Some threads were hard killed (using the .NET
"Thread.IsBackgroundThread" mechanism, which uses TerminateThread
AFAIK) in that version during normal operation, but none of them do
database writes. They could have been doing database reads though.

The whole application has been hard-killed few times when exception
happened during database manipulation though.

Best regards,
Filip Navara

On Wed, Oct 14, 2009 at 5:45 PM, McClellen, Chris
<cmcclel...@weather.com> wrote:
> Do you ever teriminate threads via TerminateThread() or run as a service 
> (where the scm will call terminateThread() once your main service thread 
> exits)?  In other words do you hard-kill your threads?
>
>
> -----Original Message-----
> From: sqlite-users-boun...@sqlite.org 
> [mailto:sqlite-users-boun...@sqlite.org] On Behalf Of Filip Navara
> Sent: Wednesday, October 14, 2009 7:15 AM
> To: General Discussion of SQLite Database
> Subject: Re: [sqlite] Corrupted database
>
> On Mon, Oct 12, 2009 at 8:56 PM, McClellen, Chris
> <cmcclel...@weather.com> wrote:
>> What is your synchronous set to?  Full?  FYI If you are using .NET data
>> providers, it is set to "Normal" by default.
>
> Normal or Off, but no power failure was involved. (Yes,
> System.Data.SQLite is used)
>
>> If it is not set to full, I have seen corruption when an application
>> crashes, or exits when a thread is in the middle of updating the db
>> (Synchronous = OFF makes corruption even easier in this case).  I have
>> seen apps that do not wait for background threads to finish before
>> termination, and without full sync on, either the db or the log gets
>> corrupted.  A corrupted log can cause problems for you db on next run
>> when recovery happens.
>
> Sounds suspiciously like our case, but still Synchronous=off is
> supposed to work in the event of application crash, hard killed
> threads and so on. Previous version of the application frequently
> forgot to close the databases on exit and did other nasty things that
> and now fixed, but none of them should cause the database to be
> corrupted.
>
> Best regards,
> Filip Navara
> _______________________________________________
> sqlite-users mailing list
> sqlite-users@sqlite.org
> http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
> _______________________________________________
> sqlite-users mailing list
> sqlite-users@sqlite.org
> http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
>
_______________________________________________
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users

Reply via email to