On 12/17/2014 11:35 PM, Scott Marlowe wrote:
> So as the title states, we're running pg 9.2.9 and slony 2.2.1
>
> If you create a parent / child table with inherit, then truncate only
> the parent table, it will work properly on the source but will
> propagate as a regular truncate on the destinations.
>
> create table parent yada;
> create table chlid inherits parent yada;
> insert rows into parent, and rows into child tables
>
> subscribe parent and child to another node...
>
> on source run "truncate only parent" and note that child table there
> still has its own rows. Check on subscriber and children tables are
> now empty.
>
> 1: Is this a known bug?
> 2: Is it planned to fix it?
>

No this isn't a known bug, this worked properly prior to 2.1 when the 
pl/sql function was used to apply the truncate on a replica (it did a 
truncate only).  The C apply trigger in 2.2 is missing the 'only'.

I think we'll need to fix this.
There is a handful of bugs that we are close to committing fixes for, 
once those are in we should do a 2.2.4 release


> We were bitten by it quite by surprise and lost a lot of data in child
> tables. Luckily we noticed before switching over from the master node
> to another and resubed the child tables.
>