Change tracking looses changes

Change tracking looses changes

Post by Dan Meie » Fri, 14 Sep 2001 04:18:09



We have a database that has change-tracking turned on.  We
have many inserts and deletes occuring on the indexed
table.  I find that if I perform many inserts/deletes very
quickly that the indexing seems to 'loose' records.  The
table record count increases, but the number of items
(Item Count) in the full-text index remains about the
same.  If I slow down the process to about 1 record every
10 seconds, then I seems to work properly.  Any ideas?
 
 
 

Change tracking looses changes

Post by Trevor Dwye » Fri, 14 Sep 2001 06:20:13


Have you checked the index the following day - after the master merge has
taken place? Are you able to query the results with the correct data or is
it simply the metadata that is incorrect?

Best Regards

Trevor Dwyer - SQL Server MVP


Quote:> We have a database that has change-tracking turned on.  We
> have many inserts and deletes occuring on the indexed
> table.  I find that if I perform many inserts/deletes very
> quickly that the indexing seems to 'loose' records.  The
> table record count increases, but the number of items
> (Item Count) in the full-text index remains about the
> same.  If I slow down the process to about 1 record every
> 10 seconds, then I seems to work properly.  Any ideas?


 
 
 

Change tracking looses changes

Post by John Kan » Fri, 14 Sep 2001 13:13:42


Dan,
You should review the NT/Win2K Application event log on the server where you
have SQL Server 2000 installed for source events "Micrsoft Search" and MssCi
to see if any related MSSearch errors are recorded durung the times that
that change tracking seems to "loose" records. Also, do these "lost" rows,
seem to return later, say after midnight? Also, if you run an Incremental or
Full Population, do you regain these "lost" rows?

Furthermore, in your situation with mass/fast update, Change Tracking with
Update Index in Background, might not be the best stragety for ensuring
accurate FTS queries. Checkout SQL 2000 BOL title "Maintaining Full-Text
Indexes":
"The method you use depends on factors such as the CPU and available memory,
the amount and rate of change of data, the amount of available disk space,
and the importance of the full-text index being current. Use these
recommendations as a guide for selecting a maintenance method.
  a.. Use change tracking with the background update index option when CPU
and memory are available, the value of an up-to-date index is high, and
immediate propagation can keep up with the rate of changes.

  b.. Use change tracking with scheduled propagation when CPU and memory can
be used at scheduled times, disk space for storing changes is available, and
changes between the scheduled times are not so significant that the
propagation takes longer than a full rebuild.

  c.. Use a full rebuild when a large percentage of records change or are
added at once. If a large percentage of records change over an extended
period of time, consider using change tracking with scheduled or background
update index.

  d.. Use an incremental rebuild when a large number, but not a large
percentage, of documents change at one time. If a large number of records
change over an extended period of time, consider using change tracking with
scheduled or background update index. "

Regards,
John
This posting is provided "AS IS" with no warranties, and confers no rights.
You assume all risk for your use. ? 2001 Microsoft Corporation.
All rights reserved.


Quote:> We have a database that has change-tracking turned on.  We
> have many inserts and deletes occuring on the indexed
> table.  I find that if I perform many inserts/deletes very
> quickly that the indexing seems to 'loose' records.  The
> table record count increases, but the number of items
> (Item Count) in the full-text index remains about the
> same.  If I slow down the process to about 1 record every
> 10 seconds, then I seems to work properly.  Any ideas?

 
 
 

1. Changing column name without loosing table date

hi,
I am required to write a DB Script that will rename a
column in a table without loosing table Data.
The name of the table is 'Gateway'. The columns that i
have in the 'Gateway' table are
GatewayId
UserId
pwd

After running the script the 'Gateway' table should have
the following column Names
GatewayId
UserID
Password

As you can see, i have changed one of the column names
from 'pwd' to 'Password'.
This change of column name should happen without loosing
the rows in the gateway table. Data values in the
old 'pwd' column should be in the new 'password' column.

How do i do this?
TIA,
Babu

2. Attend Ingres Expo - San Diego Presentation via Webinar

3. PLS help: Change field structure without loosing data?

4. NewEra Picturebutton

5. Changing SQL-order looses record pointer

6. Data Modeling Books

7. Changing column name without loosing table date

8. How to read 'Default Value' of a field in Access

9. SQL Server 2000 Loosing changes, rolling back by itselef

10. CHANGE TRACKING

11. Warning Message When Enabling Change Tracking

12. Tool to track DDL changes...

13. change tracking