Altering a table with more rows than wsrep_max_ws_rows causes cluster to break when in TOI mode

[Joseph Palermo]

This was tested on MariaDB 10.1.20 with Galera 25.3.17

Reproduce steps:

1. Create a 3 node galera cluster

2. Set wsrep_max_ws_rows to a non-zero value

3. Create a table with more than wsrep_max_ws_rows in it

4. Perform an alter table statement on that table. It will get rejected with:

   ERROR 1180 (HY000): wsrep_max_ws_rows exceeded done

5. The node you were connected to will still have the old table schema, but the other nodes have successfully applied the alter table changes.

At that point any data written to that table will cause the other nodes to terminate due to inconsistency.

[Philip Stoev]

Hello,

I was able to reproduce the problem and it seems it does not affect Galera
Cluster but is specific to MariaDB Cluster.

The code path that gets triggered is specific to MariaDB Cluster. You may
wish to open a ticket in their bug tracker. Once you have the number, please
send it in so that I can add a few observations to the ticket.

Thank you.

Philip Stoev

--
You received this message because you are subscribed to the Google Groups
"codership" group.
To unsubscribe from this group and stop receiving emails from it, send an
email to codership-tea...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Joseph Palermo]

Thanks, I've created a MariaDB bug for anybody else finding this:

https://jira.mariadb.org/browse/MDEV-11817