Multi Master - Single Slave - Configuration?

[Max Ziegler]

Hello,

i try to build up a configuration where databases (with different names) on different Servers will be syncronised (only in one direction) with one Server.
My question: is this possible with the replication feature?

overview:

Server 1 [db1]
Server 2 [db2] --> Server 4 [db1, db2, db3]
Server 3 [db3]

the arrow shows the data flow. On Server 4 only read processes will be used.

My thoughts:
I configure Server 1 - 3 as replication masters and the server 4 as slave, and use a script which switch the connection to the server each 5 minutes with the command slave stop; CHANGE MASTER TO ... slave start;
with the PARAMETER
MASTER_LOG_FILE='server<1-3>-log', RELAY_LOG_FILE='server<1-3>-relay';

But this doesnt work.
Error msg: ERROR 1380 (HY000): Failed initializing relay log position: Could not find target log during relay log initialization
error_log on slave:
110311 11:57:04 [ERROR] Error reading packet from server: Could not find first log file name in binary log index file ( server_errno=1236)
110311 11:57:04 [ERROR] Slave I/O: Got fatal error 1236 from master when reading data from binary log: 'Could not find first log file name in binary log index file', Error_code: 1236

If I remove the Parameters, he will start at position 0 at the bin-log file and re-enter all entries.

So do i have to write a script which remebers the position from the *.index file and set the parameter MASTER_LOG_POS?

Or is it easier to start 3 instances of slave-servers on server 4?

Or is there any other tool which fits better? (like clustering?)

Thanks a lot, max



Edited 1 time(s). Last edit at 03/11/2011 06:31AM by Max Ziegler.