Re: client waits for end of update operation and server proc is idle - Mailing list pgsql-general

From Matthias Apitz
Subject Re: client waits for end of update operation and server proc is idle
Date
Msg-id YIKUEg8Eo0DchREa@c720-r368166.fritz.box
Whole thread Raw
In response to Re: client waits for end of update operation and server proc is idle  (Laurenz Albe <laurenz.albe@cybertec.at>)
Responses Re: client waits for end of update operation and server proc is idle  (Matthias Apitz <guru@unixarea.de>)
List pgsql-general
El día viernes, abril 23, 2021 a las 10:48:24a. m. +0200, Laurenz Albe escribió:

> > The serverlog has around this time (sorry for German):
> >
> > 2021-04-23 05:55:23.591 CEST [2317] LOG:  unvollständige Message vom Client
> > 2021-04-23 05:55:23.593 CEST [2317] FEHLER:  Speicher aufgebraucht
> > 2021-04-23 05:55:23.593 CEST [2317] DETAIL:  Kann Zeichenkettenpuffer mit 0 Bytes nicht um 1380986158 Bytes
vergrößern.
> > 2021-04-23 05:55:23.593 CEST [2317] LOG:  konnte Daten nicht an den Client senden: Datenübergabe unterbrochen
(brokenpipe) 
> > 2021-04-23 05:55:23.593 CEST [2317] FATAL:  Verbindung zum Client wurde verloren
> > 2021-04-23 05:57:20.339 CEST [27190] LOG:  unerwartetes EOF auf Client-Verbindung mit einer offenen Transaktion
> > 2021-04-23 05:59:17.419 CEST [26827] LOG:  unerwartetes EOF auf Client-Verbindung mit einer offenen Transaktion
> >
> > What else could be checked for this? Any hints about this issue?
>
> It seems pretty clear that both the client and the server are waiting for each other.

Yes, they are both alive as processes. Both are running on the same
Linux server, i.e. no real network between. And any loss of data would
have caused perhaps the end of one (and then the other) process.
I will try to gather a TCPDUMP for this.

>
> The server is of the opinion that it has completed the query, while the client is
> still waiting for the response.
>
> I would look for the problem in the network between both.  The error messages
> (unexpected EOF on client connection, broken pipe) seem to point in the same direction.


The other 'unerwartetes EOF auf Client' (unexpected EOF client) are
logged by other PostgreSQL server processes, not invloved here, and
caused by killing other clients with 'kill -9'.

    matthias
--
Matthias Apitz, ✉ guru@unixarea.de, http://www.unixarea.de/ +49-176-38902045
Public GnuPG key: http://www.unixarea.de/key.pub
¡Con Cuba no te metas!  «»  Don't mess with Cuba!  «»  Leg Dich nicht mit Kuba an!
http://www.cubadebate.cu/noticias/2020/12/25/en-video-con-cuba-no-te-metas/



pgsql-general by date:

Previous
From: Mitar
Date:
Subject: Re: Why is writing JSONB faster than just JSON?
Next
From: Luca Ferrari
Date:
Subject: enable_seqscan to off -> initial cost 10000000000