Ignore unique key violation errors (1062) in multiwrite setup (slave_skip_errors?)

kartoffelheinzkartoffelheinz EntrantInactive User Role Participant
We have a 3-node xtradb 5.6 cluster doing multiwrites. From time to time, it happens that two nodes try to insert into a unique column at the same time, making the losing node inconstitent and having it dropped from the cluster. As a sidenote, a simple restart is rarely enough to bringt it back online with IST, most times a restart directly triggers SST, but after that, it errors out with innodb log failures and we have to delete data-directory and do another full SST. That seems to be some kind of bug or misconfiguration, but that got nothing to do with my question, I guess.

We would like to avoid these errors altogether, since the inserts into that table are not that important. With mysql master-slave replication, there is slave-skip-errors=1062 - is there an equivalent for xtradb cluster replication?

Comments

  • przemekprzemek Percona Support Engineer Percona Staff Role
    You may be experiencing this bug:
    https://bugs.launchpad.net/codership-mysql/+bug/1308016

    And no, at the moment there is no such equivalent at the time, as galera based cluster's priority goal is to guard against inconsistency.
  • kartoffelheinzkartoffelheinz Entrant Inactive User Role Participant
    Thanks for your reply.

    You might be right, the bug description sounds familiar. Do you know when there will be a fixed version available?
Sign In or Register to comment.

MySQL, InnoDB, MariaDB and MongoDB are trademarks of their respective owners.
Copyright ©2005 - 2020 Percona LLC. All rights reserved.