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

[David Relson]

At 07:42 PM 12/29/02, Matthias Andree wrote:

>Hi,
>
>I've experienced goodlist.db corruption again, a mail got stuck with
>DB_PAGE_NOTFOUND errors, but I cannot say when exactly the corruption
>has happened, I fixed it again with db_dump -r and db_load, but it's
>getting annoying. I suspect locking issues. Anyone similar issues? (Read
>below the logs)

As it happens, this morning I built totally new wordlists.  For my training 
corpus, I have the 20,000 or so messages received here since I started 
running bogofilter in early October.  I used contrib/randomtrain as I think 
it's a neat way to train.  Even ran it twice to help bogofilter "get it right".

I haven't noticed any problems of the sort you mentionned, but since I've 
kept a copy of the old wordlists I have 4 wordlists against which to run 
db_verify (two old and two new).  Here're the results:


I used the whole collection of messages received since I started running 
bogofilter in early October.  Here are the results:

$ db_verify -V
Sleepycat Software: Berkeley DB 3.3.11: (July 12, 2001)

$  for db in */*.db ; do echo $db ; db_verify $db ; done

old/goodlist.db
db_verify: Last item on page 2269 sorted greater than parent entry
db_verify: DB->verify: nic.1229/goodlist.db: DB_VERIFY_BAD: Database 
verification failed

old/spamlist.db
db_verify: Last item on page 1420 sorted greater than parent entry
db_verify: Last item on page 1449 sorted greater than parent entry
db_verify: First item on page 1420 sorted greater than parent entry
db_verify: Page 1420 linked twice
db_verify: Last item on page 32 sorted greater than parent entry
db_verify: Last item on page 423 sorted greater than parent entry
db_verify: DB->verify: nic.1229/spamlist.db: DB_VERIFY_BAD: Database 
verification failed

new/goodlist.db

new/spamlist.db