Re: [PATCH] Logical decoding of TRUNCATE

Il 23/01/18 18:13, Petr Jelinek ha scritto:
> Hi,
>
> On 23/01/18 15:38, Marco Nenciarini wrote:
>> Il 22/01/18 23:18, Petr Jelinek ha scritto:
>>> On 22/01/18 19:45, Petr Jelinek wrote:
>>>
>>> Actually on second look, I don't like the new boolean parameter much.
>>> I'd rather we didn't touch the input list and always close only
>>> relations opened inside the ExecuteTruncateGuts().
>>>
>>> It may mean more list(s) but the current interface is still not clean.
>>>
>>
>> Now ExecuteTruncateGuts unconditionally closes the relations that it
>> opens. The caller has now always the responsibility to close the
>> relations passed with the explicit_rels list.
>
> This looks good.
>
>>
>> Version 15 attached.
>>
>
> I see you still do CASCADE on the subscriber though.
>

No it doesn't. The following code in worker.c prevents that.


+     /*
+      * Even if we used CASCADE on the upstream master we explicitly
+      * default to replaying changes without further cascading.
+      * This might be later changeable with a user specified option.
+      */
+     cascade = false;

There is also a test that check it works as intended:

+ # should not cascade on replica
+ $node_publisher->safe_psql('postgres', "TRUNCATE tab_rep CASCADE");
+
+ $node_publisher->wait_for_catchup($appname);
+
+ $result = $node_subscriber->safe_psql('postgres',
+     "SELECT count(*), min(a), max(a) FROM tab_notrep_fk");
+ is($result, qq(1|-1|-1),
+     'check replicated truncate does not cascade on replica');
+
+ $result = $node_subscriber->safe_psql('postgres',
+     "SELECT nextval('seq_notrep')");
+ is($result, qq(102),
+     'check replicated truncate does not restart identities');
+

Regards,
Marco

--
Marco Nenciarini - 2ndQuadrant Italy
PostgreSQL Training, Services and Support
marco.nenciarini@2ndQuadrant.it | www.2ndQuadrant.it