Re: posgres 12 bug (partitioned table)

From: Pavel Biryukov <79166341370(at)yandex(dot)ru>
To: Andres Freund <andres(at)anarazel(dot)de>
Cc: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, Robert Haas <robertmhaas(at)gmail(dot)com>, PostgreSQL mailing lists <pgsql-bugs(at)lists(dot)postgresql(dot)org>
Subject: Re: posgres 12 bug (partitioned table)
Date: 2020-08-11 18:55:32
Message-ID: 844551597171844@mail.yandex.ru
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-bugs pgsql-hackers

<div> </div><div>I don't see a problem with "wrapping around" - the row's xmin does not change with freeze (AFAIK). It changes when the row is modified.</div><div>So event if you hold some entity (with current xmin) for a long time (enough for "wrap around") and then try to update it, it will update ok.</div><div> </div><div>"Multiple statements" possible problems is managed for us by NpgSQL :)</div><div> </div><div>-- <br />С уважением, Павел</div><div> </div><div> </div><div> </div><div>11.08.2020, 21:39, "Andres Freund" &lt;andres(at)anarazel(dot)de&gt;:</div><blockquote><p>Hi,<br /><br />On 2020-08-11 21:31:52 +0300, Pavel Biryukov wrote:</p><blockquote> Entity Framework is an ORM for .Net (and .Net Core). It has different providers<br /> for different databases (NpgSql for Postgres). It uses Optimistic concurrency.<br /> The common use case is to use xmin as "concurrency token".<br />  <br /> In code we make "var e = new Entity();", "dbContext.Add(e)" and<br /> "dbContext.SaveChanges()" (smth like that), and EF Core constructs sql for us,<br /> classical ORM;<br />  <br /> When new row is inserted, EF makes an insert with "RETURNING xmin" to keep it<br /> as concurrency token for further updates (update is made like "where id = [id]<br /> AND xmin=[xmin]" to be sure the row hasn't been updated by other clients).</blockquote><p><br />That's not really a safe use of xmin, e.g. it could have wrapped around<br />leading you to not notice a concurrent modification. Nor does it<br />properly deal with multiple statements within a transaction. Perhaps<br />those are low enough risk for you, but I don't xmin is a decent building<br />block for this kind of thing.<br /><br />Greetings,<br /><br />Andres Freund</p></blockquote>

Attachment Content-Type Size
unknown_filename text/html 1.7 KB

In response to

Responses

Browse pgsql-bugs by date

  From Date Subject
Next Message Pavel Biryukov 2020-08-11 18:59:18 Re: posgres 12 bug (partitioned table)
Previous Message Andres Freund 2020-08-11 18:39:40 Re: posgres 12 bug (partitioned table)

Browse pgsql-hackers by date

  From Date Subject
Next Message Pavel Biryukov 2020-08-11 18:59:18 Re: posgres 12 bug (partitioned table)
Previous Message Pavel Borisov 2020-08-11 18:50:46 Re: [PATCH] Covering SPGiST index