A change in production state does not take effect in regards to monitoring until a daemon restart?
I've recently noticed some funny behaviour regarding the production states and how they affect monitoring of devices after moving to Zenoss 4.2.5 Zenup SUP 732
Previously, on Zenoss 425 SUP 671 if a device is set to decommissioned (i.e. -1, or any production state value less than its zProdStateThreshold value – which defaults to 300 for devices) it would no longer monitor the device via ping or by snmp and no further events would be raised until the production state was set back to something above the zProdStateThreshold.
However, I've noticed on SUP732 that this is no longer the case and setting the production state lower than the zProdStateThreshold does not prevent monitoring or events being raised.
With further experiments I discovered that once prod state is lower than the threshold, restarting the zenping will stop ping monitoring and restarting the zenperfsnmp daemons will stop snmp monitoring.
This behaviour is also similar once re-raising the production state whilst monitoring is off. Until the ping and snmp daemons are restarted monitoring for each protocol does not begin again…
Is this a bug that has crept in during the 732 update? – I noticed that 732 has added a whole set of things around production state under zenutils and introduced a ProdStateManager.
Has anyone else noticed this behaviour or are they able to recreate it to confirm my findings?
And any ideas of how to rectify it so that reducing the production state below the zProdStateThreshold once again stops monitoring as it is supposed to?