After initialising Roundup and importing from a backup, 
we can only add 10 or so new issues before getting 
a 'cannot add item to database error' attempting to 
create another one. Thereafter, we get the error on 
every create, and must export the database, initialise 
and import the backup. Otherwise Roundup works 
beautifully.

We are on bugfix release 0.6.3 using the anydbm 
backend, and have added a small number of custom 
fields. 

According to the error message the issue is always in 
doSaveJournal() on a link action. Here's a partial 
transcript:

...back_anydbm.py in doSaveJournal(self=<back_anydbm 
instance at 72e378>, classname='priority', nodeid='5', 
action='link', params=('issue','34','priority'), 
creator=None, creation=None)

db[nodeid] = marshal.dumps(l)

db = <dbm.dbm object>, nodeid='5', global marshal = 
<module 'marshal'(built-in)>, global dumps = undefined, l 
= [('5', '20040109151940', '1', 'create', {}), ... 

Interesting that it finds dumps undefined? Anyway, many 
thanks in advance. Hope I'm using the right venue for 
reporting this sort of thing.

Regards,
Mike Durbin

Logged In: YES 
user_id=6405

Could you please attach the full error to this report. Also useful would be 
your dbinit.py module from the modified tracker.  
 
Was there any chance that someone else was using the tracker at the 
same time that the error occurred? 
 
Also, if the error is repeatable, please indicate the steps I'd need to take 
to repeat it. 

Logged In: YES 
user_id=947543

We changed the backend from anydbm to bsddb and the 
problem went away, so we appear to be in good shape now. 
Thanks just the same!

To answer your question, there is a very good chance 
someone else was using the tracker at the time the issue 
occured. 

(And sorry if I should have used the mail list for this sort of 
thing...)

Mike

Logged In: YES 
user_id=6405

Your backend switch is probably a NOOP. In all likelihood, anydbm is 
already using bsddb for its implementation. 
 
What platform are you on? 
 

Logged In: YES 
user_id=6405

Without more information, I'm going to have to close this bug. 

Logged In: NO 

Yes Richard, please close this bug. Sorry I did not 
communicate that earlier. Whether the switch to bsddb did it 
or not, the problem has not recurred. Thank you.