Complicated monitoring

[baroo]

Hi
I have the following monitoring system:

zabbix -- router1 -- DSL --- (Internet) -- DSL -- router2 -- router3

but with many sides like on the right.

For each system I create the follwing items to monitor:
(zabbix_agent i running on router2 )
- agent.ping (for monitoring availability of router2)
- some parameters (upload, download, latency, conntrack), but this is ok
- external check with ping script on zabbix_server to monitor DSL availability
- zabbix_agent UserParameter host.ping[10.0.1.2] to monitor router3

I also configure 3 triggers for each system:
- DSL depensd on Internet (internet is a trigger checked every 120sec)
(doming[xx.xx.xx.xx].nodata(180)=1)|(doming[xx.xx.xx.xx].max(180)=0)

- Router2 depends on DSL
dom:agent.ping.nodata(240)=1

- router3 depends on Router2
(dom:host.ping[10.0.1.2].nodata(300)=1)|(dom:host.ping[10.0.1.2].max(300)=0)

I`d like to send some SMS about erroes but I have the following problems :
- when my internet connection (with zabbix) goes down everything is OK becuse of the trigger depenadcies, but when connection is back Zabbix generate a lots of false informations about recovering hosts

- zabbix after some time of failure (e.g. becuse of upgrade) generates a lots of flase informations about recovering hosts becuse of nodata expressions in triggers

Thanks for any help.

Best regards
Baroo