Thread: Strange behaviour: create database hangs (v 8.4.4)

Strange behaviour: create database hangs (v 8.4.4)

From
Dale Gallagher
Date:
Hello

When running the command 'create database test' as user postgres from
within psql, the command hangs indefinitely.

I've installed 8.4.4 from source on a Slackware 13.0 box with kernel
2.6.29, GCC 4.3.3 and OpenSSL 0.9.8n with self-signed certificates.

Built with ./configure --prefix=/opt/postgresql-8.4.4 --with-openssl

Running under daemontools as follows:

setuidgid postgres postgres -i -p 8910 -l -d /srv/data/postgresql 2>&1

psql run as postgres user as follows:

psql -p 8910

There is no indication in the logfiles of errors.

Any possible reasons why this may occur? What can I do to investigate
this further?

Thanks
Dale

Re: Strange behaviour: create database hangs (v 8.4.4)

From
Dale Gallagher
Date:
On 31 May 2010 12:50, Dale Gallagher <dale.gallagher@gmail.com> wrote:
> When running the command 'create database test' as user postgres from
> within psql, the command hangs indefinitely.

I am now using strace on postgres startup and watching the filesystem
during the creation of the database. So far, it seems that files in
PGHOME/base/1234/ are appearing, but VERY slowly. Thus far it's taken
7 minutes and 92 files have been created. The command has not
completed within psql....

Could this be a filesystem corruption issue? The host is using ext4.

Re: Strange behaviour: create database hangs (v 8.4.4)

From
Dale Gallagher
Date:
On 31 May 2010 13:10, Dale Gallagher <dale.gallagher@gmail.com> wrote:
> I am now using strace on postgres startup and watching the filesystem
> during the creation of the database. So far, it seems that files in
> PGHOME/base/1234/ are appearing, but VERY slowly. Thus far it's taken
> 7 minutes and 92 files have been created. The command has not
> completed within psql....
>
> Could this be a filesystem corruption issue? The host is using ext4.

The command finally completed. It took a VERY long time. Something is
wrong, but I have no clue what.

Re: Strange behaviour: create database hangs (v 8.4.4)

From
Naomi Walker
Date:
Maybe "test" is a reserved word.

Try test1.
________________________________________
From: pgsql-admin-owner@postgresql.org [pgsql-admin-owner@postgresql.org] on behalf of Dale Gallagher
[dale.gallagher@gmail.com]
Sent: Monday, May 31, 2010 3:50 AM
To: pgsql-admin@postgresql.org
Subject: [ADMIN] Strange behaviour: create database hangs (v 8.4.4)

Hello

When running the command 'create database test' as user postgres from
within psql, the command hangs indefinitely.

I've installed 8.4.4 from source on a Slackware 13.0 box with kernel
2.6.29, GCC 4.3.3 and OpenSSL 0.9.8n with self-signed certificates.

Built with ./configure --prefix=/opt/postgresql-8.4.4 --with-openssl

Running under daemontools as follows:

setuidgid postgres postgres -i -p 8910 -l -d /srv/data/postgresql 2>&1

psql run as postgres user as follows:

psql -p 8910

There is no indication in the logfiles of errors.

Any possible reasons why this may occur? What can I do to investigate
this further?

Thanks
Dale

--
Sent via pgsql-admin mailing list (pgsql-admin@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-admin

Re: Strange behaviour: create database hangs (v 8.4.4)

From
Tom Lane
Date:
Dale Gallagher <dale.gallagher@gmail.com> writes:
> On 31 May 2010 13:10, Dale Gallagher <dale.gallagher@gmail.com> wrote:
>> I am now using strace on postgres startup and watching the filesystem
>> during the creation of the database. So far, it seems that files in
>> PGHOME/base/1234/ are appearing, but VERY slowly. Thus far it's taken
>> 7 minutes and 92 files have been created. The command has not
>> completed within psql....
>>
>> Could this be a filesystem corruption issue? The host is using ext4.

> The command finally completed. It took a VERY long time. Something is
> wrong, but I have no clue what.

Seems like you must have a kernel/filesystem issue, but I don't know
what either.  CREATE DATABASE just copies the contents of template1
using some absolutely straightforward code (look into src/port/copydir.c)
so it's entirely filesystem-limited.

You could try tracing the backend that's doing the command with
"strace -r -p <pid>" to try to get an idea of which system calls
are eating all the time.

            regards, tom lane

Re: Strange behaviour: create database hangs (v 8.4.4)

From
Jeremy Tunnell
Date:
unregister

On Mon, May 31, 2010 at 9:59 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Dale Gallagher <dale.gallagher@gmail.com> writes:
>> On 31 May 2010 13:10, Dale Gallagher <dale.gallagher@gmail.com> wrote:
>>> I am now using strace on postgres startup and watching the filesystem
>>> during the creation of the database. So far, it seems that files in
>>> PGHOME/base/1234/ are appearing, but VERY slowly. Thus far it's taken
>>> 7 minutes and 92 files have been created. The command has not
>>> completed within psql....
>>>
>>> Could this be a filesystem corruption issue? The host is using ext4.
>
>> The command finally completed. It took a VERY long time. Something is
>> wrong, but I have no clue what.
>
> Seems like you must have a kernel/filesystem issue, but I don't know
> what either.  CREATE DATABASE just copies the contents of template1
> using some absolutely straightforward code (look into src/port/copydir.c)
> so it's entirely filesystem-limited.
>
> You could try tracing the backend that's doing the command with
> "strace -r -p <pid>" to try to get an idea of which system calls
> are eating all the time.
>
>                        regards, tom lane
>
> --
> Sent via pgsql-admin mailing list (pgsql-admin@postgresql.org)
> To make changes to your subscription:
> http://www.postgresql.org/mailpref/pgsql-admin
>

Re: Strange behaviour: create database hangs (v 8.4.4)

From
Dale Gallagher
Date:
On 31 May 2010 18:59, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Seems like you must have a kernel/filesystem issue, but I don't know
> what either.  CREATE DATABASE just copies the contents of template1
> using some absolutely straightforward code (look into src/port/copydir.c)
> so it's entirely filesystem-limited.

Thanks Tom. I've confirmed that it's not related to PostgreSQL, as
installing this version on another system has zero issues. I've posted
the issue to a Slackware specific forum:

http://www.linuxquestions.org/questions/slackware-14/slackware-13-sluggish-system-possibly-filesystem-811442/