MySQL-wsrep does not replicate Innodb memcached plugin writes
Affects | Status | Importance | Assigned to | Milestone | ||
---|---|---|---|---|---|---|
MySQL patches by Codership | Status tracked in 5.6 | |||||
5.6 |
New
|
Wishlist
|
Unassigned | |||
Percona XtraDB Cluster moved to https://jira.percona.com/projects/PXC | Status tracked in 5.6 | |||||
5.5 |
Invalid
|
Undecided
|
Unassigned | |||
5.6 |
New
|
Undecided
|
Unassigned |
Bug Description
3 node cluster following example here: http://
[root@node1 ~]# mysql < /usr/share/
Set these settings: (based on http://
log-bin
innodb_
[root@node1 ~]# service mysql restart
But when I write to a row, it does not replicate:
# telnet 127.0.0.1 11211
Trying 127.0.0.1...
Connected to 127.0.0.1.
Escape character is '^]'.
get all
VALUE all 10 9
123456789
END
get AA
VALUE AA 8 12
HELLO, HELLO
END
set BB 0 0 1
0
STORED
^]
telnet> quit
Connection closed.
[root@node1 mysql]# telnet 127.0.0.1 11211
Trying 127.0.0.1...
Connected to 127.0.0.1.
Escape character is '^]'.
incr BB 1
1
incr BB 1
2
^]
telnet> ^]
?Invalid command
telnet> quit
Connection closed.
node1 mysql> select * from demo_test;
+-----+
| c1 | c2 | c3 | c4 | c5 |
+-----+
| AA | HELLO, HELLO | 8 | 0 | 0 |
| all | 123456789 | 10 | 2 | 0 |
| BB | 2 | 0 | 3 | 0 |
+-----+
3 rows in set (0.00 sec)
But the other nodes do not get the new rows or updates:
node3 mysql> select * from demo_test;
+----+-
| c1 | c2 | c3 | c4 | c5 |
+----+-
| AA | HELLO, HELLO | 8 | 0 | 0 |
+----+-
1 row in set (0.00 sec)
There are no errors in replication, the cluster otherwise behaves normally.
summary: |
- Galera does not replicate memcached plugin writes + Galera does not replicate Innodb memcached plugin writes |
summary: |
- Galera does not replicate Innodb memcached plugin writes + MySQL-wsrep does not replicate Innodb memcached plugin writes |
This is only a problem if you enable innodb_ api_enable_ binlog.
We are running an innodb memcache plugin on our cluster with binlog enabled and ws_rep replicates just fine within the Percona cluster.
We are not interested in replicating the contents of our memcache tables to the async backup servers that slave off our main cluster so we just leave innodb_ api_enable_ binlog disabled.