Re: BUG #7567: Sequences not properly replicated

Am 25.09.2012 18:08, schrieb Tom Lane:
> s.proels@pharmatechnik.de writes:
>> The following bug has been logged on the website:
>> Bug reference:      7567
>> Logged by:          Stefan Pröls
>> Email address:      s.proels@pharmatechnik.de
>> PostgreSQL version: 9.1.3
>> Operating system:   OpenSUSE 11.4 32-Bit
>> Description:
>> Scenario:
>> 1 Master, 1 Hot-Standby Slave, asynchronous streaming replication, both
>> servers PostgreSQL 9.1.3.
> The fact that the slave sees the last_value as ahead of the master is
> not a bug; you'd have the same last_value on the master if it were to
> crash and restart.  (This is due to a tradeoff made long ago to reduce
> the amount of WAL traffic created by nextval: only one WAL record is
> written per 32 nextval's, and that record claims last_value is 32 ahead
> of what it previously was.)
>
> There is a known bug, fixed in 9.1.5, associated with sequences that
> have been ALTER'd ... but that's not what you're showing here.
>
>                       regards, tom lane
>

Ah, I see. Thank you for explaining this!


Best regards,
Stefan Pröls


________________________________

PHARMATECHNIK GmbH und Co. KG
Münchner Strasse 15
D-82319 Starnberg

Sitz der Gesellschaft: Starnberg
HRA: 64434, HRB: 66369, Amtsgericht München
Geschäftsführer: Dr. Detlef Graessner, Dr. Mathias Schindl, Stephan Jörgens