Two way synch show duplcates

[pbryant23x]

I am using a two way synch job through SQLyog and for the past month and a half it has run fine. I use it to update a database from one web server to another server in the web farm. I got a call that the databases were showing duplicate or blank data after syncronization, this is the only data in the sja.log file:

No PRIMARY KEY defined in the table '`activity_log`'

No PRIMARY KEY defined in the table '`activity_log`'

No PRIMARY KEY defined in the table '`zipcodes`'

No PRIMARY KEY defined in the table '`zipcodes`'

Sync started at Thu May 12 14:00:00 2005

That is all that shows up in the sja.log file on the source server.

This has been running fine and just out of the blue started displaying duplicates or blank records. I did not see the page that was displaying this.

Should I be using a one way synch is this situation? I have the web guys updating only the source and then I let the synch job take care of updating the second server in the farm so they never actually update the content on the target server.

Thanks

[peterlaursen]

How many tables are involved in the sync ? Tables without a PK should not sync at all. I read the log-messge so that two tables do not qualify for sync (due to missing PK's) and are skipped and the rest of the tables are sync'ed

“Sync started at Thu May 12 14:00:00 2005” is the only message you'll get except for error messages.

Is the “abort on error” option set in the XML-file (I guess not!) ?

I believe the best thing you could do is to have a PK in all tables. If some tables are sync'ed and others not, I would not exclude that some transactions performed later could create empty or dublikate rows. Depends on the datastructure and what actions are performed on the data.

It's also a possibility that there isn't really empty or dublicate rows but only that the application used for reading the data “think”s there is, due to pointers pointing to the wrong row from one table to another. That could also easily be a result of an incomplete sync.

Of course there is not use to use two-way sync if you only want to write data from one server to another. But that in itself should not result in empty or dublicate rows.

I think it is not possible to say more as long as we don't know the table structure. For instance if FK's are used.

[pbryant23x]

Thank you for the quick response.

I will pass this information along to the web guys here and post what they come back with.

I never actually saw the web page in question that was displaying the duplicates and by the time I did see it everything appeared to be normal.

Thanks

[Author]

Posts