Circular replication crashes every day

[asaf halfon]

Hi,

I'm running a circular replication between 4 DBs.
Usually it works fine, but sometimes when someone manually execute something like this:
UPDATE test_table SET active=1 WHERE id IN (1,2);

the replication crashes with this error on the slaves (usually the same one):
Query caused different errors on master and slave. Error on master: message (format)='Duplicate entry '%-.192s' for key %d' error code=1062 ; Error on slave: actual message='no error', error code=0. Default database: 'test'. Query: 'UPDATE test_table SET active=1 WHERE id IN (1,2);'

All DBs are both writing and reading data.

I tried to sync them with dump but it doesn't seem to help for more than 1 day.

If I skip that query and execute it again from another server (or from the same one) - it works. (which is very weird).

any ideas why this could happen?