j-davis.com> wrote:
> 1. Torn pages -- not a problem because it's a single bit and idempotent.

Right.

> 2. PD_ALL_VISIBLE bit makes it to disk before a WAL record representing
> an action that makes the page all-visible. Depending on what action
> makes a page all-visible:
>  a. An old snapshot is released -- not a problem, because if there is a
> crash all snapshots are released.
>  b. Cleanup action on the page -- not a problem, because that will
> create a WAL record and update the page's LSN before setting the
> PD_ALL_VISIBLE.

Lazy VACUUM is the only thing that makes a page all visible.  I don't
understand the part about snapshots.

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company