[pgpool-hackers: 505] Re: [pgpool-general: 2635] Re: redhat pgpool.init script

[Ryan DeShone]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I've applied this patch to some servers and tested it. So far, it works
much better than the previous iteration in our environment. Really
appreciate that reload actually works with this patch and restarting
the VIP holder no longer fails. My only suggestion would be to place
the variable defaults (PGPOOLUSER, PGPOOLENGINE, etc.) BEFORE the block
that sources /etc/sysconfig/pgpool, thus making it so those variables
can be overridden using the typical method.

On Wed 23 Apr 2014 06:43:26 AM EDT, Yugo Nagata wrote:
> Hi Alexandru,
>
> I wrote a new pgpool.init. Could you please try this?
>
> In previous, pgpool.init uses killproc for stopping pgpool, but there are
> several problems; (1) some child processes can be left since the parent
> process restarts killed processes; (2) when pgpool has connection from
> clients, pgpool can't stop until connection closed; (3) restart() fails
> for these reasons.
>
> In the new version, "pgpool -m fast stop" is used in stop(). The problems
> above are resolved and restart() also works well, even when watchdog enabled.
> In addition, exit code according with LSB is considered and try-restart
> option is implemented.
> http://refspecs.linuxbase.org/LSB_3.1.1/LSB-Core-generic/LSB-Core-generic/iniscrptact.html
>
>
> On Mon, 17 Mar 2014 15:37:09 -0700
> Alexandru Cardaniuc <cardaniuc at gmail.com> wrote:
>
>> Hi Devrim,
>>
>> I just checked the sources from the link you provided for the pgpool.init:
>> http://svn.pgrpms.org/repo/rpm/redhat/9.3/pgpool-II/EL-6/pgpool.init
>>
>> And it's still version 2.3 over there. That doesn't account for the
>> problems described here:
>> http://www.sraoss.jp/pipermail/pgpool-hackers/2012-May/000068.html
>> and here:
>> http://lists.pgfoundry.org/pipermail/pgpool-hackers/2010-November/000407.html
>>
>> The problem seems to be also described in this bug ticket:
>> http://www.pgpool.net/mantisbt/view.php?id=12
>>
>> I can consistently reproduce it. How to reproduce it:
>> * configure pgpool-II to use watchdog with two nodes
>> say delegate_IP is on the primary node
>> * then service pgpool stop on the primary node will bring down the
>> delegate_IP interface and that would go to the secondary node, but it would
>> not remove the /var/run/pgpool_status and both socket files (for pcp 9898
>> and pgpool 9999). Hence when you try to start it later it will fail with
>> cannot bind to socket error.
>> That also makes 'service pgpool restart' fail on the node that has the
>> delegate_IP assigned at the moment.
>>
>> It works fine on the node that doesn't have delegate_IP assigned and also
>> works fine on the primary node with delegate_IP BUT only when the secondary
>> node is already down.
>>
>> Looks like it can be traced to how the 'pgpool stop' calls killproc and
>> only when watchdog with delegate_IP is configured (with ifconfig to bring
>> the interface up/down).
>>
>>
>>
>>
>>
>> On Sun, Mar 16, 2014 at 4:53 PM, Devrim GÜNDÜZ <devrim at gunduz.org> wrote:
>>
>>>
>>> Hi,
>>>
>>> Now we have the community RPM repository, RPMS are there:
>>>
>>>
>>> http://yum.postgresql.org/9.3/redhat/rhel-6-x86_64/repoview/pgpool-II-93.html
>>>
>>> This also includes the fixes there. Still, if you want to use the
>>> sources, and not the RPMs, then:
>>>
>>> http://svn.pgrpms.org/repo/rpm/redhat/9.3/pgpool-II/EL-6/
>>>
>>> is the URL for the spec file and other patches we have.
>>>
>>> Regards, Devrim
>>>
>>> On Sat, 2014-03-15 at 22:23 -0700, Alexandru Cardaniuc wrote:
>>>>
>>>>
>>>> Was reading 2012 mailing list and saw a thread about issues with
>>>> redhat
>>>> pgpool.init script. But it doesn't look like any changes have been
>>>> made to
>>>> the checked-in pgpool.init script. The last version is still the one
>>>> from 4
>>>> years ago?:
>>>> # v2.2.5 Devrim GUNDUZ <devrim at CommandPrompt.com>
>>>> # - Fix logging.
>>>>
>>>> And that's the one checked-in the latest version of pgpool sources. I
>>>> get
>>>> it when I build using pgpool-II-3.3.2.tar.gz
>>>>
>>>> Is there a newer version of the pgpool.init script that is for some
>>>> reason
>>>> not included in the latest sources tarball?
>>>
>>>
>>> --
>>> Devrim GÜNDÜZ
>>> Principal Systems Engineer @ EnterpriseDB: http://www.enterprisedb.com
>>> PostgreSQL Danışmanı/Consultant, Red Hat Certified Engineer
>>> Twitter: @DevrimGunduz , @DevrimGunduzTR
>>>
>>>
>>>
>>
>>
>> --
>> Sincerely yours,
>> Alexandru Cardaniuc
>
>
>
>
> _______________________________________________
> pgpool-hackers mailing list
> pgpool-hackers at pgpool.net
> http://www.pgpool.net/mailman/listinfo/pgpool-hackers

- --
Ryan DeShone
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.22 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQEcBAEBAgAGBQJTWSf7AAoJEJ3SEJwY0lkxkIgH/2HdX4y2WccJSsX3hxnTMF4D
s6WyU1gTXcRgMXt2vjZOJ/p4DLoKn+z5Cu/q3vrxmM8/3DXoGRi5WTF4gEQEiQPK
pBSwG7kG0Ssq3stTY6e0yWvFOonSiuQ8T182owKXvuXsbV9OEJQ4HMMRnF4fCChh
ZmYi1vwnRMBJYupQb2xhEmWullg5ugo7p2BpJ2rPesaLRv6/3ky3apsOHSfOiZle
gB1H7LxRYFynj/vKhDTEDsRUe+Zx7VXjm/2fszNP6UQ+w9OvrMbzo2iqbdwrzzGs
8LMpgziVPbY2jyvneHqhb01Fqxlze2Pv4wQ6nY580YgeSydBm1aa+LmdBEIEklA=
=vT4s
-----END PGP SIGNATURE-----