Sun Dec 9 15:09:13 PST 2007
- Previous message: [Slony1-general] Delayed replication: Explicitly creating a slony-I backlog, and then telling the slon daemon to replicate changes all at once.
- Next message: [Slony1-general] Delayed replication: Explicitly creating a slony-I backlog, and then telling the slon daemon to replicate changes all at once.
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
David, Unfortunately, changes to the database may be quite varied. We aren't talking about massive amounts of data though - in the order a few dozen rows per update per slave. Confining that to a single transaction isn't realistic, though. How would I go about applying this horrifically kludgy solution? Would the following be do-able ? : 1. Make changes to the database with no Slon daemons running, through my C++ front end. 2. Start the Slon daemon for the first subscriber using my application . Have the slon daemon update changes made to the db to its subscriber as necessary, in a reasonable space of time. 3. Shutdown the Slon daemon when it was finished. I'd need some mechanism to tell my application that it was finished, so that it'd know when to do this. I'm using the libpqxx API ( a C++ wrapper of libpq). 4. Repeat the process for some or all other subscribers, as the user may or may not require changes to be immediately replicated to all subscribers. This is clearly a bit of a hack - another option would be to use log shipping, and to transfer the log to my application (over ftp or something) as necessary. It would be once again necessary (or at least desirable) to run the Slon daemon only when doing the update - anything more would be superfluous. So I'd still have to be able to monitor the Slon daemon's progress. Could I monitor the slon daemon's progress from my MS Windows MFC C++ application, preferably without spawning DOS boxes? You may very well not be up on windows (I'm not really a windows guy myself), in which case I'd appreciate it if you could offer an equivalent solution for Linux - I'll probably be able to apply the same logic in windows, Thanks a lot, Peter -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.slony.info/pipermail/slony1-general/attachments/20071209/= e5707194/attachment.htm
- Previous message: [Slony1-general] Delayed replication: Explicitly creating a slony-I backlog, and then telling the slon daemon to replicate changes all at once.
- Next message: [Slony1-general] Delayed replication: Explicitly creating a slony-I backlog, and then telling the slon daemon to replicate changes all at once.
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Slony1-general mailing list