From: Tom Lane <tgl@sss.pgh.pa.us>
To: michele curioni <mic.curioni@googlemail.com>
cc: pgsql-sql@lists.postgresql.org
Subject: Re: Could not serialize access due to read/write dependencies among
 transactions
In-reply-to: 
 <CAPDic7m0LvwdUtpLS03QZ_gjQ08vax-DFNRqtdmTb7yb2-467Q@mail.gmail.com>
References: 
 <CAPDic7m0LvwdUtpLS03QZ_gjQ08vax-DFNRqtdmTb7yb2-467Q@mail.gmail.com>
Comments: In-reply-to michele curioni <mic.curioni@googlemail.com>
	message dated "Thu, 23 Feb 2023 21:18:20 +0000"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <719365.1677190256.1@sss.pgh.pa.us>
Date: Thu, 23 Feb 2023 17:10:56 -0500
Message-ID: <719366.1677190256@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/719366.1677190256%40sss.pgh.pa.us>
Precedence: bulk

michele curioni <mic.curioni@googlemail.com> writes:
> we have a Spring Boot application that tries to update two separate rows in
> a table in 2 separate threads, and it uses the default Transaction
> Isolation level of READ_COMMITTED.
> One of the two threads always succeeds, the other always fails with:

> org.postgresql.util.PSQLException: ERROR: could not serialize access
> due to read/write dependencies among transactions
>  [java] ERROR>  Detail: Reason code: Canceled on identification as a
> pivot, during write.
>  [java] ERROR>  Hint: The transaction might succeed if retried.

> I thought that error could only happen for SERIALIZABLE transactions,

Indeed.  I think you're mistaken to claim that the transactions are
running in READ COMMITTED.

> moreover, the two threads are trying to update different rows.

That part's not so hard to explain: serializable mode's tracking of
which rows received updates is a bit lossy.  It might be complaining
just because the updated rows were on the same page.

			regards, tom lane