Index corruption falls in the same class with database corruption, and thus 
99.9% of the time one of the following is happening.  You get database 
corruption from a database operation that does not complete.  So, what will 
interrupt the operation from completing?  Among R:base developers, I think 
everyone will agree on these things, I may forget some more obscure ones, 
but we really don't agree on which happens most often.  I will give you my 
list:

1. Power - make sure every machine on the network including hubs, routers, 
and servers has a UPS.  It does not have to be a great big one, just one 
that will take care of the brounouts and keep your machine up for 5 min when 
the power goes out.

2. Bad NIC.
3. Bad port on a hub, or bad hub
Both 2 and 3 will may cause random pack losses.

4. Other faulty hardware - Mother bord, HD....

5. User interruption.  Make sure that you don't have users closing R:base in 
the middle of a transaction, or maybee a machine is locking up randomly.

On large networks, it is like looking for a needle in a haystack.

Troy

===== Original Message from [EMAIL PROTECTED] at 9/27/01 5:28 pm
>I do lock the table & if the table won't lock, then someone has some type of
>lock (usually an edit lock) so it informs the user to have others in the
>dept to exit the table and then the user can try the program with the append
>operation again.  It appears that the index problem exists BEFORE the append
>is attempted.  Trying the append just reports that there is an indexing
>problem.  As users, throughout the day add/edit rows in the accounting
>table, the indexes will get updated, thus, the indexes for this table (as
>well as many other tables) are in a constant state of flux.
>
>I have indexes on computed columns in other tables that are accessed by
>everyone and haven't had problems with those tables.
>
>My questions again are:  What can cause the corruption of the indexes on the
>table and how can I discover which index or indexes on the table are
>corrupted.
>
>
>
>-----Original Message-----
>From: Troy Sosamon [mailto:[EMAIL PROTECTED]]
>Sent: Thursday, September 27, 2001 9:09 AM
>To: [EMAIL PROTECTED]
>Subject: RE: Index corruption
>
>
>Try putting a lock manual on the table before you do the append and then
>take it off.
>
>set lock tablename on
>
>append.....
>
>set lock tablename off
>
>Troy

Reply via email to