[Bucardo-general] Bucardo sync not running

[Michelle Sullivan]

Well you’d be checking the first to see if that is your issue... don’t just randomly do it.... look for small numbers of rows and massive dead space/rows.

If you have transactioned writes to both side (a materialized view will do the trick if running Master-Slave type setup) you could be running into cross db deadlocks....  which the deadlock resolver won’t usually get.

Lastly you might want todo like me...  split into multiple syncs.  Check your referral integrity to see if you can split the db into smaller syncs.. they’ll be faster in general.

Michelle Sullivan
http://www.mhix.org/
Sent from my iPad

> On 05 Nov 2018, at 08:14, Christophe Pettus <xof at thebuild.com> wrote:
> 
> It's a little bit fraught to do that, given that we have 300+ tables in the sync (plus about a hundred sequences), so it'll be a pretty long operation to do VACUUM FULL on all of them (plus some scripting work to generate the commands).
> 
>> On Nov 4, 2018, at 13:11, Michelle Sullivan <michelle at sorbs.net> wrote:
>> 
>> Needs todo it on the delta and track tables as I have seen table bloat slow the delta join slow to ridiculously long queries... the full will remove the bloat that a normal vacuum won’t.
>> 
>> Michelle Sullivan
>> http://www.mhix.org/
>> Sent from my iPad
>> 
>>> On 05 Nov 2018, at 02:29, David Christensen <david at endpoint.com> wrote:
>>> 
>>> Right; needs to be able to complete quickly, as VACUUM FULL will block writes, and hence transactions blocked by the trigger inserts.
>>> 
>>>> On Nov 3, 2018, at 7:01 PM, Michelle Sullivan <michelle at sorbs.net> wrote:
>>>> 
>>>> Check the delta table size... 
>>>> 
>>>> psql -U <a super user> <database>
>>>> Db=# SET SCHEMA bucardo;
>>>> Db=# SELECT COUNT(1) FROM bucardo.bucardo_<tablename>_delta;
>>>> 
>>>> ... repeat for all <tablenames> in the sync, see how many changes there are...
>>>> 
>>>> If a low number try VACUUM VERBOSE FULL ANALYZE bucardo.bucardo_<tablename>_delta; for each table.
>>>> 
>>>> Michelle Sullivan
>>>> http://www.mhix.org/
>>>> Sent from my iPad
>>>> 
>>>>> On 04 Nov 2018, at 05:27, Christophe Pettus <xof at thebuild.com> wrote:
>>>>> 
>>>>> 
>>>>> 
>>>>>> On Nov 1, 2018, at 09:35, David Christensen <david at endpoint.com> wrote:
>>>>>> 
>>>>>> So this is the same on source/target? Anything happening in the “bucardo” database?
>>>>> 
>>>>> On the source database, all the Bucardo connections are in place, but idle; same on the bucardo database.
>>>>> 
>>>>>> Is this multi-master or single master?
>>>>> 
>>>>> Single master.
>>>>> 
>>>>> Thanks!
>>>>> --
>>>>> -- Christophe Pettus
>>>>> xof at thebuild.com
>>>>> 
>>>>> _______________________________________________
>>>>> Bucardo-general mailing list
>>>>> Bucardo-general at bucardo.org
>>>>> https://mail.endcrypt.com/mailman/listinfo/bucardo-general
>>> 
>>> --
>>> David Christensen
>>> End Point Corporation
>>> david at endpoint.com
>>> 785-727-1171
>>> 
>>> 
>>> 
> 
> --
> -- Christophe Pettus
>   xof at thebuild.com
>