From: Tom Lane <tgl@sss.pgh.pa.us>
To: Robert Haas <robertmhaas@gmail.com>
cc: Andrew Dunstan <andrew@dunslane.net>,
        Alvaro Herrera <alvherre@alvh.no-ip.org>,
        David Rowley <dgrowleyml@gmail.com>, Peter Geoghegan <pg@bowt.ie>,
        pgsql-hackers@lists.postgresql.org,
        Andrew <pgsqlhackers@andrewrepp.com>
Subject: Re: pg_dump versus hash partitioning
In-reply-to: 
 <CA+TgmoaJJUZF41r-5NyEb2mMNXaT6Rf4FK=BWdRun0S758dXDw@mail.gmail.com>
References: <1376149.1675268279@sss.pgh.pa.us>
 <CA+TgmoaZ8kDKiQMtDq82Qqqgc5JAfF_UEvc2qa46QwVVLvtMDw@mail.gmail.com>
 <1412538.1675283692@sss.pgh.pa.us>
 <CA+TgmoYoSoX8bN_yzabmRDQBNxEaWmjBKPnkZPMUDpPTBy=d4A@mail.gmail.com>
 <CAH2-WzkjKCmVD9JDKHx09YC+c5kSL_1rHH0kRMyvDHRFoxgw6Q@mail.gmail.com>
 <CA+TgmoYJ+3XpW_iYapXN3gwq2MECi7GH8BxuwKFCQPOzLaa3OQ@mail.gmail.com>
 <CAH2-Wz=cMRoYHeoac7kscdDNBSYzHKgMvSHX5CFEoE66rimk8A@mail.gmail.com>
 <1425446.1675291093@sss.pgh.pa.us>
 <CAApHDvqJM5HxhZSnxESENAHLxTZ=eXKKHxeHfvVe1E8v28woDA@mail.gmail.com>
 <1440373.1675299790@sss.pgh.pa.us>
 <a7ef6212-f9fe-6826-9640-5dac54327140@dunslane.net>
 <1533086.1675349554@sss.pgh.pa.us> <1912821.1676402493@sss.pgh.pa.us>
 <CA+TgmoaJJUZF41r-5NyEb2mMNXaT6Rf4FK=BWdRun0S758dXDw@mail.gmail.com>
Comments: In-reply-to Robert Haas <robertmhaas@gmail.com>
	message dated "Mon, 27 Feb 2023 10:50:42 -0500"
MIME-Version: 1.0
Content-Type: text/plain; charset="UTF-8"
Content-ID: <369733.1677514834.1@sss.pgh.pa.us>
Content-Transfer-Encoding: 8bit
Date: Mon, 27 Feb 2023 11:20:34 -0500
Message-ID: <369734.1677514834@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/369734.1677514834%40sss.pgh.pa.us>
Precedence: bulk

Robert Haas <robertmhaas@gmail.com> writes:
> On Tue, Feb 14, 2023 at 2:21 PM Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> This made me wonder if this could be a usable solution at all, but
>> after thinking for awhile, I don't see how the claim about foreign key
>> constraints is anything but FUD.  pg_dump/pg_restore have sufficient
>> dependency logic to prevent that from happening.  I think we can just
>> drop the "or perhaps ..." clause here, and tolerate the possible
>> inefficiency as better than failing.

> Right, but isn't that dependency logic based around the fact that the
> inserts are targeting the original partition? Like, suppose partition
> A has a foreign key that is not present on partition B. A row that is
> originally in partition B gets rerouted into partition A. It must now
> satisfy the foreign key constraint when, previously, that was
> unnecessary.

Well, that's a user error not pg_dump's fault.  Particularly so for hash
partitioning, where there is no defensible reason to make the partitions
semantically different.

There could be a risk of a timing problem, namely that parallel pg_restore
tries to check an FK constraint before all the relevant data has arrived.
But in practice I don't believe that either.  We load all the data in
"data" phase and then create indexes and check FKs in "post-data" phase,
and I do not believe that parallel restore weakens that separation
(because it's enforced by a post-data boundary object in the dependencies).

			regards, tom lane