[prev in list] [next in list] [prev in thread] [next in thread] 

List:       postgresql-general
Subject:    Re: Active connections are terminated because of small wal_sender_timeout
From:       Tom Lane <tgl () sss ! pgh ! pa ! us>
Date:       2019-07-31 14:03:13
Message-ID: 23234.1564581793 () sss ! pgh ! pa ! us
[Download RAW message or body]

AYahorau@ibagroup.eu writes:
>> I do not think anybody thinks this is a bug.  Setting wal_sender_timeout
>> too small is a configuration mistake.

> Why is it a configuration mistake? This value is allowed  to be set. There 
> is no any restriction about it.

The fact that a parameter can be set does not mean that we guarantee that
every possible value will work for everybody.  As an example, if you
configure work_mem to a few TB and then start getting OOM failures because
your machine can't actually support that, it's not a bug that we let you
set the value that high.  The upper limit of what's usable is too variable
and too hard to determine, so we don't even try; it's on you to choose a
suitable setting for your situation.  Similarly, the useful range of
wal_sender_timeout is hard to predict and is likely to be quite different
for different installations.  We don't want to artificially constrain what
people can use, so the range of allowed settings *ought* to include some
values that are not practically useful in specific situations.  Those
values might be just the right ones for someone else.

			regards, tom lane


[prev in list] [next in list] [prev in thread] [next in thread]