snmpAgentDown blocks BoxDown event

[Ingeborg Hellemo]

NAV 4.2.2

We have had a couple of cases where boxes were down, but NAV belived they were still up.

Symptoms:

The box is down and does not answer to ping.

Nav Status is "Uptime Up for X days Y hours", while Availability counts down

Both times we see a SnmpAgentDown imidiately before the box is marked as down.

eventengine.log:2015-02-11 00:01:35,892 [INFO nav.eventengine.plugins.snmpagentstate.SnmpAgentStateHandler] snmpAgentState start event for examplebox; declaring down in 240 seconds (if still unresolved)
eventengine.log:2015-02-11 00:01:53,792 [INFO nav.eventengine.plugins.boxstate.BoxStateHandler] examplebox is already down, ignoring duplicate start event
eventengine.log:2015-02-11 00:05:35,897 [INFO nav.eventengine.plugins.snmpagentstate.SnmpAgentStateHandler] examplebox: Posting snmpAgentDown alert

pping.log:[2015-02-11 00:01:53] pping.py:generateEvents:145 [Notice] examplebox (10.255.42.8) marked as down.

ipdevpoll.log:2015-02-11 00:01:35,464 [10887] [WARNING plugins.snmpcheck.snmpcheck] SNMP agent down on examplebox

[Morten Brekkevold (mbrekkevold)]

fix consisting of multiple changesets merged here: https://nav.uninett.no/hg/stable/rev/b462c09dbbb9

This includes changes to snmpAgentState event handling, where snmpAgentDown alerts will be willfully held back if the netbox itself has stopped responding to ping while we were waiting for an snmpAgentUp resolve. It's kind of given that the box stops responding to snmp if its down, we don't need an extra alert for that.