Centos 7.4

What kind of "cluster" do you have? Is it a Galera cluster or the regular replication?

I believe I have a "regular" galera.
..
[galera]

1. Mandatory settings
   wsrep_on=ON
   wsrep_provider = /usr/lib64/galera/libgalera_smm.so
   wsrep_cluster_name = "ccnr-maria-cluster"
   wsrep_cluster_address = "gcomm://10.5.6.32,10.5.6.100"
   #wsrep_cluster_address = "gcomm://10.5.6.17,10.5.6.32,10.5.6.100,...?pc.wait_prim=no"
   wsrep_node_address = "10.5.6.49"
   wsrep_node_name = "whale"
   wsrep_sst_method = rsync
   log-bin = mariadb.binlog
   binlog_format = row
   default_storage_engine = InnoDB
   innodb_autoinc_lock_mode = 2
   wsrep-log-conflicts
   wsrep-debug
   ..

regards.

to be bit more specific, on the node on which I execute:
$ mysql < db db.sql
database creation/restoration is complete, but other two nodes get tables, columns but no records.
I go on a different node to exec mysql < and the same happens.

hi guys,
would you have any workarounds that one could try before you offer an "official" fix?
many thanks, L.

hi folks, would you have some update?
I do see that statements, for a database created that way: $ mysql < db db.sql are NOT replicated even with:
> update wp_options set some=some where some=is-some
as if replication does not care about such a database.

hi,
this must be some special case mysql cannot handle. I tried another database and there everything seems to be replicating okey.
It's not that I tried, tested ten different databases but the one that is not working, not replicating is wordpress database.
Is there something I can do to troubleshoot it, provide more info?
Can you test wordpress db?

How could it be that one/some db is not replicating while others are fine?

Just an update - for that wordpress db, in wordpress I used "reset db plugin" and I reset just one table: options and... ! Replications re-started immediately.
Is that not even bigger mystery? How could some data in one table stop MariaDB replication from doing that one database?

Can you provide step by step instructions how to repeat. We would need contents of db.sql and can you provide output where we can see that data is not replicated (I hope you have verified that tables are InnoDB-tables) and error log from all nodes.

I have tried to reproduce it on 10.1.31-MariaDB-debug Source distribution (aba15b864af664276261dcbbdaef35268d407af1) with Galera 25.3.22(r3787) (7715c28895faa892b5201059a904e581edc106b2) with the same commands and options as mentioned here above. I used a cluster of three nodes made by:

./mtr --suite=galera_3nodes --start-and-exit --mysqld=--wsrep-debug --mysqld=--wsrep_cluster_name="ccnr-maria-cluster"

Created DB with three such tables:

CREATE TABLE `t1` (

  `i` int(11) NOT NULL AUTO_INCREMENT,

  `j` int(11) DEFAULT NULL,

  `k` int(11) DEFAULT NULL,

  PRIMARY KEY (`i`)

) ENGINE=InnoDB AUTO_INCREMENT=61306 DEFAULT CHARSET=latin1 

Filled it with 30K rows each and restored backup successfully replicated.
It needs more info. At least "show create" for all the tables from the db and Galera lib version.