Question Re: DEFAULT TIMESTAMP & Replicating Environments in Technology Q&A

DEFAULT TIMESTAMP & Replicating Environments

Former Member — Wed, 27 Jan 2010 18:35:55 GMT

I have some table :

create table people ( id bigint not null default autoincrement, first_name varchar(50) not null, last_name varchar(50) not null, gender char(1) not null default 'M', last_update timestamp not null default timestamp );

What happens in a replicating environment if remote user 1 updates a row, and then replicates the change to the main site? Does the last_update get updated AGAIN at the main site when the change arrives and is processed?

Please tell me it doesn't.

Re: DEFAULT TIMESTAMP & Replicating Environments

VolkerBarth — Wed, 27 Jan 2010 18:45:29 GMT

Calvin,

it doesn't.

When you look at the SQL statements that are sent to the other side (by running DBREMOTE -v -o C:\\´MyLog.txt) you will notice that the receiving database gets a complete statement with the *last_update* column set. And the particular DEFAULT TIMESTAMP default will only be applied when the column's value is not set explicetely. So no need to worry, we have used this DEFAULT in a SQL Remote setup for years.

As to your new example: If the remote exeucutes the following SQL statement (say, in a client app):

UPDATE people SET last_name = 'Smith' WHERE id = 123450000

then last_update will be set automatically to current timestamp (say, to '2010-01-28 11:24:23.123456'). The transaction log will then contain something like (you could check by using DBTRAN):

UPDATE people SET last_name = 'Smith', last_update = '2010-01-28 11:24:23.123456' WHERE id = 123450000

And it's that statement that is sent to the consolidated. Obviously, the statement contains a value for the DEFAULT TIMESTAMP column, so it doesn't update *last_update* again.

Re: DEFAULT TIMESTAMP & Replicating Environments

VolkerBarth — Wed, 27 Jan 2010 18:48:45 GMT

Your example is somewhat lacking a "real" column: You wouldn't want to update the id (as a PK field) in a SQL Remote setup, otherwise last_update wouldn't be the problem:)

Re: DEFAULT TIMESTAMP & Replicating Environments

Former Member — Wed, 27 Jan 2010 18:48:59 GMT

Awesome, thanks! We're just looking to implement this default and had concerns of the affects replication might bring.

Re: DEFAULT TIMESTAMP & Replicating Environments

Former Member — Wed, 27 Jan 2010 19:40:39 GMT

Example updated.

Re: DEFAULT TIMESTAMP & Replicating Environments

VolkerBarth — Thu, 28 Jan 2010 10:06:44 GMT

Just to add: We have included 3 columns in all published tables: one "dtCreated" with DEFAULT CURRENT TIMESTAMP. one "dtLastUpdated" with DEFAULT TIMESTAMP and one "origin" with DEFAULT CURRENT PUBLISHER. That makes tracking the source of inserts/updates quite easy.