*SOLVED* A weird wordlist.db problem

[Matthias Andree]

Tom Eastman <tom at cs.otago.ac.nz> writes:

> Matthias Andree wrote:
>
>> bogoutil -d is pointless in such cases as it isn't trying to traverse
>> the database in spite of b0rked pointers, db_dump -r is the first thing
>> to try and rather safe, perhaps db_dump -R which is more aggressive and
>> can return bogus data.
>
> Hey cool!  'db_dump -r' seems to have done the trick!  I hadn't tried it
> before because I figured that it would probably suffer the same loop as
> 'bogoutil -d', and since, for some reason, my box doesn't have man pages
> for any of the db_* utilities it was all kind of guesswork as to what each
> one of them was doing.

Indeed, the SleepyCat documentation is HTML only, and in a separate
package on some systems.

In fact, it's the -r option of db_dump that salvages data from a corrupt
file - without this option, db_dump would likely goof up as bogoutil -d
did.

> db_dump *didn't* get caught in the loop, it dumped the whole thing (in a
> weird unreadable format)

It's just a header and then a hexadecimal dump of key and value, each on
a line, and an end marker.

> Running a 'bogoutil -d' on the new database dumps the whole thing nice and
> fast, and even seems to include the tokens that were on the corrupt page.

Good to hear your database is back online.

-- 
Matthias Andree