Ebuild of Postgres-8.3.8 almost a catastrophe

View: New views
6 Messages — Rating Filter:   Alert me  

Ebuild of Postgres-8.3.8 almost a catastrophe

by Konstantinos Agouros :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Hi,

usually upgrade within a minor version of postgres do not need a pg_dump_all/
and reimport. But the upgrade from 8.3.7 to 8.3.8 does as the configure option
for integer-timestamps changed and thus the newly compiled server can not read
the existing database files.

Regards,

Konstantin
--
Dipl-Inf. Konstantin Agouros aka Elwood Blues. Internet: elwood@...
Altersheimerstr. 1, 81545 Muenchen, Germany. Tel +49 89 69370185
----------------------------------------------------------------------------
"Captain, this ship will not survive the forming of the cosmos." B'Elana Torres


Re: Ebuild of Postgres-8.3.8 almost a catastrophe

by Stroller :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message


On 9 Nov 2009, at 10:25, Konstantinos Agouros wrote:
> ...
> usually upgrade within a minor version of postgres do not need a  
> pg_dump_all/
> and reimport. But the upgrade from 8.3.7 to 8.3.8 does as the  
> configure option
> for integer-timestamps changed and thus the newly compiled server  
> can not read
> the existing database files.

Thanks, dude!

This is very useful to know about.

I assume this is an upstream decision, and may be expected to apply to  
8.3.9, 8.3.10, ... also? I have no reason to "wait and see if this is  
resolved", I should be careful to do a dump and restore next time I  
upgrade?

Cheers,

Stroller.



Re: Ebuild of Postgres-8.3.8 almost a catastrophe

by Volker Armin Hemmann :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

On Montag 09 November 2009, Konstantinos Agouros wrote:

> Hi,
>
> usually upgrade within a minor version of postgres do not need a
>  pg_dump_all/ and reimport. But the upgrade from 8.3.7 to 8.3.8 does as the
>  configure option for integer-timestamps changed and thus the newly
>  compiled server can not read the existing database files.
>
> Regards,
>
> Konstantin
>

have you opened a bug report about this? Or is there one already?


Re: Ebuild of Postgres-8.3.8 almost a catastrophe

by Bugzilla from graham@gmurray.org.uk :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Stroller <stroller@...> writes:

> On 9 Nov 2009, at 10:25, Konstantinos Agouros wrote:

> I assume this is an upstream decision, and may be expected to apply to
> 8.3.9, 8.3.10, ... also? I have no reason to "wait and see if this is
> resolved", I should be careful to do a dump and restore next time I
> upgrade?

NO. It was not an upstream change. Upstream changed to integer
timestamps by default in 8.4.0, the default did not change between 8.3.7
and 8.3.8. However, in typical gentoo manner, this is configurable by
the pg-intdatetime USE flag, which is not enabled by default.


Re: Ebuild of Postgres-8.3.8 almost a catastrophe

by Konstantinos Agouros :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

In <200911091155.35236.volkerarmin@...> volkerarmin@... (Volker Armin Hemmann) writes:

>On Montag 09 November 2009, Konstantinos Agouros wrote:
>> Hi,
>>
>> usually upgrade within a minor version of postgres do not need a
>>  pg_dump_all/ and reimport. But the upgrade from 8.3.7 to 8.3.8 does as the
>>  configure option for integer-timestamps changed and thus the newly
>>  compiled server can not read the existing database files.
>>
>> Regards,
>>
>> Konstantin
>>

>have you opened a bug report about this? Or is there one already?
Nope didn't have the time. I needed to get the DB running again quick by:

ebuild .../postgresql-server-8.3.8.ebuild compile, interrupting this
call ./configure manually with --enable-integer-timestamps and then copying
the postgres binary in the approp. position after compiling. This is working.
The next upgrade if ~amd64 is enabled would be 8.4 so I should dump/restore
anyway. But I will do that in an appropriate timeslot.

Regards,

Konstantin

--
Dipl-Inf. Konstantin Agouros aka Elwood Blues. Internet: elwood@...
Altersheimerstr. 1, 81545 Muenchen, Germany. Tel +49 89 69370185
----------------------------------------------------------------------------
"Captain, this ship will not survive the forming of the cosmos." B'Elana Torres


Re: Ebuild of Postgres-8.3.8 almost a catastrophe

by Konstantinos Agouros :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

In <87r5s7q2je.fsf@...> graham@... (Graham Murray) writes:

>Stroller <stroller@...> writes:

>> On 9 Nov 2009, at 10:25, Konstantinos Agouros wrote:

>> I assume this is an upstream decision, and may be expected to apply to
>> 8.3.9, 8.3.10, ... also? I have no reason to "wait and see if this is
>> resolved", I should be careful to do a dump and restore next time I
>> upgrade?

>NO. It was not an upstream change. Upstream changed to integer
>timestamps by default in 8.4.0, the default did not change between 8.3.7
>and 8.3.8. However, in typical gentoo manner, this is configurable by
>the pg-intdatetime USE flag, which is not enabled by default.
Thanks for this bit of informationn.

--
Dipl-Inf. Konstantin Agouros aka Elwood Blues. Internet: elwood@...
Altersheimerstr. 1, 81545 Muenchen, Germany. Tel +49 89 69370185
----------------------------------------------------------------------------
"Captain, this ship will not survive the forming of the cosmos." B'Elana Torres