I've gone thru this and several other forums and am hoping a posting may get results, worth a shot or hours of eye popping code is coming....
Here is the set up, I have mutliple development servers located remotely that contain a unique schema to that site, yet all sites have the same internal table structures, mysqldump, is not an incremental or archival facility, nor is the master-slave replication inherent in 5.xxxx+.
Given, local server(ServerX) = backup server/archival master, remote = DataGenerator ServerY (Do have other 'ServerY's as well)
If ServerY is promoted master in replication relation with Server X, ALL ServerY data is transfered...GOOD!. If Server Y Crashes and has to be rebuilt, you MUST
reload ServerY's data BACK from ServerX (mysqldump or reverse the relation)but this is BAD...ServerY as the New ServerY will sync up and delete records it does not have. Clients do NOT need historical information but the historical information MUST be preserved on ServerX
I thought about mucking with the bin log offsets (and needed to switch between multiple masters), but thats bucking for trouble, so now I'm reduced to a simple program to fetch one record from one table at a time, verify its present in archive and (validate it maybe...very slow) if its not then select the record on remote and insert it in the archive
I cant be the only person in the world needing to concentrate and archive remote MySQL db's into one server, if someone has an insight or a professional or open source product that could be adapted I'd love to hear it.
Here is the set up, I have mutliple development servers located remotely that contain a unique schema to that site, yet all sites have the same internal table structures, mysqldump, is not an incremental or archival facility, nor is the master-slave replication inherent in 5.xxxx+.
Given, local server(ServerX) = backup server/archival master, remote = DataGenerator ServerY (Do have other 'ServerY's as well)
If ServerY is promoted master in replication relation with Server X, ALL ServerY data is transfered...GOOD!. If Server Y Crashes and has to be rebuilt, you MUST
reload ServerY's data BACK from ServerX (mysqldump or reverse the relation)but this is BAD...ServerY as the New ServerY will sync up and delete records it does not have. Clients do NOT need historical information but the historical information MUST be preserved on ServerX
I thought about mucking with the bin log offsets (and needed to switch between multiple masters), but thats bucking for trouble, so now I'm reduced to a simple program to fetch one record from one table at a time, verify its present in archive and (validate it maybe...very slow) if its not then select the record on remote and insert it in the archive
I cant be the only person in the world needing to concentrate and archive remote MySQL db's into one server, if someone has an insight or a professional or open source product that could be adapted I'd love to hear it.