Re: Corrupted Data ? - Mailing list pgsql-general

On 08/08/2016 09:28 AM, Ioana Danes wrote:
>
>
> On Mon, Aug 8, 2016 at 12:19 PM, Adrian Klaver
> <adrian.klaver@aklaver.com <mailto:adrian.klaver@aklaver.com>> wrote:
>
>     On 08/08/2016 09:11 AM, Ioana Danes wrote:
>
>         Hi,
>
>         I suspect I am having a case of data corruption. Here are the
>         details:
>
>         I am running postgres 9.4.8:
>
>         postgresql94-9.4.8-1PGDG.rhel7.x86_64
>         postgresql94-contrib-9.4.8-1PGDG.rhel7.x86_64
>         postgresql94-libs-9.4.8-1PGDG.rhel7.x86_64
>         postgresql94-server-9.4.8-1PGDG.rhel7.x86_64
>
>         on CentOS Linux release 7.2.1511 (Core)
>
>         This is happening in a production environment but luckily on the
>         reporting database.
>         I have a cluster of 3 databases, db1 and db2 are masters and
>         replicate
>         between each other and also replicate to db3 (db1 <-> db2, db1
>         -> db3,
>         db2 -> db3).
>         For replication I am using Bucardo.
>
>
>     I would say this is more a question for the Burcardo list:
>
>     https://mail.endcrypt.com/mailman/listinfo/bucardo-general
>     <https://mail.endcrypt.com/mailman/listinfo/bucardo-general>
>
>     I am just not seeing that replicating two masters on to a single
>     database is going to end well.
>
>
> Only one master is active at one time the other one is in stand by that
> is a topic for another discussion but in our case that works well.
>
> That was my first assumption, that it is a kind of a race condition or a
> bug on replication but I quickly ruled that out because that does not
> explain why when I filtered the table by transactionid = 75315815 it
> shows one record with transactionid 75315811...
>
> select gameplayid, transactionid, encodedplay from abrazo.matchgameplay
> where transactionid in (75315815) order by transactionid;;
>  gameplayid | transactionid | encodedplay
> ------------+---------------+--------------
>   160019271 |      75315815 | mix:9,0,9
>   160019269 |      75315815 | mix:9,8,9
>   160019267 |      75315815 | mix:9,2,2
>   160019265 |      75315815 | mix:2,2,8
>   160019263 |      *75315811 *| backup:1,9,1
>   160019261 |      75315815 | backup:2,0,9
>
> So I don't think it is a replication issue...
>

Other that, if I am following correctly, it is on the database(db3)
being replicated to. The only way db3 is getting its data is through
replication, is that correct?. On the master databases the data is correct.


--
Adrian Klaver
adrian.klaver@aklaver.com