I also see this error message in my logs. At one point we saw this in our logs for several hours, it may have been a local network disconnect, but still very strange.

It has been fixed already in one of the latest 1.8.x.

still a problem in 2.0.3

I do not see any problem here. 28800 is the default value for his parameter according to http://dev.mysql.com/doc/refman/5.0/en/server-system-variables.html#sysvar_wait_timeout. The warning messages do not harm Zabbix in any way, it re-connects automatically. Besides Zabbix 2.0.x is much quite when it comes to debug messaging.

I am closing it.

Alexei, i really see a problem here (and i'm not alone). We have wait_timeout=120 for various of reasons and are getting 'MySQL server has gone away' message about 1300 times per day. Is it really hard to find any other interesting error messages and logs are becaming useless. Is it really so big deal for you to set wait_timeout at the start of every connection?

could you please list the reasons ? it could help to evaluate how common such configuration could be

The main reason is that we have lots of MySQL users and were having big problems with connection limit with default wait_timeout configuration - many users and applications are just dump and are not closing connections correctly. There were 100s of already unused connections in sleep state. Raising the connection limit was not a solution here, we just needed to get rid of unused connections and wait_timeout was the thing which fixed that.