[zabbix] Zabbix false issues after deploying with nova network

[Claude Durocher]

Testing fuel 5.1 (fuel-community-5.1-15-2014-08-31_23-58-23). Following issues are displayed in Zabbix after new deployment with nova network vlan manager :

Nova network process is not running (all controlers nodes)
Memcache service is down (all controlers nodes)
Too many process warnings (all nodes)

In this scenario, Zabbix is checking for nova-network on controler nodes but it's actually running on compute nodes. This can be resolved by attaching "Template App OpenStack Nova Network" to compute nodes instead of controler nodes.

Memcached is running fine on controler nodes and this is a false alert due to a bad check. In item "Memcache service is running" (in "Template App Memcache") we need to change net.tcp.service[tcp,,11211] to net.tcp.listen[11211].

Too many process warnings can be eliminated by editing item "Too many processes on {HOST.NAME}" (in "Template Fuel OS Linux") and setting {Template Fuel OS Linux:proc.num[].last(0)}>500 (instead of 300).

Expected result : there should be no issues reported in Zabbix after successful deployment.

[Tomasz 'Zen' Napierala (tzn)]

This is not gonna be fixed in 5.1, so it should be documented in known issues. Workarounds included in bug report.