Re: Using SAN Splits to instantly copy a DB

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 11/15/06 14:28, Martijn van Oosterhout wrote:
> On Wed, Nov 15, 2006 at 01:41:47PM -0600, Ron Johnson wrote:
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> On 11/15/06 09:47, Jim Nasby wrote:
>>> On Nov 14, 2006, at 3:44 PM, Paul Silveira wrote:
>> [snip]
>>> Rule 2 is needed to ensure that the data files in the database are all
>>> consistent to each other. If you have a SAN/filesystem with snapshot
>>> capability (sounds like you do), then you can do that to create the copy
>>> rather than shutting the database down.
>> How does SAN-snapshot ensure transactional consistency?
>
> There is write-ahead logging to do that. It's the same machanism used
> to ensure database consistancy after a crash. When you take a snapshot
> and start a new postmaster on the snapshot, it sees what looks like a
> crashed database and recovers it to the instant it snapshotted (aka
> "crashed").

How does it know what a crashed PostgreSQL database look like?

Besides, active transactions need to be *rolled back*, not written
ahead, since half the data hasn't been sent from the computer yet.

- --
Ron Johnson, Jr.
Jefferson LA USA

Is "common sense" really valid?
For example, it is "common sense" to white-power racists that
whites are superior to blacks, and that those with brown skins
are mud people.
However, that "common sense" is obviously wrong.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (GNU/Linux)

iD8DBQFFW3nMS9HxQb37XmcRAn0vAJ996FVUuoBrtiS/B59iU5jkkJxR2ACeM5A8
BE1Bvph5pG0a+uLg9Kc/zwc=
=ZwsZ
-----END PGP SIGNATURE-----