Re: COPY with no WAL, in certain circumstances - Mailing list pgsql-patches

From Bruce Momjian
Subject Re: COPY with no WAL, in certain circumstances
Date
Msg-id 200701062024.l06KO2I22996@momjian.us
Whole thread Raw
In response to Re: COPY with no WAL, in certain circumstances  ("Joshua D. Drake" <jd@commandprompt.com>)
Responses Re: COPY with no WAL, in certain circumstances  ("Simon Riggs" <simon@2ndquadrant.com>)
List pgsql-patches
Joshua D. Drake wrote:
>
> > > BEGIN;
> > >    CREATE TABLE foo...
> > >    INSERT INTO foo VALUES ('1');
> > >    COPY foo...
> > >
> > > COMMIT;
> >
> > On ABORT, the entire table disappears, as well as the INSERT, so I don't
> > see any problem.  I assume the INSERT is WAL logged.
>
> No I don't see any problems, I am just trying to understand the
> boundaries. E.g., is there some weird limitation where if I have any
> values in the table before the copy (like the example above) that copy
> will go through WAL.
>
> Or in other words, does this patch mean that all COPY execution that is
> within a transaction will ignore WAL?

Yes, because it is possible to do in all cases.

--
  Bruce Momjian   bruce@momjian.us
  EnterpriseDB    http://www.enterprisedb.com

  + If your life is a hard drive, Christ can be your backup. +

pgsql-patches by date:

Previous
From: Bruce Momjian
Date:
Subject: Re: xlog directory at initdb time
Next
From: Bruce Momjian
Date:
Subject: Re: [HACKERS] wal_checksum = on (default) | off