Content-Type: multipart/mixed;
	boundary="------------cjZh96tZRn1u6saSBtKIaqkX"
Message-ID: <25c03b0c-1684-fc66-6aef-0ab41d824510@amazon.com>
Date: Mon, 21 Feb 2022 11:29:51 +0100
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:91.0)
 Gecko/20100101 Thunderbird/91.6.0
Subject: Re: [BUG] Logical replication failure "ERROR: could not map filenode
 "base/13237/442428" to relation OID" with catalog modifying txns
Content-Language: en-US
To: Kyotaro Horiguchi <horikyota.ntt@gmail.com>,
	<osumi.takamichi@fujitsu.com>, <sawada.mshk@gmail.com>
CC: <pgsql-hackers@lists.postgresql.org>, <minsoo@amazon.com>
References: <CAD21AoD00wV4gt-53ze+ZB8n4bqJrdH8J_UnDHddy8S2A+a25g@mail.gmail.com>
 <20211014.112128.1398650206608979125.horikyota.ntt@gmail.com>
 <OSBPR01MB48886D77AC8AD25B16C28B5CEDBD9@OSBPR01MB4888.jpnprd01.prod.outlook.com>
 <20211019.154328.1852357242791456526.horikyota.ntt@gmail.com>
From: "Drouvot, Bertrand" <bdrouvot@amazon.com>
In-Reply-To: <20211019.154328.1852357242791456526.horikyota.ntt@gmail.com>
Archived-At: <https://www.postgresql.org/message-id/25c03b0c-1684-fc66-6aef-0ab41d824510%40amazon.com>
Precedence: bulk

--------------cjZh96tZRn1u6saSBtKIaqkX
Content-Type: text/plain; charset="UTF-8"; format=flowed
Content-Transfer-Encoding: 7bit

Hi,

On 10/19/21 8:43 AM, Kyotaro Horiguchi wrote:
> At Tue, 19 Oct 2021 02:45:24 +0000, "osumi.takamichi@fujitsu.com" <osumi.takamichi@fujitsu.com> wrote in
>> On Thursday, October 14, 2021 11:21 AM Kyotaro Horiguchi <horikyota.ntt@gmail.com> wrote:
>>> It adds a call to ReorderBufferAssignChild but usually subtransactions are
>>> assigned to top level elsewherae.  Addition to that
>>> ReorderBufferCommitChild() called just later does the same thing.  We are
>>> adding the third call to the same function, which looks a bit odd.
>> It can be odd. However, we
>> have a check at the top of ReorderBufferAssignChild
>> to judge if the sub transaction is already associated or not
>> and skip the processings if it is.
> My question was why do we need to make the extra call to
> ReorerBufferCommitChild when XACT_XINFO_HAS_INVALS in spite of the
> existing call to the same fuction that unconditionally made.  It
> doesn't cost so much but also it's not free.
>
>>> And I'm not sure it is wise to mark all subtransactions as "catalog changed"
>>> always when the top transaction is XACT_XINFO_HAS_INVALS.
>> In order to avoid this,
>> can't we have a new flag (for example, in reorderbuffer struct) to check
>> if we start decoding from RUNNING_XACTS, which is similar to the first patch of [1]
>> and use it at DecodeCommit ? This still leads to some extra specific codes added
>> to DecodeCommit and this solution becomes a bit similar to other previous patches though.
> If it is somehow wrong in any sense that we add subtransactions in
> SnapBuildProcessRunningXacts (for example, we should avoid relaxing
> the assertion condition.), I think we would go another way.  Otherwise
> we don't even need that additional flag.  (But Sawadasan's recent PoC
> also needs that relaxation.)
>
> ASAICS, and unless I'm missing something (that odds are rtlatively
> high:p), we need the specially added subransactions only for the
> transactions that were running at passing the first RUNNING_XACTS,
> becuase otherwise (substantial) subtransactions are assigned to
> toplevel by the first record of the subtransaction.
>
> Before reaching consistency, DecodeCommit feeds the subtransactions to
> ReorderBufferForget individually so the subtransactions are not needed
> to be assigned to the top transaction at all. Since the
> subtransactions added by the first RUNNING_XACT are processed that
> way, we don't need in the first place to call ReorderBufferCommitChild
> for such subtransactions.
>
>> [1] - https://www.postgresql.org/message-id/81D0D8B0-E7C4-4999-B616-1E5004DBDCD2%40amazon.com
> regards.
>
> --
> Kyotaro Horiguchi
> NTT Open Source Software Center
>
Just rebased (minor change in the contrib/test_decoding/Makefile) the 
last POC version linked to the CF entry as it was failing the CF bot.

Thanks

Bertrand

--------------cjZh96tZRn1u6saSBtKIaqkX
Content-Type: text/plain; charset="UTF-8";
	name="poc_mark_all_txns_catalog_change.patch"
Content-Disposition: attachment;
	filename="poc_mark_all_txns_catalog_change.patch"
Content-Transfer-Encoding: base64
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--------------cjZh96tZRn1u6saSBtKIaqkX--