Bug #30154 removing slave information is not picked up properly
Submitted: 31 Jul 2007 19:41 Modified: 23 Aug 2007 17:42
Reporter: Monty Taylor Email Updates:
Status: Closed Impact on me:
Category:MySQL Enterprise Monitor Severity:S3 (Non-critical)
Version: OS:Any
Assigned to: Darren Oldag CPU Architecture:Any
Tags: mer 120

[31 Jul 2007 19:41] Monty Taylor
Installed NMAS and it showed that a circular replication existed where one was not intended. Went to the machine that should be the master and removed slave information from it. At this point, NMAS did not re-order the replication info, as it was not able to remove the old information that the master server had a master. Specifically, errors like the follow appeared: 

attrib 'mysql::server->Master_Log_File' is unknown, skipping collection

How to repeat:
Monitor a machine that is a slave. Then remove it's slave info (change master to master_host="", master_log_file="";) 

Suggested fix:
[3 Aug 2007 18:59] Sloan Childers
Looks like we are not removing dc_known_items that are no longer supported on re-inventory.
[11 Aug 2007 0:07] Andy Bang
Here's a workaround until this is fixed:

1) Stop the agent that is monitoring the server on which you removed the slave information.

2) Wait for the agent to show as "down" in the Dashboard (its label will turn bold red in the Servers tab).

3) Go to Settings->Manager Servers and delete that server.

4) Generate a new UUID for the agent and place it in the agent's INI file.  See section Generate a New UUID: Windows or section Generate a New UUID: Unix and Mac OS X in the online help for instructions on how to create a new UUID.

5) Start the agent.
[13 Aug 2007 22:10] Sloan Childers
r30154 trunk
[23 Aug 2007 17:42] Bill Weber
Verified that a server which becomes a slave or stops being a slave while monitored is reflected correctly in the dashboard with build