Description:
When there is a replication problem I get 3 notifications:
MySQL Enterprise Monitor Critical Alert: Slave SQL Thread Not Running (xxxx)
MySQL Enterprise Monitor Critical Alert: Slave Has Been Stopped (xxxx)
MySQL Enterprise Monitor Critical Alert: Slave Has Experienced A Replication Error (xxxx)
While each of these errors in itself it fine it is not sensible to send ALL these errors out as they are dependent on each other. I guess that the last 2 notifications are a consequence of the first one.
Ideally you only want to know about the WORST event, as the other related events are dependent on that. nagios has this type of functionality.
This reduces the number of notifications you get, and if the main problem is resolved but one of the "sub-problems" isn't THEN it makes sense to warn about the sub problem.
This starts to become an issue as the number of monitored servers increases and can end up with huge "mail bombings" causing more noise than necessary and this just becomes a distraction rather than a help.
So adding dependency or relative improtance information into the advisors might help reduce the noise that is sent out to the clients by email. The web page should of course reflect the "full picture" and show all errors.
How to repeat:
Setup hundreds of slaves and then run a command on the master which breaks replication. Wait for X slaves x 3 messages to appear in your mailbox...
Suggested fix:
See above.
