Date: Fri, 27 May 2022 10:59:24 +0900 (JST)
Message-Id: <20220527.105924.1864299458038443589.horikyota.ntt@gmail.com>
To: simseih@amazon.com
Cc: alvherre@alvh.no-ip.org, pgsql-hackers@postgresql.org
Subject: Re: [BUG] Panic due to incorrect missingContrecPtr after promotion
From: Kyotaro Horiguchi <horikyota.ntt@gmail.com>
In-Reply-To: <C7EF4CE5-0ED5-46D8-B1C4-4B805006B9B9@amazon.com>
References: <EF51690F-5006-47C4-9473-9DFAFC0B1778@amazon.com>
	<B41D8D0F-EB06-4C99-BE57-35EB3C8A7C7F@amazon.com>
	<C7EF4CE5-0ED5-46D8-B1C4-4B805006B9B9@amazon.com>
User-Agent: Mew version 6.8 on Emacs 26.1
Mime-Version: 1.0
Content-Type: Text/Plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Archived-At: 
 <https://www.postgresql.org/message-id/20220527.105924.1864299458038443589.horikyota.ntt%40gmail.com>
Precedence: bulk

At Thu, 26 May 2022 19:57:41 +0000, "Imseih (AWS), Sami" <simseih@amazon.com> wrote in 
> We see another occurrence of this bug with the last patch applied in 13.7.
> 
> After a promotion we observe the following in the logs:
...
> We think it's because VerifyOverwriteContrecord was not 
> called which is why we see this behavior. 
> 
> Are there are  other places where missingContrecPtr 
> should be invalidated, such as after a successful promotion?

The only cause known to me for EndOfLog being moved to such location
is missingContrecPtr. But if the next record is not
XLOG_OVERWRITE_CONTRECORD recovery should have stopped there. And if
XLOG_OVERWRITE_CONTRECORD is there, VerifyOverwriteContrecord should
have been called..

Could you inspect WAL files of the environment and see if the first
record of the '7/A8'th segment OVERWRITE_CONTRECORD?  I don't say that
makes some progress on this, but could be the first step.

regards.

-- 
Kyotaro Horiguchi
NTT Open Source Software Center