Bug #35648 Agent should set it's own wait_timeout
Submitted: 28 Mar 2008 14:26 Modified: 13 Jan 2009 11:09
Reporter: Johan Idrén Email Updates:
Status: Closed Impact on me:
None 
Category:MySQL Enterprise Monitor: Agent Severity:S3 (Non-critical)
Version:1.2.1.8246 OS:Any
Assigned to: Jan Kneschke CPU Architecture:Any
Triage: D3 (Medium) / R2 (Low) / E2 (Low)

[28 Mar 2008 14:26] Johan Idrén
Description:
If a system has a global wait_timeout lower than the general activity of the agent, the agent will be kicked out, the monitored server will log an error and Aborted_clients will step up one.

How to repeat:
mysql> set global wait_timeout=45;
Query OK, 0 rows affected (0.00 sec)

From log:
080328 12:16:59 [Warning] Aborted connection 154 to db: 'unconnected' user: 'root' host: 'localhost' (Got timeout reading communication packets)

Suggested fix:
Let the agent set wait_timeout to something higher than it's usual sleep time, for it's own connection
[28 Mar 2008 14:34] Johan Idrén
Can be worked around by setting wait_timeout to a higher value globally, but if it's set very low then it's probably set low to workaround other issues. Hence not a very good workaround.
[11 Oct 2008 7:52] Jan Kneschke
The fix is in 2.0.0 already.

revno: 983
committer: jan@mysql.com
branch nick: mysql-lb-clean
timestamp: Fri 2008-08-15 18:00:06 +0200
message:
  "set wait_timeout = 30min" (fixes #35648) and made the whole connection setup "fatal" to make
  sure we never operate on the connection if it isn't initialized correctly
[11 Dec 2008 0:14] Marcos Palacios
Verified fixed in agent build 2.0.0.7111.
[13 Jan 2009 11:09] Tony Bedford
An entry was added to the 2.0 changelog:

If a system had a global wait_timeout lower than the general activity of the agent, the agent was disconnected. The monitored server then logged an error and incremented Aborted_clients.