On 5/19/06, sheeri kritzer <[EMAIL PROTECTED]> wrote:
So I've seen replication break a few times due to duplicate queries.
A few times it was around a server crashing, but I thought perhaps it
was because of the crash. (ie, master sends a query, crashes, and
then tries to send the query again when it comes back up).
But in the past 16 hours, it's happened twice. Both times, no crash.
No network problems that we know of. No other query problems.
Therefore, we've deduced it's in the code -- it's trying to insert a
field with a duplicate primary key. I've sent that off to the
developers, who will hit things with sticks.
However, why does MySQL transfer over DML queries that fail? If they
have an error, shouldn't they not replicate? This seems like a very
large bug to me.
-Sheeri
(if y'all protest, I'll do some test cases to see if I can replicate
on current versions (we're using 4.1.12 as master and server,
replicating for a long time with no problems). I'm East Coast so I'm
feeling the Friday laziness).
First I would like to clarify that the Master by itself does not send
any Binlog data to the slaves, its up to the slaves to connect to the
master (using the I/O thread) get a connection id , and then read
data from the Masters Binlog.
Now coming to your case, we would like to know if you have some kind
of an automatic failover mechanism running between the master and the
slave,if so these kind of duplicate errors occur due
Partial_Network_Outages.
Could you also let us know how the master is brought back online after
it crashes, and please explain us as to what happens during a crash
(OS lock, Mysql daemon crash, etc ....)
Kishore Jalleda
http://kjalleda.googlepages.com/mysqlprojects
--
MySQL General Mailing List
For list archives: http://lists.mysql.com/mysql
To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
