Unfortunately, a user has just reported the same primary key error
message with shared cache disabled, although the freezing appears to
have been fixed.
However, when I ask the user to send me their deck, I find that:
sqlite> pragma integrity_check;
integrity_check
---------------
ok
sqlite> select id, count(id) from cards group by id having count(id) > 1;
sqlite>
Any ideas?
On Fri, Mar 20, 2009 at 10:00 AM, Damien Elmes <reso...@ichi2.net> wrote:
> Hi all,
>
> Some of my users have been reporting strange database problems
> recently, which seem to have gone away when I removed a call to
> enable_shared_cache(). The problems were noticeable in at least 3.6.1
> and 3.6.11, when using databases of 30MB+, and doing large updates
> using pysqlite.
>
> There were two distinct reported problems. One was that the program
> would just freeze, with no disk access and CPU usage, seemingly in the
> middle of a DB query, on Win32. I wasn't able to reproduce this on
> Linux.
>
> The other problem was reported by both win32 and mac users, and again
> I wasn't able to reproduce it. It resulted in errors like this:
>
> sqlalchemy.exceptions.IntegrityError: (IntegrityError) PRIMARY KEY
> must be unique 'update cards set isDue = 0 where type in (0,1,2) and
> priority = 0 and isDue = 1' {}
>
> .. which is strange, because the primary key on that table is called
> 'id' and is not affected by the update call.
>
> I also had some reports of DB corruption on OSX, but I'm not sure if
> that occurred since I upgrade to 3.6.11.
>
> One other hint is that while I'd been using shared cache mode for at
> least 6 months or more, these problems seem to have only surfaced
> recently. I'm not sure if that's due to a change in the queries I've
> been doing, or the fact that I changed the cache size to a bigger
> number, and changed the page size to 4096.
>
> Anyway, disabling the shared cache appears to have fixed the problem,
> and since my program is single threaded and has no need for the shared
> cache, it's not an issue for us anymore. But I thought it's worth
> reporting. Have there been any other instances of problems with the
> shared cache mode?
>
> Cheers,
>
> Damien
>
_______________________________________________
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
