Geoffrey wrote:
> Bill Moran wrote:
>> In response to Geoffrey <lists at serioustechnology.com>:
>>
>>> We've found where we are getting a partial replication of a set.  
>>> That is to say, about half the tables in the set are being 
>>> replicated and the rest are not.  It's like the set was partially 
>>> processed.  We see the first 15 tables in the set replicated, but 
>>> any tables after those are not.  Further, none of the sequences 
>>> defined in that set are being replicated.  It's like it got to a 
>>> table in the set and ignored everything after that.
>>>
>>> Is there max limit on the number of tables in a set?  It appears the 
>>> set stopped at 30 table entries.
>>
>> I can assure you that you can replicate over 30 tables.  We have ~150
>> in one of our databases.
>>
>> What's in the logs?  Is it possible that Slony is hung up on some sort
>> of error or other problem on one of those tables?  If you look at
>> sl_table, are all the tables listed?  What does the sl_status view
>> say?
>
> I'm digging throught the logs now.  I only see the tables that are 
> being replicated listed in sl_table.
>
>> Did this do the initial sync and then stop syncing, or did it fail to
>> do the initial sync?  What kind of activity is on the server: are the
>> slons doing anything at all?
>
> Unfortunately I just noticed this.  We are replicating 13 databases.  
> 12 have identical schemas and the same problem exists in them all.
>
> Is it possible for table ids and sequence ids to collide?  That is 
> will table id 1 collided with sequence id 1?
>
> I'm now looking at the output from when I tried to create the second 
> set, sure enough, the last entry is:
>
> <stdin>:137: Add primary keyed table public.edic
>
> This is the last table in the set that is actually being replicated.
>
>
> Yet here is an excerpt from the config file, note
>
> "pkeyedtables" => [
> .
> .
>             "edi214r", "edi990", "edi_210_qual", "edi_214_qual", "edic",
>             "esc_class", "esc_popup", "etrans", "etransc", "fuel",
> .
> .
>         ],
>
> So it's like it stopped in the middle of the set.  edic is the 30th 
> entry in that pkeyedtable entry.

Do all of the tables in "pkeyedtables" have actual primary keys?  If 
not, you need to specify a candidate unique index, under 'keyedtables' 
in slon_tools.conf if you're using altperl tools.  It's also helpful if 
you wrap the CREATE SET and  SET ADD TABLE commands in a 'try' block, so 
that if there's a problem, everything rolls back...


>
>
>


-- 
JP Fletcher
Database Administrator
Afilias Canada
voice: 416.646.3304 ext. 4123
fax: 416.646.3305
mobile: 416.561.4763
jpfletch at ca.afilias.info