[pgpool-general: 9422] Re: FATAL: failed to read kind from backend

[Tatsuo Ishii]

> On Tue, Apr 1, 2025 at 5:42 AM Tatsuo Ishii <ishii at postgresql.org> wrote:
> 
>> >> On Fri, Mar 28, 2025 at 2:59 AM Tatsuo Ishii <ishii at postgresql.org>
>> wrote:
>> >>
>> >>> >> > OK.  But what _is_ causing  the FATAL "kind mismatch" errors?
>> >>> >>
>> >>> >> Under investigation. I am failing to reproduce the problem.
>> >>> >>
>> >>> >
>> >>> > Thank you.
>> >>> >
>> >>> > The error _seems_ to be volume-related: happens more when the system
>> is
>> >>> > busy.  (But maybe a busy system has more opportunity to trigger the
>> bug.
>> >>>
>> >>> That's my thought too for now.
>> >>>
>> >>> > Would it help if I cranked up log_min_messages to DEBUG1 or even
>> DEBUG2?
>> >>>
>> >>> Yes, please enable DEBUG1.
>> >>>
>> >>
>> >> I was working on it before I saw your email. 😀   Gzip file sent
>> directly
>> >> to you, with extract for just the pid that had the error.).
>> >>
>> >> (Lord, but it generates a lot of log lines!)
>> >
>> > Thanks.
>>
>> Looking at the log file, I noticed that you are using the query cache
>> feature. To break down the issue, can you disable the query cache
>> feature?
>>
>> memory_cache_enabled = off
>> enable_shared_relcache = off
>>
>> ("enable_shared_relcache = on" uses the query cache feature even if
>> "memory_cache_enabled = off").
>>
> 
> Unfortunately, the decision has been made not to use any connection
> pooling, since the application vendor has not tested their applications
> with pgpool (or any other).
> 
> I'll keep this in my back pocket, though.

I think the issue was resolved in v4.5.6 or later.

https://git.postgresql.org/gitweb/?p=pgpool2.git;a=commit;h=6489bb0aeae49329c8cd0580cde366d918f2e4f5

Best reagards,
--
Tatsuo Ishii
SRA OSS K.K.
English: http://www.sraoss.co.jp/index_en/
Japanese:http://www.sraoss.co.jp