Re: BUG #14714: long running sessions from remote instance seems to hang some times

From: Josef Machytka <josef(dot)machytka(at)gmail(dot)com>
To: Andres Freund <andres(at)anarazel(dot)de>
Cc: pgsql-bugs(at)postgresql(dot)org
Subject: Re: BUG #14714: long running sessions from remote instance seems to hang some times
Date: 2017-07-06 09:26:18
Message-ID: CAGvVEFs3FxMKH9SQKPvszJjEBcmXe=WQqsBRgkKiq=NurEVJHw@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-bugs

Hi,
thank you for an advice - I set it and tested for several days.
Good news is we do not have "drowsy" sessions any more.
But performance of queries on huge tables went seriously down. Even with
parallel processing.

It is understandable regarding memory management but not sustainable for
the future since our tables are growing almost exponentially. In these days
we have weekly increase in size >100GB and it goes still higher. Database
now have 2,5 TB. We had to move some of our queries from PostgreSQL to
Bigquery because performance was really bad...

I checked history of this database and previously PostgreSQL 9.5 was
running on Google GCE instance with Ubuntu 14.04 now we have new instance
with Debian 8 and PostgeSQL 9.6.

On previous instance with Ubuntu we did not have all these problems at all.
I checked and Ubuntu 14.04 has by default hugepages enabled (value
“always”) + fixed setting of number of huge pages but Debian 8 has default
setting “madvise” - memory adviser.

Please has someone here some deeper experiences with differences between
Ubuntu and Debian as environment for PostgreSQL as highly used OLAP
database? Because so far it looks like switching to Debian was rather bad
choice....

Thanks

Josef

In response to

Browse pgsql-bugs by date

  From Date Subject
Next Message Ryan Murphy 2017-07-06 15:00:36 Re: Bug in Physical Replication Slots (at least 9.5)?
Previous Message Maksim Milyutin 2017-07-06 07:12:09 Re: The 'Postgre SQL Server' locked in task manager