BUG? corrupted goodlist.db again, spamlist.db fine

[Matthias Andree]

Gyepi SAM <gyepi at praxis-sw.com> writes:

>> Gyepi, can you share any ideas how we could track these down?
>
> I don't know how to track them down...
> But I just checked my  db-3.1.17 databases and they show very similar
> errors to David's databases.  I dumped and recreated both databases, using
> both bogoutil and db_dump/db_load and the problems went away. Furthermore,
> the database sizes dropped appreciably. goodlist.db went from 15M to 9.0M
> while spamlist.db went from 1.4M to 232k.

Possibly owed to the db corruption. db_dump -R might pull more data out
than db_dump -r or even db_dump.

> Both utilities created databases with exactly the same size. It seems
> like it would be a good idea to dump and recreate databases on a
> semi-regular basis.

I don't think so, but it might indeed be useful to figure WHY these
corruptions occur. I do not (yet!) believe in BDB bugs, 

>
>   db_dump goodlist.db | db_load -t btree goodlist.db.tmp
>   bogoutil -d goodlist.db | bogoutil -l goodlist.db.tmp 

Oh, is the second line important? I used mv instead of bogoutil -d/-l.

>> Incidentally, my current MAIN (non-MIME) bogofilter chokes really hard
>> on some BDB cursor c_get operation (remember it sets MALLOC_CHECK_=2,
>> which is documented as (libc.info)
>
> Hmm, that's not good.

Yup, but AFAICT, I don't see a reason why it should, it's db_open,
db_cursor, cursortype->c_get, why should *that* fail?

-- 
Matthias Andree