ping wei <[email protected]> writes: > Replication confirm is functioning as after failover replication ( properly > shutdown ), able to see table updated. > > However, if i waited a minute up to 10 minute, cut off the connection ( > shutdown slave derby service ) then connect again ( turn on slave derby > service ), i dont see any table udpate at all. ( if interval is working here > )
Hi, I think the max log shipping interval only makes sure the logs have been sent to the slave, and not that the slave has actually stored the logs to disk by that time. It sounds like you performed fail-over by shutting down the slave, and then the slave may lose transactions that it still hasn't stored to disk, even if it has the logs for those transactions in its buffers. Derby's replication service was designed to allow the slave to take over if the master goes down. In the experiment you described, it was the slave that went down rather than the master. The replication service doesn't ensure that the slave database is up to date with the master in this case. However, the master database should still have all the data. Hope this helps, -- Knut Anders
