I trying to migrate from one Datacenter with MySQL 5.5 with tokudb storage engine
to a new datacenter with MySQL 5.6 w/ tokudb storage engine
I am trying to do this by building out Slaves in new datacenter reading from Master in Old Datacenter
and when they are caught up - have a production outage for a couple minutes
Flush the master logs then switch application to read/write from old to new datacenter where I appoint one of the Slaves to be the new Master
I know the Tokudb File header structure is different in MySQL 5.5 (FH FileHeader v2.4) and 5.6 has FH v2.7 or something
but
is it possible for a 5.6 Slave database to read from a 5.5 Master ???
- or am I Doomed !! ?
Doing a mysqldump in PROD would take +1 day total down time of Prod MASTER
Copy the massive dump over to new Datacenter
Import dump which would take another day or two …
Then point the application to start using the new Datacenter …
This is not really acceptable to be down a couple day switching datacenteers!!
After several attement I was able to start the replication from 5.5 to 5.6
but I run into error
and I have to do too many GLOBAL skips_counters
so I have a feeling it’s not supported to read from Tokudb 5.5 on a 5.6 slave…
Last_SQL_Errno: 1032
Last_SQL_Error: Could not execute Update_rows_v1 event on table YSIUSER.ysi_batch_notices; Can’t find record in ‘ysi_batch_notices’, Error_code: 1032; handler error HA_ERR_KEY_NOT_FOUND; the event’s master log sjcprddbv202-binlog.023152, end_log_pos 9401388
[(none)]> stop slave;
Query OK, 0 rows affected (0.00 sec)
[(none)]> set GLOBAL SQL_SLAVE_SKIP_COUNTER = 1 ;
Query OK, 0 rows affected (0.00 sec)
[(none)]> start slave;
Query OK, 0 rows affected (0.00 sec)
Last_SQL_Errno: 1032
Last_SQL_Error: Could not execute Update_rows_v1 event on table YSIUSER.ysi_batch_notices; Can’t find record in ‘ysi_batch_notices’, Error_code: 1032; handler error HA_ERR_KEY_NOT_FOUND; the event’s master log sjcprddbv202-binlog.023152, end_log_pos 9417817