Bug #74867 False positive for "Agent Host Time Out of Sync Relative to Dashboard"

• Submitted: 14 Nov 2014 13:03
• Modified: 6 Mar 2015 12:27
• Reporter: Daniël van Eeden (OCA)
• Status: Verified
• Category: MySQL Enterprise Monitor: Advisors/Rules
• Severity: S3 (Non-critical)
• Version: 3.0.14
• OS: Any
• CPU Architecture: Any
• Tags: buffering, network, ntp, time

[14 Nov 2014 13:03] Daniël van Eeden

Description:
After a network hiccup or short network failure a "Agent Host Time Out of Sync Relative to Dashboard" event is raised.

Both the host on which the mem agent runs and the host on which the service manager runs are in sync with the same NTP server.

The reported difference is 11 minutes. I guess this is due buffering of events in the agent.

How to repeat:
Install mem agent and server and then pull the network cable.

[11 Feb 2015 10:21] MySQL Verification Team

Verified. I see this if I just shut down the tomcat server for 15 minutes and restart it. All agents on same MEM localhost so can't be out of synch.

"Investigate why the clock on the server monitored by this Agent is about 4 minutes behind the clock on the server hosting the MySQL Enterprise Service Manager."

[12 Feb 2015 16:15] Mark Matthews

Daniel,

What version of the *agent* is being used when this issue happens?

[13 Feb 2015 12:08] Daniël van Eeden

I assume 3.0.14 (as reported in the version field) was the agent version.

It looks to have affected these agent versions:
3.0.13.3026 on Windows Server
3.0.15.3058 on Solaris
3.0.13.3026 on Solaris
3.0.14.3043 on Solaris