We retrieve statistics from network appliances through SNMP, but it happens (quite often) that we find firewalls or switches without data for a while.
For example, these are logs for an ASA:
28964:20130716:011115.827 SNMP item [CISCO-MEMORY-POOL-MIB_ciscoMemoryPoolFree.1] on host [XXXX] failed: first network error, wait for 30 seconds
29054:20130716:011145.017 resuming SNMP checks on host [XXXX]: connection restored
29006:20130716:011407.531 SNMP item [CISCO-FIREWALL-MIB_cfwConnectionStatCount.protoIp.currentInUse] on host [XXXX] failed: first network error, wait for 30 seconds
28930:20130716:011413.121 SNMP item [CISCO-IPSEC-FLOW-MONITOR-MIB_cipSecGlobalInOctets.0] on host [XXXX] failed: first network error, wait for 30 seconds
29071:20130716:011443.646 resuming SNMP checks on host [XXXX]: connection restored
28985:20130716:011658.079 SNMP item [CISCO-FIREWALL-MIB_cfwHardwareStatusValue.secondaryUnit] on host [XXXX] failed: first network error, wait for 30 seconds
29055:20130716:011728.235 resuming SNMP checks on host [XXXX]: connection restored
28936:20130716:011958.734 SNMP item [CISCO-FIREWALL-MIB_cfwHardwareStatusValue.secondaryUnit] on host [XXXX] failed: first network error, wait for 30 seconds
29051:20130716:012028.469 resuming SNMP checks on host [XXXX]: connection restored
29006:20130716:012307.675 SNMP item [CISCO-FIREWALL-MIB_cfwConnectionStatCount.protoIp.currentInUse] on host [XXXX] failed: first network error, wait for 30 seconds
29065:20130716:012337.123 resuming SNMP checks on host [XXXX]: connection restored
28930:20130716:012530.537 SNMP item [CISCO-FIREWALL-MIB_cfwHardwareStatusValue.primaryUnit] on host [XXXX] failed: first network error, wait for 30 seconds
29064:20130716:012606.228 SNMP item [CISCO-FIREWALL-MIB_cfwHardwareStatusValue.secondaryUnit] on host [XXXX] failed: another network error, wait for 30 seconds
29064:20130716:012642.425 SNMP item [CISCO-FIREWALL-MIB_cfwHardwareStatusValue.secondaryUnit] on host [XXXX] failed: another network error, wait for 30 seconds
29060:20130716:012718.169 temporarily disabling SNMP checks on host [XXXX]: host unavailable

We started Zabbix server on Jul, 8 and since Jul, 16 we don't have any more data gathered via SNMP from this host.
The ASA always answers quite fast when queried via command line.

We removed VRRP-coupled appliances from SNMP monitoring, since it seems there is a problem with duplicated engine.

Is there a way to have SNMP monitoring work reliably?
And meybe be alerted when no data is gathered for a while