Обсуждение: Primary Key

On 2013-11-21 15:40, Joey Quinn wrote:

I have a table (5 columns) with approximately 670 million rows. It has had an index (unique) on an inet column from the beginning. Today I added a primary key constraint based on the same column thinking that since it already had an index, this would be a relatively quick operation. That does not appear to be case. It has gone into a "not responding" status for an hour or so now. As a point of reference, I'm using 9.3 on a 64 bit Windows Server 2008 (32 GB ram) and inserts so far have taken 6 1/2 - 7 minutes for each batch of 16.7 million rows.

Other than not creating the primary key at the beginning, did I do anything wrong? and can I reasonably expect the current operation to finish?

Joey

I'm guessing you're creating the primary key without designating your current unique index as the index to use for the constraint.
Mark your column as not null if it isn't already then do an "alter table table-name add primary key using index whatever-the-name-of-your-extant-unique-index-is".
Otherwise you're building another separate index for that constraint, which you don't need to do.

Ahhh, that's what I was missing... thank-you. (just launched, we'll see how that one goes).

On Thu, Nov 21, 2013 at 3:48 PM, Elliot <yields.falsehood@gmail.com> wrote:

On 2013-11-21 15:40, Joey Quinn wrote:

I have a table (5 columns) with approximately 670 million rows. It has had an index (unique) on an inet column from the beginning. Today I added a primary key constraint based on the same column thinking that since it already had an index, this would be a relatively quick operation. That does not appear to be case. It has gone into a "not responding" status for an hour or so now. As a point of reference, I'm using 9.3 on a 64 bit Windows Server 2008 (32 GB ram) and inserts so far have taken 6 1/2 - 7 minutes for each batch of 16.7 million rows.

Other than not creating the primary key at the beginning, did I do anything wrong? and can I reasonably expect the current operation to finish?

Joey

I'm guessing you're creating the primary key without designating your current unique index as the index to use for the constraint.
Mark your column as not null if it isn't already then do an "alter table table-name add primary key using index whatever-the-name-of-your-extant-unique-index-is".
Otherwise you're building another separate index for that constraint, which you don't need to do.

On 11/21/2013 12:40 PM, Joey Quinn wrote:

I have a table (5 columns) with approximately 670 million rows. It has
had an index (unique) on an inet column from the beginning. Today I
added a primary key constraint based on the same column thinking that
since it already had an index, this would be a relatively quick
operation. That does not appear to be case. It has gone into a "not
responding" status for an hour or so now. As a point of reference, I'm
using 9.3 on a 64 bit Windows Server 2008 (32 GB ram) and inserts so far
have taken 6 1/2 - 7 minutes for each batch of 16.7 million rows.

Other than not creating the primary key at the beginning, did I do
anything wrong? and can I reasonably expect the current operation to finish?

It will finish but it is checking that not only is it unique, but NOT NULL. You may want to cancel it and try:

select count(*) from foo where bar IS NULL;

If you have NULL in the column, you don't have a primary key

Joshua D. Drake

Joey

--
Command Prompt, Inc. - http://www.commandprompt.com/  509-416-6579
PostgreSQL Support, Training, Professional Services and Development
High Availability, Oracle Conversion, Postgres-XC, @cmdpromptinc
For my dreams of your image that blossoms
   a rose in the deeps of my heart. - W.B. Yeats