Hi AJ,

Thanks for that. We aren't doing any custom stuff with the database, so my guess is that we will be just fine going forward with this.

Thanks again for the response!

Steven

On 01/28/2015 02:37 PM, AJ McKee wrote:
Hi Steven,

I went through this process sometime ago, and like you changed to mixed mode replication for the MySQL backend.

If your underlying tables follow the pdns vanilla install, you should not encounter any problem, in fact you should see a good improvement on the performance of MySQL replication. This is assuming you have no other DB that is being replicated.

If you have customised your tables, or have a custom application accessing the DB, then it would be worthy to do a review of the query patterns (Enable query logging) to evaluate if there are any pitfalls. In particular watch out for queries that affect a large dataset within your application, however MIXED mode should determine the correct strategy to use.

An excellent write up on Statement, Row and Mixed modes is here; http://mysql.wingtiplabs.com/documentation/row639ae/configure-row-based-or-mixed-mode-replication and explains it much better than I or the MySQL docs could.

In short, I can't see any disadvantage to MIXED mode, but of course it depends on your setup.

Best of luck
AJ



On 28 January 2015 at 18:24, Steven Spencer <steven.spen...@kdsi.com <mailto:steven.spen...@kdsi.com>> wrote:

    Greetings,

    We have been using PDNS with gmysql back end for a long time and
    because of some legacy Xen server retirement projects, We have
    decided to do a forklift upgrade of our PDNS install at the same
    time. Currently, our primary and secondary DNS servers are using
    2.9.22 and we will be moving to 3.3.1-1. I've read all of the
    upgrade notes with the mysql schema changes, etc., and those items
    will be no issue to accomplish. My concern is that we are doing
    database replication using the old style replication methodology
    (not the mysql cluster methodology) and I'm wondering what, if
    anything, the change in the binlog_format from STATEMENT, which is
    what it currently is, to MIXED will do to replication.

    Our current set up is a mysql DNS master server that has the web
    front-end on it and is set up as the master for the database with
    both the primary and secondary DNS servers set as slaves to this
    machine. We also have external pdns-recursor machines that were
    recently updated to 3.6.2-1 and are working fine. Our
    implementation is on CentOS 5 (currently) and will be moving to
    CentOS 6.

    Thanks for any light that you might be able to shed on this process.

-- -- Steven G. Spencer, Network Administrator


    _______________________________________________
    Pdns-users mailing list
    Pdns-users@mailman.powerdns.com
    <mailto:Pdns-users@mailman.powerdns.com>
    http://mailman.powerdns.com/mailman/listinfo/pdns-users




--
AJ McKee
phone: +353 83 1130 545
profile: http://linkedin.com/in/ajmkee
jid: aj.mc...@druid-dns.com <mailto:aj.mc...@druid-dns.com>
blog: http://aj.mc-kee.com/
twitter: @ajmckee

_______________________________________________
Pdns-users mailing list
Pdns-users@mailman.powerdns.com
http://mailman.powerdns.com/mailman/listinfo/pdns-users

Reply via email to