This issue will be addressed in Spectrum 09.01.03.00. The following is the Release Note reference:
Symptom:The Multicast Manager may cause activation on startup to be delayed with the McastIHPerfCtrl handler causing the delays.
Resolution: The Multicast handlers will stop sending so many SNMP packets to devices that are down on startup. There should no longer be such a long activation delay on startup.
Related Issues/Questions:
Spectrum model activation hanging on Multicast models
Problem Environment:
Spectrum 09.00.00.00
Model Activation
Multicast Manager
When mdlact_debug=true is added to the $SPECROOT/SS/.vnmrc file, the following is seen in the $SPECROOT/SS/VNM.OUT file (the model handle may be different):
-----------------------------------------------------
Model 200030 is calling class McastIHPerfCtrl
-----------------------------------------------------
Causes of this problem:
On SpectroSERVER startup, McastIHPerfCtrl sends an action to McastIHPerfCol which causes probe packets to be sent to devices. Some groups in the database have over 50 sources. Each source may send 2 packets since two types of probes are tried (if the first one fails). On devices that are down, this could be over a hundred packets for one Multicast group alone. The SNMP timeout and retries thus cause at least a 9 second delay per probe and this could be over a 1000 seconds for just one group.
(Legacy KB ID CNC TS33626 )